/[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 87 by nigel, Sat Feb 24 21:41:21 2007 UTC revision 836 by ph10, Wed Dec 28 17:16:11 2011 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) 1997-2006 University of Cambridge             Copyright (c) 1997-2012 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 37  POSSIBILITY OF SUCH DAMAGE. Line 37  POSSIBILITY OF SUCH DAMAGE.
37  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
38  */  */
39    
40    #ifndef PCRE_INCLUDED
41    
42  /* 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
43  PCRE code modules. The tables are also #included by the pcretest program, which  PCRE code modules. The tables are also #included by the pcretest program, which
# Line 44  uses macros to change their names from _ Line 45  uses macros to change their names from _
45  clashes with the library. */  clashes with the library. */
46    
47    
48    #ifdef HAVE_CONFIG_H
49    #include "config.h"
50    #endif
51    
52  #include "pcre_internal.h"  #include "pcre_internal.h"
53    
54    #endif /* PCRE_INCLUDED */
55    
56  /* 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
57  the definition is next to the definition of the opcodes in pcre_internal.h. */  the definition is next to the definition of the opcodes in pcre_internal.h. */
58    
59  const uschar _pcre_OP_lengths[] = { OP_LENGTHS };  const pcre_uint8 PRIV(OP_lengths)[] = { OP_LENGTHS };
60    
61    
62    
# Line 61  const uschar _pcre_OP_lengths[] = { OP_L Line 67  const uschar _pcre_OP_lengths[] = { OP_L
67  /* 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
68  character. */  character. */
69    
70  const int _pcre_utf8_table1[] =  #if (defined SUPPORT_UTF && defined COMPILE_PCRE8) \
71      || (defined PCRE_INCLUDED && defined SUPPORT_PCRE16)
72    
73    /* These tables are also required by pcretest in 16 bit mode. */
74    
75    const int PRIV(utf8_table1)[] =
76    { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff};    { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff};
77    
78  const int _pcre_utf8_table1_size = sizeof(_pcre_utf8_table1)/sizeof(int);  const int PRIV(utf8_table1_size) = sizeof(PRIV(utf8_table1)) / sizeof(int);
79    
80  /* These are the indicator bits and the mask for the data bits to set in the  /* These are the indicator bits and the mask for the data bits to set in the
81  first byte of a character, indexed by the number of additional bytes. */  first byte of a character, indexed by the number of additional bytes. */
82    
83  const int _pcre_utf8_table2[] = { 0,    0xc0, 0xe0, 0xf0, 0xf8, 0xfc};  const int PRIV(utf8_table2)[] = { 0,    0xc0, 0xe0, 0xf0, 0xf8, 0xfc};
84  const int _pcre_utf8_table3[] = { 0xff, 0x1f, 0x0f, 0x07, 0x03, 0x01};  const int PRIV(utf8_table3)[] = { 0xff, 0x1f, 0x0f, 0x07, 0x03, 0x01};
85    
86  /* 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
87  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. */  
88    
89  const uschar _pcre_utf8_table4[] = {  const pcre_uint8 PRIV(utf8_table4)[] = {
90    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,
91    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,
92    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,
93    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 };
94    
95  /* This table translates Unicode property names into type and code values. It  #endif /* (SUPPORT_UTF && COMPILE_PCRE8) || (PCRE_INCLUDED && SUPPORT_PCRE16)*/
96  is searched by binary chop, so must be in collating sequence of name. */  
97    #ifdef SUPPORT_UTF
98    
99    /* Table to translate from particular type value to the general value. */
100    
101  const ucp_type_table _pcre_utt[] = {  const int PRIV(ucp_gentype)[] = {
102    { "Any",                 PT_ANY,  0 },    ucp_C, ucp_C, ucp_C, ucp_C, ucp_C,  /* Cc, Cf, Cn, Co, Cs */
103    { "Arabic",              PT_SC,   ucp_Arabic },    ucp_L, ucp_L, ucp_L, ucp_L, ucp_L,  /* Ll, Lu, Lm, Lo, Lt */
104    { "Armenian",            PT_SC,   ucp_Armenian },    ucp_M, ucp_M, ucp_M,                /* Mc, Me, Mn */
105    { "Bengali",             PT_SC,   ucp_Bengali },    ucp_N, ucp_N, ucp_N,                /* Nd, Nl, No */
106    { "Bopomofo",            PT_SC,   ucp_Bopomofo },    ucp_P, ucp_P, ucp_P, ucp_P, ucp_P,  /* Pc, Pd, Pe, Pf, Pi */
107    { "Braille",             PT_SC,   ucp_Braille },    ucp_P, ucp_P,                       /* Ps, Po */
108    { "Buginese",            PT_SC,   ucp_Buginese },    ucp_S, ucp_S, ucp_S, ucp_S,         /* Sc, Sk, Sm, So */
109    { "Buhid",               PT_SC,   ucp_Buhid },    ucp_Z, ucp_Z, ucp_Z                 /* Zl, Zp, Zs */
   { "C",                   PT_GC,   ucp_C },  
   { "Canadian_Aboriginal", PT_SC,   ucp_Canadian_Aboriginal },  
   { "Cc",                  PT_PC,   ucp_Cc },  
   { "Cf",                  PT_PC,   ucp_Cf },  
   { "Cherokee",            PT_SC,   ucp_Cherokee },  
   { "Cn",                  PT_PC,   ucp_Cn },  
   { "Co",                  PT_PC,   ucp_Co },  
   { "Common",              PT_SC,   ucp_Common },  
   { "Coptic",              PT_SC,   ucp_Coptic },  
   { "Cs",                  PT_PC,   ucp_Cs },  
   { "Cypriot",             PT_SC,   ucp_Cypriot },  
   { "Cyrillic",            PT_SC,   ucp_Cyrillic },  
   { "Deseret",             PT_SC,   ucp_Deseret },  
   { "Devanagari",          PT_SC,   ucp_Devanagari },  
   { "Ethiopic",            PT_SC,   ucp_Ethiopic },  
   { "Georgian",            PT_SC,   ucp_Georgian },  
   { "Glagolitic",          PT_SC,   ucp_Glagolitic },  
   { "Gothic",              PT_SC,   ucp_Gothic },  
   { "Greek",               PT_SC,   ucp_Greek },  
   { "Gujarati",            PT_SC,   ucp_Gujarati },  
   { "Gurmukhi",            PT_SC,   ucp_Gurmukhi },  
   { "Han",                 PT_SC,   ucp_Han },  
   { "Hangul",              PT_SC,   ucp_Hangul },  
   { "Hanunoo",             PT_SC,   ucp_Hanunoo },  
   { "Hebrew",              PT_SC,   ucp_Hebrew },  
   { "Hiragana",            PT_SC,   ucp_Hiragana },  
   { "Inherited",           PT_SC,   ucp_Inherited },  
   { "Kannada",             PT_SC,   ucp_Kannada },  
   { "Katakana",            PT_SC,   ucp_Katakana },  
   { "Kharoshthi",          PT_SC,   ucp_Kharoshthi },  
   { "Khmer",               PT_SC,   ucp_Khmer },  
   { "L",                   PT_GC,   ucp_L },  
   { "L&",                  PT_LAMP, 0 },  
   { "Lao",                 PT_SC,   ucp_Lao },  
   { "Latin",               PT_SC,   ucp_Latin },  
   { "Limbu",               PT_SC,   ucp_Limbu },  
   { "Linear_B",            PT_SC,   ucp_Linear_B },  
   { "Ll",                  PT_PC,   ucp_Ll },  
   { "Lm",                  PT_PC,   ucp_Lm },  
   { "Lo",                  PT_PC,   ucp_Lo },  
   { "Lt",                  PT_PC,   ucp_Lt },  
   { "Lu",                  PT_PC,   ucp_Lu },  
   { "M",                   PT_GC,   ucp_M },  
   { "Malayalam",           PT_SC,   ucp_Malayalam },  
   { "Mc",                  PT_PC,   ucp_Mc },  
   { "Me",                  PT_PC,   ucp_Me },  
   { "Mn",                  PT_PC,   ucp_Mn },  
   { "Mongolian",           PT_SC,   ucp_Mongolian },  
   { "Myanmar",             PT_SC,   ucp_Myanmar },  
   { "N",                   PT_GC,   ucp_N },  
   { "Nd",                  PT_PC,   ucp_Nd },  
   { "New_Tai_Lue",         PT_SC,   ucp_New_Tai_Lue },  
   { "Nl",                  PT_PC,   ucp_Nl },  
   { "No",                  PT_PC,   ucp_No },  
   { "Ogham",               PT_SC,   ucp_Ogham },  
   { "Old_Italic",          PT_SC,   ucp_Old_Italic },  
   { "Old_Persian",         PT_SC,   ucp_Old_Persian },  
   { "Oriya",               PT_SC,   ucp_Oriya },  
   { "Osmanya",             PT_SC,   ucp_Osmanya },  
   { "P",                   PT_GC,   ucp_P },  
   { "Pc",                  PT_PC,   ucp_Pc },  
   { "Pd",                  PT_PC,   ucp_Pd },  
   { "Pe",                  PT_PC,   ucp_Pe },  
   { "Pf",                  PT_PC,   ucp_Pf },  
   { "Pi",                  PT_PC,   ucp_Pi },  
   { "Po",                  PT_PC,   ucp_Po },  
   { "Ps",                  PT_PC,   ucp_Ps },  
   { "Runic",               PT_SC,   ucp_Runic },  
   { "S",                   PT_GC,   ucp_S },  
   { "Sc",                  PT_PC,   ucp_Sc },  
   { "Shavian",             PT_SC,   ucp_Shavian },  
   { "Sinhala",             PT_SC,   ucp_Sinhala },  
   { "Sk",                  PT_PC,   ucp_Sk },  
   { "Sm",                  PT_PC,   ucp_Sm },  
   { "So",                  PT_PC,   ucp_So },  
   { "Syloti_Nagri",        PT_SC,   ucp_Syloti_Nagri },  
   { "Syriac",              PT_SC,   ucp_Syriac },  
   { "Tagalog",             PT_SC,   ucp_Tagalog },  
   { "Tagbanwa",            PT_SC,   ucp_Tagbanwa },  
   { "Tai_Le",              PT_SC,   ucp_Tai_Le },  
   { "Tamil",               PT_SC,   ucp_Tamil },  
   { "Telugu",              PT_SC,   ucp_Telugu },  
   { "Thaana",              PT_SC,   ucp_Thaana },  
   { "Thai",                PT_SC,   ucp_Thai },  
   { "Tibetan",             PT_SC,   ucp_Tibetan },  
   { "Tifinagh",            PT_SC,   ucp_Tifinagh },  
   { "Ugaritic",            PT_SC,   ucp_Ugaritic },  
   { "Yi",                  PT_SC,   ucp_Yi },  
   { "Z",                   PT_GC,   ucp_Z },  
   { "Zl",                  PT_PC,   ucp_Zl },  
   { "Zp",                  PT_PC,   ucp_Zp },  
   { "Zs",                  PT_PC,   ucp_Zs }  
110  };  };
111    
112  const int _pcre_utt_size = sizeof(_pcre_utt)/sizeof(ucp_type_table);  #ifdef SUPPORT_JIT
113    /* This table reverses PRIV(ucp_gentype). We can save the cost
114    of a memory load. */
115    
116    const int PRIV(ucp_typerange)[] = {
117      ucp_Cc, ucp_Cs,
118      ucp_Ll, ucp_Lu,
119      ucp_Mc, ucp_Mn,
120      ucp_Nd, ucp_No,
121      ucp_Pc, ucp_Ps,
122      ucp_Sc, ucp_So,
123      ucp_Zl, ucp_Zs,
124    };
125    #endif /* SUPPORT_JIT */
126    
127    /* The pcre_utt[] table below translates Unicode property names into type and
128    code values. It is searched by binary chop, so must be in collating sequence of
129    name. Originally, the table contained pointers to the name strings in the first
130    field of each entry. However, that leads to a large number of relocations when
131    a shared library is dynamically loaded. A significant reduction is made by
132    putting all the names into a single, large string and then using offsets in the
133    table itself. Maintenance is more error-prone, but frequent changes to this
134    data are unlikely.
135    
136    July 2008: There is now a script called maint/GenerateUtt.py that can be used
137    to generate this data automatically instead of maintaining it by hand.
138    
139    The script was updated in March 2009 to generate a new EBCDIC-compliant
140    version. Like all other character and string literals that are compared against
141    the regular expression pattern, we must use STR_ macros instead of literal
142    strings to make sure that UTF-8 support works on EBCDIC platforms. */
143    
144    #define STRING_Any0 STR_A STR_n STR_y "\0"
145    #define STRING_Arabic0 STR_A STR_r STR_a STR_b STR_i STR_c "\0"
146    #define STRING_Armenian0 STR_A STR_r STR_m STR_e STR_n STR_i STR_a STR_n "\0"
147    #define STRING_Avestan0 STR_A STR_v STR_e STR_s STR_t STR_a STR_n "\0"
148    #define STRING_Balinese0 STR_B STR_a STR_l STR_i STR_n STR_e STR_s STR_e "\0"
149    #define STRING_Bamum0 STR_B STR_a STR_m STR_u STR_m "\0"
150    #define STRING_Batak0 STR_B STR_a STR_t STR_a STR_k "\0"
151    #define STRING_Bengali0 STR_B STR_e STR_n STR_g STR_a STR_l STR_i "\0"
152    #define STRING_Bopomofo0 STR_B STR_o STR_p STR_o STR_m STR_o STR_f STR_o "\0"
153    #define STRING_Brahmi0 STR_B STR_r STR_a STR_h STR_m STR_i "\0"
154    #define STRING_Braille0 STR_B STR_r STR_a STR_i STR_l STR_l STR_e "\0"
155    #define STRING_Buginese0 STR_B STR_u STR_g STR_i STR_n STR_e STR_s STR_e "\0"
156    #define STRING_Buhid0 STR_B STR_u STR_h STR_i STR_d "\0"
157    #define STRING_C0 STR_C "\0"
158    #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"
159    #define STRING_Carian0 STR_C STR_a STR_r STR_i STR_a STR_n "\0"
160    #define STRING_Cc0 STR_C STR_c "\0"
161    #define STRING_Cf0 STR_C STR_f "\0"
162    #define STRING_Cham0 STR_C STR_h STR_a STR_m "\0"
163    #define STRING_Cherokee0 STR_C STR_h STR_e STR_r STR_o STR_k STR_e STR_e "\0"
164    #define STRING_Cn0 STR_C STR_n "\0"
165    #define STRING_Co0 STR_C STR_o "\0"
166    #define STRING_Common0 STR_C STR_o STR_m STR_m STR_o STR_n "\0"
167    #define STRING_Coptic0 STR_C STR_o STR_p STR_t STR_i STR_c "\0"
168    #define STRING_Cs0 STR_C STR_s "\0"
169    #define STRING_Cuneiform0 STR_C STR_u STR_n STR_e STR_i STR_f STR_o STR_r STR_m "\0"
170    #define STRING_Cypriot0 STR_C STR_y STR_p STR_r STR_i STR_o STR_t "\0"
171    #define STRING_Cyrillic0 STR_C STR_y STR_r STR_i STR_l STR_l STR_i STR_c "\0"
172    #define STRING_Deseret0 STR_D STR_e STR_s STR_e STR_r STR_e STR_t "\0"
173    #define STRING_Devanagari0 STR_D STR_e STR_v STR_a STR_n STR_a STR_g STR_a STR_r STR_i "\0"
174    #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"
175    #define STRING_Ethiopic0 STR_E STR_t STR_h STR_i STR_o STR_p STR_i STR_c "\0"
176    #define STRING_Georgian0 STR_G STR_e STR_o STR_r STR_g STR_i STR_a STR_n "\0"
177    #define STRING_Glagolitic0 STR_G STR_l STR_a STR_g STR_o STR_l STR_i STR_t STR_i STR_c "\0"
178    #define STRING_Gothic0 STR_G STR_o STR_t STR_h STR_i STR_c "\0"
179    #define STRING_Greek0 STR_G STR_r STR_e STR_e STR_k "\0"
180    #define STRING_Gujarati0 STR_G STR_u STR_j STR_a STR_r STR_a STR_t STR_i "\0"
181    #define STRING_Gurmukhi0 STR_G STR_u STR_r STR_m STR_u STR_k STR_h STR_i "\0"
182    #define STRING_Han0 STR_H STR_a STR_n "\0"
183    #define STRING_Hangul0 STR_H STR_a STR_n STR_g STR_u STR_l "\0"
184    #define STRING_Hanunoo0 STR_H STR_a STR_n STR_u STR_n STR_o STR_o "\0"
185    #define STRING_Hebrew0 STR_H STR_e STR_b STR_r STR_e STR_w "\0"
186    #define STRING_Hiragana0 STR_H STR_i STR_r STR_a STR_g STR_a STR_n STR_a "\0"
187    #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"
188    #define STRING_Inherited0 STR_I STR_n STR_h STR_e STR_r STR_i STR_t STR_e STR_d "\0"
189    #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"
190    #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"
191    #define STRING_Javanese0 STR_J STR_a STR_v STR_a STR_n STR_e STR_s STR_e "\0"
192    #define STRING_Kaithi0 STR_K STR_a STR_i STR_t STR_h STR_i "\0"
193    #define STRING_Kannada0 STR_K STR_a STR_n STR_n STR_a STR_d STR_a "\0"
194    #define STRING_Katakana0 STR_K STR_a STR_t STR_a STR_k STR_a STR_n STR_a "\0"
195    #define STRING_Kayah_Li0 STR_K STR_a STR_y STR_a STR_h STR_UNDERSCORE STR_L STR_i "\0"
196    #define STRING_Kharoshthi0 STR_K STR_h STR_a STR_r STR_o STR_s STR_h STR_t STR_h STR_i "\0"
197    #define STRING_Khmer0 STR_K STR_h STR_m STR_e STR_r "\0"
198    #define STRING_L0 STR_L "\0"
199    #define STRING_L_AMPERSAND0 STR_L STR_AMPERSAND "\0"
200    #define STRING_Lao0 STR_L STR_a STR_o "\0"
201    #define STRING_Latin0 STR_L STR_a STR_t STR_i STR_n "\0"
202    #define STRING_Lepcha0 STR_L STR_e STR_p STR_c STR_h STR_a "\0"
203    #define STRING_Limbu0 STR_L STR_i STR_m STR_b STR_u "\0"
204    #define STRING_Linear_B0 STR_L STR_i STR_n STR_e STR_a STR_r STR_UNDERSCORE STR_B "\0"
205    #define STRING_Lisu0 STR_L STR_i STR_s STR_u "\0"
206    #define STRING_Ll0 STR_L STR_l "\0"
207    #define STRING_Lm0 STR_L STR_m "\0"
208    #define STRING_Lo0 STR_L STR_o "\0"
209    #define STRING_Lt0 STR_L STR_t "\0"
210    #define STRING_Lu0 STR_L STR_u "\0"
211    #define STRING_Lycian0 STR_L STR_y STR_c STR_i STR_a STR_n "\0"
212    #define STRING_Lydian0 STR_L STR_y STR_d STR_i STR_a STR_n "\0"
213    #define STRING_M0 STR_M "\0"
214    #define STRING_Malayalam0 STR_M STR_a STR_l STR_a STR_y STR_a STR_l STR_a STR_m "\0"
215    #define STRING_Mandaic0 STR_M STR_a STR_n STR_d STR_a STR_i STR_c "\0"
216    #define STRING_Mc0 STR_M STR_c "\0"
217    #define STRING_Me0 STR_M STR_e "\0"
218    #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"
219    #define STRING_Mn0 STR_M STR_n "\0"
220    #define STRING_Mongolian0 STR_M STR_o STR_n STR_g STR_o STR_l STR_i STR_a STR_n "\0"
221    #define STRING_Myanmar0 STR_M STR_y STR_a STR_n STR_m STR_a STR_r "\0"
222    #define STRING_N0 STR_N "\0"
223    #define STRING_Nd0 STR_N STR_d "\0"
224    #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"
225    #define STRING_Nko0 STR_N STR_k STR_o "\0"
226    #define STRING_Nl0 STR_N STR_l "\0"
227    #define STRING_No0 STR_N STR_o "\0"
228    #define STRING_Ogham0 STR_O STR_g STR_h STR_a STR_m "\0"
229    #define STRING_Ol_Chiki0 STR_O STR_l STR_UNDERSCORE STR_C STR_h STR_i STR_k STR_i "\0"
230    #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"
231    #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"
232    #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"
233    #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"
234    #define STRING_Oriya0 STR_O STR_r STR_i STR_y STR_a "\0"
235    #define STRING_Osmanya0 STR_O STR_s STR_m STR_a STR_n STR_y STR_a "\0"
236    #define STRING_P0 STR_P "\0"
237    #define STRING_Pc0 STR_P STR_c "\0"
238    #define STRING_Pd0 STR_P STR_d "\0"
239    #define STRING_Pe0 STR_P STR_e "\0"
240    #define STRING_Pf0 STR_P STR_f "\0"
241    #define STRING_Phags_Pa0 STR_P STR_h STR_a STR_g STR_s STR_UNDERSCORE STR_P STR_a "\0"
242    #define STRING_Phoenician0 STR_P STR_h STR_o STR_e STR_n STR_i STR_c STR_i STR_a STR_n "\0"
243    #define STRING_Pi0 STR_P STR_i "\0"
244    #define STRING_Po0 STR_P STR_o "\0"
245    #define STRING_Ps0 STR_P STR_s "\0"
246    #define STRING_Rejang0 STR_R STR_e STR_j STR_a STR_n STR_g "\0"
247    #define STRING_Runic0 STR_R STR_u STR_n STR_i STR_c "\0"
248    #define STRING_S0 STR_S "\0"
249    #define STRING_Samaritan0 STR_S STR_a STR_m STR_a STR_r STR_i STR_t STR_a STR_n "\0"
250    #define STRING_Saurashtra0 STR_S STR_a STR_u STR_r STR_a STR_s STR_h STR_t STR_r STR_a "\0"
251    #define STRING_Sc0 STR_S STR_c "\0"
252    #define STRING_Shavian0 STR_S STR_h STR_a STR_v STR_i STR_a STR_n "\0"
253    #define STRING_Sinhala0 STR_S STR_i STR_n STR_h STR_a STR_l STR_a "\0"
254    #define STRING_Sk0 STR_S STR_k "\0"
255    #define STRING_Sm0 STR_S STR_m "\0"
256    #define STRING_So0 STR_S STR_o "\0"
257    #define STRING_Sundanese0 STR_S STR_u STR_n STR_d STR_a STR_n STR_e STR_s STR_e "\0"
258    #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"
259    #define STRING_Syriac0 STR_S STR_y STR_r STR_i STR_a STR_c "\0"
260    #define STRING_Tagalog0 STR_T STR_a STR_g STR_a STR_l STR_o STR_g "\0"
261    #define STRING_Tagbanwa0 STR_T STR_a STR_g STR_b STR_a STR_n STR_w STR_a "\0"
262    #define STRING_Tai_Le0 STR_T STR_a STR_i STR_UNDERSCORE STR_L STR_e "\0"
263    #define STRING_Tai_Tham0 STR_T STR_a STR_i STR_UNDERSCORE STR_T STR_h STR_a STR_m "\0"
264    #define STRING_Tai_Viet0 STR_T STR_a STR_i STR_UNDERSCORE STR_V STR_i STR_e STR_t "\0"
265    #define STRING_Tamil0 STR_T STR_a STR_m STR_i STR_l "\0"
266    #define STRING_Telugu0 STR_T STR_e STR_l STR_u STR_g STR_u "\0"
267    #define STRING_Thaana0 STR_T STR_h STR_a STR_a STR_n STR_a "\0"
268    #define STRING_Thai0 STR_T STR_h STR_a STR_i "\0"
269    #define STRING_Tibetan0 STR_T STR_i STR_b STR_e STR_t STR_a STR_n "\0"
270    #define STRING_Tifinagh0 STR_T STR_i STR_f STR_i STR_n STR_a STR_g STR_h "\0"
271    #define STRING_Ugaritic0 STR_U STR_g STR_a STR_r STR_i STR_t STR_i STR_c "\0"
272    #define STRING_Vai0 STR_V STR_a STR_i "\0"
273    #define STRING_Xan0 STR_X STR_a STR_n "\0"
274    #define STRING_Xps0 STR_X STR_p STR_s "\0"
275    #define STRING_Xsp0 STR_X STR_s STR_p "\0"
276    #define STRING_Xwd0 STR_X STR_w STR_d "\0"
277    #define STRING_Yi0 STR_Y STR_i "\0"
278    #define STRING_Z0 STR_Z "\0"
279    #define STRING_Zl0 STR_Z STR_l "\0"
280    #define STRING_Zp0 STR_Z STR_p "\0"
281    #define STRING_Zs0 STR_Z STR_s "\0"
282    
283    const char PRIV(utt_names)[] =
284      STRING_Any0
285      STRING_Arabic0
286      STRING_Armenian0
287      STRING_Avestan0
288      STRING_Balinese0
289      STRING_Bamum0
290      STRING_Batak0
291      STRING_Bengali0
292      STRING_Bopomofo0
293      STRING_Brahmi0
294      STRING_Braille0
295      STRING_Buginese0
296      STRING_Buhid0
297      STRING_C0
298      STRING_Canadian_Aboriginal0
299      STRING_Carian0
300      STRING_Cc0
301      STRING_Cf0
302      STRING_Cham0
303      STRING_Cherokee0
304      STRING_Cn0
305      STRING_Co0
306      STRING_Common0
307      STRING_Coptic0
308      STRING_Cs0
309      STRING_Cuneiform0
310      STRING_Cypriot0
311      STRING_Cyrillic0
312      STRING_Deseret0
313      STRING_Devanagari0
314      STRING_Egyptian_Hieroglyphs0
315      STRING_Ethiopic0
316      STRING_Georgian0
317      STRING_Glagolitic0
318      STRING_Gothic0
319      STRING_Greek0
320      STRING_Gujarati0
321      STRING_Gurmukhi0
322      STRING_Han0
323      STRING_Hangul0
324      STRING_Hanunoo0
325      STRING_Hebrew0
326      STRING_Hiragana0
327      STRING_Imperial_Aramaic0
328      STRING_Inherited0
329      STRING_Inscriptional_Pahlavi0
330      STRING_Inscriptional_Parthian0
331      STRING_Javanese0
332      STRING_Kaithi0
333      STRING_Kannada0
334      STRING_Katakana0
335      STRING_Kayah_Li0
336      STRING_Kharoshthi0
337      STRING_Khmer0
338      STRING_L0
339      STRING_L_AMPERSAND0
340      STRING_Lao0
341      STRING_Latin0
342      STRING_Lepcha0
343      STRING_Limbu0
344      STRING_Linear_B0
345      STRING_Lisu0
346      STRING_Ll0
347      STRING_Lm0
348      STRING_Lo0
349      STRING_Lt0
350      STRING_Lu0
351      STRING_Lycian0
352      STRING_Lydian0
353      STRING_M0
354      STRING_Malayalam0
355      STRING_Mandaic0
356      STRING_Mc0
357      STRING_Me0
358      STRING_Meetei_Mayek0
359      STRING_Mn0
360      STRING_Mongolian0
361      STRING_Myanmar0
362      STRING_N0
363      STRING_Nd0
364      STRING_New_Tai_Lue0
365      STRING_Nko0
366      STRING_Nl0
367      STRING_No0
368      STRING_Ogham0
369      STRING_Ol_Chiki0
370      STRING_Old_Italic0
371      STRING_Old_Persian0
372      STRING_Old_South_Arabian0
373      STRING_Old_Turkic0
374      STRING_Oriya0
375      STRING_Osmanya0
376      STRING_P0
377      STRING_Pc0
378      STRING_Pd0
379      STRING_Pe0
380      STRING_Pf0
381      STRING_Phags_Pa0
382      STRING_Phoenician0
383      STRING_Pi0
384      STRING_Po0
385      STRING_Ps0
386      STRING_Rejang0
387      STRING_Runic0
388      STRING_S0
389      STRING_Samaritan0
390      STRING_Saurashtra0
391      STRING_Sc0
392      STRING_Shavian0
393      STRING_Sinhala0
394      STRING_Sk0
395      STRING_Sm0
396      STRING_So0
397      STRING_Sundanese0
398      STRING_Syloti_Nagri0
399      STRING_Syriac0
400      STRING_Tagalog0
401      STRING_Tagbanwa0
402      STRING_Tai_Le0
403      STRING_Tai_Tham0
404      STRING_Tai_Viet0
405      STRING_Tamil0
406      STRING_Telugu0
407      STRING_Thaana0
408      STRING_Thai0
409      STRING_Tibetan0
410      STRING_Tifinagh0
411      STRING_Ugaritic0
412      STRING_Vai0
413      STRING_Xan0
414      STRING_Xps0
415      STRING_Xsp0
416      STRING_Xwd0
417      STRING_Yi0
418      STRING_Z0
419      STRING_Zl0
420      STRING_Zp0
421      STRING_Zs0;
422    
423    const ucp_type_table PRIV(utt)[] = {
424      {   0, PT_ANY, 0 },
425      {   4, PT_SC, ucp_Arabic },
426      {  11, PT_SC, ucp_Armenian },
427      {  20, PT_SC, ucp_Avestan },
428      {  28, PT_SC, ucp_Balinese },
429      {  37, PT_SC, ucp_Bamum },
430      {  43, PT_SC, ucp_Batak },
431      {  49, PT_SC, ucp_Bengali },
432      {  57, PT_SC, ucp_Bopomofo },
433      {  66, PT_SC, ucp_Brahmi },
434      {  73, PT_SC, ucp_Braille },
435      {  81, PT_SC, ucp_Buginese },
436      {  90, PT_SC, ucp_Buhid },
437      {  96, PT_GC, ucp_C },
438      {  98, PT_SC, ucp_Canadian_Aboriginal },
439      { 118, PT_SC, ucp_Carian },
440      { 125, PT_PC, ucp_Cc },
441      { 128, PT_PC, ucp_Cf },
442      { 131, PT_SC, ucp_Cham },
443      { 136, PT_SC, ucp_Cherokee },
444      { 145, PT_PC, ucp_Cn },
445      { 148, PT_PC, ucp_Co },
446      { 151, PT_SC, ucp_Common },
447      { 158, PT_SC, ucp_Coptic },
448      { 165, PT_PC, ucp_Cs },
449      { 168, PT_SC, ucp_Cuneiform },
450      { 178, PT_SC, ucp_Cypriot },
451      { 186, PT_SC, ucp_Cyrillic },
452      { 195, PT_SC, ucp_Deseret },
453      { 203, PT_SC, ucp_Devanagari },
454      { 214, PT_SC, ucp_Egyptian_Hieroglyphs },
455      { 235, PT_SC, ucp_Ethiopic },
456      { 244, PT_SC, ucp_Georgian },
457      { 253, PT_SC, ucp_Glagolitic },
458      { 264, PT_SC, ucp_Gothic },
459      { 271, PT_SC, ucp_Greek },
460      { 277, PT_SC, ucp_Gujarati },
461      { 286, PT_SC, ucp_Gurmukhi },
462      { 295, PT_SC, ucp_Han },
463      { 299, PT_SC, ucp_Hangul },
464      { 306, PT_SC, ucp_Hanunoo },
465      { 314, PT_SC, ucp_Hebrew },
466      { 321, PT_SC, ucp_Hiragana },
467      { 330, PT_SC, ucp_Imperial_Aramaic },
468      { 347, PT_SC, ucp_Inherited },
469      { 357, PT_SC, ucp_Inscriptional_Pahlavi },
470      { 379, PT_SC, ucp_Inscriptional_Parthian },
471      { 402, PT_SC, ucp_Javanese },
472      { 411, PT_SC, ucp_Kaithi },
473      { 418, PT_SC, ucp_Kannada },
474      { 426, PT_SC, ucp_Katakana },
475      { 435, PT_SC, ucp_Kayah_Li },
476      { 444, PT_SC, ucp_Kharoshthi },
477      { 455, PT_SC, ucp_Khmer },
478      { 461, PT_GC, ucp_L },
479      { 463, PT_LAMP, 0 },
480      { 466, PT_SC, ucp_Lao },
481      { 470, PT_SC, ucp_Latin },
482      { 476, PT_SC, ucp_Lepcha },
483      { 483, PT_SC, ucp_Limbu },
484      { 489, PT_SC, ucp_Linear_B },
485      { 498, PT_SC, ucp_Lisu },
486      { 503, PT_PC, ucp_Ll },
487      { 506, PT_PC, ucp_Lm },
488      { 509, PT_PC, ucp_Lo },
489      { 512, PT_PC, ucp_Lt },
490      { 515, PT_PC, ucp_Lu },
491      { 518, PT_SC, ucp_Lycian },
492      { 525, PT_SC, ucp_Lydian },
493      { 532, PT_GC, ucp_M },
494      { 534, PT_SC, ucp_Malayalam },
495      { 544, PT_SC, ucp_Mandaic },
496      { 552, PT_PC, ucp_Mc },
497      { 555, PT_PC, ucp_Me },
498      { 558, PT_SC, ucp_Meetei_Mayek },
499      { 571, PT_PC, ucp_Mn },
500      { 574, PT_SC, ucp_Mongolian },
501      { 584, PT_SC, ucp_Myanmar },
502      { 592, PT_GC, ucp_N },
503      { 594, PT_PC, ucp_Nd },
504      { 597, PT_SC, ucp_New_Tai_Lue },
505      { 609, PT_SC, ucp_Nko },
506      { 613, PT_PC, ucp_Nl },
507      { 616, PT_PC, ucp_No },
508      { 619, PT_SC, ucp_Ogham },
509      { 625, PT_SC, ucp_Ol_Chiki },
510      { 634, PT_SC, ucp_Old_Italic },
511      { 645, PT_SC, ucp_Old_Persian },
512      { 657, PT_SC, ucp_Old_South_Arabian },
513      { 675, PT_SC, ucp_Old_Turkic },
514      { 686, PT_SC, ucp_Oriya },
515      { 692, PT_SC, ucp_Osmanya },
516      { 700, PT_GC, ucp_P },
517      { 702, PT_PC, ucp_Pc },
518      { 705, PT_PC, ucp_Pd },
519      { 708, PT_PC, ucp_Pe },
520      { 711, PT_PC, ucp_Pf },
521      { 714, PT_SC, ucp_Phags_Pa },
522      { 723, PT_SC, ucp_Phoenician },
523      { 734, PT_PC, ucp_Pi },
524      { 737, PT_PC, ucp_Po },
525      { 740, PT_PC, ucp_Ps },
526      { 743, PT_SC, ucp_Rejang },
527      { 750, PT_SC, ucp_Runic },
528      { 756, PT_GC, ucp_S },
529      { 758, PT_SC, ucp_Samaritan },
530      { 768, PT_SC, ucp_Saurashtra },
531      { 779, PT_PC, ucp_Sc },
532      { 782, PT_SC, ucp_Shavian },
533      { 790, PT_SC, ucp_Sinhala },
534      { 798, PT_PC, ucp_Sk },
535      { 801, PT_PC, ucp_Sm },
536      { 804, PT_PC, ucp_So },
537      { 807, PT_SC, ucp_Sundanese },
538      { 817, PT_SC, ucp_Syloti_Nagri },
539      { 830, PT_SC, ucp_Syriac },
540      { 837, PT_SC, ucp_Tagalog },
541      { 845, PT_SC, ucp_Tagbanwa },
542      { 854, PT_SC, ucp_Tai_Le },
543      { 861, PT_SC, ucp_Tai_Tham },
544      { 870, PT_SC, ucp_Tai_Viet },
545      { 879, PT_SC, ucp_Tamil },
546      { 885, PT_SC, ucp_Telugu },
547      { 892, PT_SC, ucp_Thaana },
548      { 899, PT_SC, ucp_Thai },
549      { 904, PT_SC, ucp_Tibetan },
550      { 912, PT_SC, ucp_Tifinagh },
551      { 921, PT_SC, ucp_Ugaritic },
552      { 930, PT_SC, ucp_Vai },
553      { 934, PT_ALNUM, 0 },
554      { 938, PT_PXSPACE, 0 },
555      { 942, PT_SPACE, 0 },
556      { 946, PT_WORD, 0 },
557      { 950, PT_SC, ucp_Yi },
558      { 953, PT_GC, ucp_Z },
559      { 955, PT_PC, ucp_Zl },
560      { 958, PT_PC, ucp_Zp },
561      { 961, PT_PC, ucp_Zs }
562    };
563    
564    const int PRIV(utt_size) = sizeof(PRIV(utt)) / sizeof(ucp_type_table);
565    
566    #endif /* SUPPORT_UTF */
567    
568  /* End of pcre_tables.c */  /* End of pcre_tables.c */

Legend:
Removed from v.87  
changed lines
  Added in v.836

  ViewVC Help
Powered by ViewVC 1.1.5