/[pcre]/code/trunk/pcre_tables.c
ViewVC logotype

Diff of /code/trunk/pcre_tables.c

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 77 by nigel, Sat Feb 24 21:40:45 2007 UTC revision 507 by ph10, Wed Mar 10 16:08:01 2010 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2005 University of Cambridge             Copyright (c) 1997-2009 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 39  POSSIBILITY OF SUCH DAMAGE. Line 39  POSSIBILITY OF SUCH DAMAGE.
39    
40    
41  /* This module contains some fixed tables that are used by more than one of the  /* This module contains some fixed tables that are used by more than one of the
42  PCRE code modules. */  PCRE code modules. The tables are also #included by the pcretest program, which
43    uses macros to change their names from _pcre_xxx to xxxx, thereby avoiding name
44    clashes with the library. */
45    
46    
47    #ifdef HAVE_CONFIG_H
48    #include "config.h"
49    #endif
50    
51  #include "pcre_internal.h"  #include "pcre_internal.h"
52    
53    
54  /* Table of sizes for the fixed-length opcodes. It's defined in a macro so that  /* Table of sizes for the fixed-length opcodes. It's defined in a macro so that
55  the definition is next to the definition of the opcodes in internal.h. */  the definition is next to the definition of the opcodes in pcre_internal.h. */
56    
57  const uschar _pcre_OP_lengths[] = { OP_LENGTHS };  const uschar _pcre_OP_lengths[] = { OP_LENGTHS };
58    
# Line 59  const uschar _pcre_OP_lengths[] = { OP_L Line 65  const uschar _pcre_OP_lengths[] = { OP_L
65  /* These are the breakpoints for different numbers of bytes in a UTF-8  /* These are the breakpoints for different numbers of bytes in a UTF-8
66  character. */  character. */
67    
68    #ifdef SUPPORT_UTF8
69    
70  const int _pcre_utf8_table1[] =  const int _pcre_utf8_table1[] =
71    { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff};    { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff};
72    
# Line 70  first byte of a character, indexed by th Line 78  first byte of a character, indexed by th
78  const int _pcre_utf8_table2[] = { 0,    0xc0, 0xe0, 0xf0, 0xf8, 0xfc};  const int _pcre_utf8_table2[] = { 0,    0xc0, 0xe0, 0xf0, 0xf8, 0xfc};
79  const int _pcre_utf8_table3[] = { 0xff, 0x1f, 0x0f, 0x07, 0x03, 0x01};  const int _pcre_utf8_table3[] = { 0xff, 0x1f, 0x0f, 0x07, 0x03, 0x01};
80    
81  /* Table of the number of extra characters, indexed by the first character  /* Table of the number of extra bytes, indexed by the first byte masked with
82  masked with 0x3f. The highest number for a valid UTF-8 character is in fact  0x3f. The highest number for a valid UTF-8 first byte is in fact 0x3d. */
 0x3d. */  
83    
84  const uschar _pcre_utf8_table4[] = {  const uschar _pcre_utf8_table4[] = {
85    1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,    1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
# Line 80  const uschar _pcre_utf8_table4[] = { Line 87  const uschar _pcre_utf8_table4[] = {
87    2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,    2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
88    3,3,3,3,3,3,3,3,4,4,4,4,5,5,5,5 };    3,3,3,3,3,3,3,3,4,4,4,4,5,5,5,5 };
89    
90  /* This table translates Unicode property names into code values for the  /* Table to translate from particular type value to the general value. */
91  ucp_findchar() function. It is used by pcretest as well as by the library  
92  functions. */  const int _pcre_ucp_gentype[] = {
93      ucp_C, ucp_C, ucp_C, ucp_C, ucp_C,  /* Cc, Cf, Cn, Co, Cs */
94      ucp_L, ucp_L, ucp_L, ucp_L, ucp_L,  /* Ll, Lu, Lm, Lo, Lt */
95      ucp_M, ucp_M, ucp_M,                /* Mc, Me, Mn */
96      ucp_N, ucp_N, ucp_N,                /* Nd, Nl, No */
97      ucp_P, ucp_P, ucp_P, ucp_P, ucp_P,  /* Pc, Pd, Pe, Pf, Pi */
98      ucp_P, ucp_P,                       /* Ps, Po */
99      ucp_S, ucp_S, ucp_S, ucp_S,         /* Sc, Sk, Sm, So */
100      ucp_Z, ucp_Z, ucp_Z                 /* Zl, Zp, Zs */
101    };
102    
103    /* The pcre_utt[] table below translates Unicode property names into type and
104    code values. It is searched by binary chop, so must be in collating sequence of
105    name. Originally, the table contained pointers to the name strings in the first
106    field of each entry. However, that leads to a large number of relocations when
107    a shared library is dynamically loaded. A significant reduction is made by
108    putting all the names into a single, large string and then using offsets in the
109    table itself. Maintenance is more error-prone, but frequent changes to this
110    data are unlikely.
111    
112    July 2008: There is now a script called maint/GenerateUtt.py that can be used
113    to generate this data instead of maintaining it entirely by hand.
114    
115    The script was updated in March 2009 to generate a new EBCDIC-compliant
116    version. Like all other character and string literals that are compared against
117    the regular expression pattern, we must use STR_ macros instead of literal
118    strings to make sure that UTF-8 support works on EBCDIC platforms. */
119    
120    #define STRING_Any0 STR_A STR_n STR_y "\0"
121    #define STRING_Arabic0 STR_A STR_r STR_a STR_b STR_i STR_c "\0"
122    #define STRING_Armenian0 STR_A STR_r STR_m STR_e STR_n STR_i STR_a STR_n "\0"
123    #define STRING_Avestan0 STR_A STR_v STR_e STR_s STR_t STR_a STR_n "\0"
124    #define STRING_Balinese0 STR_B STR_a STR_l STR_i STR_n STR_e STR_s STR_e "\0"
125    #define STRING_Bamum0 STR_B STR_a STR_m STR_u STR_m "\0"
126    #define STRING_Bengali0 STR_B STR_e STR_n STR_g STR_a STR_l STR_i "\0"
127    #define STRING_Bopomofo0 STR_B STR_o STR_p STR_o STR_m STR_o STR_f STR_o "\0"
128    #define STRING_Braille0 STR_B STR_r STR_a STR_i STR_l STR_l STR_e "\0"
129    #define STRING_Buginese0 STR_B STR_u STR_g STR_i STR_n STR_e STR_s STR_e "\0"
130    #define STRING_Buhid0 STR_B STR_u STR_h STR_i STR_d "\0"
131    #define STRING_C0 STR_C "\0"
132    #define STRING_Canadian_Aboriginal0 STR_C STR_a STR_n STR_a STR_d STR_i STR_a STR_n STR_UNDERSCORE STR_A STR_b STR_o STR_r STR_i STR_g STR_i STR_n STR_a STR_l "\0"
133    #define STRING_Carian0 STR_C STR_a STR_r STR_i STR_a STR_n "\0"
134    #define STRING_Cc0 STR_C STR_c "\0"
135    #define STRING_Cf0 STR_C STR_f "\0"
136    #define STRING_Cham0 STR_C STR_h STR_a STR_m "\0"
137    #define STRING_Cherokee0 STR_C STR_h STR_e STR_r STR_o STR_k STR_e STR_e "\0"
138    #define STRING_Cn0 STR_C STR_n "\0"
139    #define STRING_Co0 STR_C STR_o "\0"
140    #define STRING_Common0 STR_C STR_o STR_m STR_m STR_o STR_n "\0"
141    #define STRING_Coptic0 STR_C STR_o STR_p STR_t STR_i STR_c "\0"
142    #define STRING_Cs0 STR_C STR_s "\0"
143    #define STRING_Cuneiform0 STR_C STR_u STR_n STR_e STR_i STR_f STR_o STR_r STR_m "\0"
144    #define STRING_Cypriot0 STR_C STR_y STR_p STR_r STR_i STR_o STR_t "\0"
145    #define STRING_Cyrillic0 STR_C STR_y STR_r STR_i STR_l STR_l STR_i STR_c "\0"
146    #define STRING_Deseret0 STR_D STR_e STR_s STR_e STR_r STR_e STR_t "\0"
147    #define STRING_Devanagari0 STR_D STR_e STR_v STR_a STR_n STR_a STR_g STR_a STR_r STR_i "\0"
148    #define STRING_Egyptian_Hieroglyphs0 STR_E STR_g STR_y STR_p STR_t STR_i STR_a STR_n STR_UNDERSCORE STR_H STR_i STR_e STR_r STR_o STR_g STR_l STR_y STR_p STR_h STR_s "\0"
149    #define STRING_Ethiopic0 STR_E STR_t STR_h STR_i STR_o STR_p STR_i STR_c "\0"
150    #define STRING_Georgian0 STR_G STR_e STR_o STR_r STR_g STR_i STR_a STR_n "\0"
151    #define STRING_Glagolitic0 STR_G STR_l STR_a STR_g STR_o STR_l STR_i STR_t STR_i STR_c "\0"
152    #define STRING_Gothic0 STR_G STR_o STR_t STR_h STR_i STR_c "\0"
153    #define STRING_Greek0 STR_G STR_r STR_e STR_e STR_k "\0"
154    #define STRING_Gujarati0 STR_G STR_u STR_j STR_a STR_r STR_a STR_t STR_i "\0"
155    #define STRING_Gurmukhi0 STR_G STR_u STR_r STR_m STR_u STR_k STR_h STR_i "\0"
156    #define STRING_Han0 STR_H STR_a STR_n "\0"
157    #define STRING_Hangul0 STR_H STR_a STR_n STR_g STR_u STR_l "\0"
158    #define STRING_Hanunoo0 STR_H STR_a STR_n STR_u STR_n STR_o STR_o "\0"
159    #define STRING_Hebrew0 STR_H STR_e STR_b STR_r STR_e STR_w "\0"
160    #define STRING_Hiragana0 STR_H STR_i STR_r STR_a STR_g STR_a STR_n STR_a "\0"
161    #define STRING_Imperial_Aramaic0 STR_I STR_m STR_p STR_e STR_r STR_i STR_a STR_l STR_UNDERSCORE STR_A STR_r STR_a STR_m STR_a STR_i STR_c "\0"
162    #define STRING_Inherited0 STR_I STR_n STR_h STR_e STR_r STR_i STR_t STR_e STR_d "\0"
163    #define STRING_Inscriptional_Pahlavi0 STR_I STR_n STR_s STR_c STR_r STR_i STR_p STR_t STR_i STR_o STR_n STR_a STR_l STR_UNDERSCORE STR_P STR_a STR_h STR_l STR_a STR_v STR_i "\0"
164    #define STRING_Inscriptional_Parthian0 STR_I STR_n STR_s STR_c STR_r STR_i STR_p STR_t STR_i STR_o STR_n STR_a STR_l STR_UNDERSCORE STR_P STR_a STR_r STR_t STR_h STR_i STR_a STR_n "\0"
165    #define STRING_Javanese0 STR_J STR_a STR_v STR_a STR_n STR_e STR_s STR_e "\0"
166    #define STRING_Kaithi0 STR_K STR_a STR_i STR_t STR_h STR_i "\0"
167    #define STRING_Kannada0 STR_K STR_a STR_n STR_n STR_a STR_d STR_a "\0"
168    #define STRING_Katakana0 STR_K STR_a STR_t STR_a STR_k STR_a STR_n STR_a "\0"
169    #define STRING_Kayah_Li0 STR_K STR_a STR_y STR_a STR_h STR_UNDERSCORE STR_L STR_i "\0"
170    #define STRING_Kharoshthi0 STR_K STR_h STR_a STR_r STR_o STR_s STR_h STR_t STR_h STR_i "\0"
171    #define STRING_Khmer0 STR_K STR_h STR_m STR_e STR_r "\0"
172    #define STRING_L0 STR_L "\0"
173    #define STRING_L_AMPERSAND0 STR_L STR_AMPERSAND "\0"
174    #define STRING_Lao0 STR_L STR_a STR_o "\0"
175    #define STRING_Latin0 STR_L STR_a STR_t STR_i STR_n "\0"
176    #define STRING_Lepcha0 STR_L STR_e STR_p STR_c STR_h STR_a "\0"
177    #define STRING_Limbu0 STR_L STR_i STR_m STR_b STR_u "\0"
178    #define STRING_Linear_B0 STR_L STR_i STR_n STR_e STR_a STR_r STR_UNDERSCORE STR_B "\0"
179    #define STRING_Lisu0 STR_L STR_i STR_s STR_u "\0"
180    #define STRING_Ll0 STR_L STR_l "\0"
181    #define STRING_Lm0 STR_L STR_m "\0"
182    #define STRING_Lo0 STR_L STR_o "\0"
183    #define STRING_Lt0 STR_L STR_t "\0"
184    #define STRING_Lu0 STR_L STR_u "\0"
185    #define STRING_Lycian0 STR_L STR_y STR_c STR_i STR_a STR_n "\0"
186    #define STRING_Lydian0 STR_L STR_y STR_d STR_i STR_a STR_n "\0"
187    #define STRING_M0 STR_M "\0"
188    #define STRING_Malayalam0 STR_M STR_a STR_l STR_a STR_y STR_a STR_l STR_a STR_m "\0"
189    #define STRING_Mc0 STR_M STR_c "\0"
190    #define STRING_Me0 STR_M STR_e "\0"
191    #define STRING_Meetei_Mayek0 STR_M STR_e STR_e STR_t STR_e STR_i STR_UNDERSCORE STR_M STR_a STR_y STR_e STR_k "\0"
192    #define STRING_Mn0 STR_M STR_n "\0"
193    #define STRING_Mongolian0 STR_M STR_o STR_n STR_g STR_o STR_l STR_i STR_a STR_n "\0"
194    #define STRING_Myanmar0 STR_M STR_y STR_a STR_n STR_m STR_a STR_r "\0"
195    #define STRING_N0 STR_N "\0"
196    #define STRING_Nd0 STR_N STR_d "\0"
197    #define STRING_New_Tai_Lue0 STR_N STR_e STR_w STR_UNDERSCORE STR_T STR_a STR_i STR_UNDERSCORE STR_L STR_u STR_e "\0"
198    #define STRING_Nko0 STR_N STR_k STR_o "\0"
199    #define STRING_Nl0 STR_N STR_l "\0"
200    #define STRING_No0 STR_N STR_o "\0"
201    #define STRING_Ogham0 STR_O STR_g STR_h STR_a STR_m "\0"
202    #define STRING_Ol_Chiki0 STR_O STR_l STR_UNDERSCORE STR_C STR_h STR_i STR_k STR_i "\0"
203    #define STRING_Old_Italic0 STR_O STR_l STR_d STR_UNDERSCORE STR_I STR_t STR_a STR_l STR_i STR_c "\0"
204    #define STRING_Old_Persian0 STR_O STR_l STR_d STR_UNDERSCORE STR_P STR_e STR_r STR_s STR_i STR_a STR_n "\0"
205    #define STRING_Old_South_Arabian0 STR_O STR_l STR_d STR_UNDERSCORE STR_S STR_o STR_u STR_t STR_h STR_UNDERSCORE STR_A STR_r STR_a STR_b STR_i STR_a STR_n "\0"
206    #define STRING_Old_Turkic0 STR_O STR_l STR_d STR_UNDERSCORE STR_T STR_u STR_r STR_k STR_i STR_c "\0"
207    #define STRING_Oriya0 STR_O STR_r STR_i STR_y STR_a "\0"
208    #define STRING_Osmanya0 STR_O STR_s STR_m STR_a STR_n STR_y STR_a "\0"
209    #define STRING_P0 STR_P "\0"
210    #define STRING_Pc0 STR_P STR_c "\0"
211    #define STRING_Pd0 STR_P STR_d "\0"
212    #define STRING_Pe0 STR_P STR_e "\0"
213    #define STRING_Pf0 STR_P STR_f "\0"
214    #define STRING_Phags_Pa0 STR_P STR_h STR_a STR_g STR_s STR_UNDERSCORE STR_P STR_a "\0"
215    #define STRING_Phoenician0 STR_P STR_h STR_o STR_e STR_n STR_i STR_c STR_i STR_a STR_n "\0"
216    #define STRING_Pi0 STR_P STR_i "\0"
217    #define STRING_Po0 STR_P STR_o "\0"
218    #define STRING_Ps0 STR_P STR_s "\0"
219    #define STRING_Rejang0 STR_R STR_e STR_j STR_a STR_n STR_g "\0"
220    #define STRING_Runic0 STR_R STR_u STR_n STR_i STR_c "\0"
221    #define STRING_S0 STR_S "\0"
222    #define STRING_Samaritan0 STR_S STR_a STR_m STR_a STR_r STR_i STR_t STR_a STR_n "\0"
223    #define STRING_Saurashtra0 STR_S STR_a STR_u STR_r STR_a STR_s STR_h STR_t STR_r STR_a "\0"
224    #define STRING_Sc0 STR_S STR_c "\0"
225    #define STRING_Shavian0 STR_S STR_h STR_a STR_v STR_i STR_a STR_n "\0"
226    #define STRING_Sinhala0 STR_S STR_i STR_n STR_h STR_a STR_l STR_a "\0"
227    #define STRING_Sk0 STR_S STR_k "\0"
228    #define STRING_Sm0 STR_S STR_m "\0"
229    #define STRING_So0 STR_S STR_o "\0"
230    #define STRING_Sundanese0 STR_S STR_u STR_n STR_d STR_a STR_n STR_e STR_s STR_e "\0"
231    #define STRING_Syloti_Nagri0 STR_S STR_y STR_l STR_o STR_t STR_i STR_UNDERSCORE STR_N STR_a STR_g STR_r STR_i "\0"
232    #define STRING_Syriac0 STR_S STR_y STR_r STR_i STR_a STR_c "\0"
233    #define STRING_Tagalog0 STR_T STR_a STR_g STR_a STR_l STR_o STR_g "\0"
234    #define STRING_Tagbanwa0 STR_T STR_a STR_g STR_b STR_a STR_n STR_w STR_a "\0"
235    #define STRING_Tai_Le0 STR_T STR_a STR_i STR_UNDERSCORE STR_L STR_e "\0"
236    #define STRING_Tai_Tham0 STR_T STR_a STR_i STR_UNDERSCORE STR_T STR_h STR_a STR_m "\0"
237    #define STRING_Tai_Viet0 STR_T STR_a STR_i STR_UNDERSCORE STR_V STR_i STR_e STR_t "\0"
238    #define STRING_Tamil0 STR_T STR_a STR_m STR_i STR_l "\0"
239    #define STRING_Telugu0 STR_T STR_e STR_l STR_u STR_g STR_u "\0"
240    #define STRING_Thaana0 STR_T STR_h STR_a STR_a STR_n STR_a "\0"
241    #define STRING_Thai0 STR_T STR_h STR_a STR_i "\0"
242    #define STRING_Tibetan0 STR_T STR_i STR_b STR_e STR_t STR_a STR_n "\0"
243    #define STRING_Tifinagh0 STR_T STR_i STR_f STR_i STR_n STR_a STR_g STR_h "\0"
244    #define STRING_Ugaritic0 STR_U STR_g STR_a STR_r STR_i STR_t STR_i STR_c "\0"
245    #define STRING_Vai0 STR_V STR_a STR_i "\0"
246    #define STRING_Yi0 STR_Y STR_i "\0"
247    #define STRING_Z0 STR_Z "\0"
248    #define STRING_Zl0 STR_Z STR_l "\0"
249    #define STRING_Zp0 STR_Z STR_p "\0"
250    #define STRING_Zs0 STR_Z STR_s "\0"
251    
252    const char _pcre_utt_names[] =
253      STRING_Any0
254      STRING_Arabic0
255      STRING_Armenian0
256      STRING_Avestan0
257      STRING_Balinese0
258      STRING_Bamum0
259      STRING_Bengali0
260      STRING_Bopomofo0
261      STRING_Braille0
262      STRING_Buginese0
263      STRING_Buhid0
264      STRING_C0
265      STRING_Canadian_Aboriginal0
266      STRING_Carian0
267      STRING_Cc0
268      STRING_Cf0
269      STRING_Cham0
270      STRING_Cherokee0
271      STRING_Cn0
272      STRING_Co0
273      STRING_Common0
274      STRING_Coptic0
275      STRING_Cs0
276      STRING_Cuneiform0
277      STRING_Cypriot0
278      STRING_Cyrillic0
279      STRING_Deseret0
280      STRING_Devanagari0
281      STRING_Egyptian_Hieroglyphs0
282      STRING_Ethiopic0
283      STRING_Georgian0
284      STRING_Glagolitic0
285      STRING_Gothic0
286      STRING_Greek0
287      STRING_Gujarati0
288      STRING_Gurmukhi0
289      STRING_Han0
290      STRING_Hangul0
291      STRING_Hanunoo0
292      STRING_Hebrew0
293      STRING_Hiragana0
294      STRING_Imperial_Aramaic0
295      STRING_Inherited0
296      STRING_Inscriptional_Pahlavi0
297      STRING_Inscriptional_Parthian0
298      STRING_Javanese0
299      STRING_Kaithi0
300      STRING_Kannada0
301      STRING_Katakana0
302      STRING_Kayah_Li0
303      STRING_Kharoshthi0
304      STRING_Khmer0
305      STRING_L0
306      STRING_L_AMPERSAND0
307      STRING_Lao0
308      STRING_Latin0
309      STRING_Lepcha0
310      STRING_Limbu0
311      STRING_Linear_B0
312      STRING_Lisu0
313      STRING_Ll0
314      STRING_Lm0
315      STRING_Lo0
316      STRING_Lt0
317      STRING_Lu0
318      STRING_Lycian0
319      STRING_Lydian0
320      STRING_M0
321      STRING_Malayalam0
322      STRING_Mc0
323      STRING_Me0
324      STRING_Meetei_Mayek0
325      STRING_Mn0
326      STRING_Mongolian0
327      STRING_Myanmar0
328      STRING_N0
329      STRING_Nd0
330      STRING_New_Tai_Lue0
331      STRING_Nko0
332      STRING_Nl0
333      STRING_No0
334      STRING_Ogham0
335      STRING_Ol_Chiki0
336      STRING_Old_Italic0
337      STRING_Old_Persian0
338      STRING_Old_South_Arabian0
339      STRING_Old_Turkic0
340      STRING_Oriya0
341      STRING_Osmanya0
342      STRING_P0
343      STRING_Pc0
344      STRING_Pd0
345      STRING_Pe0
346      STRING_Pf0
347      STRING_Phags_Pa0
348      STRING_Phoenician0
349      STRING_Pi0
350      STRING_Po0
351      STRING_Ps0
352      STRING_Rejang0
353      STRING_Runic0
354      STRING_S0
355      STRING_Samaritan0
356      STRING_Saurashtra0
357      STRING_Sc0
358      STRING_Shavian0
359      STRING_Sinhala0
360      STRING_Sk0
361      STRING_Sm0
362      STRING_So0
363      STRING_Sundanese0
364      STRING_Syloti_Nagri0
365      STRING_Syriac0
366      STRING_Tagalog0
367      STRING_Tagbanwa0
368      STRING_Tai_Le0
369      STRING_Tai_Tham0
370      STRING_Tai_Viet0
371      STRING_Tamil0
372      STRING_Telugu0
373      STRING_Thaana0
374      STRING_Thai0
375      STRING_Tibetan0
376      STRING_Tifinagh0
377      STRING_Ugaritic0
378      STRING_Vai0
379      STRING_Yi0
380      STRING_Z0
381      STRING_Zl0
382      STRING_Zp0
383      STRING_Zs0;
384    
385  const ucp_type_table _pcre_utt[] = {  const ucp_type_table _pcre_utt[] = {
386    { "C",  128 + ucp_C },    {   0, PT_ANY, 0 },
387    { "Cc", ucp_Cc },    {   4, PT_SC, ucp_Arabic },
388    { "Cf", ucp_Cf },    {  11, PT_SC, ucp_Armenian },
389    { "Cn", ucp_Cn },    {  20, PT_SC, ucp_Avestan },
390    { "Co", ucp_Co },    {  28, PT_SC, ucp_Balinese },
391    { "Cs", ucp_Cs },    {  37, PT_SC, ucp_Bamum },
392    { "L",  128 + ucp_L },    {  43, PT_SC, ucp_Bengali },
393    { "Ll", ucp_Ll },    {  51, PT_SC, ucp_Bopomofo },
394    { "Lm", ucp_Lm },    {  60, PT_SC, ucp_Braille },
395    { "Lo", ucp_Lo },    {  68, PT_SC, ucp_Buginese },
396    { "Lt", ucp_Lt },    {  77, PT_SC, ucp_Buhid },
397    { "Lu", ucp_Lu },    {  83, PT_GC, ucp_C },
398    { "M",  128 + ucp_M },    {  85, PT_SC, ucp_Canadian_Aboriginal },
399    { "Mc", ucp_Mc },    { 105, PT_SC, ucp_Carian },
400    { "Me", ucp_Me },    { 112, PT_PC, ucp_Cc },
401    { "Mn", ucp_Mn },    { 115, PT_PC, ucp_Cf },
402    { "N",  128 + ucp_N },    { 118, PT_SC, ucp_Cham },
403    { "Nd", ucp_Nd },    { 123, PT_SC, ucp_Cherokee },
404    { "Nl", ucp_Nl },    { 132, PT_PC, ucp_Cn },
405    { "No", ucp_No },    { 135, PT_PC, ucp_Co },
406    { "P",  128 + ucp_P },    { 138, PT_SC, ucp_Common },
407    { "Pc", ucp_Pc },    { 145, PT_SC, ucp_Coptic },
408    { "Pd", ucp_Pd },    { 152, PT_PC, ucp_Cs },
409    { "Pe", ucp_Pe },    { 155, PT_SC, ucp_Cuneiform },
410    { "Pf", ucp_Pf },    { 165, PT_SC, ucp_Cypriot },
411    { "Pi", ucp_Pi },    { 173, PT_SC, ucp_Cyrillic },
412    { "Po", ucp_Po },    { 182, PT_SC, ucp_Deseret },
413    { "Ps", ucp_Ps },    { 190, PT_SC, ucp_Devanagari },
414    { "S",  128 + ucp_S },    { 201, PT_SC, ucp_Egyptian_Hieroglyphs },
415    { "Sc", ucp_Sc },    { 222, PT_SC, ucp_Ethiopic },
416    { "Sk", ucp_Sk },    { 231, PT_SC, ucp_Georgian },
417    { "Sm", ucp_Sm },    { 240, PT_SC, ucp_Glagolitic },
418    { "So", ucp_So },    { 251, PT_SC, ucp_Gothic },
419    { "Z",  128 + ucp_Z },    { 258, PT_SC, ucp_Greek },
420    { "Zl", ucp_Zl },    { 264, PT_SC, ucp_Gujarati },
421    { "Zp", ucp_Zp },    { 273, PT_SC, ucp_Gurmukhi },
422    { "Zs", ucp_Zs }    { 282, PT_SC, ucp_Han },
423      { 286, PT_SC, ucp_Hangul },
424      { 293, PT_SC, ucp_Hanunoo },
425      { 301, PT_SC, ucp_Hebrew },
426      { 308, PT_SC, ucp_Hiragana },
427      { 317, PT_SC, ucp_Imperial_Aramaic },
428      { 334, PT_SC, ucp_Inherited },
429      { 344, PT_SC, ucp_Inscriptional_Pahlavi },
430      { 366, PT_SC, ucp_Inscriptional_Parthian },
431      { 389, PT_SC, ucp_Javanese },
432      { 398, PT_SC, ucp_Kaithi },
433      { 405, PT_SC, ucp_Kannada },
434      { 413, PT_SC, ucp_Katakana },
435      { 422, PT_SC, ucp_Kayah_Li },
436      { 431, PT_SC, ucp_Kharoshthi },
437      { 442, PT_SC, ucp_Khmer },
438      { 448, PT_GC, ucp_L },
439      { 450, PT_LAMP, 0 },
440      { 453, PT_SC, ucp_Lao },
441      { 457, PT_SC, ucp_Latin },
442      { 463, PT_SC, ucp_Lepcha },
443      { 470, PT_SC, ucp_Limbu },
444      { 476, PT_SC, ucp_Linear_B },
445      { 485, PT_SC, ucp_Lisu },
446      { 490, PT_PC, ucp_Ll },
447      { 493, PT_PC, ucp_Lm },
448      { 496, PT_PC, ucp_Lo },
449      { 499, PT_PC, ucp_Lt },
450      { 502, PT_PC, ucp_Lu },
451      { 505, PT_SC, ucp_Lycian },
452      { 512, PT_SC, ucp_Lydian },
453      { 519, PT_GC, ucp_M },
454      { 521, PT_SC, ucp_Malayalam },
455      { 531, PT_PC, ucp_Mc },
456      { 534, PT_PC, ucp_Me },
457      { 537, PT_SC, ucp_Meetei_Mayek },
458      { 550, PT_PC, ucp_Mn },
459      { 553, PT_SC, ucp_Mongolian },
460      { 563, PT_SC, ucp_Myanmar },
461      { 571, PT_GC, ucp_N },
462      { 573, PT_PC, ucp_Nd },
463      { 576, PT_SC, ucp_New_Tai_Lue },
464      { 588, PT_SC, ucp_Nko },
465      { 592, PT_PC, ucp_Nl },
466      { 595, PT_PC, ucp_No },
467      { 598, PT_SC, ucp_Ogham },
468      { 604, PT_SC, ucp_Ol_Chiki },
469      { 613, PT_SC, ucp_Old_Italic },
470      { 624, PT_SC, ucp_Old_Persian },
471      { 636, PT_SC, ucp_Old_South_Arabian },
472      { 654, PT_SC, ucp_Old_Turkic },
473      { 665, PT_SC, ucp_Oriya },
474      { 671, PT_SC, ucp_Osmanya },
475      { 679, PT_GC, ucp_P },
476      { 681, PT_PC, ucp_Pc },
477      { 684, PT_PC, ucp_Pd },
478      { 687, PT_PC, ucp_Pe },
479      { 690, PT_PC, ucp_Pf },
480      { 693, PT_SC, ucp_Phags_Pa },
481      { 702, PT_SC, ucp_Phoenician },
482      { 713, PT_PC, ucp_Pi },
483      { 716, PT_PC, ucp_Po },
484      { 719, PT_PC, ucp_Ps },
485      { 722, PT_SC, ucp_Rejang },
486      { 729, PT_SC, ucp_Runic },
487      { 735, PT_GC, ucp_S },
488      { 737, PT_SC, ucp_Samaritan },
489      { 747, PT_SC, ucp_Saurashtra },
490      { 758, PT_PC, ucp_Sc },
491      { 761, PT_SC, ucp_Shavian },
492      { 769, PT_SC, ucp_Sinhala },
493      { 777, PT_PC, ucp_Sk },
494      { 780, PT_PC, ucp_Sm },
495      { 783, PT_PC, ucp_So },
496      { 786, PT_SC, ucp_Sundanese },
497      { 796, PT_SC, ucp_Syloti_Nagri },
498      { 809, PT_SC, ucp_Syriac },
499      { 816, PT_SC, ucp_Tagalog },
500      { 824, PT_SC, ucp_Tagbanwa },
501      { 833, PT_SC, ucp_Tai_Le },
502      { 840, PT_SC, ucp_Tai_Tham },
503      { 849, PT_SC, ucp_Tai_Viet },
504      { 858, PT_SC, ucp_Tamil },
505      { 864, PT_SC, ucp_Telugu },
506      { 871, PT_SC, ucp_Thaana },
507      { 878, PT_SC, ucp_Thai },
508      { 883, PT_SC, ucp_Tibetan },
509      { 891, PT_SC, ucp_Tifinagh },
510      { 900, PT_SC, ucp_Ugaritic },
511      { 909, PT_SC, ucp_Vai },
512      { 913, PT_SC, ucp_Yi },
513      { 916, PT_GC, ucp_Z },
514      { 918, PT_PC, ucp_Zl },
515      { 921, PT_PC, ucp_Zp },
516      { 924, PT_PC, ucp_Zs }
517  };  };
518    
519  const int _pcre_utt_size = sizeof(_pcre_utt)/sizeof(ucp_type_table);  const int _pcre_utt_size = sizeof(_pcre_utt)/sizeof(ucp_type_table);
520    
521    #endif  /* SUPPORT_UTF8 */
522    
523  /* End of pcre_tables.c */  /* End of pcre_tables.c */
Legend:
Removed from v.77  
changed lines
  Added in v.507
webmaster@exim.org
 ViewVC Help
Powered by ViewVC 1.1.12