;;;-*-Mode: LISP; Package: CCL -*-
;;;
;;;   Copyright (C) 2009 Clozure Associates
;;;   Copyright (C) 1994-2001 Digitool, Inc
;;;   This file is part of Clozure CL.  
;;;
;;;   Clozure CL is licensed under the terms of the Lisp Lesser GNU Public
;;;   License , known as the LLGPL and distributed with Clozure CL as the
;;;   file "LICENSE".  The LLGPL consists of a preamble and the LGPL,
;;;   which is distributed with Clozure CL as the file "LGPL".  Where these
;;;   conflict, the preamble takes precedence.  
;;;
;;;   Clozure CL is referenced in the preamble as the "LIBRARY."
;;;
;;;   The LLGPL is also available online at
;;;   http://opensource.franz.com/preamble.html

;;; READ and related functions.

(in-package "CCL")

(eval-when (:compile-toplevel :execute)
  (defconstant readtable-case-keywords '((:upcase . 1) (:downcase . 2) (:preserve . 0)
                                         (:invert . -1) (:studly . -2)))
  (defmacro readtable-case-keywords () `',readtable-case-keywords))

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

(defmethod make-load-form ((ref package-ref) &optional env)
  (declare (ignore env))
  `(register-package-ref ',(package-ref.name ref)))

(defmethod print-object ((ref package-ref) stream)
  (print-unreadable-object (ref stream :type t :identity t)
    (format stream "for ~s [~s]" (package-ref.name ref) (package-ref.pkg ref))))

;;; Maps character names to characters
(defvar *name->char* (make-hash-table :test #'equalp))
;;; Maps characters to (canonical) character names.
(defvar *char->name* (make-hash-table :test #'eql))

;;; This isn't thread-safe.  If the user really wants to register character
;;; names from multiple threads, they should do their own locking.
(defun register-character-name (name char)
  (setf (gethash name *name->char*) char)    
  (unless (gethash char *char->name*)
    (setf (gethash char *char->name*) name)))

(dolist (pair '(
                ;; Standard character names
                ("Newline" .  #\012) ("Space" . #\040)
                ;; Semi-standard character names
                ("Rubout" . #\177) ("Page" . #\014) ("Tab" . #\011)
                ("Backspace" . #\010) ("Return" . #\015) ("Linefeed" . #\012)
                ;; Other character names.  (When available, standard
                ;; names should be used for printing in preference to
                ;; any non-standard names.)
                ("Null" . #\000) ("Nul" . #\000)
		("SOH" . #\001)
		("STX" . #\002)
		("ETX" . #\003)
		("EOT" . #\004)
		("ENQ" . #\005)
		("ACK" . #\006)
                ("Bell"  . #\007) ("BEL" . #\007)       ; ^G , used by Franz (and others with bells.)
                ("Delete" . #\010) ("BS" . #\010)
		("HT" . #\011)
                ("LF" . #\012) ("NL" . #\012)
                ("PageUp" . #\013) ("VT" . #\013)
                ("PageDown" . #\014) ("Formfeed" . #\014) ("FF" . #\014) ("NP" . #\014)
                ("CR" . #\015)
		("SO" . #\016)
		("SI" . #\017)
                ("Sub" . #\032)
                ("ESC" .  #\033) ("Escape" . #\033) ("Clear" .  #\033)
                ("Altmode" .  #\033) ("ALT" .  #\033)
                ("Fs" . #\034)
                ("Gs" . #\035)
                ("Rs" . #\036)
                ("Us" . #\037)
                ("DEL" . #\177)("ForwardDelete" . #\177)
                ("No-Break_Space" . #\u+00a0)
                ("Inverted_Exclamation_Mark" . #\u+00a1)
                ("Cent_Sign" . #\u+00a2)
                ("Pound_Sign" . #\u+00a3)
                ("Currency_Sign" . #\u+00a4)
                ("Yen_Sign" . #\u+00a5)
                ("Broken_Bar" . #\u+00a6)
                ("Section_Sign" . #\u+00a7)
                ("Diaeresis" . #\u+00a8)
                ("Copyright_Sign" . #\u+00a9)
                ("Feminine_Ordinal_Indicator" . #\u+00aa)
                ("Left-Pointing_Double_Angle_Quotation_Mark" . #\u+00ab)
                ("Not_Sign" . #\u+00ac)
                ("Soft_Hyphen" . #\u+00ad)
                ("Registered_Sign" . #\u+00ae)
                ("Macron" . #\u+00af)
                ("Degree_Sign" . #\u+00b0)
                ("Plus-Minus_Sign" . #\u+00b1)
                ("Superscript_Two" . #\u+00b2)
                ("Superscript_Three" . #\u+00b3)
                ("Acute_Accent" . #\u+00b4)
                ("Micro_Sign" . #\u+00b5)
                ("Pilcrow_Sign" . #\u+00b6)
                ("Middle_Dot" . #\u+00b7)
                ("Cedilla" . #\u+00b8)
                ("Superscript_One" . #\u+00b9)
                ("Masculine_Ordinal_Indicator" . #\u+00ba)
                ("Right-Pointing_Double_Angle_Quotation_Mark" . #\u+00bb)
                ("Vulgar_Fraction_One_Quarter" . #\u+00bc)
                ("Vulgar_Fraction_One_Half" . #\u+00bd)
                ("Vulgar_Fraction_Three_Quarters" . #\u+00be)
                ("Inverted_Question_Mark" . #\u+00bf)
                ("Latin_Capital_Letter_A_With_Grave" . #\u+00c0)
                ("Latin_Capital_Letter_A_With_Acute" . #\u+00c1)
                ("Latin_Capital_Letter_A_With_Circumflex" . #\u+00c2)
                ("Latin_Capital_Letter_A_With_Tilde" . #\u+00c3)
                ("Latin_Capital_Letter_A_With_Diaeresis" . #\u+00c4)
                ("Latin_Capital_Letter_A_With_Ring_Above" . #\u+00c5)
                ("Latin_Capital_Letter_Ae" . #\u+00c6)
                ("Latin_Capital_Letter_C_With_Cedilla" . #\u+00c7)
                ("Latin_Capital_Letter_E_With_Grave" . #\u+00c8)
                ("Latin_Capital_Letter_E_With_Acute" . #\u+00c9)
                ("Latin_Capital_Letter_E_With_Circumflex" . #\u+00ca)
                ("Latin_Capital_Letter_E_With_Diaeresis" . #\u+00cb)
                ("Latin_Capital_Letter_I_With_Grave" . #\u+00cc)
                ("Latin_Capital_Letter_I_With_Acute" . #\u+00cd)
                ("Latin_Capital_Letter_I_With_Circumflex" . #\u+00ce)
                ("Latin_Capital_Letter_I_With_Diaeresis" . #\u+00cf)
                ("Latin_Capital_Letter_Eth" . #\u+00d0)
                ("Latin_Capital_Letter_N_With_Tilde" . #\u+00d1)
                ("Latin_Capital_Letter_O_With_Grave" . #\u+00d2)
                ("Latin_Capital_Letter_O_With_Acute" . #\u+00d3)
                ("Latin_Capital_Letter_O_With_Circumflex" . #\u+00d4)
                ("Latin_Capital_Letter_O_With_Tilde" . #\u+00d5)
                ("Latin_Capital_Letter_O_With_Diaeresis" . #\u+00d6)
                ("Multiplication_Sign" . #\u+00d7)
                ("Latin_Capital_Letter_O_With_Stroke" . #\u+00d8)
                ("Latin_Capital_Letter_U_With_Grave" . #\u+00d9)
                ("Latin_Capital_Letter_U_With_Acute" . #\u+00da)
                ("Latin_Capital_Letter_U_With_Circumflex" . #\u+00db)
                ("Latin_Capital_Letter_U_With_Diaeresis" . #\u+00dc)
                ("Latin_Capital_Letter_Y_With_Acute" . #\u+00dd)
                ("Latin_Capital_Letter_Thorn" . #\u+00de)
                ("Latin_Small_Letter_Sharp_S" . #\u+00df)
                ("Latin_Small_Letter_A_With_Grave" . #\u+00e0)
                ("Latin_Small_Letter_A_With_Acute" . #\u+00e1)
                ("Latin_Small_Letter_A_With_Circumflex" . #\u+00e2)
                ("Latin_Small_Letter_A_With_Tilde" . #\u+00e3)
                ("Latin_Small_Letter_A_With_Diaeresis" . #\u+00e4)
                ("Latin_Small_Letter_A_With_Ring_Above" . #\u+00e5)
                ("Latin_Small_Letter_Ae" . #\u+00e6)
                ("Latin_Small_Letter_C_With_Cedilla" . #\u+00e7)
                ("Latin_Small_Letter_E_With_Grave" . #\u+00e8)
                ("Latin_Small_Letter_E_With_Acute" . #\u+00e9)
                ("Latin_Small_Letter_E_With_Circumflex" . #\u+00ea)
                ("Latin_Small_Letter_E_With_Diaeresis" . #\u+00eb)
                ("Latin_Small_Letter_I_With_Grave" . #\u+00ec)
                ("Latin_Small_Letter_I_With_Acute" . #\u+00ed)
                ("Latin_Small_Letter_I_With_Circumflex" . #\u+00ee)
                ("Latin_Small_Letter_I_With_Diaeresis" . #\u+00ef)
                ("Latin_Small_Letter_Eth" . #\u+00f0)
                ("Latin_Small_Letter_N_With_Tilde" . #\u+00f1)
                ("Latin_Small_Letter_O_With_Grave" . #\u+00f2)
                ("Latin_Small_Letter_O_With_Acute" . #\u+00f3)
                ("Latin_Small_Letter_O_With_Circumflex" . #\u+00f4)
                ("Latin_Small_Letter_O_With_Tilde" . #\u+00f5)
                ("Latin_Small_Letter_O_With_Diaeresis" . #\u+00f6)
                ("Division_Sign" . #\u+00f7)
                ("Latin_Small_Letter_O_With_Stroke" . #\u+00f8)
                ("Latin_Small_Letter_U_With_Grave" . #\u+00f9)
                ("Latin_Small_Letter_U_With_Acute" . #\u+00fa)
                ("Latin_Small_Letter_U_With_Circumflex" . #\u+00fb)
                ("Latin_Small_Letter_U_With_Diaeresis" . #\u+00fc)
                ("Latin_Small_Letter_Y_With_Acute" . #\u+00fd)
                ("Latin_Small_Letter_Thorn" . #\u+00fe)
                ("Latin_Small_Letter_Y_With_Diaeresis" . #\u+00ff)
                ("Latin_Capital_Letter_A_With_Macron" . #\u+0100)
                ("Latin_Small_Letter_A_With_Macron" . #\u+0101)
                ("Latin_Capital_Letter_A_With_Breve" . #\u+0102)
                ("Latin_Small_Letter_A_With_Breve" . #\u+0103)
                ("Latin_Capital_Letter_A_With_Ogonek" . #\u+0104)
                ("Latin_Small_Letter_A_With_Ogonek" . #\u+0105)
                ("Latin_Capital_Letter_C_With_Acute" . #\u+0106)
                ("Latin_Small_Letter_C_With_Acute" . #\u+0107)
                ("Latin_Capital_Letter_C_With_Circumflex" . #\u+0108)
                ("Latin_Small_Letter_C_With_Circumflex" . #\u+0109)
                ("Latin_Capital_Letter_C_With_Dot_Above" . #\u+010a)
                ("Latin_Small_Letter_C_With_Dot_Above" . #\u+010b)
                ("Latin_Capital_Letter_C_With_Caron" . #\u+010c)
                ("Latin_Small_Letter_C_With_Caron" . #\u+010d)
                ("Latin_Capital_Letter_D_With_Caron" . #\u+010e)
                ("Latin_Small_Letter_D_With_Caron" . #\u+010f)
                ("Latin_Capital_Letter_D_With_Stroke" . #\u+0110)
                ("Latin_Small_Letter_D_With_Stroke" . #\u+0111)
                ("Latin_Capital_Letter_E_With_Macron" . #\u+0112)
                ("Latin_Small_Letter_E_With_Macron" . #\u+0113)
                ("Latin_Capital_Letter_E_With_Breve" . #\u+0114)
                ("Latin_Small_Letter_E_With_Breve" . #\u+0115)
                ("Latin_Capital_Letter_E_With_Dot_Above" . #\u+0116)
                ("Latin_Small_Letter_E_With_Dot_Above" . #\u+0117)
                ("Latin_Capital_Letter_E_With_Ogonek" . #\u+0118)
                ("Latin_Small_Letter_E_With_Ogonek" . #\u+0119)
                ("Latin_Capital_Letter_E_With_Caron" . #\u+011a)
                ("Latin_Small_Letter_E_With_Caron" . #\u+011b)
                ("Latin_Capital_Letter_G_With_Circumflex" . #\u+011c)
                ("Latin_Small_Letter_G_With_Circumflex" . #\u+011d)
                ("Latin_Capital_Letter_G_With_Breve" . #\u+011e)
                ("Latin_Small_Letter_G_With_Breve" . #\u+011f)
                ("Latin_Capital_Letter_G_With_Dot_Above" . #\u+0120)
                ("Latin_Small_Letter_G_With_Dot_Above" . #\u+0121)
                ("Latin_Capital_Letter_G_With_Cedilla" . #\u+0122)
                ("Latin_Small_Letter_G_With_Cedilla" . #\u+0123)
                ("Latin_Capital_Letter_H_With_Circumflex" . #\u+0124)
                ("Latin_Small_Letter_H_With_Circumflex" . #\u+0125)
                ("Latin_Capital_Letter_H_With_Stroke" . #\u+0126)
                ("Latin_Small_Letter_H_With_Stroke" . #\u+0127)
                ("Latin_Capital_Letter_I_With_Tilde" . #\u+0128)
                ("Latin_Small_Letter_I_With_Tilde" . #\u+0129)
                ("Latin_Capital_Letter_I_With_Macron" . #\u+012a)
                ("Latin_Small_Letter_I_With_Macron" . #\u+012b)
                ("Latin_Capital_Letter_I_With_Breve" . #\u+012c)
                ("Latin_Small_Letter_I_With_Breve" . #\u+012d)
                ("Latin_Capital_Letter_I_With_Ogonek" . #\u+012e)
                ("Latin_Small_Letter_I_With_Ogonek" . #\u+012f)
                ("Latin_Capital_Letter_I_With_Dot_Above" . #\u+0130)
                ("Latin_Small_Letter_Dotless_I" . #\u+0131)
                ("Latin_Capital_Ligature_Ij" . #\u+0132)
                ("Latin_Small_Ligature_Ij" . #\u+0133)
                ("Latin_Capital_Letter_J_With_Circumflex" . #\u+0134)
                ("Latin_Small_Letter_J_With_Circumflex" . #\u+0135)
                ("Latin_Capital_Letter_K_With_Cedilla" . #\u+0136)
                ("Latin_Small_Letter_K_With_Cedilla" . #\u+0137)
                ("Latin_Small_Letter_Kra" . #\u+0138)
                ("Latin_Capital_Letter_L_With_Acute" . #\u+0139)
                ("Latin_Small_Letter_L_With_Acute" . #\u+013a)
                ("Latin_Capital_Letter_L_With_Cedilla" . #\u+013b)
                ("Latin_Small_Letter_L_With_Cedilla" . #\u+013c)
                ("Latin_Capital_Letter_L_With_Caron" . #\u+013d)
                ("Latin_Small_Letter_L_With_Caron" . #\u+013e)
                ("Latin_Capital_Letter_L_With_Middle_Dot" . #\u+013f)
                ("Latin_Small_Letter_L_With_Middle_Dot" . #\u+0140)
                ("Latin_Capital_Letter_L_With_Stroke" . #\u+0141)
                ("Latin_Small_Letter_L_With_Stroke" . #\u+0142)
                ("Latin_Capital_Letter_N_With_Acute" . #\u+0143)
                ("Latin_Small_Letter_N_With_Acute" . #\u+0144)
                ("Latin_Capital_Letter_N_With_Cedilla" . #\u+0145)
                ("Latin_Small_Letter_N_With_Cedilla" . #\u+0146)
                ("Latin_Capital_Letter_N_With_Caron" . #\u+0147)
                ("Latin_Small_Letter_N_With_Caron" . #\u+0148)
                ("Latin_Small_Letter_N_Preceded_By_Apostrophe" . #\u+0149)
                ("Latin_Capital_Letter_Eng" . #\u+014a)
                ("Latin_Small_Letter_Eng" . #\u+014b)
                ("Latin_Capital_Letter_O_With_Macron" . #\u+014c)
                ("Latin_Small_Letter_O_With_Macron" . #\u+014d)
                ("Latin_Capital_Letter_O_With_Breve" . #\u+014e)
                ("Latin_Small_Letter_O_With_Breve" . #\u+014f)
                ("Latin_Capital_Letter_O_With_Double_Acute" . #\u+0150)
                ("Latin_Small_Letter_O_With_Double_Acute" . #\u+0151)
                ("Latin_Capital_Ligature_Oe" . #\u+0152)
                ("Latin_Small_Ligature_Oe" . #\u+0153)
                ("Latin_Capital_Letter_R_With_Acute" . #\u+0154)
                ("Latin_Small_Letter_R_With_Acute" . #\u+0155)
                ("Latin_Capital_Letter_R_With_Cedilla" . #\u+0156)
                ("Latin_Small_Letter_R_With_Cedilla" . #\u+0157)
                ("Latin_Capital_Letter_R_With_Caron" . #\u+0158)
                ("Latin_Small_Letter_R_With_Caron" . #\u+0159)
                ("Latin_Capital_Letter_S_With_Acute" . #\u+015a)
                ("Latin_Small_Letter_S_With_Acute" . #\u+015b)
                ("Latin_Capital_Letter_S_With_Circumflex" . #\u+015c)
                ("Latin_Small_Letter_S_With_Circumflex" . #\u+015d)
                ("Latin_Capital_Letter_S_With_Cedilla" . #\u+015e)
                ("Latin_Small_Letter_S_With_Cedilla" . #\u+015f)
                ("Latin_Capital_Letter_S_With_Caron" . #\u+0160)
                ("Latin_Small_Letter_S_With_Caron" . #\u+0161)
                ("Latin_Capital_Letter_T_With_Cedilla" . #\u+0162)
                ("Latin_Small_Letter_T_With_Cedilla" . #\u+0163)
                ("Latin_Capital_Letter_T_With_Caron" . #\u+0164)
                ("Latin_Small_Letter_T_With_Caron" . #\u+0165)
                ("Latin_Capital_Letter_T_With_Stroke" . #\u+0166)
                ("Latin_Small_Letter_T_With_Stroke" . #\u+0167)
                ("Latin_Capital_Letter_U_With_Tilde" . #\u+0168)
                ("Latin_Small_Letter_U_With_Tilde" . #\u+0169)
                ("Latin_Capital_Letter_U_With_Macron" . #\u+016a)
                ("Latin_Small_Letter_U_With_Macron" . #\u+016b)
                ("Latin_Capital_Letter_U_With_Breve" . #\u+016c)
                ("Latin_Small_Letter_U_With_Breve" . #\u+016d)
                ("Latin_Capital_Letter_U_With_Ring_Above" . #\u+016e)
                ("Latin_Small_Letter_U_With_Ring_Above" . #\u+016f)
                ("Latin_Capital_Letter_U_With_Double_Acute" . #\u+0170)
                ("Latin_Small_Letter_U_With_Double_Acute" . #\u+0171)
                ("Latin_Capital_Letter_U_With_Ogonek" . #\u+0172)
                ("Latin_Small_Letter_U_With_Ogonek" . #\u+0173)
                ("Latin_Capital_Letter_W_With_Circumflex" . #\u+0174)
                ("Latin_Small_Letter_W_With_Circumflex" . #\u+0175)
                ("Latin_Capital_Letter_Y_With_Circumflex" . #\u+0176)
                ("Latin_Small_Letter_Y_With_Circumflex" . #\u+0177)
                ("Latin_Capital_Letter_Y_With_Diaeresis" . #\u+0178)
                ("Latin_Capital_Letter_Z_With_Acute" . #\u+0179)
                ("Latin_Small_Letter_Z_With_Acute" . #\u+017a)
                ("Latin_Capital_Letter_Z_With_Dot_Above" . #\u+017b)
                ("Latin_Small_Letter_Z_With_Dot_Above" . #\u+017c)
                ("Latin_Capital_Letter_Z_With_Caron" . #\u+017d)
                ("Latin_Small_Letter_Z_With_Caron" . #\u+017e)
                ("Latin_Small_Letter_Long_S" . #\u+017f)
                ("Latin_Small_Letter_B_With_Stroke" . #\u+0180)
                ("Latin_Capital_Letter_B_With_Hook" . #\u+0181)
                ("Latin_Capital_Letter_B_With_Topbar" . #\u+0182)
                ("Latin_Small_Letter_B_With_Topbar" . #\u+0183)
                ("Latin_Capital_Letter_Tone_Six" . #\u+0184)
                ("Latin_Small_Letter_Tone_Six" . #\u+0185)
                ("Latin_Capital_Letter_Open_O" . #\u+0186)
                ("Latin_Capital_Letter_C_With_Hook" . #\u+0187)
                ("Latin_Small_Letter_C_With_Hook" . #\u+0188)
                ("Latin_Capital_Letter_African_D" . #\u+0189)
                ("Latin_Capital_Letter_D_With_Hook" . #\u+018a)
                ("Latin_Capital_Letter_D_With_Topbar" . #\u+018b)
                ("Latin_Small_Letter_D_With_Topbar" . #\u+018c)
                ("Latin_Small_Letter_Turned_Delta" . #\u+018d)
                ("Latin_Capital_Letter_Reversed_E" . #\u+018e)
                ("Latin_Capital_Letter_Schwa" . #\u+018f)
                ("Latin_Capital_Letter_Open_E" . #\u+0190)
                ("Latin_Capital_Letter_F_With_Hook" . #\u+0191)
                ("Latin_Small_Letter_F_With_Hook" . #\u+0192)
                ("Latin_Capital_Letter_G_With_Hook" . #\u+0193)
                ("Latin_Capital_Letter_Gamma" . #\u+0194)
                ("Latin_Small_Letter_Hv" . #\u+0195)
                ("Latin_Capital_Letter_Iota" . #\u+0196)
                ("Latin_Capital_Letter_I_With_Stroke" . #\u+0197)
                ("Latin_Capital_Letter_K_With_Hook" . #\u+0198)
                ("Latin_Small_Letter_K_With_Hook" . #\u+0199)
                ("Latin_Small_Letter_L_With_Bar" . #\u+019a)
                ("Latin_Small_Letter_Lambda_With_Stroke" . #\u+019b)
                ("Latin_Capital_Letter_Turned_M" . #\u+019c)
                ("Latin_Capital_Letter_N_With_Left_Hook" . #\u+019d)
                ("Latin_Small_Letter_N_With_Long_Right_Leg" . #\u+019e)
                ("Latin_Capital_Letter_O_With_Middle_Tilde" . #\u+019f)
                ("Latin_Capital_Letter_O_With_Horn" . #\u+01a0)
                ("Latin_Small_Letter_O_With_Horn" . #\u+01a1)
                ("Latin_Capital_Letter_Oi" . #\u+01a2)
                ("Latin_Small_Letter_Oi" . #\u+01a3)
                ("Latin_Capital_Letter_P_With_Hook" . #\u+01a4)
                ("Latin_Small_Letter_P_With_Hook" . #\u+01a5)
                ("Latin_Letter_Yr" . #\u+01a6)
                ("Latin_Capital_Letter_Tone_Two" . #\u+01a7)
                ("Latin_Small_Letter_Tone_Two" . #\u+01a8)
                ("Latin_Capital_Letter_Esh" . #\u+01a9)
                ("Latin_Letter_Reversed_Esh_Loop" . #\u+01aa)
                ("Latin_Small_Letter_T_With_Palatal_Hook" . #\u+01ab)
                ("Latin_Capital_Letter_T_With_Hook" . #\u+01ac)
                ("Latin_Small_Letter_T_With_Hook" . #\u+01ad)
                ("Latin_Capital_Letter_T_With_Retroflex_Hook" . #\u+01ae)
                ("Latin_Capital_Letter_U_With_Horn" . #\u+01af)
                ("Latin_Small_Letter_U_With_Horn" . #\u+01b0)
                ("Latin_Capital_Letter_Upsilon" . #\u+01b1)
                ("Latin_Capital_Letter_V_With_Hook" . #\u+01b2)
                ("Latin_Capital_Letter_Y_With_Hook" . #\u+01b3)
                ("Latin_Small_Letter_Y_With_Hook" . #\u+01b4)
                ("Latin_Capital_Letter_Z_With_Stroke" . #\u+01b5)
                ("Latin_Small_Letter_Z_With_Stroke" . #\u+01b6)
                ("Latin_Capital_Letter_Ezh" . #\u+01b7)
                ("Latin_Capital_Letter_Ezh_Reversed" . #\u+01b8)
                ("Latin_Small_Letter_Ezh_Reversed" . #\u+01b9)
                ("Latin_Small_Letter_Ezh_With_Tail" . #\u+01ba)
                ("Latin_Letter_Two_With_Stroke" . #\u+01bb)
                ("Latin_Capital_Letter_Tone_Five" . #\u+01bc)
                ("Latin_Small_Letter_Tone_Five" . #\u+01bd)
                ("Latin_Letter_Inverted_Glottal_Stop_With_Stroke" . #\u+01be)
                ("Latin_Letter_Wynn" . #\u+01bf)
                ("Latin_Letter_Dental_Click" . #\u+01c0)
                ("Latin_Letter_Lateral_Click" . #\u+01c1)
                ("Latin_Letter_Alveolar_Click" . #\u+01c2)
                ("Latin_Letter_Retroflex_Click" . #\u+01c3)
                ("Latin_Capital_Letter_Dz_With_Caron" . #\u+01c4)
                ("Latin_Capital_Letter_D_With_Small_Letter_Z_With_Caron" . #\u+01c5)
                ("Latin_Small_Letter_Dz_With_Caron" . #\u+01c6)
                ("Latin_Capital_Letter_Lj" . #\u+01c7)
                ("Latin_Capital_Letter_L_With_Small_Letter_J" . #\u+01c8)
                ("Latin_Small_Letter_Lj" . #\u+01c9)
                ("Latin_Capital_Letter_Nj" . #\u+01ca)
                ("Latin_Capital_Letter_N_With_Small_Letter_J" . #\u+01cb)
                ("Latin_Small_Letter_Nj" . #\u+01cc)
                ("Latin_Capital_Letter_A_With_Caron" . #\u+01cd)
                ("Latin_Small_Letter_A_With_Caron" . #\u+01ce)
                ("Latin_Capital_Letter_I_With_Caron" . #\u+01cf)
                ("Latin_Small_Letter_I_With_Caron" . #\u+01d0)
                ("Latin_Capital_Letter_O_With_Caron" . #\u+01d1)
                ("Latin_Small_Letter_O_With_Caron" . #\u+01d2)
                ("Latin_Capital_Letter_U_With_Caron" . #\u+01d3)
                ("Latin_Small_Letter_U_With_Caron" . #\u+01d4)
                ("Latin_Capital_Letter_U_With_Diaeresis_And_Macron" . #\u+01d5)
                ("Latin_Small_Letter_U_With_Diaeresis_And_Macron" . #\u+01d6)
                ("Latin_Capital_Letter_U_With_Diaeresis_And_Acute" . #\u+01d7)
                ("Latin_Small_Letter_U_With_Diaeresis_And_Acute" . #\u+01d8)
                ("Latin_Capital_Letter_U_With_Diaeresis_And_Caron" . #\u+01d9)
                ("Latin_Small_Letter_U_With_Diaeresis_And_Caron" . #\u+01da)
                ("Latin_Capital_Letter_U_With_Diaeresis_And_Grave" . #\u+01db)
                ("Latin_Small_Letter_U_With_Diaeresis_And_Grave" . #\u+01dc)
                ("Latin_Small_Letter_Turned_E" . #\u+01dd)
                ("Latin_Capital_Letter_A_With_Diaeresis_And_Macron" . #\u+01de)
                ("Latin_Small_Letter_A_With_Diaeresis_And_Macron" . #\u+01df)
                ("Latin_Capital_Letter_A_With_Dot_Above_And_Macron" . #\u+01e0)
                ("Latin_Small_Letter_A_With_Dot_Above_And_Macron" . #\u+01e1)
                ("Latin_Capital_Letter_Ae_With_Macron" . #\u+01e2)
                ("Latin_Small_Letter_Ae_With_Macron" . #\u+01e3)
                ("Latin_Capital_Letter_G_With_Stroke" . #\u+01e4)
                ("Latin_Small_Letter_G_With_Stroke" . #\u+01e5)
                ("Latin_Capital_Letter_G_With_Caron" . #\u+01e6)
                ("Latin_Small_Letter_G_With_Caron" . #\u+01e7)
                ("Latin_Capital_Letter_K_With_Caron" . #\u+01e8)
                ("Latin_Small_Letter_K_With_Caron" . #\u+01e9)
                ("Latin_Capital_Letter_O_With_Ogonek" . #\u+01ea)
                ("Latin_Small_Letter_O_With_Ogonek" . #\u+01eb)
                ("Latin_Capital_Letter_O_With_Ogonek_And_Macron" . #\u+01ec)
                ("Latin_Small_Letter_O_With_Ogonek_And_Macron" . #\u+01ed)
                ("Latin_Capital_Letter_Ezh_With_Caron" . #\u+01ee)
                ("Latin_Small_Letter_Ezh_With_Caron" . #\u+01ef)
                ("Latin_Small_Letter_J_With_Caron" . #\u+01f0)
                ("Latin_Capital_Letter_Dz" . #\u+01f1)
                ("Latin_Capital_Letter_D_With_Small_Letter_Z" . #\u+01f2)
                ("Latin_Small_Letter_Dz" . #\u+01f3)
                ("Latin_Capital_Letter_G_With_Acute" . #\u+01f4)
                ("Latin_Small_Letter_G_With_Acute" . #\u+01f5)
                ("Latin_Capital_Letter_Hwair" . #\u+01f6)
                ("Latin_Capital_Letter_Wynn" . #\u+01f7)
                ("Latin_Capital_Letter_N_With_Grave" . #\u+01f8)
                ("Latin_Small_Letter_N_With_Grave" . #\u+01f9)
                ("Latin_Capital_Letter_A_With_Ring_Above_And_Acute" . #\u+01fa)
                ("Latin_Small_Letter_A_With_Ring_Above_And_Acute" . #\u+01fb)
                ("Latin_Capital_Letter_Ae_With_Acute" . #\u+01fc)
                ("Latin_Small_Letter_Ae_With_Acute" . #\u+01fd)
                ("Latin_Capital_Letter_O_With_Stroke_And_Acute" . #\u+01fe)
                ("Latin_Small_Letter_O_With_Stroke_And_Acute" . #\u+01ff)
                ("Latin_Capital_Letter_A_With_Double_Grave" . #\u+0200)
                ("Latin_Small_Letter_A_With_Double_Grave" . #\u+0201)
                ("Latin_Capital_Letter_A_With_Inverted_Breve" . #\u+0202)
                ("Latin_Small_Letter_A_With_Inverted_Breve" . #\u+0203)
                ("Latin_Capital_Letter_E_With_Double_Grave" . #\u+0204)
                ("Latin_Small_Letter_E_With_Double_Grave" . #\u+0205)
                ("Latin_Capital_Letter_E_With_Inverted_Breve" . #\u+0206)
                ("Latin_Small_Letter_E_With_Inverted_Breve" . #\u+0207)
                ("Latin_Capital_Letter_I_With_Double_Grave" . #\u+0208)
                ("Latin_Small_Letter_I_With_Double_Grave" . #\u+0209)
                ("Latin_Capital_Letter_I_With_Inverted_Breve" . #\u+020a)
                ("Latin_Small_Letter_I_With_Inverted_Breve" . #\u+020b)
                ("Latin_Capital_Letter_O_With_Double_Grave" . #\u+020c)
                ("Latin_Small_Letter_O_With_Double_Grave" . #\u+020d)
                ("Latin_Capital_Letter_O_With_Inverted_Breve" . #\u+020e)
                ("Latin_Small_Letter_O_With_Inverted_Breve" . #\u+020f)
                ("Latin_Capital_Letter_R_With_Double_Grave" . #\u+0210)
                ("Latin_Small_Letter_R_With_Double_Grave" . #\u+0211)
                ("Latin_Capital_Letter_R_With_Inverted_Breve" . #\u+0212)
                ("Latin_Small_Letter_R_With_Inverted_Breve" . #\u+0213)
                ("Latin_Capital_Letter_U_With_Double_Grave" . #\u+0214)
                ("Latin_Small_Letter_U_With_Double_Grave" . #\u+0215)
                ("Latin_Capital_Letter_U_With_Inverted_Breve" . #\u+0216)
                ("Latin_Small_Letter_U_With_Inverted_Breve" . #\u+0217)
                ("Latin_Capital_Letter_S_With_Comma_Below" . #\u+0218)
                ("Latin_Small_Letter_S_With_Comma_Below" . #\u+0219)
                ("Latin_Capital_Letter_T_With_Comma_Below" . #\u+021a)
                ("Latin_Small_Letter_T_With_Comma_Below" . #\u+021b)
                ("Latin_Capital_Letter_Yogh" . #\u+021c)
                ("Latin_Small_Letter_Yogh" . #\u+021d)
                ("Latin_Capital_Letter_H_With_Caron" . #\u+021e)
                ("Latin_Small_Letter_H_With_Caron" . #\u+021f)
                ("Latin_Capital_Letter_N_With_Long_Right_Leg" . #\u+0220)
                ("Latin_Small_Letter_D_With_Curl" . #\u+0221)
                ("Latin_Capital_Letter_Ou" . #\u+0222)
                ("Latin_Small_Letter_Ou" . #\u+0223)
                ("Latin_Capital_Letter_Z_With_Hook" . #\u+0224)
                ("Latin_Small_Letter_Z_With_Hook" . #\u+0225)
                ("Latin_Capital_Letter_A_With_Dot_Above" . #\u+0226)
                ("Latin_Small_Letter_A_With_Dot_Above" . #\u+0227)
                ("Latin_Capital_Letter_E_With_Cedilla" . #\u+0228)
                ("Latin_Small_Letter_E_With_Cedilla" . #\u+0229)
                ("Latin_Capital_Letter_O_With_Diaeresis_And_Macron" . #\u+022a)
                ("Latin_Small_Letter_O_With_Diaeresis_And_Macron" . #\u+022b)
                ("Latin_Capital_Letter_O_With_Tilde_And_Macron" . #\u+022c)
                ("Latin_Small_Letter_O_With_Tilde_And_Macron" . #\u+022d)
                ("Latin_Capital_Letter_O_With_Dot_Above" . #\u+022e)
                ("Latin_Small_Letter_O_With_Dot_Above" . #\u+022f)
                ("Latin_Capital_Letter_O_With_Dot_Above_And_Macron" . #\u+0230)
                ("Latin_Small_Letter_O_With_Dot_Above_And_Macron" . #\u+0231)
                ("Latin_Capital_Letter_Y_With_Macron" . #\u+0232)
                ("Latin_Small_Letter_Y_With_Macron" . #\u+0233)
                ("Latin_Small_Letter_L_With_Curl" . #\u+0234)
                ("Latin_Small_Letter_N_With_Curl" . #\u+0235)
                ("Latin_Small_Letter_T_With_Curl" . #\u+0236)
                ("Latin_Small_Letter_Dotless_J" . #\u+0237)
                ("Latin_Small_Letter_Db_Digraph" . #\u+0238)
                ("Latin_Small_Letter_Qp_Digraph" . #\u+0239)
                ("Latin_Capital_Letter_A_With_Stroke" . #\u+023a)
                ("Latin_Capital_Letter_C_With_Stroke" . #\u+023b)
                ("Latin_Small_Letter_C_With_Stroke" . #\u+023c)
                ("Latin_Capital_Letter_L_With_Bar" . #\u+023d)
                ("Latin_Capital_Letter_T_With_Diagonal_Stroke" . #\u+023e)
                ("Latin_Small_Letter_S_With_Swash_Tail" . #\u+023f)
                ("Latin_Small_Letter_Z_With_Swash_Tail" . #\u+0240)
                ("Latin_Capital_Letter_Glottal_Stop" . #\u+0241)
                ("Latin_Small_Letter_Glottal_Stop" . #\u+0242)
                ("Latin_Capital_Letter_B_With_Stroke" . #\u+0243)
                ("Latin_Capital_Letter_U_Bar" . #\u+0244)
                ("Latin_Capital_Letter_Turned_V" . #\u+0245)
                ("Latin_Capital_Letter_E_With_Stroke" . #\u+0246)
                ("Latin_Small_Letter_E_With_Stroke" . #\u+0247)
                ("Latin_Capital_Letter_J_With_Stroke" . #\u+0248)
                ("Latin_Small_Letter_J_With_Stroke" . #\u+0249)
                ("Latin_Capital_Letter_Small_Q_With_Hook_Tail" . #\u+024a)
                ("Latin_Small_Letter_Q_With_Hook_Tail" . #\u+024b)
                ("Latin_Capital_Letter_R_With_Stroke" . #\u+024c)
                ("Latin_Small_Letter_R_With_Stroke" . #\u+024d)
                ("Latin_Capital_Letter_Y_With_Stroke" . #\u+024e)
                ("Latin_Small_Letter_Y_With_Stroke" . #\u+024f)
                ("Latin_Small_Letter_Turned_A" . #\u+0250)
                ("Latin_Small_Letter_Alpha" . #\u+0251)
                ("Latin_Small_Letter_Turned_Alpha" . #\u+0252)
                ("Latin_Small_Letter_B_With_Hook" . #\u+0253)
                ("Latin_Small_Letter_Open_O" . #\u+0254)
                ("Latin_Small_Letter_C_With_Curl" . #\u+0255)
                ("Latin_Small_Letter_D_With_Tail" . #\u+0256)
                ("Latin_Small_Letter_D_With_Hook" . #\u+0257)
                ("Latin_Small_Letter_Reversed_E" . #\u+0258)
                ("Latin_Small_Letter_Schwa" . #\u+0259)
                ("Latin_Small_Letter_Schwa_With_Hook" . #\u+025a)
                ("Latin_Small_Letter_Open_E" . #\u+025b)
                ("Latin_Small_Letter_Reversed_Open_E" . #\u+025c)
                ("Latin_Small_Letter_Reversed_Open_E_With_Hook" . #\u+025d)
                ("Latin_Small_Letter_Closed_Reversed_Open_E" . #\u+025e)
                ("Latin_Small_Letter_Dotless_J_With_Stroke" . #\u+025f)
                ("Latin_Small_Letter_G_With_Hook" . #\u+0260)
                ("Latin_Small_Letter_Script_G" . #\u+0261)
                ("Latin_Letter_Small_Capital_G" . #\u+0262)
                ("Latin_Small_Letter_Gamma" . #\u+0263)
                ("Latin_Small_Letter_Rams_Horn" . #\u+0264)
                ("Latin_Small_Letter_Turned_H" . #\u+0265)
                ("Latin_Small_Letter_H_With_Hook" . #\u+0266)
                ("Latin_Small_Letter_Heng_With_Hook" . #\u+0267)
                ("Latin_Small_Letter_I_With_Stroke" . #\u+0268)
                ("Latin_Small_Letter_Iota" . #\u+0269)
                ("Latin_Letter_Small_Capital_I" . #\u+026a)
                ("Latin_Small_Letter_L_With_Middle_Tilde" . #\u+026b)
                ("Latin_Small_Letter_L_With_Belt" . #\u+026c)
                ("Latin_Small_Letter_L_With_Retroflex_Hook" . #\u+026d)
                ("Latin_Small_Letter_Lezh" . #\u+026e)
                ("Latin_Small_Letter_Turned_M" . #\u+026f)
                ("Latin_Small_Letter_Turned_M_With_Long_Leg" . #\u+0270)
                ("Latin_Small_Letter_M_With_Hook" . #\u+0271)
                ("Latin_Small_Letter_N_With_Left_Hook" . #\u+0272)
                ("Latin_Small_Letter_N_With_Retroflex_Hook" . #\u+0273)
                ("Latin_Letter_Small_Capital_N" . #\u+0274)
                ("Latin_Small_Letter_Barred_O" . #\u+0275)
                ("Latin_Letter_Small_Capital_Oe" . #\u+0276)
                ("Latin_Small_Letter_Closed_Omega" . #\u+0277)
                ("Latin_Small_Letter_Phi" . #\u+0278)
                ("Latin_Small_Letter_Turned_R" . #\u+0279)
                ("Latin_Small_Letter_Turned_R_With_Long_Leg" . #\u+027a)
                ("Latin_Small_Letter_Turned_R_With_Hook" . #\u+027b)
                ("Latin_Small_Letter_R_With_Long_Leg" . #\u+027c)
                ("Latin_Small_Letter_R_With_Tail" . #\u+027d)
                ("Latin_Small_Letter_R_With_Fishhook" . #\u+027e)
                ("Latin_Small_Letter_Reversed_R_With_Fishhook" . #\u+027f)
                ("Latin_Letter_Small_Capital_R" . #\u+0280)
                ("Latin_Letter_Small_Capital_Inverted_R" . #\u+0281)
                ("Latin_Small_Letter_S_With_Hook" . #\u+0282)
                ("Latin_Small_Letter_Esh" . #\u+0283)
                ("Latin_Small_Letter_Dotless_J_With_Stroke_And_Hook" . #\u+0284)
                ("Latin_Small_Letter_Squat_Reversed_Esh" . #\u+0285)
                ("Latin_Small_Letter_Esh_With_Curl" . #\u+0286)
                ("Latin_Small_Letter_Turned_T" . #\u+0287)
                ("Latin_Small_Letter_T_With_Retroflex_Hook" . #\u+0288)
                ("Latin_Small_Letter_U_Bar" . #\u+0289)
                ("Latin_Small_Letter_Upsilon" . #\u+028a)
                ("Latin_Small_Letter_V_With_Hook" . #\u+028b)
                ("Latin_Small_Letter_Turned_V" . #\u+028c)
                ("Latin_Small_Letter_Turned_W" . #\u+028d)
                ("Latin_Small_Letter_Turned_Y" . #\u+028e)
                ("Latin_Letter_Small_Capital_Y" . #\u+028f)
                ("Latin_Small_Letter_Z_With_Retroflex_Hook" . #\u+0290)
                ("Latin_Small_Letter_Z_With_Curl" . #\u+0291)
                ("Latin_Small_Letter_Ezh" . #\u+0292)
                ("Latin_Small_Letter_Ezh_With_Curl" . #\u+0293)
                ("Latin_Letter_Glottal_Stop" . #\u+0294)
                ("Latin_Letter_Pharyngeal_Voiced_Fricative" . #\u+0295)
                ("Latin_Letter_Inverted_Glottal_Stop" . #\u+0296)
                ("Latin_Letter_Stretched_C" . #\u+0297)
                ("Latin_Letter_Bilabial_Click" . #\u+0298)
                ("Latin_Letter_Small_Capital_B" . #\u+0299)
                ("Latin_Small_Letter_Closed_Open_E" . #\u+029a)
                ("Latin_Letter_Small_Capital_G_With_Hook" . #\u+029b)
                ("Latin_Letter_Small_Capital_H" . #\u+029c)
                ("Latin_Small_Letter_J_With_Crossed-Tail" . #\u+029d)
                ("Latin_Small_Letter_Turned_K" . #\u+029e)
                ("Latin_Letter_Small_Capital_L" . #\u+029f)
                ("Latin_Small_Letter_Q_With_Hook" . #\u+02a0)
                ("Latin_Letter_Glottal_Stop_With_Stroke" . #\u+02a1)
                ("Latin_Letter_Reversed_Glottal_Stop_With_Stroke" . #\u+02a2)
                ("Latin_Small_Letter_Dz_Digraph" . #\u+02a3)
                ("Latin_Small_Letter_Dezh_Digraph" . #\u+02a4)
                ("Latin_Small_Letter_Dz_Digraph_With_Curl" . #\u+02a5)
                ("Latin_Small_Letter_Ts_Digraph" . #\u+02a6)
                ("Latin_Small_Letter_Tesh_Digraph" . #\u+02a7)
                ("Latin_Small_Letter_Tc_Digraph_With_Curl" . #\u+02a8)
                ("Latin_Small_Letter_Feng_Digraph" . #\u+02a9)
                ("Latin_Small_Letter_Ls_Digraph" . #\u+02aa)
                ("Latin_Small_Letter_Lz_Digraph" . #\u+02ab)
                ("Latin_Letter_Bilabial_Percussive" . #\u+02ac)
                ("Latin_Letter_Bidental_Percussive" . #\u+02ad)
                ("Latin_Small_Letter_Turned_H_With_Fishhook" . #\u+02ae)
                ("Latin_Small_Letter_Turned_H_With_Fishhook_And_Tail" . #\u+02af)
                ("Modifier_Letter_Small_H" . #\u+02b0)
                ("Modifier_Letter_Small_H_With_Hook" . #\u+02b1)
                ("Modifier_Letter_Small_J" . #\u+02b2)
                ("Modifier_Letter_Small_R" . #\u+02b3)
                ("Modifier_Letter_Small_Turned_R" . #\u+02b4)
                ("Modifier_Letter_Small_Turned_R_With_Hook" . #\u+02b5)
                ("Modifier_Letter_Small_Capital_Inverted_R" . #\u+02b6)
                ("Modifier_Letter_Small_W" . #\u+02b7)
                ("Modifier_Letter_Small_Y" . #\u+02b8)
                ("Modifier_Letter_Prime" . #\u+02b9)
                ("Modifier_Letter_Double_Prime" . #\u+02ba)
                ("Modifier_Letter_Turned_Comma" . #\u+02bb)
                ("Modifier_Letter_Apostrophe" . #\u+02bc)
                ("Modifier_Letter_Reversed_Comma" . #\u+02bd)
                ("Modifier_Letter_Right_Half_Ring" . #\u+02be)
                ("Modifier_Letter_Left_Half_Ring" . #\u+02bf)
                ("Modifier_Letter_Glottal_Stop" . #\u+02c0)
                ("Modifier_Letter_Reversed_Glottal_Stop" . #\u+02c1)
                ("Modifier_Letter_Left_Arrowhead" . #\u+02c2)
                ("Modifier_Letter_Right_Arrowhead" . #\u+02c3)
                ("Modifier_Letter_Up_Arrowhead" . #\u+02c4)
                ("Modifier_Letter_Down_Arrowhead" . #\u+02c5)
                ("Modifier_Letter_Circumflex_Accent" . #\u+02c6)
                ("Caron" . #\u+02c7)
                ("Modifier_Letter_Vertical_Line" . #\u+02c8)
                ("Modifier_Letter_Macron" . #\u+02c9)
                ("Modifier_Letter_Acute_Accent" . #\u+02ca)
                ("Modifier_Letter_Grave_Accent" . #\u+02cb)
                ("Modifier_Letter_Low_Vertical_Line" . #\u+02cc)
                ("Modifier_Letter_Low_Macron" . #\u+02cd)
                ("Modifier_Letter_Low_Grave_Accent" . #\u+02ce)
                ("Modifier_Letter_Low_Acute_Accent" . #\u+02cf)
                ("Modifier_Letter_Triangular_Colon" . #\u+02d0)
                ("Modifier_Letter_Half_Triangular_Colon" . #\u+02d1)
                ("Modifier_Letter_Centred_Right_Half_Ring" . #\u+02d2)
                ("Modifier_Letter_Centred_Left_Half_Ring" . #\u+02d3)
                ("Modifier_Letter_Up_Tack" . #\u+02d4)
                ("Modifier_Letter_Down_Tack" . #\u+02d5)
                ("Modifier_Letter_Plus_Sign" . #\u+02d6)
                ("Modifier_Letter_Minus_Sign" . #\u+02d7)
                ("Breve" . #\u+02d8)
                ("Dot_Above" . #\u+02d9)
                ("Ring_Above" . #\u+02da)
                ("Ogonek" . #\u+02db)
                ("Small_Tilde" . #\u+02dc)
                ("Double_Acute_Accent" . #\u+02dd)
                ("Modifier_Letter_Rhotic_Hook" . #\u+02de)
                ("Modifier_Letter_Cross_Accent" . #\u+02df)
                ("Modifier_Letter_Small_Gamma" . #\u+02e0)
                ("Modifier_Letter_Small_L" . #\u+02e1)
                ("Modifier_Letter_Small_S" . #\u+02e2)
                ("Modifier_Letter_Small_X" . #\u+02e3)
                ("Modifier_Letter_Small_Reversed_Glottal_Stop" . #\u+02e4)
                ("Modifier_Letter_Extra-High_Tone_Bar" . #\u+02e5)
                ("Modifier_Letter_High_Tone_Bar" . #\u+02e6)
                ("Modifier_Letter_Mid_Tone_Bar" . #\u+02e7)
                ("Modifier_Letter_Low_Tone_Bar" . #\u+02e8)
                ("Modifier_Letter_Extra-Low_Tone_Bar" . #\u+02e9)
                ("Modifier_Letter_Yin_Departing_Tone_Mark" . #\u+02ea)
                ("Modifier_Letter_Yang_Departing_Tone_Mark" . #\u+02eb)
                ("Modifier_Letter_Voicing" . #\u+02ec)
                ("Modifier_Letter_Unaspirated" . #\u+02ed)
                ("Modifier_Letter_Double_Apostrophe" . #\u+02ee)
                ("Modifier_Letter_Low_Down_Arrowhead" . #\u+02ef)
                ("Modifier_Letter_Low_Up_Arrowhead" . #\u+02f0)
                ("Modifier_Letter_Low_Left_Arrowhead" . #\u+02f1)
                ("Modifier_Letter_Low_Right_Arrowhead" . #\u+02f2)
                ("Modifier_Letter_Low_Ring" . #\u+02f3)
                ("Modifier_Letter_Middle_Grave_Accent" . #\u+02f4)
                ("Modifier_Letter_Middle_Double_Grave_Accent" . #\u+02f5)
                ("Modifier_Letter_Middle_Double_Acute_Accent" . #\u+02f6)
                ("Modifier_Letter_Low_Tilde" . #\u+02f7)
                ("Modifier_Letter_Raised_Colon" . #\u+02f8)
                ("Modifier_Letter_Begin_High_Tone" . #\u+02f9)
                ("Modifier_Letter_End_High_Tone" . #\u+02fa)
                ("Modifier_Letter_Begin_Low_Tone" . #\u+02fb)
                ("Modifier_Letter_End_Low_Tone" . #\u+02fc)
                ("Modifier_Letter_Shelf" . #\u+02fd)
                ("Modifier_Letter_Open_Shelf" . #\u+02fe)
                ("Modifier_Letter_Low_Left_Arrow" . #\u+02ff)
                ("Combining_Grave_Accent" . #\u+0300)
                ("Combining_Acute_Accent" . #\u+0301)
                ("Combining_Circumflex_Accent" . #\u+0302)
                ("Combining_Tilde" . #\u+0303)
                ("Combining_Macron" . #\u+0304)
                ("Combining_Overline" . #\u+0305)
                ("Combining_Breve" . #\u+0306)
                ("Combining_Dot_Above" . #\u+0307)
                ("Combining_Diaeresis" . #\u+0308)
                ("Combining_Hook_Above" . #\u+0309)
                ("Combining_Ring_Above" . #\u+030a)
                ("Combining_Double_Acute_Accent" . #\u+030b)
                ("Combining_Caron" . #\u+030c)
                ("Combining_Vertical_Line_Above" . #\u+030d)
                ("Combining_Double_Vertical_Line_Above" . #\u+030e)
                ("Combining_Double_Grave_Accent" . #\u+030f)
                ("Combining_Candrabindu" . #\u+0310)
                ("Combining_Inverted_Breve" . #\u+0311)
                ("Combining_Turned_Comma_Above" . #\u+0312)
                ("Combining_Comma_Above" . #\u+0313)
                ("Combining_Reversed_Comma_Above" . #\u+0314)
                ("Combining_Comma_Above_Right" . #\u+0315)
                ("Combining_Grave_Accent_Below" . #\u+0316)
                ("Combining_Acute_Accent_Below" . #\u+0317)
                ("Combining_Left_Tack_Below" . #\u+0318)
                ("Combining_Right_Tack_Below" . #\u+0319)
                ("Combining_Left_Angle_Above" . #\u+031a)
                ("Combining_Horn" . #\u+031b)
                ("Combining_Left_Half_Ring_Below" . #\u+031c)
                ("Combining_Up_Tack_Below" . #\u+031d)
                ("Combining_Down_Tack_Below" . #\u+031e)
                ("Combining_Plus_Sign_Below" . #\u+031f)
                ("Combining_Minus_Sign_Below" . #\u+0320)
                ("Combining_Palatalized_Hook_Below" . #\u+0321)
                ("Combining_Retroflex_Hook_Below" . #\u+0322)
                ("Combining_Dot_Below" . #\u+0323)
                ("Combining_Diaeresis_Below" . #\u+0324)
                ("Combining_Ring_Below" . #\u+0325)
                ("Combining_Comma_Below" . #\u+0326)
                ("Combining_Cedilla" . #\u+0327)
                ("Combining_Ogonek" . #\u+0328)
                ("Combining_Vertical_Line_Below" . #\u+0329)
                ("Combining_Bridge_Below" . #\u+032a)
                ("Combining_Inverted_Double_Arch_Below" . #\u+032b)
                ("Combining_Caron_Below" . #\u+032c)
                ("Combining_Circumflex_Accent_Below" . #\u+032d)
                ("Combining_Breve_Below" . #\u+032e)
                ("Combining_Inverted_Breve_Below" . #\u+032f)
                ("Combining_Tilde_Below" . #\u+0330)
                ("Combining_Macron_Below" . #\u+0331)
                ("Combining_Low_Line" . #\u+0332)
                ("Combining_Double_Low_Line" . #\u+0333)
                ("Combining_Tilde_Overlay" . #\u+0334)
                ("Combining_Short_Stroke_Overlay" . #\u+0335)
                ("Combining_Long_Stroke_Overlay" . #\u+0336)
                ("Combining_Short_Solidus_Overlay" . #\u+0337)
                ("Combining_Long_Solidus_Overlay" . #\u+0338)
                ("Combining_Right_Half_Ring_Below" . #\u+0339)
                ("Combining_Inverted_Bridge_Below" . #\u+033a)
                ("Combining_Square_Below" . #\u+033b)
                ("Combining_Seagull_Below" . #\u+033c)
                ("Combining_X_Above" . #\u+033d)
                ("Combining_Vertical_Tilde" . #\u+033e)
                ("Combining_Double_Overline" . #\u+033f)
                ("Combining_Grave_Tone_Mark" . #\u+0340)
                ("Combining_Acute_Tone_Mark" . #\u+0341)
                ("Combining_Greek_Perispomeni" . #\u+0342)
                ("Combining_Greek_Koronis" . #\u+0343)
                ("Combining_Greek_Dialytika_Tonos" . #\u+0344)
                ("Combining_Greek_Ypogegrammeni" . #\u+0345)
                ("Combining_Bridge_Above" . #\u+0346)
                ("Combining_Equals_Sign_Below" . #\u+0347)
                ("Combining_Double_Vertical_Line_Below" . #\u+0348)
                ("Combining_Left_Angle_Below" . #\u+0349)
                ("Combining_Not_Tilde_Above" . #\u+034a)
                ("Combining_Homothetic_Above" . #\u+034b)
                ("Combining_Almost_Equal_To_Above" . #\u+034c)
                ("Combining_Left_Right_Arrow_Below" . #\u+034d)
                ("Combining_Upwards_Arrow_Below" . #\u+034e)
                ("Combining_Grapheme_Joiner" . #\u+034f)
                ("Combining_Right_Arrowhead_Above" . #\u+0350)
                ("Combining_Left_Half_Ring_Above" . #\u+0351)
                ("Combining_Fermata" . #\u+0352)
                ("Combining_X_Below" . #\u+0353)
                ("Combining_Left_Arrowhead_Below" . #\u+0354)
                ("Combining_Right_Arrowhead_Below" . #\u+0355)
                ("Combining_Right_Arrowhead_And_Up_Arrowhead_Below" . #\u+0356)
                ("Combining_Right_Half_Ring_Above" . #\u+0357)
                ("Combining_Dot_Above_Right" . #\u+0358)
                ("Combining_Asterisk_Below" . #\u+0359)
                ("Combining_Double_Ring_Below" . #\u+035a)
                ("Combining_Zigzag_Above" . #\u+035b)
                ("Combining_Double_Breve_Below" . #\u+035c)
                ("Combining_Double_Breve" . #\u+035d)
                ("Combining_Double_Macron" . #\u+035e)
                ("Combining_Double_Macron_Below" . #\u+035f)
                ("Combining_Double_Tilde" . #\u+0360)
                ("Combining_Double_Inverted_Breve" . #\u+0361)
                ("Combining_Double_Rightwards_Arrow_Below" . #\u+0362)
                ("Combining_Latin_Small_Letter_A" . #\u+0363)
                ("Combining_Latin_Small_Letter_E" . #\u+0364)
                ("Combining_Latin_Small_Letter_I" . #\u+0365)
                ("Combining_Latin_Small_Letter_O" . #\u+0366)
                ("Combining_Latin_Small_Letter_U" . #\u+0367)
                ("Combining_Latin_Small_Letter_C" . #\u+0368)
                ("Combining_Latin_Small_Letter_D" . #\u+0369)
                ("Combining_Latin_Small_Letter_H" . #\u+036a)
                ("Combining_Latin_Small_Letter_M" . #\u+036b)
                ("Combining_Latin_Small_Letter_R" . #\u+036c)
                ("Combining_Latin_Small_Letter_T" . #\u+036d)
                ("Combining_Latin_Small_Letter_V" . #\u+036e)
                ("Combining_Latin_Small_Letter_X" . #\u+036f)
                ("Greek_Numeral_Sign" . #\u+0374)
                ("Greek_Lower_Numeral_Sign" . #\u+0375)
                ("Greek_Ypogegrammeni" . #\u+037a)
                ("Greek_Small_Reversed_Lunate_Sigma_Symbol" . #\u+037b)
                ("Greek_Small_Dotted_Lunate_Sigma_Symbol" . #\u+037c)
                ("Greek_Small_Reversed_Dotted_Lunate_Sigma_Symbol" . #\u+037d)
                ("Greek_Question_Mark" . #\u+037e)
                ("Greek_Tonos" . #\u+0384)
                ("Greek_Dialytika_Tonos" . #\u+0385)
                ("Greek_Capital_Letter_Alpha_With_Tonos" . #\u+0386)
                ("Greek_Ano_Teleia" . #\u+0387)
                ("Greek_Capital_Letter_Epsilon_With_Tonos" . #\u+0388)
                ("Greek_Capital_Letter_Eta_With_Tonos" . #\u+0389)
                ("Greek_Capital_Letter_Iota_With_Tonos" . #\u+038a)
                ("Greek_Capital_Letter_Omicron_With_Tonos" . #\u+038c)
                ("Greek_Capital_Letter_Upsilon_With_Tonos" . #\u+038e)
                ("Greek_Capital_Letter_Omega_With_Tonos" . #\u+038f)
                ("Greek_Small_Letter_Iota_With_Dialytika_And_Tonos" . #\u+0390)
                ("Greek_Capital_Letter_Alpha" . #\u+0391)
                ("Greek_Capital_Letter_Beta" . #\u+0392)
                ("Greek_Capital_Letter_Gamma" . #\u+0393)
                ("Greek_Capital_Letter_Delta" . #\u+0394)
                ("Greek_Capital_Letter_Epsilon" . #\u+0395)
                ("Greek_Capital_Letter_Zeta" . #\u+0396)
                ("Greek_Capital_Letter_Eta" . #\u+0397)
                ("Greek_Capital_Letter_Theta" . #\u+0398)
                ("Greek_Capital_Letter_Iota" . #\u+0399)
                ("Greek_Capital_Letter_Kappa" . #\u+039a)
                ("Greek_Capital_Letter_Lamda" . #\u+039b)
                ("Greek_Capital_Letter_Mu" . #\u+039c)
                ("Greek_Capital_Letter_Nu" . #\u+039d)
                ("Greek_Capital_Letter_Xi" . #\u+039e)
                ("Greek_Capital_Letter_Omicron" . #\u+039f)
                ("Greek_Capital_Letter_Pi" . #\u+03a0)
                ("Greek_Capital_Letter_Rho" . #\u+03a1)
                ("Greek_Capital_Letter_Sigma" . #\u+03a3)
                ("Greek_Capital_Letter_Tau" . #\u+03a4)
                ("Greek_Capital_Letter_Upsilon" . #\u+03a5)
                ("Greek_Capital_Letter_Phi" . #\u+03a6)
                ("Greek_Capital_Letter_Chi" . #\u+03a7)
                ("Greek_Capital_Letter_Psi" . #\u+03a8)
                ("Greek_Capital_Letter_Omega" . #\u+03a9)
                ("Greek_Capital_Letter_Iota_With_Dialytika" . #\u+03aa)
                ("Greek_Capital_Letter_Upsilon_With_Dialytika" . #\u+03ab)
                ("Greek_Small_Letter_Alpha_With_Tonos" . #\u+03ac)
                ("Greek_Small_Letter_Epsilon_With_Tonos" . #\u+03ad)
                ("Greek_Small_Letter_Eta_With_Tonos" . #\u+03ae)
                ("Greek_Small_Letter_Iota_With_Tonos" . #\u+03af)
                ("Greek_Small_Letter_Upsilon_With_Dialytika_And_Tonos" . #\u+03b0)
                ("Greek_Small_Letter_Alpha" . #\u+03b1)
                ("Greek_Small_Letter_Beta" . #\u+03b2)
                ("Greek_Small_Letter_Gamma" . #\u+03b3)
                ("Greek_Small_Letter_Delta" . #\u+03b4)
                ("Greek_Small_Letter_Epsilon" . #\u+03b5)
                ("Greek_Small_Letter_Zeta" . #\u+03b6)
                ("Greek_Small_Letter_Eta" . #\u+03b7)
                ("Greek_Small_Letter_Theta" . #\u+03b8)
                ("Greek_Small_Letter_Iota" . #\u+03b9)
                ("Greek_Small_Letter_Kappa" . #\u+03ba)
                ("Greek_Small_Letter_Lamda" . #\u+03bb)
                ("Greek_Small_Letter_Mu" . #\u+03bc)
                ("Greek_Small_Letter_Nu" . #\u+03bd)
                ("Greek_Small_Letter_Xi" . #\u+03be)
                ("Greek_Small_Letter_Omicron" . #\u+03bf)
                ("Greek_Small_Letter_Pi" . #\u+03c0)
                ("Greek_Small_Letter_Rho" . #\u+03c1)
                ("Greek_Small_Letter_Final_Sigma" . #\u+03c2)
                ("Greek_Small_Letter_Sigma" . #\u+03c3)
                ("Greek_Small_Letter_Tau" . #\u+03c4)
                ("Greek_Small_Letter_Upsilon" . #\u+03c5)
                ("Greek_Small_Letter_Phi" . #\u+03c6)
                ("Greek_Small_Letter_Chi" . #\u+03c7)
                ("Greek_Small_Letter_Psi" . #\u+03c8)
                ("Greek_Small_Letter_Omega" . #\u+03c9)
                ("Greek_Small_Letter_Iota_With_Dialytika" . #\u+03ca)
                ("Greek_Small_Letter_Upsilon_With_Dialytika" . #\u+03cb)
                ("Greek_Small_Letter_Omicron_With_Tonos" . #\u+03cc)
                ("Greek_Small_Letter_Upsilon_With_Tonos" . #\u+03cd)
                ("Greek_Small_Letter_Omega_With_Tonos" . #\u+03ce)
                ("Greek_Beta_Symbol" . #\u+03d0)
                ("Greek_Theta_Symbol" . #\u+03d1)
                ("Greek_Upsilon_With_Hook_Symbol" . #\u+03d2)
                ("Greek_Upsilon_With_Acute_And_Hook_Symbol" . #\u+03d3)
                ("Greek_Upsilon_With_Diaeresis_And_Hook_Symbol" . #\u+03d4)
                ("Greek_Phi_Symbol" . #\u+03d5)
                ("Greek_Pi_Symbol" . #\u+03d6)
                ("Greek_Kai_Symbol" . #\u+03d7)
                ("Greek_Letter_Archaic_Koppa" . #\u+03d8)
                ("Greek_Small_Letter_Archaic_Koppa" . #\u+03d9)
                ("Greek_Letter_Stigma" . #\u+03da)
                ("Greek_Small_Letter_Stigma" . #\u+03db)
                ("Greek_Letter_Digamma" . #\u+03dc)
                ("Greek_Small_Letter_Digamma" . #\u+03dd)
                ("Greek_Letter_Koppa" . #\u+03de)
                ("Greek_Small_Letter_Koppa" . #\u+03df)
                ("Greek_Letter_Sampi" . #\u+03e0)
                ("Greek_Small_Letter_Sampi" . #\u+03e1)
                ("Coptic_Capital_Letter_Shei" . #\u+03e2)
                ("Coptic_Small_Letter_Shei" . #\u+03e3)
                ("Coptic_Capital_Letter_Fei" . #\u+03e4)
                ("Coptic_Small_Letter_Fei" . #\u+03e5)
                ("Coptic_Capital_Letter_Khei" . #\u+03e6)
                ("Coptic_Small_Letter_Khei" . #\u+03e7)
                ("Coptic_Capital_Letter_Hori" . #\u+03e8)
                ("Coptic_Small_Letter_Hori" . #\u+03e9)
                ("Coptic_Capital_Letter_Gangia" . #\u+03ea)
                ("Coptic_Small_Letter_Gangia" . #\u+03eb)
                ("Coptic_Capital_Letter_Shima" . #\u+03ec)
                ("Coptic_Small_Letter_Shima" . #\u+03ed)
                ("Coptic_Capital_Letter_Dei" . #\u+03ee)
                ("Coptic_Small_Letter_Dei" . #\u+03ef)
                ("Greek_Kappa_Symbol" . #\u+03f0)
                ("Greek_Rho_Symbol" . #\u+03f1)
                ("Greek_Lunate_Sigma_Symbol" . #\u+03f2)
                ("Greek_Letter_Yot" . #\u+03f3)
                ("Greek_Capital_Theta_Symbol" . #\u+03f4)
                ("Greek_Lunate_Epsilon_Symbol" . #\u+03f5)
                ("Greek_Reversed_Lunate_Epsilon_Symbol" . #\u+03f6)
                ("Greek_Capital_Letter_Sho" . #\u+03f7)
                ("Greek_Small_Letter_Sho" . #\u+03f8)
                ("Greek_Capital_Lunate_Sigma_Symbol" . #\u+03f9)
                ("Greek_Capital_Letter_San" . #\u+03fa)
                ("Greek_Small_Letter_San" . #\u+03fb)
                ("Greek_Rho_With_Stroke_Symbol" . #\u+03fc)
                ("Greek_Capital_Reversed_Lunate_Sigma_Symbol" . #\u+03fd)
                ("Greek_Capital_Dotted_Lunate_Sigma_Symbol" . #\u+03fe)
                ("Greek_Capital_Reversed_Dotted_Lunate_Sigma_Symbol" . #\u+03ff)
                ("Cyrillic_Capital_Letter_Ie_With_Grave" . #\u+0400)
                ("Cyrillic_Capital_Letter_Io" . #\u+0401)
                ("Cyrillic_Capital_Letter_Dje" . #\u+0402)
                ("Cyrillic_Capital_Letter_Gje" . #\u+0403)
                ("Cyrillic_Capital_Letter_Ukrainian_Ie" . #\u+0404)
                ("Cyrillic_Capital_Letter_Dze" . #\u+0405)
                ("Cyrillic_Capital_Letter_Byelorussian-Ukrainian_I" . #\u+0406)
                ("Cyrillic_Capital_Letter_Yi" . #\u+0407)
                ("Cyrillic_Capital_Letter_Je" . #\u+0408)
                ("Cyrillic_Capital_Letter_Lje" . #\u+0409)
                ("Cyrillic_Capital_Letter_Nje" . #\u+040a)
                ("Cyrillic_Capital_Letter_Tshe" . #\u+040b)
                ("Cyrillic_Capital_Letter_Kje" . #\u+040c)
                ("Cyrillic_Capital_Letter_I_With_Grave" . #\u+040d)
                ("Cyrillic_Capital_Letter_Short_U" . #\u+040e)
                ("Cyrillic_Capital_Letter_Dzhe" . #\u+040f)
                ("Cyrillic_Capital_Letter_A" . #\u+0410)
                ("Cyrillic_Capital_Letter_Be" . #\u+0411)
                ("Cyrillic_Capital_Letter_Ve" . #\u+0412)
                ("Cyrillic_Capital_Letter_Ghe" . #\u+0413)
                ("Cyrillic_Capital_Letter_De" . #\u+0414)
                ("Cyrillic_Capital_Letter_Ie" . #\u+0415)
                ("Cyrillic_Capital_Letter_Zhe" . #\u+0416)
                ("Cyrillic_Capital_Letter_Ze" . #\u+0417)
                ("Cyrillic_Capital_Letter_I" . #\u+0418)
                ("Cyrillic_Capital_Letter_Short_I" . #\u+0419)
                ("Cyrillic_Capital_Letter_Ka" . #\u+041a)
                ("Cyrillic_Capital_Letter_El" . #\u+041b)
                ("Cyrillic_Capital_Letter_Em" . #\u+041c)
                ("Cyrillic_Capital_Letter_En" . #\u+041d)
                ("Cyrillic_Capital_Letter_O" . #\u+041e)
                ("Cyrillic_Capital_Letter_Pe" . #\u+041f)
                ("Cyrillic_Capital_Letter_Er" . #\u+0420)
                ("Cyrillic_Capital_Letter_Es" . #\u+0421)
                ("Cyrillic_Capital_Letter_Te" . #\u+0422)
                ("Cyrillic_Capital_Letter_U" . #\u+0423)
                ("Cyrillic_Capital_Letter_Ef" . #\u+0424)
                ("Cyrillic_Capital_Letter_Ha" . #\u+0425)
                ("Cyrillic_Capital_Letter_Tse" . #\u+0426)
                ("Cyrillic_Capital_Letter_Che" . #\u+0427)
                ("Cyrillic_Capital_Letter_Sha" . #\u+0428)
                ("Cyrillic_Capital_Letter_Shcha" . #\u+0429)
                ("Cyrillic_Capital_Letter_Hard_Sign" . #\u+042a)
                ("Cyrillic_Capital_Letter_Yeru" . #\u+042b)
                ("Cyrillic_Capital_Letter_Soft_Sign" . #\u+042c)
                ("Cyrillic_Capital_Letter_E" . #\u+042d)
                ("Cyrillic_Capital_Letter_Yu" . #\u+042e)
                ("Cyrillic_Capital_Letter_Ya" . #\u+042f)
                ("Cyrillic_Small_Letter_A" . #\u+0430)
                ("Cyrillic_Small_Letter_Be" . #\u+0431)
                ("Cyrillic_Small_Letter_Ve" . #\u+0432)
                ("Cyrillic_Small_Letter_Ghe" . #\u+0433)
                ("Cyrillic_Small_Letter_De" . #\u+0434)
                ("Cyrillic_Small_Letter_Ie" . #\u+0435)
                ("Cyrillic_Small_Letter_Zhe" . #\u+0436)
                ("Cyrillic_Small_Letter_Ze" . #\u+0437)
                ("Cyrillic_Small_Letter_I" . #\u+0438)
                ("Cyrillic_Small_Letter_Short_I" . #\u+0439)
                ("Cyrillic_Small_Letter_Ka" . #\u+043a)
                ("Cyrillic_Small_Letter_El" . #\u+043b)
                ("Cyrillic_Small_Letter_Em" . #\u+043c)
                ("Cyrillic_Small_Letter_En" . #\u+043d)
                ("Cyrillic_Small_Letter_O" . #\u+043e)
                ("Cyrillic_Small_Letter_Pe" . #\u+043f)
                ("Cyrillic_Small_Letter_Er" . #\u+0440)
                ("Cyrillic_Small_Letter_Es" . #\u+0441)
                ("Cyrillic_Small_Letter_Te" . #\u+0442)
                ("Cyrillic_Small_Letter_U" . #\u+0443)
                ("Cyrillic_Small_Letter_Ef" . #\u+0444)
                ("Cyrillic_Small_Letter_Ha" . #\u+0445)
                ("Cyrillic_Small_Letter_Tse" . #\u+0446)
                ("Cyrillic_Small_Letter_Che" . #\u+0447)
                ("Cyrillic_Small_Letter_Sha" . #\u+0448)
                ("Cyrillic_Small_Letter_Shcha" . #\u+0449)
                ("Cyrillic_Small_Letter_Hard_Sign" . #\u+044a)
                ("Cyrillic_Small_Letter_Yeru" . #\u+044b)
                ("Cyrillic_Small_Letter_Soft_Sign" . #\u+044c)
                ("Cyrillic_Small_Letter_E" . #\u+044d)
                ("Cyrillic_Small_Letter_Yu" . #\u+044e)
                ("Cyrillic_Small_Letter_Ya" . #\u+044f)
                ("Cyrillic_Small_Letter_Ie_With_Grave" . #\u+0450)
                ("Cyrillic_Small_Letter_Io" . #\u+0451)
                ("Cyrillic_Small_Letter_Dje" . #\u+0452)
                ("Cyrillic_Small_Letter_Gje" . #\u+0453)
                ("Cyrillic_Small_Letter_Ukrainian_Ie" . #\u+0454)
                ("Cyrillic_Small_Letter_Dze" . #\u+0455)
                ("Cyrillic_Small_Letter_Byelorussian-Ukrainian_I" . #\u+0456)
                ("Cyrillic_Small_Letter_Yi" . #\u+0457)
                ("Cyrillic_Small_Letter_Je" . #\u+0458)
                ("Cyrillic_Small_Letter_Lje" . #\u+0459)
                ("Cyrillic_Small_Letter_Nje" . #\u+045a)
                ("Cyrillic_Small_Letter_Tshe" . #\u+045b)
                ("Cyrillic_Small_Letter_Kje" . #\u+045c)
                ("Cyrillic_Small_Letter_I_With_Grave" . #\u+045d)
                ("Cyrillic_Small_Letter_Short_U" . #\u+045e)
                ("Cyrillic_Small_Letter_Dzhe" . #\u+045f)
                ("Cyrillic_Capital_Letter_Omega" . #\u+0460)
                ("Cyrillic_Small_Letter_Omega" . #\u+0461)
                ("Cyrillic_Capital_Letter_Yat" . #\u+0462)
                ("Cyrillic_Small_Letter_Yat" . #\u+0463)
                ("Cyrillic_Capital_Letter_Iotified_E" . #\u+0464)
                ("Cyrillic_Small_Letter_Iotified_E" . #\u+0465)
                ("Cyrillic_Capital_Letter_Little_Yus" . #\u+0466)
                ("Cyrillic_Small_Letter_Little_Yus" . #\u+0467)
                ("Cyrillic_Capital_Letter_Iotified_Little_Yus" . #\u+0468)
                ("Cyrillic_Small_Letter_Iotified_Little_Yus" . #\u+0469)
                ("Cyrillic_Capital_Letter_Big_Yus" . #\u+046a)
                ("Cyrillic_Small_Letter_Big_Yus" . #\u+046b)
                ("Cyrillic_Capital_Letter_Iotified_Big_Yus" . #\u+046c)
                ("Cyrillic_Small_Letter_Iotified_Big_Yus" . #\u+046d)
                ("Cyrillic_Capital_Letter_Ksi" . #\u+046e)
                ("Cyrillic_Small_Letter_Ksi" . #\u+046f)
                ("Cyrillic_Capital_Letter_Psi" . #\u+0470)
                ("Cyrillic_Small_Letter_Psi" . #\u+0471)
                ("Cyrillic_Capital_Letter_Fita" . #\u+0472)
                ("Cyrillic_Small_Letter_Fita" . #\u+0473)
                ("Cyrillic_Capital_Letter_Izhitsa" . #\u+0474)
                ("Cyrillic_Small_Letter_Izhitsa" . #\u+0475)
                ("Cyrillic_Capital_Letter_Izhitsa_With_Double_Grave_Accent" . #\u+0476)
                ("Cyrillic_Small_Letter_Izhitsa_With_Double_Grave_Accent" . #\u+0477)
                ("Cyrillic_Capital_Letter_Uk" . #\u+0478)
                ("Cyrillic_Small_Letter_Uk" . #\u+0479)
                ("Cyrillic_Capital_Letter_Round_Omega" . #\u+047a)
                ("Cyrillic_Small_Letter_Round_Omega" . #\u+047b)
                ("Cyrillic_Capital_Letter_Omega_With_Titlo" . #\u+047c)
                ("Cyrillic_Small_Letter_Omega_With_Titlo" . #\u+047d)
                ("Cyrillic_Capital_Letter_Ot" . #\u+047e)
                ("Cyrillic_Small_Letter_Ot" . #\u+047f)
                ("Cyrillic_Capital_Letter_Koppa" . #\u+0480)
                ("Cyrillic_Small_Letter_Koppa" . #\u+0481)
                ("Cyrillic_Thousands_Sign" . #\u+0482)
                ("Combining_Cyrillic_Titlo" . #\u+0483)
                ("Combining_Cyrillic_Palatalization" . #\u+0484)
                ("Combining_Cyrillic_Dasia_Pneumata" . #\u+0485)
                ("Combining_Cyrillic_Psili_Pneumata" . #\u+0486)
                ("Combining_Cyrillic_Hundred_Thousands_Sign" . #\u+0488)
                ("Combining_Cyrillic_Millions_Sign" . #\u+0489)
                ("Cyrillic_Capital_Letter_Short_I_With_Tail" . #\u+048a)
                ("Cyrillic_Small_Letter_Short_I_With_Tail" . #\u+048b)
                ("Cyrillic_Capital_Letter_Semisoft_Sign" . #\u+048c)
                ("Cyrillic_Small_Letter_Semisoft_Sign" . #\u+048d)
                ("Cyrillic_Capital_Letter_Er_With_Tick" . #\u+048e)
                ("Cyrillic_Small_Letter_Er_With_Tick" . #\u+048f)
                ("Cyrillic_Capital_Letter_Ghe_With_Upturn" . #\u+0490)
                ("Cyrillic_Small_Letter_Ghe_With_Upturn" . #\u+0491)
                ("Cyrillic_Capital_Letter_Ghe_With_Stroke" . #\u+0492)
                ("Cyrillic_Small_Letter_Ghe_With_Stroke" . #\u+0493)
                ("Cyrillic_Capital_Letter_Ghe_With_Middle_Hook" . #\u+0494)
                ("Cyrillic_Small_Letter_Ghe_With_Middle_Hook" . #\u+0495)
                ("Cyrillic_Capital_Letter_Zhe_With_Descender" . #\u+0496)
                ("Cyrillic_Small_Letter_Zhe_With_Descender" . #\u+0497)
                ("Cyrillic_Capital_Letter_Ze_With_Descender" . #\u+0498)
                ("Cyrillic_Small_Letter_Ze_With_Descender" . #\u+0499)
                ("Cyrillic_Capital_Letter_Ka_With_Descender" . #\u+049a)
                ("Cyrillic_Small_Letter_Ka_With_Descender" . #\u+049b)
                ("Cyrillic_Capital_Letter_Ka_With_Vertical_Stroke" . #\u+049c)
                ("Cyrillic_Small_Letter_Ka_With_Vertical_Stroke" . #\u+049d)
                ("Cyrillic_Capital_Letter_Ka_With_Stroke" . #\u+049e)
                ("Cyrillic_Small_Letter_Ka_With_Stroke" . #\u+049f)
                ("Cyrillic_Capital_Letter_Bashkir_Ka" . #\u+04a0)
                ("Cyrillic_Small_Letter_Bashkir_Ka" . #\u+04a1)
                ("Cyrillic_Capital_Letter_En_With_Descender" . #\u+04a2)
                ("Cyrillic_Small_Letter_En_With_Descender" . #\u+04a3)
                ("Cyrillic_Capital_Ligature_En_Ghe" . #\u+04a4)
                ("Cyrillic_Small_Ligature_En_Ghe" . #\u+04a5)
                ("Cyrillic_Capital_Letter_Pe_With_Middle_Hook" . #\u+04a6)
                ("Cyrillic_Small_Letter_Pe_With_Middle_Hook" . #\u+04a7)
                ("Cyrillic_Capital_Letter_Abkhasian_Ha" . #\u+04a8)
                ("Cyrillic_Small_Letter_Abkhasian_Ha" . #\u+04a9)
                ("Cyrillic_Capital_Letter_Es_With_Descender" . #\u+04aa)
                ("Cyrillic_Small_Letter_Es_With_Descender" . #\u+04ab)
                ("Cyrillic_Capital_Letter_Te_With_Descender" . #\u+04ac)
                ("Cyrillic_Small_Letter_Te_With_Descender" . #\u+04ad)
                ("Cyrillic_Capital_Letter_Straight_U" . #\u+04ae)
                ("Cyrillic_Small_Letter_Straight_U" . #\u+04af)
                ("Cyrillic_Capital_Letter_Straight_U_With_Stroke" . #\u+04b0)
                ("Cyrillic_Small_Letter_Straight_U_With_Stroke" . #\u+04b1)
                ("Cyrillic_Capital_Letter_Ha_With_Descender" . #\u+04b2)
                ("Cyrillic_Small_Letter_Ha_With_Descender" . #\u+04b3)
                ("Cyrillic_Capital_Ligature_Te_Tse" . #\u+04b4)
                ("Cyrillic_Small_Ligature_Te_Tse" . #\u+04b5)
                ("Cyrillic_Capital_Letter_Che_With_Descender" . #\u+04b6)
                ("Cyrillic_Small_Letter_Che_With_Descender" . #\u+04b7)
                ("Cyrillic_Capital_Letter_Che_With_Vertical_Stroke" . #\u+04b8)
                ("Cyrillic_Small_Letter_Che_With_Vertical_Stroke" . #\u+04b9)
                ("Cyrillic_Capital_Letter_Shha" . #\u+04ba)
                ("Cyrillic_Small_Letter_Shha" . #\u+04bb)
                ("Cyrillic_Capital_Letter_Abkhasian_Che" . #\u+04bc)
                ("Cyrillic_Small_Letter_Abkhasian_Che" . #\u+04bd)
                ("Cyrillic_Capital_Letter_Abkhasian_Che_With_Descender" . #\u+04be)
                ("Cyrillic_Small_Letter_Abkhasian_Che_With_Descender" . #\u+04bf)
                ("Cyrillic_Letter_Palochka" . #\u+04c0)
                ("Cyrillic_Capital_Letter_Zhe_With_Breve" . #\u+04c1)
                ("Cyrillic_Small_Letter_Zhe_With_Breve" . #\u+04c2)
                ("Cyrillic_Capital_Letter_Ka_With_Hook" . #\u+04c3)
                ("Cyrillic_Small_Letter_Ka_With_Hook" . #\u+04c4)
                ("Cyrillic_Capital_Letter_El_With_Tail" . #\u+04c5)
                ("Cyrillic_Small_Letter_El_With_Tail" . #\u+04c6)
                ("Cyrillic_Capital_Letter_En_With_Hook" . #\u+04c7)
                ("Cyrillic_Small_Letter_En_With_Hook" . #\u+04c8)
                ("Cyrillic_Capital_Letter_En_With_Tail" . #\u+04c9)
                ("Cyrillic_Small_Letter_En_With_Tail" . #\u+04ca)
                ("Cyrillic_Capital_Letter_Khakassian_Che" . #\u+04cb)
                ("Cyrillic_Small_Letter_Khakassian_Che" . #\u+04cc)
                ("Cyrillic_Capital_Letter_Em_With_Tail" . #\u+04cd)
                ("Cyrillic_Small_Letter_Em_With_Tail" . #\u+04ce)
                ("Cyrillic_Small_Letter_Palochka" . #\u+04cf)
                ("Cyrillic_Capital_Letter_A_With_Breve" . #\u+04d0)
                ("Cyrillic_Small_Letter_A_With_Breve" . #\u+04d1)
                ("Cyrillic_Capital_Letter_A_With_Diaeresis" . #\u+04d2)
                ("Cyrillic_Small_Letter_A_With_Diaeresis" . #\u+04d3)
                ("Cyrillic_Capital_Ligature_A_Ie" . #\u+04d4)
                ("Cyrillic_Small_Ligature_A_Ie" . #\u+04d5)
                ("Cyrillic_Capital_Letter_Ie_With_Breve" . #\u+04d6)
                ("Cyrillic_Small_Letter_Ie_With_Breve" . #\u+04d7)
                ("Cyrillic_Capital_Letter_Schwa" . #\u+04d8)
                ("Cyrillic_Small_Letter_Schwa" . #\u+04d9)
                ("Cyrillic_Capital_Letter_Schwa_With_Diaeresis" . #\u+04da)
                ("Cyrillic_Small_Letter_Schwa_With_Diaeresis" . #\u+04db)
                ("Cyrillic_Capital_Letter_Zhe_With_Diaeresis" . #\u+04dc)
                ("Cyrillic_Small_Letter_Zhe_With_Diaeresis" . #\u+04dd)
                ("Cyrillic_Capital_Letter_Ze_With_Diaeresis" . #\u+04de)
                ("Cyrillic_Small_Letter_Ze_With_Diaeresis" . #\u+04df)
                ("Cyrillic_Capital_Letter_Abkhasian_Dze" . #\u+04e0)
                ("Cyrillic_Small_Letter_Abkhasian_Dze" . #\u+04e1)
                ("Cyrillic_Capital_Letter_I_With_Macron" . #\u+04e2)
                ("Cyrillic_Small_Letter_I_With_Macron" . #\u+04e3)
                ("Cyrillic_Capital_Letter_I_With_Diaeresis" . #\u+04e4)
                ("Cyrillic_Small_Letter_I_With_Diaeresis" . #\u+04e5)
                ("Cyrillic_Capital_Letter_O_With_Diaeresis" . #\u+04e6)
                ("Cyrillic_Small_Letter_O_With_Diaeresis" . #\u+04e7)
                ("Cyrillic_Capital_Letter_Barred_O" . #\u+04e8)
                ("Cyrillic_Small_Letter_Barred_O" . #\u+04e9)
                ("Cyrillic_Capital_Letter_Barred_O_With_Diaeresis" . #\u+04ea)
                ("Cyrillic_Small_Letter_Barred_O_With_Diaeresis" . #\u+04eb)
                ("Cyrillic_Capital_Letter_E_With_Diaeresis" . #\u+04ec)
                ("Cyrillic_Small_Letter_E_With_Diaeresis" . #\u+04ed)
                ("Cyrillic_Capital_Letter_U_With_Macron" . #\u+04ee)
                ("Cyrillic_Small_Letter_U_With_Macron" . #\u+04ef)
                ("Cyrillic_Capital_Letter_U_With_Diaeresis" . #\u+04f0)
                ("Cyrillic_Small_Letter_U_With_Diaeresis" . #\u+04f1)
                ("Cyrillic_Capital_Letter_U_With_Double_Acute" . #\u+04f2)
                ("Cyrillic_Small_Letter_U_With_Double_Acute" . #\u+04f3)
                ("Cyrillic_Capital_Letter_Che_With_Diaeresis" . #\u+04f4)
                ("Cyrillic_Small_Letter_Che_With_Diaeresis" . #\u+04f5)
                ("Cyrillic_Capital_Letter_Ghe_With_Descender" . #\u+04f6)
                ("Cyrillic_Small_Letter_Ghe_With_Descender" . #\u+04f7)
                ("Cyrillic_Capital_Letter_Yeru_With_Diaeresis" . #\u+04f8)
                ("Cyrillic_Small_Letter_Yeru_With_Diaeresis" . #\u+04f9)
                ("Cyrillic_Capital_Letter_Ghe_With_Stroke_And_Hook" . #\u+04fa)
                ("Cyrillic_Small_Letter_Ghe_With_Stroke_And_Hook" . #\u+04fb)
                ("Cyrillic_Capital_Letter_Ha_With_Hook" . #\u+04fc)
                ("Cyrillic_Small_Letter_Ha_With_Hook" . #\u+04fd)
                ("Cyrillic_Capital_Letter_Ha_With_Stroke" . #\u+04fe)
                ("Cyrillic_Small_Letter_Ha_With_Stroke" . #\u+04ff)
                ("Cyrillic_Capital_Letter_Komi_De" . #\u+0500)
                ("Cyrillic_Small_Letter_Komi_De" . #\u+0501)
                ("Cyrillic_Capital_Letter_Komi_Dje" . #\u+0502)
                ("Cyrillic_Small_Letter_Komi_Dje" . #\u+0503)
                ("Cyrillic_Capital_Letter_Komi_Zje" . #\u+0504)
                ("Cyrillic_Small_Letter_Komi_Zje" . #\u+0505)
                ("Cyrillic_Capital_Letter_Komi_Dzje" . #\u+0506)
                ("Cyrillic_Small_Letter_Komi_Dzje" . #\u+0507)
                ("Cyrillic_Capital_Letter_Komi_Lje" . #\u+0508)
                ("Cyrillic_Small_Letter_Komi_Lje" . #\u+0509)
                ("Cyrillic_Capital_Letter_Komi_Nje" . #\u+050a)
                ("Cyrillic_Small_Letter_Komi_Nje" . #\u+050b)
                ("Cyrillic_Capital_Letter_Komi_Sje" . #\u+050c)
                ("Cyrillic_Small_Letter_Komi_Sje" . #\u+050d)
                ("Cyrillic_Capital_Letter_Komi_Tje" . #\u+050e)
                ("Cyrillic_Small_Letter_Komi_Tje" . #\u+050f)
                ("Cyrillic_Capital_Letter_Reversed_Ze" . #\u+0510)
                ("Cyrillic_Small_Letter_Reversed_Ze" . #\u+0511)
                ("Cyrillic_Capital_Letter_El_With_Hook" . #\u+0512)
                ("Cyrillic_Small_Letter_El_With_Hook" . #\u+0513)
                ("Armenian_Capital_Letter_Ayb" . #\u+0531)
                ("Armenian_Capital_Letter_Ben" . #\u+0532)
                ("Armenian_Capital_Letter_Gim" . #\u+0533)
                ("Armenian_Capital_Letter_Da" . #\u+0534)
                ("Armenian_Capital_Letter_Ech" . #\u+0535)
                ("Armenian_Capital_Letter_Za" . #\u+0536)
                ("Armenian_Capital_Letter_Eh" . #\u+0537)
                ("Armenian_Capital_Letter_Et" . #\u+0538)
                ("Armenian_Capital_Letter_To" . #\u+0539)
                ("Armenian_Capital_Letter_Zhe" . #\u+053a)
                ("Armenian_Capital_Letter_Ini" . #\u+053b)
                ("Armenian_Capital_Letter_Liwn" . #\u+053c)
                ("Armenian_Capital_Letter_Xeh" . #\u+053d)
                ("Armenian_Capital_Letter_Ca" . #\u+053e)
                ("Armenian_Capital_Letter_Ken" . #\u+053f)
                ("Armenian_Capital_Letter_Ho" . #\u+0540)
                ("Armenian_Capital_Letter_Ja" . #\u+0541)
                ("Armenian_Capital_Letter_Ghad" . #\u+0542)
                ("Armenian_Capital_Letter_Cheh" . #\u+0543)
                ("Armenian_Capital_Letter_Men" . #\u+0544)
                ("Armenian_Capital_Letter_Yi" . #\u+0545)
                ("Armenian_Capital_Letter_Now" . #\u+0546)
                ("Armenian_Capital_Letter_Sha" . #\u+0547)
                ("Armenian_Capital_Letter_Vo" . #\u+0548)
                ("Armenian_Capital_Letter_Cha" . #\u+0549)
                ("Armenian_Capital_Letter_Peh" . #\u+054a)
                ("Armenian_Capital_Letter_Jheh" . #\u+054b)
                ("Armenian_Capital_Letter_Ra" . #\u+054c)
                ("Armenian_Capital_Letter_Seh" . #\u+054d)
                ("Armenian_Capital_Letter_Vew" . #\u+054e)
                ("Armenian_Capital_Letter_Tiwn" . #\u+054f)
                ("Armenian_Capital_Letter_Reh" . #\u+0550)
                ("Armenian_Capital_Letter_Co" . #\u+0551)
                ("Armenian_Capital_Letter_Yiwn" . #\u+0552)
                ("Armenian_Capital_Letter_Piwr" . #\u+0553)
                ("Armenian_Capital_Letter_Keh" . #\u+0554)
                ("Armenian_Capital_Letter_Oh" . #\u+0555)
                ("Armenian_Capital_Letter_Feh" . #\u+0556)
                ("Armenian_Modifier_Letter_Left_Half_Ring" . #\u+0559)
                ("Armenian_Apostrophe" . #\u+055a)
                ("Armenian_Emphasis_Mark" . #\u+055b)
                ("Armenian_Exclamation_Mark" . #\u+055c)
                ("Armenian_Comma" . #\u+055d)
                ("Armenian_Question_Mark" . #\u+055e)
                ("Armenian_Abbreviation_Mark" . #\u+055f)
                ("Armenian_Small_Letter_Ayb" . #\u+0561)
                ("Armenian_Small_Letter_Ben" . #\u+0562)
                ("Armenian_Small_Letter_Gim" . #\u+0563)
                ("Armenian_Small_Letter_Da" . #\u+0564)
                ("Armenian_Small_Letter_Ech" . #\u+0565)
                ("Armenian_Small_Letter_Za" . #\u+0566)
                ("Armenian_Small_Letter_Eh" . #\u+0567)
                ("Armenian_Small_Letter_Et" . #\u+0568)
                ("Armenian_Small_Letter_To" . #\u+0569)
                ("Armenian_Small_Letter_Zhe" . #\u+056a)
                ("Armenian_Small_Letter_Ini" . #\u+056b)
                ("Armenian_Small_Letter_Liwn" . #\u+056c)
                ("Armenian_Small_Letter_Xeh" . #\u+056d)
                ("Armenian_Small_Letter_Ca" . #\u+056e)
                ("Armenian_Small_Letter_Ken" . #\u+056f)
                ("Armenian_Small_Letter_Ho" . #\u+0570)
                ("Armenian_Small_Letter_Ja" . #\u+0571)
                ("Armenian_Small_Letter_Ghad" . #\u+0572)
                ("Armenian_Small_Letter_Cheh" . #\u+0573)
                ("Armenian_Small_Letter_Men" . #\u+0574)
                ("Armenian_Small_Letter_Yi" . #\u+0575)
                ("Armenian_Small_Letter_Now" . #\u+0576)
                ("Armenian_Small_Letter_Sha" . #\u+0577)
                ("Armenian_Small_Letter_Vo" . #\u+0578)
                ("Armenian_Small_Letter_Cha" . #\u+0579)
                ("Armenian_Small_Letter_Peh" . #\u+057a)
                ("Armenian_Small_Letter_Jheh" . #\u+057b)
                ("Armenian_Small_Letter_Ra" . #\u+057c)
                ("Armenian_Small_Letter_Seh" . #\u+057d)
                ("Armenian_Small_Letter_Vew" . #\u+057e)
                ("Armenian_Small_Letter_Tiwn" . #\u+057f)
                ("Armenian_Small_Letter_Reh" . #\u+0580)
                ("Armenian_Small_Letter_Co" . #\u+0581)
                ("Armenian_Small_Letter_Yiwn" . #\u+0582)
                ("Armenian_Small_Letter_Piwr" . #\u+0583)
                ("Armenian_Small_Letter_Keh" . #\u+0584)
                ("Armenian_Small_Letter_Oh" . #\u+0585)
                ("Armenian_Small_Letter_Feh" . #\u+0586)
                ("Armenian_Small_Ligature_Ech_Yiwn" . #\u+0587)
                ("Armenian_Full_Stop" . #\u+0589)
                ("Armenian_Hyphen" . #\u+058a)
                ("Hebrew_Accent_Etnahta" . #\u+0591)
                ("Hebrew_Accent_Segol" . #\u+0592)
                ("Hebrew_Accent_Shalshelet" . #\u+0593)
                ("Hebrew_Accent_Zaqef_Qatan" . #\u+0594)
                ("Hebrew_Accent_Zaqef_Gadol" . #\u+0595)
                ("Hebrew_Accent_Tipeha" . #\u+0596)
                ("Hebrew_Accent_Revia" . #\u+0597)
                ("Hebrew_Accent_Zarqa" . #\u+0598)
                ("Hebrew_Accent_Pashta" . #\u+0599)
                ("Hebrew_Accent_Yetiv" . #\u+059a)
                ("Hebrew_Accent_Tevir" . #\u+059b)
                ("Hebrew_Accent_Geresh" . #\u+059c)
                ("Hebrew_Accent_Geresh_Muqdam" . #\u+059d)
                ("Hebrew_Accent_Gershayim" . #\u+059e)
                ("Hebrew_Accent_Qarney_Para" . #\u+059f)
                ("Hebrew_Accent_Telisha_Gedola" . #\u+05a0)
                ("Hebrew_Accent_Pazer" . #\u+05a1)
                ("Hebrew_Accent_Atnah_Hafukh" . #\u+05a2)
                ("Hebrew_Accent_Munah" . #\u+05a3)
                ("Hebrew_Accent_Mahapakh" . #\u+05a4)
                ("Hebrew_Accent_Merkha" . #\u+05a5)
                ("Hebrew_Accent_Merkha_Kefula" . #\u+05a6)
                ("Hebrew_Accent_Darga" . #\u+05a7)
                ("Hebrew_Accent_Qadma" . #\u+05a8)
                ("Hebrew_Accent_Telisha_Qetana" . #\u+05a9)
                ("Hebrew_Accent_Yerah_Ben_Yomo" . #\u+05aa)
                ("Hebrew_Accent_Ole" . #\u+05ab)
                ("Hebrew_Accent_Iluy" . #\u+05ac)
                ("Hebrew_Accent_Dehi" . #\u+05ad)
                ("Hebrew_Accent_Zinor" . #\u+05ae)
                ("Hebrew_Mark_Masora_Circle" . #\u+05af)
                ("Hebrew_Point_Sheva" . #\u+05b0)
                ("Hebrew_Point_Hataf_Segol" . #\u+05b1)
                ("Hebrew_Point_Hataf_Patah" . #\u+05b2)
                ("Hebrew_Point_Hataf_Qamats" . #\u+05b3)
                ("Hebrew_Point_Hiriq" . #\u+05b4)
                ("Hebrew_Point_Tsere" . #\u+05b5)
                ("Hebrew_Point_Segol" . #\u+05b6)
                ("Hebrew_Point_Patah" . #\u+05b7)
                ("Hebrew_Point_Qamats" . #\u+05b8)
                ("Hebrew_Point_Holam" . #\u+05b9)
                ("Hebrew_Point_Holam_Haser_For_Vav" . #\u+05ba)
                ("Hebrew_Point_Qubuts" . #\u+05bb)
                ("Hebrew_Point_Dagesh_Or_Mapiq" . #\u+05bc)
                ("Hebrew_Point_Meteg" . #\u+05bd)
                ("Hebrew_Punctuation_Maqaf" . #\u+05be)
                ("Hebrew_Point_Rafe" . #\u+05bf)
                ("Hebrew_Punctuation_Paseq" . #\u+05c0)
                ("Hebrew_Point_Shin_Dot" . #\u+05c1)
                ("Hebrew_Point_Sin_Dot" . #\u+05c2)
                ("Hebrew_Punctuation_Sof_Pasuq" . #\u+05c3)
                ("Hebrew_Mark_Upper_Dot" . #\u+05c4)
                ("Hebrew_Mark_Lower_Dot" . #\u+05c5)
                ("Hebrew_Punctuation_Nun_Hafukha" . #\u+05c6)
                ("Hebrew_Point_Qamats_Qatan" . #\u+05c7)
                ("Hebrew_Letter_Alef" . #\u+05d0)
                ("Hebrew_Letter_Bet" . #\u+05d1)
                ("Hebrew_Letter_Gimel" . #\u+05d2)
                ("Hebrew_Letter_Dalet" . #\u+05d3)
                ("Hebrew_Letter_He" . #\u+05d4)
                ("Hebrew_Letter_Vav" . #\u+05d5)
                ("Hebrew_Letter_Zayin" . #\u+05d6)
                ("Hebrew_Letter_Het" . #\u+05d7)
                ("Hebrew_Letter_Tet" . #\u+05d8)
                ("Hebrew_Letter_Yod" . #\u+05d9)
                ("Hebrew_Letter_Final_Kaf" . #\u+05da)
                ("Hebrew_Letter_Kaf" . #\u+05db)
                ("Hebrew_Letter_Lamed" . #\u+05dc)
                ("Hebrew_Letter_Final_Mem" . #\u+05dd)
                ("Hebrew_Letter_Mem" . #\u+05de)
                ("Hebrew_Letter_Final_Nun" . #\u+05df)
                ("Hebrew_Letter_Nun" . #\u+05e0)
                ("Hebrew_Letter_Samekh" . #\u+05e1)
                ("Hebrew_Letter_Ayin" . #\u+05e2)
                ("Hebrew_Letter_Final_Pe" . #\u+05e3)
                ("Hebrew_Letter_Pe" . #\u+05e4)
                ("Hebrew_Letter_Final_Tsadi" . #\u+05e5)
                ("Hebrew_Letter_Tsadi" . #\u+05e6)
                ("Hebrew_Letter_Qof" . #\u+05e7)
                ("Hebrew_Letter_Resh" . #\u+05e8)
                ("Hebrew_Letter_Shin" . #\u+05e9)
                ("Hebrew_Letter_Tav" . #\u+05ea)
                ("Hebrew_Ligature_Yiddish_Double_Vav" . #\u+05f0)
                ("Hebrew_Ligature_Yiddish_Vav_Yod" . #\u+05f1)
                ("Hebrew_Ligature_Yiddish_Double_Yod" . #\u+05f2)
                ("Hebrew_Punctuation_Geresh" . #\u+05f3)
                ("Hebrew_Punctuation_Gershayim" . #\u+05f4)
                ("Arabic_Number_Sign" . #\u+0600)
                ("Arabic_Sign_Sanah" . #\u+0601)
                ("Arabic_Footnote_Marker" . #\u+0602)
                ("Arabic_Sign_Safha" . #\u+0603)
                ("Afghani_Sign" . #\u+060b)
                ("Arabic_Comma" . #\u+060c)
                ("Arabic_Date_Separator" . #\u+060d)
                ("Arabic_Poetic_Verse_Sign" . #\u+060e)
                ("Arabic_Sign_Misra" . #\u+060f)
                ("Arabic_Sign_Sallallahou_Alayhe_Wassallam" . #\u+0610)
                ("Arabic_Sign_Alayhe_Assallam" . #\u+0611)
                ("Arabic_Sign_Rahmatullah_Alayhe" . #\u+0612)
                ("Arabic_Sign_Radi_Allahou_Anhu" . #\u+0613)
                ("Arabic_Sign_Takhallus" . #\u+0614)
                ("Arabic_Small_High_Tah" . #\u+0615)
                ("Arabic_Semicolon" . #\u+061b)
                ("Arabic_Triple_Dot_Punctuation_Mark" . #\u+061e)
                ("Arabic_Question_Mark" . #\u+061f)
                ("Arabic_Letter_Hamza" . #\u+0621)
                ("Arabic_Letter_Alef_With_Madda_Above" . #\u+0622)
                ("Arabic_Letter_Alef_With_Hamza_Above" . #\u+0623)
                ("Arabic_Letter_Waw_With_Hamza_Above" . #\u+0624)
                ("Arabic_Letter_Alef_With_Hamza_Below" . #\u+0625)
                ("Arabic_Letter_Yeh_With_Hamza_Above" . #\u+0626)
                ("Arabic_Letter_Alef" . #\u+0627)
                ("Arabic_Letter_Beh" . #\u+0628)
                ("Arabic_Letter_Teh_Marbuta" . #\u+0629)
                ("Arabic_Letter_Teh" . #\u+062a)
                ("Arabic_Letter_Theh" . #\u+062b)
                ("Arabic_Letter_Jeem" . #\u+062c)
                ("Arabic_Letter_Hah" . #\u+062d)
                ("Arabic_Letter_Khah" . #\u+062e)
                ("Arabic_Letter_Dal" . #\u+062f)
                ("Arabic_Letter_Thal" . #\u+0630)
                ("Arabic_Letter_Reh" . #\u+0631)
                ("Arabic_Letter_Zain" . #\u+0632)
                ("Arabic_Letter_Seen" . #\u+0633)
                ("Arabic_Letter_Sheen" . #\u+0634)
                ("Arabic_Letter_Sad" . #\u+0635)
                ("Arabic_Letter_Dad" . #\u+0636)
                ("Arabic_Letter_Tah" . #\u+0637)
                ("Arabic_Letter_Zah" . #\u+0638)
                ("Arabic_Letter_Ain" . #\u+0639)
                ("Arabic_Letter_Ghain" . #\u+063a)
                ("Arabic_Tatweel" . #\u+0640)
                ("Arabic_Letter_Feh" . #\u+0641)
                ("Arabic_Letter_Qaf" . #\u+0642)
                ("Arabic_Letter_Kaf" . #\u+0643)
                ("Arabic_Letter_Lam" . #\u+0644)
                ("Arabic_Letter_Meem" . #\u+0645)
                ("Arabic_Letter_Noon" . #\u+0646)
                ("Arabic_Letter_Heh" . #\u+0647)
                ("Arabic_Letter_Waw" . #\u+0648)
                ("Arabic_Letter_Alef_Maksura" . #\u+0649)
                ("Arabic_Letter_Yeh" . #\u+064a)
                ("Arabic_Fathatan" . #\u+064b)
                ("Arabic_Dammatan" . #\u+064c)
                ("Arabic_Kasratan" . #\u+064d)
                ("Arabic_Fatha" . #\u+064e)
                ("Arabic_Damma" . #\u+064f)
                ("Arabic_Kasra" . #\u+0650)
                ("Arabic_Shadda" . #\u+0651)
                ("Arabic_Sukun" . #\u+0652)
                ("Arabic_Maddah_Above" . #\u+0653)
                ("Arabic_Hamza_Above" . #\u+0654)
                ("Arabic_Hamza_Below" . #\u+0655)
                ("Arabic_Subscript_Alef" . #\u+0656)
                ("Arabic_Inverted_Damma" . #\u+0657)
                ("Arabic_Mark_Noon_Ghunna" . #\u+0658)
                ("Arabic_Zwarakay" . #\u+0659)
                ("Arabic_Vowel_Sign_Small_V_Above" . #\u+065a)
                ("Arabic_Vowel_Sign_Inverted_Small_V_Above" . #\u+065b)
                ("Arabic_Vowel_Sign_Dot_Below" . #\u+065c)
                ("Arabic_Reversed_Damma" . #\u+065d)
                ("Arabic_Fatha_With_Two_Dots" . #\u+065e)
                ("Arabic-Indic_Digit_Zero" . #\u+0660)
                ("Arabic-Indic_Digit_One" . #\u+0661)
                ("Arabic-Indic_Digit_Two" . #\u+0662)
                ("Arabic-Indic_Digit_Three" . #\u+0663)
                ("Arabic-Indic_Digit_Four" . #\u+0664)
                ("Arabic-Indic_Digit_Five" . #\u+0665)
                ("Arabic-Indic_Digit_Six" . #\u+0666)
                ("Arabic-Indic_Digit_Seven" . #\u+0667)
                ("Arabic-Indic_Digit_Eight" . #\u+0668)
                ("Arabic-Indic_Digit_Nine" . #\u+0669)
                ("Arabic_Percent_Sign" . #\u+066a)
                ("Arabic_Decimal_Separator" . #\u+066b)
                ("Arabic_Thousands_Separator" . #\u+066c)
                ("Arabic_Five_Pointed_Star" . #\u+066d)
                ("Arabic_Letter_Dotless_Beh" . #\u+066e)
                ("Arabic_Letter_Dotless_Qaf" . #\u+066f)
                ("Arabic_Letter_Superscript_Alef" . #\u+0670)
                ("Arabic_Letter_Alef_Wasla" . #\u+0671)
                ("Arabic_Letter_Alef_With_Wavy_Hamza_Above" . #\u+0672)
                ("Arabic_Letter_Alef_With_Wavy_Hamza_Below" . #\u+0673)
                ("Arabic_Letter_High_Hamza" . #\u+0674)
                ("Arabic_Letter_High_Hamza_Alef" . #\u+0675)
                ("Arabic_Letter_High_Hamza_Waw" . #\u+0676)
                ("Arabic_Letter_U_With_Hamza_Above" . #\u+0677)
                ("Arabic_Letter_High_Hamza_Yeh" . #\u+0678)
                ("Arabic_Letter_Tteh" . #\u+0679)
                ("Arabic_Letter_Tteheh" . #\u+067a)
                ("Arabic_Letter_Beeh" . #\u+067b)
                ("Arabic_Letter_Teh_With_Ring" . #\u+067c)
                ("Arabic_Letter_Teh_With_Three_Dots_Above_Downwards" . #\u+067d)
                ("Arabic_Letter_Peh" . #\u+067e)
                ("Arabic_Letter_Teheh" . #\u+067f)
                ("Arabic_Letter_Beheh" . #\u+0680)
                ("Arabic_Letter_Hah_With_Hamza_Above" . #\u+0681)
                ("Arabic_Letter_Hah_With_Two_Dots_Vertical_Above" . #\u+0682)
                ("Arabic_Letter_Nyeh" . #\u+0683)
                ("Arabic_Letter_Dyeh" . #\u+0684)
                ("Arabic_Letter_Hah_With_Three_Dots_Above" . #\u+0685)
                ("Arabic_Letter_Tcheh" . #\u+0686)
                ("Arabic_Letter_Tcheheh" . #\u+0687)
                ("Arabic_Letter_Ddal" . #\u+0688)
                ("Arabic_Letter_Dal_With_Ring" . #\u+0689)
                ("Arabic_Letter_Dal_With_Dot_Below" . #\u+068a)
                ("Arabic_Letter_Dal_With_Dot_Below_And_Small_Tah" . #\u+068b)
                ("Arabic_Letter_Dahal" . #\u+068c)
                ("Arabic_Letter_Ddahal" . #\u+068d)
                ("Arabic_Letter_Dul" . #\u+068e)
                ("Arabic_Letter_Dal_With_Three_Dots_Above_Downwards" . #\u+068f)
                ("Arabic_Letter_Dal_With_Four_Dots_Above" . #\u+0690)
                ("Arabic_Letter_Rreh" . #\u+0691)
                ("Arabic_Letter_Reh_With_Small_V" . #\u+0692)
                ("Arabic_Letter_Reh_With_Ring" . #\u+0693)
                ("Arabic_Letter_Reh_With_Dot_Below" . #\u+0694)
                ("Arabic_Letter_Reh_With_Small_V_Below" . #\u+0695)
                ("Arabic_Letter_Reh_With_Dot_Below_And_Dot_Above" . #\u+0696)
                ("Arabic_Letter_Reh_With_Two_Dots_Above" . #\u+0697)
                ("Arabic_Letter_Jeh" . #\u+0698)
                ("Arabic_Letter_Reh_With_Four_Dots_Above" . #\u+0699)
                ("Arabic_Letter_Seen_With_Dot_Below_And_Dot_Above" . #\u+069a)
                ("Arabic_Letter_Seen_With_Three_Dots_Below" . #\u+069b)
                ("Arabic_Letter_Seen_With_Three_Dots_Below_And_Three_Dots_Above" . #\u+069c)
                ("Arabic_Letter_Sad_With_Two_Dots_Below" . #\u+069d)
                ("Arabic_Letter_Sad_With_Three_Dots_Above" . #\u+069e)
                ("Arabic_Letter_Tah_With_Three_Dots_Above" . #\u+069f)
                ("Arabic_Letter_Ain_With_Three_Dots_Above" . #\u+06a0)
                ("Arabic_Letter_Dotless_Feh" . #\u+06a1)
                ("Arabic_Letter_Feh_With_Dot_Moved_Below" . #\u+06a2)
                ("Arabic_Letter_Feh_With_Dot_Below" . #\u+06a3)
                ("Arabic_Letter_Veh" . #\u+06a4)
                ("Arabic_Letter_Feh_With_Three_Dots_Below" . #\u+06a5)
                ("Arabic_Letter_Peheh" . #\u+06a6)
                ("Arabic_Letter_Qaf_With_Dot_Above" . #\u+06a7)
                ("Arabic_Letter_Qaf_With_Three_Dots_Above" . #\u+06a8)
                ("Arabic_Letter_Keheh" . #\u+06a9)
                ("Arabic_Letter_Swash_Kaf" . #\u+06aa)
                ("Arabic_Letter_Kaf_With_Ring" . #\u+06ab)
                ("Arabic_Letter_Kaf_With_Dot_Above" . #\u+06ac)
                ("Arabic_Letter_Ng" . #\u+06ad)
                ("Arabic_Letter_Kaf_With_Three_Dots_Below" . #\u+06ae)
                ("Arabic_Letter_Gaf" . #\u+06af)
                ("Arabic_Letter_Gaf_With_Ring" . #\u+06b0)
                ("Arabic_Letter_Ngoeh" . #\u+06b1)
                ("Arabic_Letter_Gaf_With_Two_Dots_Below" . #\u+06b2)
                ("Arabic_Letter_Gueh" . #\u+06b3)
                ("Arabic_Letter_Gaf_With_Three_Dots_Above" . #\u+06b4)
                ("Arabic_Letter_Lam_With_Small_V" . #\u+06b5)
                ("Arabic_Letter_Lam_With_Dot_Above" . #\u+06b6)
                ("Arabic_Letter_Lam_With_Three_Dots_Above" . #\u+06b7)
                ("Arabic_Letter_Lam_With_Three_Dots_Below" . #\u+06b8)
                ("Arabic_Letter_Noon_With_Dot_Below" . #\u+06b9)
                ("Arabic_Letter_Noon_Ghunna" . #\u+06ba)
                ("Arabic_Letter_Rnoon" . #\u+06bb)
                ("Arabic_Letter_Noon_With_Ring" . #\u+06bc)
                ("Arabic_Letter_Noon_With_Three_Dots_Above" . #\u+06bd)
                ("Arabic_Letter_Heh_Doachashmee" . #\u+06be)
                ("Arabic_Letter_Tcheh_With_Dot_Above" . #\u+06bf)
                ("Arabic_Letter_Heh_With_Yeh_Above" . #\u+06c0)
                ("Arabic_Letter_Heh_Goal" . #\u+06c1)
                ("Arabic_Letter_Heh_Goal_With_Hamza_Above" . #\u+06c2)
                ("Arabic_Letter_Teh_Marbuta_Goal" . #\u+06c3)
                ("Arabic_Letter_Waw_With_Ring" . #\u+06c4)
                ("Arabic_Letter_Kirghiz_Oe" . #\u+06c5)
                ("Arabic_Letter_Oe" . #\u+06c6)
                ("Arabic_Letter_U" . #\u+06c7)
                ("Arabic_Letter_Yu" . #\u+06c8)
                ("Arabic_Letter_Kirghiz_Yu" . #\u+06c9)
                ("Arabic_Letter_Waw_With_Two_Dots_Above" . #\u+06ca)
                ("Arabic_Letter_Ve" . #\u+06cb)
                ("Arabic_Letter_Farsi_Yeh" . #\u+06cc)
                ("Arabic_Letter_Yeh_With_Tail" . #\u+06cd)
                ("Arabic_Letter_Yeh_With_Small_V" . #\u+06ce)
                ("Arabic_Letter_Waw_With_Dot_Above" . #\u+06cf)
                ("Arabic_Letter_E" . #\u+06d0)
                ("Arabic_Letter_Yeh_With_Three_Dots_Below" . #\u+06d1)
                ("Arabic_Letter_Yeh_Barree" . #\u+06d2)
                ("Arabic_Letter_Yeh_Barree_With_Hamza_Above" . #\u+06d3)
                ("Arabic_Full_Stop" . #\u+06d4)
                ("Arabic_Letter_Ae" . #\u+06d5)
                ("Arabic_Small_High_Ligature_Sad_With_Lam_With_Alef_Maksura" . #\u+06d6)
                ("Arabic_Small_High_Ligature_Qaf_With_Lam_With_Alef_Maksura" . #\u+06d7)
                ("Arabic_Small_High_Meem_Initial_Form" . #\u+06d8)
                ("Arabic_Small_High_Lam_Alef" . #\u+06d9)
                ("Arabic_Small_High_Jeem" . #\u+06da)
                ("Arabic_Small_High_Three_Dots" . #\u+06db)
                ("Arabic_Small_High_Seen" . #\u+06dc)
                ("Arabic_End_Of_Ayah" . #\u+06dd)
                ("Arabic_Start_Of_Rub_El_Hizb" . #\u+06de)
                ("Arabic_Small_High_Rounded_Zero" . #\u+06df)
                ("Arabic_Small_High_Upright_Rectangular_Zero" . #\u+06e0)
                ("Arabic_Small_High_Dotless_Head_Of_Khah" . #\u+06e1)
                ("Arabic_Small_High_Meem_Isolated_Form" . #\u+06e2)
                ("Arabic_Small_Low_Seen" . #\u+06e3)
                ("Arabic_Small_High_Madda" . #\u+06e4)
                ("Arabic_Small_Waw" . #\u+06e5)
                ("Arabic_Small_Yeh" . #\u+06e6)
                ("Arabic_Small_High_Yeh" . #\u+06e7)
                ("Arabic_Small_High_Noon" . #\u+06e8)
                ("Arabic_Place_Of_Sajdah" . #\u+06e9)
                ("Arabic_Empty_Centre_Low_Stop" . #\u+06ea)
                ("Arabic_Empty_Centre_High_Stop" . #\u+06eb)
                ("Arabic_Rounded_High_Stop_With_Filled_Centre" . #\u+06ec)
                ("Arabic_Small_Low_Meem" . #\u+06ed)
                ("Arabic_Letter_Dal_With_Inverted_V" . #\u+06ee)
                ("Arabic_Letter_Reh_With_Inverted_V" . #\u+06ef)
                ("Extended_Arabic-Indic_Digit_Zero" . #\u+06f0)
                ("Extended_Arabic-Indic_Digit_One" . #\u+06f1)
                ("Extended_Arabic-Indic_Digit_Two" . #\u+06f2)
                ("Extended_Arabic-Indic_Digit_Three" . #\u+06f3)
                ("Extended_Arabic-Indic_Digit_Four" . #\u+06f4)
                ("Extended_Arabic-Indic_Digit_Five" . #\u+06f5)
                ("Extended_Arabic-Indic_Digit_Six" . #\u+06f6)
                ("Extended_Arabic-Indic_Digit_Seven" . #\u+06f7)
                ("Extended_Arabic-Indic_Digit_Eight" . #\u+06f8)
                ("Extended_Arabic-Indic_Digit_Nine" . #\u+06f9)
                ("Arabic_Letter_Sheen_With_Dot_Below" . #\u+06fa)
                ("Arabic_Letter_Dad_With_Dot_Below" . #\u+06fb)
                ("Arabic_Letter_Ghain_With_Dot_Below" . #\u+06fc)
                ("Arabic_Sign_Sindhi_Ampersand" . #\u+06fd)
                ("Arabic_Sign_Sindhi_Postposition_Men" . #\u+06fe)
                ("Arabic_Letter_Heh_With_Inverted_V" . #\u+06ff)
                ("Syriac_End_Of_Paragraph" . #\u+0700)
                ("Syriac_Supralinear_Full_Stop" . #\u+0701)
                ("Syriac_Sublinear_Full_Stop" . #\u+0702)
                ("Syriac_Supralinear_Colon" . #\u+0703)
                ("Syriac_Sublinear_Colon" . #\u+0704)
                ("Syriac_Horizontal_Colon" . #\u+0705)
                ("Syriac_Colon_Skewed_Left" . #\u+0706)
                ("Syriac_Colon_Skewed_Right" . #\u+0707)
                ("Syriac_Supralinear_Colon_Skewed_Left" . #\u+0708)
                ("Syriac_Sublinear_Colon_Skewed_Right" . #\u+0709)
                ("Syriac_Contraction" . #\u+070a)
                ("Syriac_Harklean_Obelus" . #\u+070b)
                ("Syriac_Harklean_Metobelus" . #\u+070c)
                ("Syriac_Harklean_Asteriscus" . #\u+070d)
                ("Syriac_Abbreviation_Mark" . #\u+070f)
                ("Syriac_Letter_Alaph" . #\u+0710)
                ("Syriac_Letter_Superscript_Alaph" . #\u+0711)
                ("Syriac_Letter_Beth" . #\u+0712)
                ("Syriac_Letter_Gamal" . #\u+0713)
                ("Syriac_Letter_Gamal_Garshuni" . #\u+0714)
                ("Syriac_Letter_Dalath" . #\u+0715)
                ("Syriac_Letter_Dotless_Dalath_Rish" . #\u+0716)
                ("Syriac_Letter_He" . #\u+0717)
                ("Syriac_Letter_Waw" . #\u+0718)
                ("Syriac_Letter_Zain" . #\u+0719)
                ("Syriac_Letter_Heth" . #\u+071a)
                ("Syriac_Letter_Teth" . #\u+071b)
                ("Syriac_Letter_Teth_Garshuni" . #\u+071c)
                ("Syriac_Letter_Yudh" . #\u+071d)
                ("Syriac_Letter_Yudh_He" . #\u+071e)
                ("Syriac_Letter_Kaph" . #\u+071f)
                ("Syriac_Letter_Lamadh" . #\u+0720)
                ("Syriac_Letter_Mim" . #\u+0721)
                ("Syriac_Letter_Nun" . #\u+0722)
                ("Syriac_Letter_Semkath" . #\u+0723)
                ("Syriac_Letter_Final_Semkath" . #\u+0724)
                ("Syriac_Letter_E" . #\u+0725)
                ("Syriac_Letter_Pe" . #\u+0726)
                ("Syriac_Letter_Reversed_Pe" . #\u+0727)
                ("Syriac_Letter_Sadhe" . #\u+0728)
                ("Syriac_Letter_Qaph" . #\u+0729)
                ("Syriac_Letter_Rish" . #\u+072a)
                ("Syriac_Letter_Shin" . #\u+072b)
                ("Syriac_Letter_Taw" . #\u+072c)
                ("Syriac_Letter_Persian_Bheth" . #\u+072d)
                ("Syriac_Letter_Persian_Ghamal" . #\u+072e)
                ("Syriac_Letter_Persian_Dhalath" . #\u+072f)
                ("Syriac_Pthaha_Above" . #\u+0730)
                ("Syriac_Pthaha_Below" . #\u+0731)
                ("Syriac_Pthaha_Dotted" . #\u+0732)
                ("Syriac_Zqapha_Above" . #\u+0733)
                ("Syriac_Zqapha_Below" . #\u+0734)
                ("Syriac_Zqapha_Dotted" . #\u+0735)
                ("Syriac_Rbasa_Above" . #\u+0736)
                ("Syriac_Rbasa_Below" . #\u+0737)
                ("Syriac_Dotted_Zlama_Horizontal" . #\u+0738)
                ("Syriac_Dotted_Zlama_Angular" . #\u+0739)
                ("Syriac_Hbasa_Above" . #\u+073a)
                ("Syriac_Hbasa_Below" . #\u+073b)
                ("Syriac_Hbasa-Esasa_Dotted" . #\u+073c)
                ("Syriac_Esasa_Above" . #\u+073d)
                ("Syriac_Esasa_Below" . #\u+073e)
                ("Syriac_Rwaha" . #\u+073f)
                ("Syriac_Feminine_Dot" . #\u+0740)
                ("Syriac_Qushshaya" . #\u+0741)
                ("Syriac_Rukkakha" . #\u+0742)
                ("Syriac_Two_Vertical_Dots_Above" . #\u+0743)
                ("Syriac_Two_Vertical_Dots_Below" . #\u+0744)
                ("Syriac_Three_Dots_Above" . #\u+0745)
                ("Syriac_Three_Dots_Below" . #\u+0746)
                ("Syriac_Oblique_Line_Above" . #\u+0747)
                ("Syriac_Oblique_Line_Below" . #\u+0748)
                ("Syriac_Music" . #\u+0749)
                ("Syriac_Barrekh" . #\u+074a)
                ("Syriac_Letter_Sogdian_Zhain" . #\u+074d)
                ("Syriac_Letter_Sogdian_Khaph" . #\u+074e)
                ("Syriac_Letter_Sogdian_Fe" . #\u+074f)
                ("Arabic_Letter_Beh_With_Three_Dots_Horizontally_Below" . #\u+0750)
                ("Arabic_Letter_Beh_With_Dot_Below_And_Three_Dots_Above" . #\u+0751)
                ("Arabic_Letter_Beh_With_Three_Dots_Pointing_Upwards_Below" . #\u+0752)
                ("Arabic_Letter_Beh_With_Three_Dots_Pointing_Upwards_Below_And_Two_Dots_Above" . #\u+0753)
                ("Arabic_Letter_Beh_With_Two_Dots_Below_And_Dot_Above" . #\u+0754)
                ("Arabic_Letter_Beh_With_Inverted_Small_V_Below" . #\u+0755)
                ("Arabic_Letter_Beh_With_Small_V" . #\u+0756)
                ("Arabic_Letter_Hah_With_Two_Dots_Above" . #\u+0757)
                ("Arabic_Letter_Hah_With_Three_Dots_Pointing_Upwards_Below" . #\u+0758)
                ("Arabic_Letter_Dal_With_Two_Dots_Vertically_Below_And_Small_Tah" . #\u+0759)
                ("Arabic_Letter_Dal_With_Inverted_Small_V_Below" . #\u+075a)
                ("Arabic_Letter_Reh_With_Stroke" . #\u+075b)
                ("Arabic_Letter_Seen_With_Four_Dots_Above" . #\u+075c)
                ("Arabic_Letter_Ain_With_Two_Dots_Above" . #\u+075d)
                ("Arabic_Letter_Ain_With_Three_Dots_Pointing_Downwards_Above" . #\u+075e)
                ("Arabic_Letter_Ain_With_Two_Dots_Vertically_Above" . #\u+075f)
                ("Arabic_Letter_Feh_With_Two_Dots_Below" . #\u+0760)
                ("Arabic_Letter_Feh_With_Three_Dots_Pointing_Upwards_Below" . #\u+0761)
                ("Arabic_Letter_Keheh_With_Dot_Above" . #\u+0762)
                ("Arabic_Letter_Keheh_With_Three_Dots_Above" . #\u+0763)
                ("Arabic_Letter_Keheh_With_Three_Dots_Pointing_Upwards_Below" . #\u+0764)
                ("Arabic_Letter_Meem_With_Dot_Above" . #\u+0765)
                ("Arabic_Letter_Meem_With_Dot_Below" . #\u+0766)
                ("Arabic_Letter_Noon_With_Two_Dots_Below" . #\u+0767)
                ("Arabic_Letter_Noon_With_Small_Tah" . #\u+0768)
                ("Arabic_Letter_Noon_With_Small_V" . #\u+0769)
                ("Arabic_Letter_Lam_With_Bar" . #\u+076a)
                ("Arabic_Letter_Reh_With_Two_Dots_Vertically_Above" . #\u+076b)
                ("Arabic_Letter_Reh_With_Hamza_Above" . #\u+076c)
                ("Arabic_Letter_Seen_With_Two_Dots_Vertically_Above" . #\u+076d)
                ("Thaana_Letter_Haa" . #\u+0780)
                ("Thaana_Letter_Shaviyani" . #\u+0781)
                ("Thaana_Letter_Noonu" . #\u+0782)
                ("Thaana_Letter_Raa" . #\u+0783)
                ("Thaana_Letter_Baa" . #\u+0784)
                ("Thaana_Letter_Lhaviyani" . #\u+0785)
                ("Thaana_Letter_Kaafu" . #\u+0786)
                ("Thaana_Letter_Alifu" . #\u+0787)
                ("Thaana_Letter_Vaavu" . #\u+0788)
                ("Thaana_Letter_Meemu" . #\u+0789)
                ("Thaana_Letter_Faafu" . #\u+078a)
                ("Thaana_Letter_Dhaalu" . #\u+078b)
                ("Thaana_Letter_Thaa" . #\u+078c)
                ("Thaana_Letter_Laamu" . #\u+078d)
                ("Thaana_Letter_Gaafu" . #\u+078e)
                ("Thaana_Letter_Gnaviyani" . #\u+078f)
                ("Thaana_Letter_Seenu" . #\u+0790)
                ("Thaana_Letter_Daviyani" . #\u+0791)
                ("Thaana_Letter_Zaviyani" . #\u+0792)
                ("Thaana_Letter_Taviyani" . #\u+0793)
                ("Thaana_Letter_Yaa" . #\u+0794)
                ("Thaana_Letter_Paviyani" . #\u+0795)
                ("Thaana_Letter_Javiyani" . #\u+0796)
                ("Thaana_Letter_Chaviyani" . #\u+0797)
                ("Thaana_Letter_Ttaa" . #\u+0798)
                ("Thaana_Letter_Hhaa" . #\u+0799)
                ("Thaana_Letter_Khaa" . #\u+079a)
                ("Thaana_Letter_Thaalu" . #\u+079b)
                ("Thaana_Letter_Zaa" . #\u+079c)
                ("Thaana_Letter_Sheenu" . #\u+079d)
                ("Thaana_Letter_Saadhu" . #\u+079e)
                ("Thaana_Letter_Daadhu" . #\u+079f)
                ("Thaana_Letter_To" . #\u+07a0)
                ("Thaana_Letter_Zo" . #\u+07a1)
                ("Thaana_Letter_Ainu" . #\u+07a2)
                ("Thaana_Letter_Ghainu" . #\u+07a3)
                ("Thaana_Letter_Qaafu" . #\u+07a4)
                ("Thaana_Letter_Waavu" . #\u+07a5)
                ("Thaana_Abafili" . #\u+07a6)
                ("Thaana_Aabaafili" . #\u+07a7)
                ("Thaana_Ibifili" . #\u+07a8)
                ("Thaana_Eebeefili" . #\u+07a9)
                ("Thaana_Ubufili" . #\u+07aa)
                ("Thaana_Ooboofili" . #\u+07ab)
                ("Thaana_Ebefili" . #\u+07ac)
                ("Thaana_Eybeyfili" . #\u+07ad)
                ("Thaana_Obofili" . #\u+07ae)
                ("Thaana_Oaboafili" . #\u+07af)
                ("Thaana_Sukun" . #\u+07b0)
                ("Thaana_Letter_Naa" . #\u+07b1)
                ("Nko_Digit_Zero" . #\u+07c0)
                ("Nko_Digit_One" . #\u+07c1)
                ("Nko_Digit_Two" . #\u+07c2)
                ("Nko_Digit_Three" . #\u+07c3)
                ("Nko_Digit_Four" . #\u+07c4)
                ("Nko_Digit_Five" . #\u+07c5)
                ("Nko_Digit_Six" . #\u+07c6)
                ("Nko_Digit_Seven" . #\u+07c7)
                ("Nko_Digit_Eight" . #\u+07c8)
                ("Nko_Digit_Nine" . #\u+07c9)
                ("Nko_Letter_A" . #\u+07ca)
                ("Nko_Letter_Ee" . #\u+07cb)
                ("Nko_Letter_I" . #\u+07cc)
                ("Nko_Letter_E" . #\u+07cd)
                ("Nko_Letter_U" . #\u+07ce)
                ("Nko_Letter_Oo" . #\u+07cf)
                ("Nko_Letter_O" . #\u+07d0)
                ("Nko_Letter_Dagbasinna" . #\u+07d1)
                ("Nko_Letter_N" . #\u+07d2)
                ("Nko_Letter_Ba" . #\u+07d3)
                ("Nko_Letter_Pa" . #\u+07d4)
                ("Nko_Letter_Ta" . #\u+07d5)
                ("Nko_Letter_Ja" . #\u+07d6)
                ("Nko_Letter_Cha" . #\u+07d7)
                ("Nko_Letter_Da" . #\u+07d8)
                ("Nko_Letter_Ra" . #\u+07d9)
                ("Nko_Letter_Rra" . #\u+07da)
                ("Nko_Letter_Sa" . #\u+07db)
                ("Nko_Letter_Gba" . #\u+07dc)
                ("Nko_Letter_Fa" . #\u+07dd)
                ("Nko_Letter_Ka" . #\u+07de)
                ("Nko_Letter_La" . #\u+07df)
                ("Nko_Letter_Na_Woloso" . #\u+07e0)
                ("Nko_Letter_Ma" . #\u+07e1)
                ("Nko_Letter_Nya" . #\u+07e2)
                ("Nko_Letter_Na" . #\u+07e3)
                ("Nko_Letter_Ha" . #\u+07e4)
                ("Nko_Letter_Wa" . #\u+07e5)
                ("Nko_Letter_Ya" . #\u+07e6)
                ("Nko_Letter_Nya_Woloso" . #\u+07e7)
                ("Nko_Letter_Jona_Ja" . #\u+07e8)
                ("Nko_Letter_Jona_Cha" . #\u+07e9)
                ("Nko_Letter_Jona_Ra" . #\u+07ea)
                ("Nko_Combining_Short_High_Tone" . #\u+07eb)
                ("Nko_Combining_Short_Low_Tone" . #\u+07ec)
                ("Nko_Combining_Short_Rising_Tone" . #\u+07ed)
                ("Nko_Combining_Long_Descending_Tone" . #\u+07ee)
                ("Nko_Combining_Long_High_Tone" . #\u+07ef)
                ("Nko_Combining_Long_Low_Tone" . #\u+07f0)
                ("Nko_Combining_Long_Rising_Tone" . #\u+07f1)
                ("Nko_Combining_Nasalization_Mark" . #\u+07f2)
                ("Nko_Combining_Double_Dot_Above" . #\u+07f3)
                ("Nko_High_Tone_Apostrophe" . #\u+07f4)
                ("Nko_Low_Tone_Apostrophe" . #\u+07f5)
                ("Nko_Symbol_Oo_Dennen" . #\u+07f6)
                ("Nko_Symbol_Gbakurunen" . #\u+07f7)
                ("Nko_Comma" . #\u+07f8)
                ("Nko_Exclamation_Mark" . #\u+07f9)
                ("Nko_Lajanyalan" . #\u+07fa)
		("Figure_Space" . #\u+2007)
		("Zero_Width_Space" . #\u+200b)
                ("Line_Separator" . #\u+2028)
                ("Paragraph_Separator" . #\u+2029)
                ("Replacement_Character" . #\u+fffd)
                ("Skull_And_Crossbones" . #\u+2620)))
  (destructuring-bind (name . char) pair
    (register-character-name name char)))



;;;(NAME-CHAR name)
;;;If name has an entry in the *NAME->CHAR*, return first such entry.
;;;Otherwise, if it consists of one char, return it.
;;;Otherwise, if it consists of two chars, the first of which  is ^,
;;; return %code-char(c xor 64), where c is the uppercased second char.
;;;Otherwise, if it starts with the prefix "u+" or "U+" followed by
;;; hex digits, the number denoted by those hex digits is interpreted as the
;;; unicode code of the character; if this value is less than
;;; CHAR-CODE-LIMIT, CODE-CHAR of that value is returned.
;;;Otherwise, if it consists of octal digits, the number denoted by
;;;  those octal digits is interpreted as per the U+ case above.
;;;Otherwise return NIL.


(defun name-char (name)
  "Given an argument acceptable to STRING, NAME-CHAR returns a character
  whose name is that string, if one exists. Otherwise, NIL is returned."
  (if (characterp name)
    name
    (let* ((name (string name)))
      (let* ((namelen (length name)))
        (declare (fixnum namelen))
        (or (gethash name *name->char*)
            (if (= namelen 1)
              (char name 0)
              (if (and (= namelen 2) (eq (char name 0) #\^))
                (let* ((c1 (char-code (char-upcase (char name 1)))))
                  (if (and (>= c1 64) (< c1 96))
                    (code-char (the fixnum (logxor (the fixnum c1) #x40)))))
                (let* ((n 0)
                       (start 1))
                  (declare (fixnum start))
                  (or
                   (if (and (> namelen 1)
                            (or (eql (char name 0) #\U)
                                (eql (char name 0) #\u))
                            (or (= namelen 2)
                                (progn
                                  (when (eql (char name 1) #\+)
                                    (incf start))
                                  t)))
                     (do* ((i start (1+ i)))
                          ((= i namelen) (if (< n char-code-limit)
                                           (code-char n)))
                       (declare (fixnum i))
                       (let* ((pos (position (char-upcase (char name i))
                                             "0123456789ABCDEF")))
                         (if pos
                           (setq n (logior (ash n 4) pos))
                           (progn
                             (setq n 0)
                             (return))))))
                   (dotimes (i namelen (if (< n char-code-limit)
                                         (code-char n)))
                     (let* ((code (the fixnum (- (the fixnum (char-code (char name i)))
                                                 (char-code #\0)))))
                       (declare (fixnum code))
                       (if (and (>= code 0)
                                (<= code 7))
                         (setq n (logior code (the fixnum (ash n 3))))
                         (return)))))))))))))

(eval-when (:compile-toplevel :load-toplevel :execute)
(defconstant wsp #.(let ((str (make-string 6  :element-type 'base-char)))
                      (set-schar str 0 #\Space)
                      (set-schar str 1 #\^I)
                      (set-schar str 2 #\^L)
                      (set-schar str 3 #\^@)
                      (set-schar str 4 #\^J)
                      (set-schar str 5 (code-char #xa0))
                      str))

(defconstant wsp&cr #.(let ((str (make-string 7 :element-type 'base-char)))
                        (set-schar str 0 #\Space)
                        (set-schar str 1 #\^M)
                        (set-schar str 2 #\^I)
                        (set-schar str 3 #\^L)
                        (set-schar str 4 #\^@)
                        (set-schar str 5 #\^J)
                        (set-schar str 0 #\Space)
                        (set-schar str 6 (code-char #xa0))
                        str))
)

(defun whitespacep (char)
  (eql $cht_wsp (%character-attribute char (rdtab.ttab *readtable*))))
	   
	 

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;			Readtables					;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

;;; Readtable = istructure with data [1] type-table and [2] macro-char-alist
;;; Type-table is a 256 byte ivector with a type byte for each char.
;;; macro-char-alist is a list of (char . defn).  The defn is either a
;;; cons of (#'read-dispatch . macro-char-alist) for
;;; dispatch macros, or it is a function or a symbol to call for simple macros.

(defun readtablep (object) (istruct-typep object 'readtable)) 

(defun readtable-arg (object)
  (if (null object) (setq object *readtable*))
  (unless (istruct-typep object 'readtable)
    (report-bad-arg object 'readtable))
  object)

(eval-when (:compile-toplevel :execute)
(def-accessors %svref
  token.string
  token.ipos
  token.opos
  token.len
)

(defmacro with-token-buffer ((name) &body body &environment env)
  (multiple-value-bind (body decls) (parse-body body env nil)
    `(let* ((,name (vector (%get-token-string 16) 0 0 16 nil)))
       (declare (dynamic-extent ,name))
       (unwind-protect
         (locally ,@decls ,@body)
         (%return-token-string ,name)))))
)

(defun read-dispatch (stream char)
  (let* ((info (cdr (assq char (rdtab.alist *readtable*)))))
    (with-token-buffer (tb)
      (let* ((subchar nil)
             (numarg nil))
        (loop
            (if (digit-char-p (setq subchar (%read-char-no-eof stream)))
                (%add-char-to-token subchar tb)
                (return (setq subchar (char-upcase subchar) 
                              numarg (%token-to-number tb 10)))))
        (let* ((dispfun (cdr (assq subchar (cdr info)))))     ; <== WAS char
          (if dispfun
              (funcall dispfun stream subchar numarg)
              (signal-reader-error stream "Undefined character ~S in a ~S dispatch macro." subchar char)))))))

;;; This -really- gets initialized later in the file
(defvar %standard-readtable%
  (let* ((ttab (make-sparse-vector char-code-limit '(unsigned-byte 8) $cht_cnst))
         (macs `((#\# . (,#'read-dispatch))))
         (case :upcase))
    (dotimes (ch (1+ (char-code #\Space)))
      (setf (sparse-vector-ref ttab ch) $cht_wsp))
    (setf (sparse-vector-ref ttab #xa0) $cht_wsp)
    (setf (sparse-vector-ref ttab (char-code #\\)) $cht_sesc)
    (setf (sparse-vector-ref ttab (char-code #\|)) $cht_mesc)
    (setf (sparse-vector-ref ttab (char-code #\#)) $cht_ntmac)
    (setf (sparse-vector-ref ttab (char-code #\Backspace)) $cht_ill)
    (setf (sparse-vector-ref ttab (char-code #\Rubout)) $cht_ill)
    (%istruct 'readtable ttab macs case)))

(defvar %initial-readtable%)
(setq *readtable* %standard-readtable%)
(def-standard-initial-binding *readtable* )
(queue-fixup (setq %initial-readtable% (copy-readtable *readtable*)))

(defun copy-readtable (&optional (from *readtable*) to)
  (setq from (if from (readtable-arg from)  %standard-readtable%))
  (let* ((fttab (rdtab.ttab from)))
    (setq to (if to 
               (readtable-arg to)
               (%istruct 'readtable
                         (copy-sparse-vector fttab)
                         nil (rdtab.case from))))
    (setf (rdtab.alist to) (copy-tree (rdtab.alist from)))
    (setf (rdtab.case to) (rdtab.case from))
    to))

(declaim (inline %character-attribute))

(defun %character-attribute (char attrtab)
  (declare (character char)
           (optimize (speed 3) (safety 0)))
  (let* ((code (char-code char)))
    (declare (fixnum code))
    (sparse-vector-ref attrtab code)))

(defun %set-character-attribute (char readtable attr)
  (let* ((code (char-code char))
         (attrtab (rdtab.ttab readtable)))
    (declare (type (mod #x110000) code))
    (setf (sparse-vector-ref attrtab code) attr)))


;;; returns: (values attrib <aux-info>), where
;;;           <aux-info> = (char . fn), if terminating macro
;;;                      = (char . (fn . dispatch-alist)), if dispatching macro
;;;                      = nil otherwise


(defun %get-readtable-char (char &optional (readtable *readtable*))
  (setq char (require-type char 'character))
  (let* ((attr (%character-attribute char (rdtab.ttab readtable))))
    (declare (fixnum attr))
    (values attr (if (logbitp $cht_macbit attr) (assoc char (rdtab.alist readtable))))))


(defun set-syntax-from-char (to-char from-char &optional to-readtable from-readtable)
  "Causes the syntax of TO-CHAR to be the same as FROM-CHAR in the
  optional readtable (defaults to the current readtable). The
  FROM-TABLE defaults to the standard Lisp readtable when NIL."
  (setq to-char (require-type to-char 'base-char))
  (setq from-char (require-type from-char 'base-char))
  (setq to-readtable (readtable-arg to-readtable))
  (setq from-readtable (readtable-arg (or from-readtable %initial-readtable%)))
  (multiple-value-bind (from-attr from-info) (%get-readtable-char from-char from-readtable)
    (let* ((new-tree (copy-tree (cdr from-info)))
           (old-to-info (nth-value 1 (%get-readtable-char to-char to-readtable))))
      (without-interrupts
       (if from-info
         (if old-to-info
           (setf (cdr old-to-info) new-tree)
           (push (cons to-char new-tree) (rdtab.alist to-readtable)))
         (if old-to-info
           (setf (rdtab.alist to-readtable) (delq old-to-info (rdtab.alist to-readtable)))))
       (%set-character-attribute to-char
                                 to-readtable
                                 (if (and (= from-attr $cht_cnst)
                                          (member to-char '(#\Newline #\Linefeed #\Page #\Return
                                                            #\Space #\Tab #\Backspace #\Rubout)))
                                   $cht_ill
                                   from-attr)))
      t)))

(defun get-macro-character (char &optional readtable)
  "Return the function associated with the specified CHAR which is a macro
  character, or NIL if there is no such function. As a second value, return
  T if CHAR is a macro character which is non-terminating, i.e. which can
  be embedded in a symbol name."
  (setq readtable (readtable-arg readtable))
  (multiple-value-bind (attr info) (%get-readtable-char char readtable)
    (declare (fixnum attr) (list info))
    (let* ((def (cdr info)))
      (values (if (consp def) (car def) def)
              (= attr $cht_ntmac)))))

(defun set-macro-character (char fn &optional non-terminating-p readtable)
  "Causes CHAR to be a macro character which invokes FUNCTION when seen
   by the reader. The NON-TERMINATINGP flag can be used to make the macro
   character non-terminating, i.e. embeddable in a symbol name."
  (setq char (require-type char 'base-char))
  (setq readtable (readtable-arg readtable))
  (when fn
    (unless (or (symbolp fn) (functionp fn))
      (setq fn (require-type fn '(or symbol function)))))
  (let* ((info (nth-value 1 (%get-readtable-char char readtable))))
    (declare (list info))
    (without-interrupts
     (%set-character-attribute char readtable
                               (if (null fn) $cht_cnst (if non-terminating-p $cht_ntmac $cht_tmac)))
     (if (and (null fn) info)
       (setf (rdtab.alist readtable) (delete info (rdtab.alist readtable) :test #'eq)) 
       (if (null info)
         (push (cons char fn) (rdtab.alist readtable))
         (let* ((def (cdr info)))
           (if (atom def)
             (setf (cdr info) fn)         ; Non-dispatching
             (setf (car def) fn))))))     ; Dispatching
    t))

(defun readtable-case (readtable)
  (unless (istruct-typep readtable 'readtable)
    (report-bad-arg readtable 'readtable))
  (let* ((case (rdtab.case (readtable-arg readtable))))
    (if (symbolp case)
      case
      (%car (rassoc case (readtable-case-keywords) :test #'eq)))))

(defun %set-readtable-case (readtable case)
  (unless (istruct-typep readtable 'readtable)
    (report-bad-arg readtable 'readtable))
  (check-type case (member :upcase :downcase :preserve :invert))
  (setf (rdtab.case (readtable-arg readtable)) case))
  
(defsetf readtable-case %set-readtable-case)

(defun make-dispatch-macro-character (char &optional non-terminating-p readtable)
  "Cause CHAR to become a dispatching macro character in readtable (which
   defaults to the current readtable). If NON-TERMINATING-P, the char will
   be non-terminating."
  (setq readtable (readtable-arg readtable))
  (setq char (require-type char 'base-char))
  (let* ((info (nth-value 1 (%get-readtable-char char readtable))))
    (declare (list info))
    (without-interrupts
     (%set-character-attribute char readtable
           (if non-terminating-p $cht_ntmac $cht_tmac))
     (if info
       (rplacd (cdr info) nil)
       (push (cons char (cons #'read-dispatch nil)) (rdtab.alist readtable)))))
  t)

(defun get-dispatch-macro-character (disp-ch sub-ch &optional (readtable *readtable*))
  "Return the macro character function for SUB-CHAR under DISP-CHAR
   or NIL if there is no associated function."
  (setq readtable (readtable-arg (or readtable %initial-readtable%)))
  (setq disp-ch (require-type disp-ch 'base-char))
  (setq sub-ch (char-upcase (require-type sub-ch 'base-char)))
  (unless (digit-char-p sub-ch 10)
    (let* ((def (cdr (nth-value 1 (%get-readtable-char disp-ch readtable)))))
      (if (consp def)
        (cdr (assq sub-ch (cdr def)))
        (error "~A is not a dispatching macro character in ~s ." disp-ch readtable)))))

(defun set-dispatch-macro-character (disp-ch sub-ch fn &optional readtable)
  "Cause FUNCTION to be called whenever the reader reads DISP-CHAR
   followed by SUB-CHAR."
  (setq readtable (readtable-arg readtable))
  (setq disp-ch (require-type disp-ch 'base-char))
  (setq sub-ch (char-upcase (require-type sub-ch 'base-char)))
  (when (digit-char-p sub-ch 10)
    (error "subchar can't be a decimal digit - ~a ." sub-ch))
  (let* ((info (nth-value 1 (%get-readtable-char disp-ch readtable)))
         (def (cdr info)))
    (declare (list info))
    (unless (consp def)
      (error "~A is not a dispatching macro character in ~s ." disp-ch readtable))
    (let* ((alist (cdr def))
           (pair (assq sub-ch alist)))
      (if pair
        (setf (cdr pair) fn)
        (push (cons sub-ch fn) (cdr def))))
    t))


;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;				Reader					;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

(def-standard-initial-binding *read-eval* t "When nil, #. signals an error.")
(def-standard-initial-binding *read-default-float-format* 'single-float)

(def-standard-initial-binding *read-suppress* nil
  "Suppress most interpreting in the reader when T.")

(def-standard-initial-binding *read-base* 10.
  "the radix that Lisp reads numbers in")

(defvar %read-objects% nil)
(defvar %keep-whitespace% nil)




(def-standard-initial-binding %token-strings% (%cons-pool nil))


(defun %return-token-string (token)
  (let* ((str (token.string token))
         (pool %token-strings%))
    (setf (token.string token) nil)
    (without-interrupts
     (setf (pool.data pool)
           (cheap-cons str (pool.data pool))))))

;;;Look for an exact match, else create a simple-string.
(defun %get-token-string (len)
  (declare (fixnum len))
  (without-interrupts
   (do* ((pool %token-strings%)
         (head (cons nil (pool.data pool)))
         (prev head next)
         (next (cdr prev) (cdr next)))
        ((null next)
         (make-string len :element-type 'base-char))
     (declare (dynamic-extent head)
              (list head prev next))
     (let* ((s (car next)))
       (when (= len (length s))
         (rplacd prev (cdr next))
         (setf (pool.data pool) (cdr head))
         (free-cons next)
         (return s))))))

(defun %extend-token-string (token)
  (let* ((old-string (token.string token))
         (old-length (token.len token)))
    (declare (fixnum old-length))
    (let* ((new-length (the fixnum (ash old-length 1)))
           (new-string (%get-token-string new-length)))
      (dotimes (i old-length)
        (setf (%schar new-string i)
              (%schar old-string i)))
      (%return-token-string token)
      (setf (token.string token) new-string
            (token.len token) new-length)
      token)))

(defun %add-char-to-token (char token)
  (let* ((len (token.len token))
         (opos (token.opos token)))
    (declare (fixnum len opos))
    (when (= opos len)
      (%extend-token-string token))
    (setf (token.opos token) (the fixnum (1+ opos))
          (%schar (token.string token) opos) char)))

(defun %string-from-token (token)
  (let* ((opos (token.opos token))
         (ipos (token.ipos token))
         (tstr (token.string token))
         (len (the fixnum (- opos ipos)))
         (string (make-string len :element-type 'base-char)))
    (do* ((k 0 (1+ k))
          (i ipos (1+ i)))
         ((= i opos) string)
      (declare (fixnum i k))
      (setf (%schar string k) (%schar tstr i)))))

(defun %next-token-char (token)
  (let* ((ipos (token.ipos token)))
    (declare (fixnum ipos))
    (when (< ipos (the fixnum (token.opos token)))
      (setf (token.ipos token) (the fixnum (1+ ipos)))
      (%schar (token.string token) ipos))))
      
(defun input-stream-arg (stream)
  (cond ((null stream) *standard-input*)
        ((eq stream t) *terminal-io*)
        ;Otherwise, let ASK complain...
        (t stream)))

(defun %read-char-no-eof (stream)
  (read-char stream))

(defun %next-char-and-attr (stream &optional (attrtab (rdtab.ttab *readtable*)))
  (let* ((ch (read-char stream nil :eof)))
    (if (eq ch :eof)
      (values nil nil)
      (values ch (%character-attribute ch attrtab)))))

(defun %next-non-whitespace-char-and-attr (stream)
  (let* ((attrtab (rdtab.ttab *readtable*)))
    (loop
      (multiple-value-bind (ch attr) (%next-char-and-attr stream attrtab)
        (if (null ch)
          (return (values nil nil))
          (unless (eql attr $cht_wsp)
            (return (values ch attr))))))))

(defun %next-char-and-attr-no-eof (stream &optional (attrtab (rdtab.ttab *readtable*)))
  (let* ((ch (%read-char-no-eof stream)))
    (values ch (%character-attribute ch attrtab))))

(defun %next-non-whitespace-char-and-attr-no-eof (stream)
  (let* ((attrtab (rdtab.ttab *readtable*)))
    (loop
      (multiple-value-bind (ch attr) (%next-char-and-attr-no-eof stream attrtab)
        (declare (fixnum attr))
        (unless (= attr $cht_wsp)
          (return (values ch attr)))))))

;;; "escapes" is a list of escaped character positions, in reverse order
(defun %casify-token (token escapes)
  (let* ((case (readtable-case *readtable*))
         (opos (token.opos token))
         (string (token.string token)))
    (declare (fixnum opos))
    (if (and (null escapes) (eq case :upcase))          ; Most common case, pardon the pun
      ; %strup is faster - boot probs tho
      (dotimes (i opos)
        (setf (%schar string i) (char-upcase (%schar string i))))
      (unless (eq case :preserve)
        (when (eq case :invert)
          (let* ((lower-seen nil)
                 (upper-seen nil))
            (do* ((i (1- opos) (1- i))
                  (esclist escapes)
                  (nextesc (if esclist (pop esclist) -1)))
                 ((< i 0) (if upper-seen (unless lower-seen (setq case :downcase))
                                         (when lower-seen (setq case :upcase))))
              (declare (fixnum i nextesc))
              (if (= nextesc i)
                (setq nextesc (if esclist (pop esclist) -1))
                (let* ((ch (%schar string i)))
                  (if (upper-case-p ch)
                    (setq upper-seen t)
                    (if (lower-case-p ch)
                      (setq lower-seen t))))))))
        (if (eq case :upcase)
          (do* ((i (1- opos) (1- i))
                  (nextesc (if escapes (pop escapes) -1)))
               ((< i 0))
            (declare (fixnum i nextesc))
            (if (= nextesc i)
                (setq nextesc (if escapes (pop escapes) -1))
                (setf (%schar string i) (char-upcase (%schar string i)))))
          (if (eq case :downcase)
            (do* ((i (1- opos) (1- i))
                  (nextesc (if escapes (pop escapes) -1)))
               ((< i 0))
            (declare (fixnum i nextesc))
            (if (= nextesc i)
                (setq nextesc (if escapes (pop escapes) -1))
                (setf (%schar string i) (char-downcase (%schar string i)))))))))))

;;; MCL's reader has historically treated ||:foo as a reference to the
;;; symbol FOO in the package which has the null string as its name.
;;; Some other implementations treat it as a keyword.  This takes an
;;; argument indicating whether or not something was "seen" before the
;;; first colon was read, even if that thing caused no characters to
;;; be added to the token.

(defun %token-package (token colonpos seenbeforecolon stream)
  (declare (ignorable stream))
  (if colonpos
    (if (and (eql colonpos 0) (not seenbeforecolon))
      *keyword-package*
      (let* ((string (token.string token)))
        (or (%find-pkg string colonpos)
            (subseq string 0 colonpos)
            #+nomore
            (signal-reader-error stream "Reference to unknown package ~s." (subseq string 0 colonpos)))))
    *package*))

;;; Returns 4 values: reversed list of escaped character positions,
;;; explicit package (if unescaped ":" or "::") or nil, t iff any
;;; non-dot, non-escaped chars in token, and t if either no explicit
;;; package or "::"

(defun %collect-xtoken (token stream 1stchar)
  (let* ((escapes ())
         (nondots nil)
         (explicit-package *read-suppress*)
         (double-colon t)
         (multi-escaped nil))
    (do* ((attrtab (rdtab.ttab *readtable*))
          (char 1stchar (read-char stream nil :eof )))
         ((eq char :eof))
      (flet ((add-note-escape-pos (char token escapes)
               (push (token.opos token) escapes)
               (%add-char-to-token char token)
               escapes))
        (let* ((attr (%character-attribute char attrtab)))
          (declare (fixnum attr))
          (when (or (= attr $cht_tmac)
                    (= attr $cht_wsp))
            (when (or (not (= attr $cht_wsp)) %keep-whitespace%)
              (unread-char char stream))
            (return ))
          (if (= attr $cht_ill)
              (signal-reader-error stream "Illegal character ~S." char)
              (if (= attr $cht_sesc)
                  (setq nondots t 
                        escapes (add-note-escape-pos (%read-char-no-eof stream) token escapes))
                  (if (= attr $cht_mesc)
                      (progn 
                        (setq nondots t)
                        (loop
                            (multiple-value-bind (nextchar nextattr) (%next-char-and-attr-no-eof stream attrtab)
                              (declare (fixnum nextattr))
                              (if (= nextattr $cht_mesc) 
                                  (return (setq multi-escaped t))
                                  (if (= nextattr $cht_sesc)
                                      (setq escapes (add-note-escape-pos (%read-char-no-eof stream) token escapes))
                            (setq escapes (add-note-escape-pos nextchar token escapes)))))))
                  (let* ((opos (token.opos token)))         ; Add char to token, note 1st colonpos
                    (declare (fixnum opos))
                    (if (and (eq char #\:)       ; (package-delimiter-p char ?)
                             (not explicit-package))
                      (let* ((nextch (%read-char-no-eof stream)))
                        (if (eq nextch #\:)
                          (setq double-colon t)
                          (progn
			    (unread-char nextch stream)
                            (setq double-colon nil)))
                        (%casify-token token escapes)
                        (setq explicit-package (%token-package token opos nondots stream)
                              nondots t
                              escapes nil)
                        (setf (token.opos token) 0))
                      (progn
                        (unless (eq char #\.) (setq nondots t))
                        (%add-char-to-token char token))))))))))
        (values (or escapes multi-escaped) (if *read-suppress* nil explicit-package) nondots double-colon)))
          
(defun %validate-radix (radix)
  (if (and (typep radix 'fixnum)
           (>= (the fixnum radix) 2)
           (<= (the fixnum radix) 36))
    radix
    (progn
      (check-type radix (integer 2 36))
      radix)))

(defun %token-to-number (token radix &optional no-rat)
  (new-numtoken (token.string token) (token.ipos token) (token.opos token) radix no-rat))

;;; If we're allowed to have a single "." in this context, DOT-OK is some distinguished
;;; value that's returned to the caller when exactly one dot is present.
(defun %parse-token (stream firstchar dot-ok)
  (with-token-buffer (tb)
    (multiple-value-bind (escapes explicit-package nondots double-colon) (%collect-xtoken tb stream firstchar)
      (unless *read-suppress* 
        (let* ((string (token.string tb))
               (len (token.opos tb)))
          (declare (fixnum len))
          (if (not nondots)
            (if (= len 1)
              (or dot-ok
                  (signal-reader-error stream "Dot context error in ~s." (%string-from-token tb)))
              (signal-reader-error stream "Illegal symbol syntax in ~s." (%string-from-token tb)))
            ;; Something other than a buffer full of dots.  Thank god.
            (let* ((num (if (null escapes)
                            (handler-case
                                (%token-to-number tb (%validate-radix *read-base*))
                              (arithmetic-error (c)
                                (error 'impossible-number
                                       :stream stream
                                       :token (%string-from-token tb)
                                       :condition c))))))
              (if (and num (not explicit-package))
                num
                (if (and (zerop len) (null escapes))
                  (%err-disp $XBADSYM)
                  (progn                  ; Muck with readtable case of extended token.
                    (%casify-token tb (unless (atom escapes) escapes))
                    (let* ((pkg (if explicit-package (pkg-arg explicit-package) *package*)))
                      (if (or double-colon (eq pkg *keyword-package*))
                        (with-package-lock (pkg)
			  (multiple-value-bind (symbol access internal-offset external-offset)
			      (%find-symbol string len pkg)
			    (if access
			      symbol
			      (%add-symbol (%string-from-token tb) pkg internal-offset external-offset))))
                        (multiple-value-bind (found symbol) (%get-htab-symbol string len (pkg.etab pkg))
                          (if found
                            symbol
                            (let* ((token (%string-from-token tb))
                                   (symbol (find-symbol token pkg)))
                              (with-simple-restart (continue
                                                    "~:[Create and use the internal symbol ~a::~a~;Use the internal symbol ~:*~s~]"
                                                    symbol (package-name pkg) token)
                                (%err-disp $XNOESYM token pkg))
                              (or symbol (intern token pkg)))))))))))))))))
                    
#|
(defun %parse-token-test (string &key dot-ok (case (readtable-case *readtable*)))
  (let* ((stream (make-string-input-stream string))
         (oldcase (readtable-case *readtable*)))
    (unwind-protect
      (progn
        (setf (readtable-case *readtable*) case) 
        (%parse-token stream (read-char stream t) dot-ok))
      (setf (readtable-case *readtable*) oldcase))))

(%parse-token-test "ABC")
(%parse-token-test "TRAPS::_DEBUGGER")
(%parse-token-test "3.14159")
(ignore-errors (%parse-token-test "BAD-PACKAGE:WORSE-SYMBOL"))
(ignore-errors (%parse-token-test "CCL::"))
(%parse-token-test "TRAPS::_debugger" :case :preserve)
(%parse-token-test ":foo")
|#

;;; firstchar must not be whitespace.
;;; People who think that there's so much overhead in all of
;;; this (multiple-value-list, etc.) should probably consider
;;; rewriting those parts of the CLOS and I/O code that make
;;; using things like READ-CHAR impractical...

;;; mb: the reason multiple-value-list is used here is that we need to distunguish between the
;;; recursive parse call returning (values nil) and (values).
(defun %parse-expression (stream firstchar dot-ok)
  (let* ((readtable *readtable*)
         (attrtab (rdtab.ttab readtable))
         (attr (%character-attribute firstchar attrtab))
         (start-pos (stream-position stream)))
    (declare (fixnum attr))
    (when (eql attr $cht_ill)
      (signal-reader-error stream "Illegal character ~S." firstchar))
    (let* ((vals (multiple-value-list 
                     (if (not (logbitp $cht_macbit attr))
                       (%parse-token stream firstchar dot-ok)
                       (let* ((def (cdr (assq firstchar (rdtab.alist readtable)))))
                         (cond ((null def))
                               ((atom def)
                                (funcall def stream firstchar))
                               #+no     ; include if %initial-readtable% broken (see above)
                               ((and (consp (car def))
                                     (eq (caar def) 'function))
                                (funcall (cadar def) stream firstchar))
                               ((functionp (car def))
                                (funcall (car def) stream firstchar))
                               (t (error "Bogus default dispatch fn: ~S" (car def)) nil))))))
           (end-pos (and start-pos (stream-position stream))))
      (declare (dynamic-extent vals)
               (list vals))
      (if (null vals)
        (values nil nil)
        (destructuring-bind (form &optional nested-source-notes)
                            vals
          ;; Can't really trust random reader macros to return source notes...
          (unless (and (consp nested-source-notes)
                       (source-note-p (car nested-source-notes)))
            (setq nested-source-notes nil))
          (values form
                  t
                  (and start-pos
                       (record-source-note :form form
                                           :stream stream
                                           :start-pos (1- start-pos)
                                           :end-pos end-pos
                                           :subform-notes nested-source-notes))))))))

#|
(defun %parse-expression-test (string)
  (let* ((stream (make-string-input-stream string)))
    (%parse-expression stream (read-char stream t) nil)))

(%parse-expression-test ";hello")
(%parse-expression-test "#'cdr")
(%parse-expression-test "#+foo 1 2")

|#

(defun %read-list-expression (stream dot-ok &optional (termch #\)))
  (loop
      (let* ((firstch (%next-non-whitespace-char-and-attr-no-eof stream)))
        (if (eq firstch termch)
            (return (values nil nil nil))
            (multiple-value-bind (val val-p source-info)
                (%parse-expression stream firstch dot-ok)
              (if val-p
                  (return (values val t source-info))))))))

(defun read-list (stream &optional nodots (termch #\)))
  (let* ((dot-ok (cons nil nil))
         (head (cons nil nil))
         (tail head)
         (source-note-list nil))
    (declare (dynamic-extent dot-ok head)
             (list head tail))
    (if nodots (setq dot-ok nil))
    (multiple-value-bind (firstform firstform-p firstform-source-note)
        (%read-list-expression stream dot-ok termch)
      (when firstform-source-note
        (push firstform-source-note source-note-list))
      (when firstform-p
        (if (and dot-ok (eq firstform dot-ok))       ; just read a dot
            (signal-reader-error stream "Dot context error."))
        (rplacd tail (setq tail (cons firstform nil)))
        (loop
          (multiple-value-bind (nextform nextform-p nextform-source-note)
              (%read-list-expression stream dot-ok termch)
            (when nextform-source-note
              (push nextform-source-note source-note-list))
            (if (not nextform-p) (return))
            (if (and dot-ok (eq nextform dot-ok))    ; just read a dot
                (if (multiple-value-bind (lastform lastform-p lastform-source-note)
                        (%read-list-expression stream nil termch)
                      (when lastform-source-note
                        (push lastform-source-note source-note-list))
                      (and lastform-p
                           (progn (rplacd tail lastform)
                                  (not (nth-value 1 (%read-list-expression stream nil termch))))))
                    (return)
                    (signal-reader-error stream "Dot context error."))
              (rplacd tail (setq tail (cons nextform nil))))))))
    (values (cdr head) source-note-list)))

#|
(defun read-list-test (string &optional nodots)
  (read-list (make-string-input-stream string) nodots))

(read-list-test ")")
(read-list-test "a b c)" t)
(read-list-test "a b ;hello
c)" t)

|#

(set-macro-character
 #\(
 #'(lambda (stream ignore)
     (declare (ignore ignore))
     (read-list stream nil #\))))

(set-macro-character 
 #\' 
 (nfunction |'-reader| 
            (lambda (stream ignore)
              (declare (ignore ignore))
              (multiple-value-bind (form source-note)
                  (read-internal stream t nil t)
                (values `(quote ,form) (and source-note (list source-note)))))))

(defparameter *alternate-line-terminator*
    #+darwin-target #\Return
    #-darwin-target nil
    "This variable is currently only used by the standard reader macro
function for #\; (single-line comments); that function reads successive
characters until EOF, a #\NewLine is read, or a character EQL to the value
of *alternate-line-terminator* is read. In Clozure CL for Darwin, the value
of this variable is initially #\Return ; in Clozure CL for other OSes, it's
initially NIL.")
	     
(set-macro-character
 #\;
 (nfunction |;-reader|
            (lambda (stream ignore)
              (declare (ignore ignore))
              (let* ((ch nil))
                (loop 
                    (if (or (eq :eof (setq ch (read-char stream nil :eof)))
                            (eq ch #\NewLine)
			    (eq ch *alternate-line-terminator*))
                        (return (values))))))))

(defun read-string (stream termch)
  (with-token-buffer (tb)
    (do* ((attrs (rdtab.ttab *readtable*))
          (ch (%read-char-no-eof stream)
              (%read-char-no-eof stream)))
         ((eq ch termch)
          (%string-from-token tb))
      (if (= (the fixnum (%character-attribute ch attrs)) $CHT_SESC)
          (setq ch (%read-char-no-eof stream)))
      (%add-char-to-token ch tb))))

(set-macro-character #\" #'read-string)

(defparameter *ignore-extra-close-parenthesis* nil)

(set-macro-character 
 #\)
 #'(lambda (stream ch)
     (let* ((pos (if (typep stream 'file-stream)
                     (file-position stream))))
       (if *ignore-extra-close-parenthesis*
           (warn "Ignoring extra \"~c\" ~@[near position ~d~] on ~s ." ch pos stream)
           (signal-reader-error stream "Unmatched ')' ~@[near position ~d~]." pos)))))




(eval-when (:load-toplevel)             ; But not when mousing around!
  (make-dispatch-macro-character #\# t))


(set-dispatch-macro-character
 #\#
 #\(
 (nfunction 
  |#(-reader| 
  (lambda (stream subchar numarg)
    (declare (ignore subchar))
    (if (or (null numarg) *read-suppress*)
      (multiple-value-bind (lst notes) (read-list stream t)
        (let* ((len (length lst))
               (vec (make-array len)))
          (declare (list lst) (fixnum len) (simple-vector vec))
          (dotimes (i len)
            (setf (svref vec i) (pop lst)))
          (values vec notes)))
      (locally
          (declare (fixnum numarg))
        (do* ((vec (make-array numarg))
              (notes ())
              (lastform)
              (i 0 (1+ i)))
            ((multiple-value-bind (form form-p source-info)
                 (%read-list-expression stream nil)
               (if form-p
                 (progn
                   (setq lastform form)
                   (when source-info (push source-info notes)))
                 (unless (= i numarg)
                   (if (= i 0) 
                     (%err-disp $XARROOB -1 vec)
                     (do* ((j i (1+ j)))
                         ((= j numarg))
                       (declare (fixnum j))
                       (setf (svref vec j) lastform)))))
               (not form-p))
               (values vec notes))
          (declare (fixnum i))
          (setf (svref vec i) lastform)))))))

(defun %read-rational (stream subchar radix)
  (declare (ignore subchar))
  (with-token-buffer (tb)
      (multiple-value-bind (escapes xpackage)
                           (%collect-xtoken tb stream (%next-non-whitespace-char-and-attr-no-eof stream))
        (unless *read-suppress*
          (let* ((val (%token-to-number tb radix)))
          (or (and (null escapes)
                   (null xpackage)
                   (typep val 'rational)
                   val)
              (%err-disp $xbadnum)))))))

(defun require-numarg (subchar numarg)
  (or numarg *read-suppress*
      (error "Numeric argument required for #~A reader macro ." subchar)))

(defun require-no-numarg (subchar numarg)
  (if (and numarg (not *read-suppress*))
      (error "Spurious numeric argument in #~D~A reader macro ." numarg subchar)))

(defun read-eval (stream subchar numarg)
  (require-no-numarg subchar numarg)
  (if *read-eval*
    (let* ((exp (%read-list-expression stream nil)))
      (unless *read-suppress*
        (eval exp)))
    (signal-reader-error stream "#. reader macro invoked when ~S is false ."
                         '*read-eval*)))

(set-dispatch-macro-character 
 #\# 
 #\C
 #'(lambda (stream char arg)
     (require-no-numarg char arg )
     (multiple-value-bind (form note) (read-internal stream t nil t)
       (values (unless *read-suppress* (apply #'complex form)) (and note (list note))))))

(set-dispatch-macro-character 
 #\#
 #\.
 #'read-eval)

;;; Read a valid, non-numeric token string from stream; *READ-SUPPRESS*
;;; is known to be false.
(defun read-symbol-token (stream)
  (multiple-value-bind (firstch attr) (%next-non-whitespace-char-and-attr-no-eof stream)
    (declare (fixnum attr))
    (with-token-buffer (tb)
      (if (or (= attr $CHT_ILL)
              (logbitp $cht_macbit attr)
              (multiple-value-bind (escapes explicit-package nondots) (%collect-xtoken tb stream firstch)
                (declare (ignore nondots))
                (%casify-token tb (unless (atom escapes) escapes))
                (or explicit-package
                    (and (not escapes)
                         (%token-to-number tb (%validate-radix *read-base*))))))
        (%err-disp $XBADSYM)
        (%string-from-token tb)))))

(set-dispatch-macro-character
 #\#
 #\:
 #'(lambda (stream subchar numarg)
     (require-no-numarg subchar numarg)
     (if (not *read-suppress*)
       (make-symbol (read-symbol-token stream))
       (progn
         (%read-list-expression stream nil)
         nil))))

(set-dispatch-macro-character 
 #\# 
 #\b
 #'(lambda (stream subchar numarg)
     (require-no-numarg subchar numarg)
     (%read-rational stream subchar 2)))

(set-dispatch-macro-character 
 #\# 
 #\o
 #'(lambda (stream subchar numarg)
     (require-no-numarg subchar numarg)
     (%read-rational stream subchar 8)))

(set-dispatch-macro-character 
 #\# 
 #\x
 #'(lambda (stream subchar numarg)
     (require-no-numarg subchar numarg)
     (%read-rational stream subchar 16)))

(set-dispatch-macro-character 
 #\# 
 #\r
 #'(lambda (stream subchar numarg)
     (unless *read-suppress*
       (require-numarg subchar numarg)
       (check-type numarg (integer 2 36)))
     (%read-rational stream subchar numarg)))

(set-dispatch-macro-character
 #\#
 #\'
 (nfunction |#'-reader| 
            (lambda (stream subchar numarg)
              (require-no-numarg subchar numarg)
              (multiple-value-bind (form note) (read-internal stream t nil t)
                (values `(function ,form) (and note (list note)))))))

(set-dispatch-macro-character
 #\#
 #\|
 (nfunction |#\|-reader| 
            (lambda (stream subchar numarg)
              (require-no-numarg subchar numarg)
              (do* ((lastch nil ch)
                    (ch )
                    (level 1))
                   ((= level 0) (values))
                (declare (fixnum level))
                (setq ch (%read-char-no-eof stream))
                (if (and (eq ch #\|)
                         (eq lastch #\#))
                    (progn 
                      (setq ch nil)
                      (incf level))
                    (if (and (eq ch #\#)
                             (eq lastch #\|))
                        (progn 
                          (setq ch nil)
                          (decf level))))))))

(defun %unreadable (stream description)
  (signal-reader-error stream "~S encountered." description))

(set-dispatch-macro-character
 #\#
 #\<
 #'(lambda (stream &rest ignore)
     (declare (ignore ignore))
     (%unreadable stream "#<")))

(dolist (ch '(#\null #\tab #\linefeed #\page #\return #\space #\312))
  (set-dispatch-macro-character
   #\#
   ch
   #'(lambda (stream &rest ignore)
       (declare (ignore ignore))
       (%unreadable stream "#<whitespace>"))))

(set-dispatch-macro-character
 #\#
 #\)
 #'(lambda (stream &rest ignore)
     (declare (ignore ignore))
     (%unreadable stream "#)")))

(set-dispatch-macro-character
 #\#
 #\\
 #'(lambda (stream subchar numarg)
     (require-no-numarg subchar numarg)
     (with-token-buffer (tb)
       (%collect-xtoken tb stream #\\)
       (unless *read-suppress*
         (let* ((str (%string-from-token tb)))
           (or (name-char str)
               (error "Unknown character name - \"~a\" ." str)))))))


     
;;;Since some built-in read macros used to use internal reader entry points
;;;for efficiency, we couldn't reliably offer a protocol for stream-dependent
;;;recursive reading.  So recursive reads always get done via tyi's, and streams
;;;only get to intercept toplevel reads.
(defun read (&optional stream (eof-error-p t) eof-value recursive-p)
  (declare (resident))
  ;; just return the first value of read-internal
  (values (read-internal stream eof-error-p eof-value recursive-p)))

(defun read-internal (stream eof-error-p eof-value recursive-p)
  (setq stream (input-stream-arg stream))
  (if recursive-p
    (%read-form stream (if eof-error-p 0) nil)
    (let ((%read-objects% nil) (%keep-whitespace% nil))
      (%read-form stream (if eof-error-p 0) eof-value))))

(defun read-preserving-whitespace (&optional stream (eof-error-p t) eof-value recursive-p)
  "Read from STREAM and return the value read, preserving any whitespace
   that followed the object."
  (setq stream (input-stream-arg stream))
  (values
    (if recursive-p
      (%read-form stream 0 nil)
      (let ((%read-objects% nil) (%keep-whitespace% t))
        (%read-form stream (if eof-error-p 0) eof-value)))))


(defun read-delimited-list (char &optional stream recursive-p)
  "Read Lisp values from INPUT-STREAM until the next character after a
   value's representation is CHAR, and return the objects as a list."
  (setq char (require-type char 'character))
  (setq stream (input-stream-arg stream))
  (values
   (let ((%keep-whitespace% nil))
     (if recursive-p
       (%read-form stream char nil)
       (let ((%read-objects% nil))
         (%read-form stream char nil))))))

(defun read-conditional (stream subchar int)
  (declare (ignore int))
  (cond ((eq subchar (read-feature stream))
         (multiple-value-bind (form note) (read-internal stream t nil t)
           (values form (and note (list note)))))
        (t (let* ((*read-suppress* t))
             (read stream t nil t)
             (values)))))

(defun read-feature (stream)
  (let* ((f (let* ((*package* *keyword-package*))
              (read stream t nil t))))
    (labels ((eval-feature (form)
               (cond ((atom form) 
                      (member form *features*))
                     ((eq (car form) :not) 
                      (not (eval-feature (cadr form))))
                     ((eq (car form) :and) 
                      (dolist (subform (cdr form) t)
                        (unless (eval-feature subform) (return))))
                     ((eq (car form) :or) 
                      (dolist (subform (cdr form) nil)
                        (when (eval-feature subform) (return t))))
                     (t (%err-disp $XRDFEATURE form)))))
      (if (eval-feature f) #\+ #\-))))

(set-dispatch-macro-character #\# #\+ #'read-conditional)
(set-dispatch-macro-character #\# #\- #'read-conditional)

(defun %read-form (stream arg eof-val)
  "Read a lisp form from STREAM

arg=0 : read form, error if eof
arg=nil : read form, eof-val if eof.
arg=char : read delimited list"
  (declare (resident))
  (check-type *readtable* readtable)
  (check-type *package* package)
  (if (and arg (not (eq arg 0)))
      (read-list stream nil arg)
      (loop
        (let* ((ch (%next-non-whitespace-char-and-attr stream)))
          (if (null ch)
            (if arg 
              (error 'end-of-file :stream stream)
              (return eof-val))
            (multiple-value-bind (form form-p source-note)
                (%parse-expression stream ch nil)
              (when form-p
                (return
                 (values (if *read-suppress* nil form)
                         source-note)))))))))

;;;Until load backquote...
(set-macro-character #\`
  #'(lambda (stream char) (declare (ignore stream)) (%err-disp $xbadmac char)))
(set-macro-character #\, (get-macro-character #\`))



(set-dispatch-macro-character #\# #\P
 (qlfun |#P-reader| (stream char flags &aux (invalid-string "Invalid flags (~S) for pathname ~S"))
   (declare (ignore char))
   (when (null flags) (setq flags 0))
   (unless (memq flags '(0 1 2 3 4))
     (unless *read-suppress* (report-bad-arg flags '(integer 0 4))))
   (multiple-value-bind (path note) (read-internal stream t nil t)
     (unless *read-suppress*
       (unless (stringp path) (report-bad-arg path 'string))
       (setq path (pathname path))
       (when (%ilogbitp 0 flags)
         (when (%pathname-type path) (error invalid-string flags path))
         (setf (%pathname-type path) :unspecific))
       (when (%ilogbitp 1 flags)
         (when (%pathname-name path) (error invalid-string flags path))
         (setf (%pathname-name path) ""))
       (values path (and note (list note)))))))


;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

(defstruct (source-note (:conc-name "SOURCE-NOTE.") (:constructor %make-source-note))
  ;; For an inner source form, the source-note of the outer source form.
  ;; For outer source note, octets
  source
  filename
  ;; start and end file positions (NOT characters positions)
  file-range)

(defun make-source-note (&key filename start-pos end-pos source)
  (%make-source-note :filename filename
                     :file-range (encode-file-range start-pos end-pos)
                     :source source))

(defmethod print-object ((sn source-note) stream)
  (print-unreadable-object (sn stream :type t :identity nil)
    (print-source-note sn stream)))

(defun print-source-note (sn stream)
  (let* ((file (source-note-filename sn))
         (text (ignore-errors (source-note-text sn))))
    (when file
      ;; Should fix this when record the name.
      (when (eq (pathname-version file) :newest)
	(setq file (namestring (make-pathname :version nil :defaults file)))))
    (when text
      (setq text (string-sans-most-whitespace text 121))
      (when (> (length text) 120)
        (setq text (concatenate 'string (subseq text 0 120) "..."))))
    (if file
      (format stream "~s:~s-~s ~s" file
	      (source-note-start-pos sn) (source-note-end-pos sn)
	      text)
      (format stream "Interactive ~s" text))))

(defun source-note-filename (source)
  (if (source-note-p source)
    (source-note.filename source)
    ;;  else null or a pathname, as in record-source-file
    source))

(defun (setf source-note-filename) (filename source-note)
  (setf (source-note.filename (require-type source-note 'source-note)) filename))

;; Since source notes are optional, it simplifies a lot of code
;; to have these accessors allow NIL.

(defun source-note-source (source-note)
  (when source-note
    (source-note.source (require-type source-note 'source-note))))

(defun source-note-file-range (source-note)
  (when source-note
    (source-note.file-range (require-type source-note 'source-note))))

(defun source-note-start-pos (source-note)
  (let ((range (source-note-file-range source-note)))
    (when range
      (if (consp range) (car range) (ash range -14)))))

(defun source-note-end-pos (source-note)
  (let ((range (source-note-file-range source-note)))
    (when range
      (if (consp range) (cdr range) (+ (ash range -14) (logand range #x3FFF))))))

(defun encode-file-range (start-pos end-pos)
  (let ((len (- end-pos start-pos)))
    (if (< len (ash 1 14))
      (+ (ash start-pos 14) len)
      (cons start-pos end-pos))))

(defun decode-file-range (range)
  (when range
    (if (consp range)
      (values (car range) (cdr range))
      (let ((start-pos (ash range -14)))
        (values start-pos (+ start-pos (logand range #x3FFF)))))))

(defun source-note-text (source-note &optional start end)
  (when source-note
    (let* ((source (source-note-source source-note))
	   (start-pos (source-note-start-pos source-note))
	   (end-pos (source-note-end-pos source-note))
	   (start (or start start-pos))
	   (end (or end end-pos)))
      (etypecase source
	(source-note
         (assert (<= (source-note-start-pos source) start end (source-note-end-pos source)))
         (source-note-text source start end))
	((simple-array (unsigned-byte 8) (*))
         (decf start start-pos)
         (decf end start-pos)
         (assert (and (<= 0 start end (length source))))
         (decode-string-from-octets source :start start :end end :external-format :utf-8))
	(null source)))))

(defun source-note-toplevel-note (source-note)
  (when source-note
    (loop for source = (source-note-source source-note)
          while (source-note-p source)
          do (setq source-note source))
    source-note))

(defvar *recording-source-streams* ())

(defun record-source-note (&key form stream start-pos end-pos subform-notes)
  (let ((recording (assq stream *recording-source-streams*)))
    (when (and recording (not *read-suppress*))
      (destructuring-bind (map file-name stream-offset) (cdr recording)
        (let* ((prev (gethash form map))
               (note (make-source-note :filename file-name
                                       :start-pos (+ stream-offset start-pos)
                                       :end-pos (+ stream-offset end-pos))))
          (setf (gethash form map)
                (cond ((null prev) note)
                      ((consp prev) (cons note prev))
                      (t (list note prev))))
          (loop for subnote in subform-notes
            do (when (source-note-source subnote) (error "Subnote ~s already owned?" subnote))
            do (setf (source-note.source subnote) note))
          note)))))

(defun read-recording-source (stream &key eofval file-name start-offset map save-source-text)
  "Read a top-level form, perhaps recording source locations.
If MAP is NIL, just reads a form as if by READ.
If MAP is non-NIL, returns a second value of a source-note object describing the form.
In addition, if MAP is a hash table, it gets filled with source-note's for all
non-atomic nested subforms."
  (when (null start-offset) (setq start-offset 0))
  (typecase map
    (null (values (read-internal stream nil eofval nil) nil))
    (hash-table
       (let* ((stream (recording-input-stream stream))
	      (recording (list stream map file-name start-offset))
              (*recording-source-streams* (cons recording *recording-source-streams*)))
         (declare (dynamic-extent recording *recording-source-streams*))
         (multiple-value-bind (form source-note) (read-internal stream nil eofval nil)
           (when (and source-note (not (eq form eofval)))
             (assert (null (source-note-source source-note)))
             (loop for form being the hash-key using (hash-value note) of map
                   do (cond ((eq note source-note) nil)
                            ;; Remove entries with multiple source notes, which can happen
                            ;; for atoms.  If we can't tell which instance we mean, then we
                            ;; don't have useful source info.
                            ((listp note) (remhash form map))
                            ((loop for p = note then (source-note-source p) while (source-note-p p)
                                   thereis (eq p source-note))
                             ;; Flatten the backpointers so each subnote points directly
                             ;; to the toplevel note.
                             (setf (source-note.source note) source-note))))
             (when save-source-text
               (setf (source-note.source source-note)
                     (fetch-octets-from-stream stream
                                               (- (source-note-start-pos source-note)
                                                  start-offset)
                                               (- (source-note-end-pos source-note)
                                                  start-offset)))))
           (values form source-note))))
    (T ;; not clear if this is ever useful
       (let* ((start-pos (stream-position stream))
              (form (read-internal stream nil eofval nil))
              (end-pos (and start-pos (neq form eofval) (stream-position stream)))
              (source-note (and end-pos
                                (make-source-note :filename file-name
                                                  :start-pos (+ start-offset start-pos)
                                                  :end-pos (+ start-offset end-pos)))))
         (when (and source-note save-source-text)
           (setf (source-note.source source-note) (fetch-octets-from-stream stream start-pos end-pos)))
         (values form source-note)))))

(defmethod fetch-octets-from-stream ((stream input-stream) start-offset end-offset)
  ;; We basically want to read the bytes between two positions, but there is no
  ;; direct interface for that.  So we let the stream decode and then we re-encode.
  ;; (Just as well, since otherwise we'd have to remember the file's encoding).
  (declare (fixnum start-offset))
  (when (< start-offset end-offset)
    (let* ((cur-pos (stream-position stream))
           (noctets (- end-offset start-offset))
           (vec (make-array noctets :element-type '(unsigned-byte 8)))
           (index 0)
           (crlfp (eq :crlf
                      (cdr (assoc (external-format-line-termination
                                   (stream-external-format stream))
                                  *canonical-line-termination-conventions*)))))
      (declare (type fixnum end-offset noctets index)
               (type (simple-array (unsigned-byte 8) (*)) vec))
      (macrolet ((out (code)
                   `(progn
                      (setf (aref vec index) ,code)
                      (when (eql (incf index) noctets) (return)))))
        (stream-position stream start-offset)
        (loop
          (let ((code (char-code (read-char stream))))
            (declare (fixnum code))
            (cond ((< code #x80)
                   (when (and crlfp (= code (char-code #\NewLine)))
                     (out (char-code #\Return)))
                   (out code))
                  ((< code #x800)
                   (out (logior #xc0 (ldb (byte 5 6) code)))
                   (out (logior #x80 (ldb (byte 6 0) code))))
                  ((< code #x10000)
                   (out (logior #xe0 (ldb (byte 4 12) code)))
                   (out (logior #x80 (ldb (byte 6 6) code)))
                   (out (logior #x80 (ldb (byte 6 0) code))))
                  (t
                   (out (logior #xf0 (ldb (byte 3 18) code)))
                   (out (logior #xe0 (ldb (byte 6 12) code)))
                   (out (logior #x80 (ldb (byte 6 6) code)))
                   (out (logior #x80 (ldb (byte 6 0) code))))))))
      (stream-position stream cur-pos)
      vec)))

(defun ensure-source-note-text (source-note &key (if-does-not-exist nil))
  "Fetch source text from file if don't have it"
  (setq if-does-not-exist (require-type if-does-not-exist '(member :error nil)))
  (if source-note
    (let ((source (source-note-source source-note))
	  (filename (source-note-filename source-note)))
      (etypecase source
	(null
	 (if filename
	   (with-open-file (stream filename :if-does-not-exist if-does-not-exist)
	     (when stream
	       (let ((start (source-note-start-pos source-note))
		     (end (source-note-end-pos source-note))
		     (len (file-length stream)))
		 (if (<= end len)
		     (setf (source-note.source source-note)
			   (fetch-octets-from-stream stream start end))
		     (when if-does-not-exist
		       (error 'simple-file-error :pathname filename
			      :error-type "File ~s changed since source info recorded"))))))
	   (when if-does-not-exist
	     (error "Missing source text in internative source note"))))
	(source-note
	 (ensure-source-note-text source))
	((simple-array (unsigned-byte 8) (*))
	 source)))
    (when if-does-not-exist
      (error "Missing source note"))))


;; This can be called explicitly by macros that do more complicated transforms
(defun note-source-transformation (original new)
  (nx-note-source-transformation original new))


;;; Wrapper stream for recording source of non-random-access streams.
(defclass recording-character-input-stream (fundamental-stream character-input-stream)
  ((input-stream :initarg :input-stream)
   (string :initform (make-array 1024 :element-type 'character :fill-pointer 0 :adjustable t))))

(defmethod stream-element-type ((s recording-character-input-stream))
  (with-slots (input-stream) s
    (stream-element-type input-stream)))

(defmethod stream-read-char ((s recording-character-input-stream))
  (with-slots (input-stream string) s
    (let ((char (stream-read-char input-stream)))
      (when (and char (neq char :eof))
	(vector-push-extend char string))
      char)))

(defmethod stream-read-char-no-hang ((s recording-character-input-stream))
  (with-slots (input-stream string) s
    (let ((char (stream-read-char-no-hang input-stream)))
      (when (and char (neq char :eof))
	(vector-push-extend char string))
      char)))

(defmethod stream-peek-char ((s recording-character-input-stream))
  (with-slots (input-stream) s
    (stream-peek-char input-stream)))

(defmethod stream-listen ((s recording-character-input-stream))
  (with-slots (input-stream) s
    (stream-listen input-stream)))

(defmethod stream-read-line ((s recording-character-input-stream))
  (generic-read-line s))

(defmethod stream-read-list ((s recording-character-input-stream) list count)
  (generic-character-read-list s list count))

(defmethod stream-read-vector ((s recording-character-input-stream) vector start end)
  (generic-character-read-vector s vector start end))

(defmethod stream-unread-char ((s recording-character-input-stream) char)
  (with-slots (input-stream string) s
    (vector-pop string)    ;; Error if no characters read since last reset.
    (stream-unread-char input-stream char)))

(defmethod stream-eofp ((s recording-character-input-stream))
  (with-slots (input-stream) s
    (stream-eofp input-stream)))

(defmethod stream-clear-input ((s recording-character-input-stream))
  (with-slots (input-stream) s
    (stream-clear-input input-stream)))

(defmethod stream-position ((s recording-character-input-stream) &optional newpos)
  (with-slots (string) s
    (unless newpos
      (fill-pointer string))))

(defun recording-input-stream (stream)
  (let ((pos (stream-position stream)))
    (if (and pos (stream-position stream pos))
      stream
      (make-instance 'recording-character-input-stream :input-stream stream))))

(defmethod fetch-octets-from-stream ((s recording-character-input-stream) start-offset end-offset)
  (declare (fixnum start-offset end-offset))
  (with-slots (string) s
    (when (< start-offset end-offset)
      (let* ((sstring (array-data-and-offset string))
	     (noctets (loop for i fixnum from start-offset below end-offset
			 as code fixnum = (%char-code (%schar sstring i))
			 sum (cond ((< code #x80) 1)
				   ((< code #x800) 2)
				   ((< code #x10000) 3)
				   (t 4))
			 of-type fixnum))
	     (vec (make-array noctets :element-type '(unsigned-byte 8)))
	     (index 0))
	(declare (type fixnum noctets index)
		 (type simple-base-string sstring)
		 (type (simple-array (unsigned-byte 8) (*)) vec))
	(macrolet ((out (octet) `(progn
				   (setf (aref vec index) ,octet)
				   (incf index))))
	  (loop for i fixnum from start-offset below end-offset
	     as code fixnum = (%char-code (%schar sstring i))
	     do (cond ((< code #x80)
		       (out code))
		      ((< code #x800)
		       (out (logior #xc0 (ldb (byte 5 6) code)))
		       (out (logior #x80 (ldb (byte 6 0) code))))
		      ((< code #x10000)
		       (out (logior #xe0 (ldb (byte 4 12) code)))
		       (out (logior #x80 (ldb (byte 6 6) code)))
		       (out (logior #x80 (ldb (byte 6 0) code))))
		      (t
		       (out (logior #xf0 (ldb (byte 3 18) code)))
		       (out (logior #xe0 (ldb (byte 6 12) code)))
		       (out (logior #x80 (ldb (byte 6 6) code)))
		       (out (logior #x80 (ldb (byte 6 0) code)))))))
	(setf (fill-pointer string) 0) ;; reset
	vec))))




; end