Annotation of parser3/src/main/pa_string.C, revision 1.221
1.45 paf 1: /** @file
1.174 paf 2: Parser: string class. @see untalength_t.C.
1.46 paf 3:
1.219 misha 4: Copyright (c) 2001-2009 ArtLebedev Group (http://www.artlebedev.com)
1.138 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.164 paf 6: */
1.46 paf 7:
1.221 ! misha 8: static const char * const IDENT_STRING_C="$Date: 2009-04-15 07:44:25 $";
1.70 paf 9:
1.12 paf 10: #include "pa_string.h"
1.22 paf 11: #include "pa_exception.h"
1.61 paf 12: #include "pa_table.h"
1.101 parser 13: #include "pa_dictionary.h"
1.132 paf 14: #include "pa_charset.h"
1.60 paf 15:
1.215 misha 16: #include "pcre.h"
17:
1.185 paf 18: const String String::Empty;
19:
1.193 paf 20: int pa_atoi(const char* str, const String* problem_source) {
21: if(!str)
22: return 0;
23:
1.199 paf 24: while(*str && isspace((unsigned char)*str))
1.193 paf 25: str++;
26: if(!*str)
27: return 0;
28:
29: int result;
30: char *error_pos;
1.200 paf 31: bool negative=false;
32: if(str[0]=='-') {
33: negative=true;
34: str++;
35: } else if(str[0]=='+') {
36: str++;
37: }
1.193 paf 38: // 0xABC
39: if(str[0]=='0')
40: if(str[1]=='x' || str[1]=='X')
41: result=(int)(unsigned long)strtol(str, &error_pos, 0);
1.197 paf 42: else {
43: // skip leading 0000, to disable octal interpretation
44: do str++; while(*str=='0');
45: result=(int)strtol(str, &error_pos, 0);
46: }
1.193 paf 47: else
48: result=(int)strtol(str, &error_pos, 0);
1.200 paf 49: if(negative)
50: result=-result;
1.193 paf 51:
52: while(char c=*error_pos++)
1.199 paf 53: if(!isspace((unsigned char)c))
1.193 paf 54: throw Exception("number.format",
55: problem_source,
56: problem_source?"invalid number (int)": "'%s' is invalid number (int)", str);
57:
58: return result;
59: }
60:
61: double pa_atod(const char* str, const String* problem_source) {
62: if(!str)
63: return 0;
64:
1.199 paf 65: while(*str && isspace((unsigned char)*str))
1.193 paf 66: str++;
67: if(!*str)
68: return 0;
69:
70: double result;
71: char *error_pos;
1.200 paf 72: bool negative=false;
73: if(str[0]=='-') {
74: negative=true;
75: str++;
76: } else if(str[0]=='+') {
77: str++;
78: }
1.193 paf 79: // 0xABC
80: if(str[0]=='0')
81: if(str[1]=='x' || str[1]=='X')
82: result=(double)(unsigned long)strtol(str, &error_pos, 0);
1.200 paf 83: else {
84: // skip leading 0000, to disable octal interpretation
85: do str++; while(*str=='0');
86: result=(double)strtod(str, &error_pos);
87: }
1.193 paf 88: else
89: result=(double)strtod(str, &error_pos);
1.200 paf 90: if(negative)
91: result=-result;
1.193 paf 92:
93: while(char c=*error_pos++)
1.199 paf 94: if(!isspace((unsigned char)c))
1.193 paf 95: throw Exception("number.format",
96: problem_source,
97: problem_source?"invalid number (double)": "'%s' is invalid number (double)", str);
98:
99: return result;
100: }
101:
1.176 paf 102: // cord lib extension
103:
104: #ifndef DOXYGEN
105: typedef struct {
106: ssize_t countdown;
1.204 paf 107: int target; /* Character we're looking for */
1.176 paf 108: } chr_data;
109: #endif
110: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
111: {
112: register chr_data * d = (chr_data *)client_data;
113:
114: if (d -> countdown<=0) return(2);
115: d -> countdown -= size;
116: if (c > d -> target) return(1);
117: return(0);
118: }
119: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
120: {
121: chr_data d;
122:
123: d.countdown = n;
124: d.target = c;
125: return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/);
126: }
127:
1.187 paf 128: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 129: {
130: int* result=(int*)client_data;
131: (*result)++;
132: return(0); // 0=continue
133: }
134: size_t CORD_block_count(CORD x)
135: {
136: size_t result=0;
137: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
138: return result;
139: }
140:
1.174 paf 141: // helpers
1.139 paf 142:
1.174 paf 143: /// String::match uses this as replace & global search table columns
1.139 paf 144:
1.174 paf 145: const int MAX_MATCH_GROUPS=100;
1.139 paf 146:
1.174 paf 147: class String_match_table_template_columns: public ArrayString {
148: public:
149: String_match_table_template_columns() {
150: *this+=new String("prematch");
151: *this+=new String("match");
152: *this+=new String("postmatch");
153: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.176 paf 154: *this+=new String(String::Body::Format(1+i), String::L_CLEAN);
1.174 paf 155: }
156: }
157: };
158:
159: Table string_match_table_template(new String_match_table_template_columns);
160:
1.176 paf 161: // String::Body methods
1.140 paf 162:
1.176 paf 163: String::Body String::Body::Format(int value) {
1.174 paf 164: char local[MAX_NUMBER];
165: size_t length=snprintf(local, MAX_NUMBER, "%d", value);
1.176 paf 166: return String::Body(pa_strdup(local, length), length);
1.120 paf 167: }
168:
1.195 paf 169: String::Body String::Body::trim(String::Trim_kind kind, const char* chars,
170: size_t* out_start, size_t* out_length) const {
171: size_t our_length=length();
172: if(!our_length)
173: return *this;
174: if(!chars)
175: chars=" \t\n"; // white space
176:
177: size_t start=0;
178: size_t end=our_length;
1.196 paf 179: // from left...
180: if(kind!=TRIM_END) {
1.195 paf 181: CORD_pos pos; set_pos(pos, 0);
182: while(true) {
183: char c=CORD_pos_fetch(pos);
184: if(strchr(chars, c)) {
185: if(++start==our_length)
186: return 0; // all chars are empty, just return empty string
187: } else
188: break;
189:
190: CORD_next(pos);
191: }
192: }
1.196 paf 193: // from right..
194: if(kind!=TRIM_START) {
195: CORD_pos pos; set_pos(pos, end-1);
196: while(true) {
197: char c=CORD_pos_fetch(pos);
198: if(strchr(chars, c)) {
199: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
200: return 0; // all chars are empty, just return empty string
201: } else
202: break;
203:
204: CORD_prev(pos);
205: }
206: }
1.195 paf 207:
208: if(start==0 && end==our_length) // nobody moved a thing
209: return *this;
210:
211: if(out_start)
212: *out_start=start;
213: size_t new_length=end-start;
214: if(out_length)
215: *out_length=new_length;
216:
217: return mid(start, new_length);
218: }
219:
1.174 paf 220: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
221: uint& result=*static_cast<uint*>(client_data);
222: generic_hash_code(result, c);
223: return 0;
224: }
225: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
226: uint& result=*static_cast<uint*>(client_data);
227: generic_hash_code(result, s);
228: return 0;
229: };
1.176 paf 230: uint String::Body::hash_code() const {
1.174 paf 231: uint result=0;
1.220 misha 232: if (body && CORD_IS_STRING(body)){
233: generic_hash_code(result, body);
234: } else {
235: CORD_iter5(body, 0,
236: CORD_batched_iter_fn_generic_hash_code,
237: CORD_batched_iter_fn_generic_hash_code, &result);
238: }
1.120 paf 239: return result;
1.94 parser 240: }
241:
1.174 paf 242: // String methods
243:
244: String::String(const char* cstr, size_t helper_length, bool tainted): body(CORD_EMPTY) {
245: append_help_length(cstr, helper_length, tainted?L_TAINTED:L_CLEAN);
1.115 paf 246: }
1.174 paf 247: String::String(const String::C cstr, bool tainted): body(CORD_EMPTY) {
248: append_know_length(cstr.str, cstr.length, tainted?L_TAINTED:L_CLEAN);
1.5 paf 249: }
1.28 paf 250:
1.174 paf 251: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
252: if(!known_length)
1.9 paf 253: return *this;
1.122 paf 254:
1.176 paf 255: // first: langs
256: langs.append(body, lang, known_length);
257: // next: letters themselves
1.174 paf 258: body.append_know_length(str, known_length);
1.1 paf 259:
1.174 paf 260: ASSERT_STRING_INVARIANT(*this);
1.1 paf 261: return *this;
262: }
1.174 paf 263: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
264: if(!str)
265: return *this;
266: size_t known_length=helper_length?helper_length:strlen(str);
267: if(!known_length)
268: return *this;
1.1 paf 269:
1.174 paf 270: return append_know_length(str, known_length, lang);
1.5 paf 271: }
1.174 paf 272: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
273: size_t known_length=helper_length?helper_length:strlen(str);
274: if(!known_length)
275: return *this;
1.5 paf 276:
1.176 paf 277: // first: langs
278: langs.append(body, lang, known_length);
279: // next: letters themselves
1.174 paf 280: body.append_strdup_know_length(str, known_length);
1.33 paf 281:
1.174 paf 282: ASSERT_STRING_INVARIANT(*this);
283: return *this;
1.5 paf 284: }
1.46 paf 285:
1.221 ! misha 286: int CORD_batched_len(const char* s, size_t* len){
! 287: (*len) += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s));
! 288: return 0;
! 289: }
! 290:
! 291: // can be called only for IS_FUNCTION(CORD) which are used only in Lang
! 292: int CORD_batched_len(const char, size_t *len){
! 293: (*len)++;
1.220 misha 294: return 0;
295: }
296:
1.210 misha 297: size_t String::length(Charset& charset) const {
298: if(charset.isUTF8()){
1.220 misha 299: size_t len=0;
1.221 ! misha 300: body.for_each(CORD_batched_len, CORD_batched_len, &len);
1.220 misha 301: return len;
1.210 misha 302: } else
303: return body.length();
304: }
305:
1.174 paf 306: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
307: String& String::mid(size_t substr_begin, size_t substr_end) const {
308: String& result=*new String;
309:
310: size_t self_length=length();
311: substr_begin=min(substr_begin, self_length);
312: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 313: size_t substr_length=substr_end-substr_begin;
314: if(!substr_length)
1.107 parser 315: return result;
1.53 paf 316:
1.176 paf 317: // first: their langs
318: result.langs.append(result.body, langs, substr_begin, substr_length);
319: // next: letters themselves
320: result.body=body.mid(substr_begin, substr_length);
1.174 paf 321:
322: ASSERT_STRING_INVARIANT(result);
1.53 paf 323: return result;
1.54 paf 324: }
325:
1.211 misha 326: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
327: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 328: String& result=*new String;
329:
1.211 misha 330: size_t self_length=(helper_length)?helper_length:length(charset);
331:
332: if(!self_length)
333: return result;
334:
335: from=min(min(to, from), self_length);
1.210 misha 336: to=min(max(to, from), self_length);
1.211 misha 337:
1.210 misha 338: size_t substr_length=to-from;
1.211 misha 339:
1.210 misha 340: if(!substr_length)
341: return result;
342:
343: if(charset.isUTF8()){
344: const XMLByte* srcPtr=(const XMLByte*)cstrm();
1.211 misha 345: const XMLByte* srcEnd=srcPtr+body.length();
1.210 misha 346:
1.212 misha 347: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.210 misha 348: from=getUTF8BytePos(srcPtr, srcEnd, from);
349: substr_length=getUTF8BytePos(srcPtr+from, srcEnd, substr_length);
350: if(!substr_length)
351: return result;
352: }
353:
354: // first: their langs
355: result.langs.append(result.body, langs, from, substr_length);
356: // next: letters themselves
357: result.body=body.mid(from, substr_length);
358:
359: ASSERT_STRING_INVARIANT(result);
360: return result;
361: }
362:
1.176 paf 363: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 364: size_t substr_length=substr.length();
365: while(true) {
366: size_t substr_begin=body.pos(substr, this_offset);
367:
368: if(substr_begin==CORD_NOT_FOUND)
369: return STRING_NOT_FOUND;
1.174 paf 370:
1.183 paf 371: if(langs.check_lang(lang, substr_begin, substr_length))
372: return substr_begin;
373:
374: this_offset=substr_begin+substr_length;
375: }
1.58 paf 376: }
377:
1.174 paf 378: size_t String::pos(const String& substr,
379: size_t this_offset, Language lang) const {
380: return pos(substr.body, this_offset, lang);
1.60 paf 381: }
382:
1.210 misha 383: size_t String::pos(Charset& charset, const String& substr,
384: size_t this_offset, Language lang) const {
385:
1.212 misha 386: size_t result;
1.210 misha 387: if(charset.isUTF8()){
388: const XMLByte* srcPtr=(const XMLByte*)cstrm();
1.212 misha 389: const XMLByte* srcEnd=srcPtr+body.length();
390:
391: // convert 'this_offset' from 'characters' to 'bytes'
392: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
393:
394: result=pos(substr.body, this_offset, lang);
395: if(result==CORD_NOT_FOUND)
396: return STRING_NOT_FOUND;
397:
398: // convert 'result' from 'bytes' to 'characters'
399: result=getUTF8CharPos(srcPtr, srcEnd, result);
400: } else {
401: result=pos(substr.body, this_offset, lang);
402: if(result==CORD_NOT_FOUND)
403: return STRING_NOT_FOUND;
1.210 misha 404: }
405:
406: return result;
407: }
408:
1.174 paf 409: void String::split(ArrayString& result,
410: size_t& pos_after,
411: const char* delim,
412: Language lang, int limit) const {
413: size_t self_length=length();
414: if(size_t delim_length=strlen(delim)) {
1.186 paf 415: size_t pos_before;
1.60 paf 416: // while we have 'delim'...
1.174 paf 417: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 418: result+=&mid(pos_after, pos_before);
1.174 paf 419: pos_after=pos_before+delim_length;
1.60 paf 420: }
421: // last piece
1.174 paf 422: if(pos_after<self_length && limit) {
423: result+=&mid(pos_after, self_length);
424: pos_after=self_length;
1.60 paf 425: }
426: } else { // empty delim
427: result+=this;
1.174 paf 428: pos_after+=self_length;
1.60 paf 429: }
430: }
431:
1.174 paf 432: void String::split(ArrayString& result,
433: size_t& pos_after,
434: const String& delim, Language lang,
435: int limit) const {
1.140 paf 436: if(!delim.is_empty()) {
1.186 paf 437: size_t pos_before;
1.60 paf 438: // while we have 'delim'...
1.174 paf 439: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 440: result+=&mid(pos_after, pos_before);
1.174 paf 441: pos_after=pos_before+delim.length();
1.60 paf 442: }
443: // last piece
1.174 paf 444: if(pos_after<length() && limit) {
445: result+=&mid(pos_after, length());
446: pos_after=length();
1.60 paf 447: }
448: } else { // empty delim
449: result+=this;
1.174 paf 450: pos_after+=length();
1.60 paf 451: }
1.61 paf 452: }
453:
1.209 misha 454: enum Match_feature {
455: MF_NEED_PRE_POST_MATCH = 0x01,
456: MF_JUST_COUNT_MATCHES = 0x02
457: };
458:
459: static void regex_options(const String* options, int* result, int* match_features){
1.63 paf 460: struct Regex_option {
1.216 misha 461: const char* key;
462: const char* keyAlt;
1.209 misha 463: int clear;
464: int set;
1.63 paf 465: int *result;
1.209 misha 466: int flag;
1.63 paf 467: } regex_option[]={
1.189 paf 468: {"i", "I", 0, PCRE_CASELESS, result, 0}, // a=A
1.216 misha 469: {"s", "S", 0, PCRE_DOTALL, result, 0}, // ^\n\n$ [default]
1.189 paf 470: {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result, 0}, // ^aaa\n$^bbb\n$
1.216 misha 471: {"x", 0, 0, PCRE_EXTENDED, result, 0}, // whitespace in regex ignored
472: {"U", 0, 0, PCRE_UNGREEDY, result, 0}, // ungreedy patterns (greedy by default)
1.189 paf 473: {"g", "G", 0, 1, result+1, 0}, // many rows
1.209 misha 474: {"'", 0, 0, 0, 0, MF_NEED_PRE_POST_MATCH},
1.216 misha 475: {"n", 0, 0, 0, 0, MF_JUST_COUNT_MATCHES},
1.189 paf 476: {0, 0, 0, 0, 0, 0}
1.63 paf 477: };
1.171 paf 478: result[0]=PCRE_EXTRA | PCRE_DOTALL | PCRE_DOLLAR_ENDONLY;
1.63 paf 479: result[1]=0;
480:
1.174 paf 481: if(options && !options->is_empty())
1.216 misha 482: for(Regex_option *o=regex_option; o->key; o++)
1.209 misha 483: if(
1.216 misha 484: options->pos(o->key)!=STRING_NOT_FOUND
485: || (o->keyAlt && options->pos(o->keyAlt)!=STRING_NOT_FOUND)
1.209 misha 486: ){
487: if(o->flag){
488: (*match_features) |= o->flag;
489: } else {
1.154 paf 490: *o->result &= ~o->clear;
491: *o->result |= o->set;
492: }
1.63 paf 493: }
494: }
495:
1.174 paf 496: Table* String::match(Charset& source_charset,
497: const String& regexp,
498: const String* options,
499: Row_action row_action, void *info,
1.209 misha 500: int& matches_count) const {
1.140 paf 501: if(regexp.is_empty())
1.218 misha 502: throw Exception(PARSER_RUNTIME,
1.174 paf 503: 0,
1.73 paf 504: "regexp is empty");
1.154 paf 505:
1.205 paf 506: const char* pattern=regexp.cstr(String::L_UNSPECIFIED); // fix any tainted with L_REGEX
1.174 paf 507: const char* errptr;
1.62 paf 508: int erroffset;
1.209 misha 509: int option_bits[2]={0};
510: int match_features=0;
511: regex_options(options, option_bits, &match_features);
512: bool need_pre_post_match=(match_features & MF_NEED_PRE_POST_MATCH) != 0;
513: bool just_count_matches=(match_features & MF_JUST_COUNT_MATCHES) != 0;
1.174 paf 514: bool global=option_bits[1]!=0;
1.215 misha 515:
1.219 misha 516: if(source_charset.isUTF8()){
517: // @todo (for UTF-8): check string & pattern and use PCRE_NO_UTF8_CHECK option
518: option_bits[0]|=PCRE_UTF8;
519: }
1.215 misha 520:
521: pcre *code=pcre_compile(pattern, option_bits[0],
1.62 paf 522: &errptr, &erroffset,
1.174 paf 523: source_charset.pcre_tables);
1.62 paf 524:
1.67 paf 525: if(!code)
1.218 misha 526: throw Exception(PCRE_EXCEPTION_TYPE,
1.174 paf 527: ®exp.mid(erroffset, regexp.length()),
1.74 paf 528: "regular expression syntax error - %s", errptr);
1.62 paf 529:
1.215 misha 530: int subpatterns=pcre_info(code, 0, 0);
1.174 paf 531: if(subpatterns<0) {
1.215 misha 532: pcre_free(code);
1.218 misha 533: throw Exception(PCRE_EXCEPTION_TYPE,
1.174 paf 534: ®exp,
1.76 paf 535: "pcre_info error (%d)",
1.174 paf 536: subpatterns);
1.63 paf 537: }
538:
1.174 paf 539: const char* subject=cstr();
540: size_t subject_length=strlen(subject);
541: const int oveclength=(1/*match*/+MAX_MATCH_GROUPS)*3;
542: int ovector[oveclength];
1.155 paf 543:
544: // create table
1.173 paf 545: Table::Action_options table_options;
1.174 paf 546: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 547:
1.64 paf 548: int exec_option_bits=0;
1.154 paf 549: int prestart=0;
550: int poststart=0;
1.174 paf 551: int postfinish=length();
1.63 paf 552: while(true) {
1.215 misha 553: int exec_substrings=pcre_exec(code, 0,
1.174 paf 554: subject, subject_length, prestart,
555: exec_option_bits, ovector, oveclength);
1.63 paf 556:
557: if(exec_substrings==PCRE_ERROR_NOMATCH) {
1.215 misha 558: pcre_free(code);
1.174 paf 559: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
1.209 misha 560: return just_count_matches ? 0 : &table;
1.63 paf 561: }
562:
563: if(exec_substrings<0) {
1.215 misha 564: pcre_free(code);
1.218 misha 565: throw Exception(PCRE_EXCEPTION_TYPE,
1.174 paf 566: ®exp,
1.217 misha 567: print_pcre_exec_error_text(exec_substrings),
1.63 paf 568: exec_substrings);
569: }
570:
1.154 paf 571: int prefinish=ovector[0];
572: poststart=ovector[1];
1.174 paf 573: ArrayString* row=new ArrayString;
574: if(need_pre_post_match) {
575: *row+=&mid(0, prefinish); // .prematch column value
576: *row+=&mid(prefinish, poststart); // .match
577: *row+=&mid(poststart, postfinish); // .postmatch
578: } else {
1.185 paf 579: *row+=&Empty; // .prematch column value
580: *row+=&Empty; // .match
581: *row+=&Empty; // .postmatch
1.174 paf 582: }
1.63 paf 583:
584: for(int i=1; i<exec_substrings; i++) {
1.69 paf 585: // -1:-1 case handled peacefully by mid() itself
1.174 paf 586: *row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value
1.63 paf 587: }
588:
1.209 misha 589: matches_count++;
1.174 paf 590: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 591:
1.219 misha 592: if(!global || prestart==poststart) { // last step
1.215 misha 593: pcre_free(code);
1.174 paf 594: row_action(table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info);
1.209 misha 595: return just_count_matches ? 0 : &table;
1.63 paf 596: }
1.154 paf 597: prestart=poststart;
1.63 paf 598:
599: /*
600: if(option_bits[0] & PCRE_MULTILINE)
1.64 paf 601: exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL
1.63 paf 602: */
603: }
1.82 parser 604: }
605:
1.174 paf 606: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
607: String& result=*new String();
608: if(is_empty())
609: return result;
610:
611: char* new_cstr=cstrm();
1.216 misha 612:
1.181 paf 613: if(source_charset.isUTF8()) {
1.216 misha 614: size_t new_cstr_len=length();
1.181 paf 615: switch(kind) {
616: case CC_UPPER:
1.192 paf 617: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 618: break;
619: case CC_LOWER:
1.192 paf 620: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 621: break;
622: default:
623: assert(!"unknown change case kind");
624: break; // never
625: }
626:
627: } else {
628: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 629:
1.181 paf 630: const unsigned char *a;
631: const unsigned char *b;
632: switch(kind) {
633: case CC_UPPER:
634: a=tables+lcc_offset;
635: b=tables+fcc_offset;
636: break;
637: case CC_LOWER:
638: a=tables+lcc_offset;
639: b=0;
640: break;
641: default:
642: assert(!"unknown change case kind");
643: a=b=0; // calm, compiler
644: break; // never
645: }
646:
1.192 paf 647: char *dest=new_cstr;
1.181 paf 648: unsigned char index;
1.190 paf 649: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 650: unsigned char c=a[index];
651: if(b)
652: c=b[c];
653:
654: *dest++=(char)c;
655: }
1.174 paf 656: }
1.176 paf 657: result.langs=langs;
1.174 paf 658: result.body=new_cstr;
1.89 parser 659:
1.101 parser 660: return result;
661: }
662:
1.213 misha 663: const String& String::escape(Charset& source_charset) const {
664: if(is_empty())
665: return *this;
666:
667: return Charset::escape(*this, source_charset);
668: }
669:
1.174 paf 670: const String& String::replace(const Dictionary& dict) const {
671: String& result=*new String();
672: const char* old_cstr=cstr();
673: const char* prematch_begin=old_cstr;
674:
675: const char* current=old_cstr;
676: while(*current) {
1.184 paf 677: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
1.174 paf 678: // prematch
679: if(size_t prematch_length=current-prematch_begin) {
1.179 paf 680: result.langs.append(result.body, langs, prematch_begin-old_cstr, prematch_length);
1.174 paf 681: result.body.append_strdup_know_length(prematch_begin, prematch_length);
1.101 parser 682: }
683:
1.174 paf 684: // match
685: // skip 'a' in 'current'; move prematch_begin
1.184 paf 686: current+=subst.from_length; prematch_begin=current;
1.174 paf 687:
1.184 paf 688: if(const String* b=subst.to) // are there any b?
1.174 paf 689: result<<*b;
690: } else // simply advance
691: current++;
692: }
1.156 paf 693:
1.174 paf 694: // postmatch
695: if(size_t postmatch_length=current-prematch_begin) {
1.179 paf 696: result.langs.append(result.body, langs, prematch_begin-old_cstr, postmatch_length);
1.174 paf 697: result.body.append_strdup_know_length(prematch_begin, postmatch_length);
698: }
1.156 paf 699:
1.174 paf 700: ASSERT_STRING_INVARIANT(result);
1.82 parser 701: return result;
1.61 paf 702: }
1.113 parser 703:
1.180 paf 704: static int serialize_body_char(char c, char** cur) {
705: *((*cur)++)=c;
706: return 0; // 0=continue
707: };
1.174 paf 708: static int serialize_body_piece(const char* s, char** cur) {
709: size_t length=strlen(s);
710: memcpy(*cur, s, length); *cur+=length;
1.178 paf 711: return 0; // 0=continue
1.174 paf 712: };
1.178 paf 713: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
714: // lang
1.191 paf 715: **cur=alang; (*cur)++;
716: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 717: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
718:
719: return 0; // 0=continue
720: }
1.174 paf 721: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 722: size_t fragments_count=langs.count();
1.202 paf 723: size_t body_length=body.length();
1.174 paf 724: size_t buf_length=
1.178 paf 725: prolog_length //1
726: +sizeof(size_t) //2
1.202 paf 727: +body_length //3
728: +1 // 4 for zero terminator used in deserialize
729: +sizeof(size_t) //5
730: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
731:
1.174 paf 732: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
733:
734: // 1: prolog
735: char *cur=result.str+prolog_length;
1.202 paf 736: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
737: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
738: // 3: letters
739: body.for_each(serialize_body_char, serialize_body_piece, &cur);
740: // 4: zero terminator
741: *cur++=0;
742: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 743: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 744: // 6: lang info
1.178 paf 745: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 746:
1.174 paf 747: return result;
1.113 parser 748: }
1.202 paf 749: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
750: size_t in_buf=buf_size;
751: if(in_buf<=prolog_size)
1.148 paf 752: return false;
1.202 paf 753: in_buf-=prolog_size;
1.135 paf 754:
1.174 paf 755: // 1: prolog
1.202 paf 756: const char* cur=(const char* )buf+prolog_size;
1.113 parser 757:
1.207 paf 758: // 2: chars.count
1.202 paf 759: size_t body_length;
760: if(in_buf<sizeof(body_length)) // body.length don't fit?
761: return false;
762: // [WARNING: not cast, addresses must be %4=0 on sparc]
763: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
764: in_buf-=sizeof(body_length);
765:
766: if(in_buf<body_length+1) // letters+terminator don't fit?
767: return false;
768: // 4: zero terminator
769: if(cur[body_length] != 0) // in place?
770: return false;
771: // 3: letters
772: body=String::Body(cur, body_length);
773: cur+=body_length+1;
774: in_buf-=body_length+1;
775:
776: // 5: langs.count
1.191 paf 777: size_t fragments_count;
1.202 paf 778: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 779: return false;
1.191 paf 780: // [WARNING: not cast, addresses must be %4=0 on sparc]
781: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 782: in_buf-=sizeof(fragments_count);
1.174 paf 783:
784: if(fragments_count) {
1.202 paf 785: // 6: lang info
1.174 paf 786: size_t total_length=0;
787: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 788: char lang;
789: size_t fragment_length;
790: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 791: if(in_buf<piece_length) // lang+length
1.174 paf 792: return false;
793:
1.191 paf 794: // lang
795: lang=*cur++;
796: // length [WARNING: not cast, addresses must be %4=0 on sparc]
797: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
798:
1.206 paf 799: size_t combined_length=total_length+fragment_length;
800: if(combined_length>body_length)
801: return false; // file curruption
1.191 paf 802: // uchar needed to prevent propagating 0x80 bit to upper bytes
803: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 804: total_length=combined_length;
1.202 paf 805: in_buf-=piece_length;
1.174 paf 806: }
1.128 paf 807:
1.202 paf 808: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 809: return false;
1.174 paf 810: }
1.202 paf 811: if(in_buf!=0) // some strange extra bytes
812: return false;
1.113 parser 813:
1.174 paf 814: ASSERT_STRING_INVARIANT(*this);
1.148 paf 815: return true;
1.176 paf 816: }
817:
818: const char* String::Body::v() const {
819: return CORD_to_const_char_star(body);
820: }
1.201 paf 821: void String::Body::dump() const {
822: CORD_dump(body);
823: }
824:
1.176 paf 825: const char* String::Languages::v() const {
1.177 paf 826: if(opt.is_not_just_lang)
1.176 paf 827: return CORD_to_const_char_star(langs);
828: else
829: return (const char*)&langs;
830: }
1.201 paf 831: void String::Languages::dump() const {
832: if(opt.is_not_just_lang)
833: CORD_dump(langs);
834: else
835: puts((const char*)&langs);
836: }
1.176 paf 837: const char* String::v() const {
1.198 paf 838: const uint LIMIT_VIEW=20;
1.176 paf 839: char* buf=(char*)malloc(MAX_STRING);
840: const char*body_view=body.v();
841: const char*langs_view=langs.v();
842: snprintf(buf, MAX_STRING,
1.178 paf 843: "%d:%.*s%s} "
1.176 paf 844: "{%d:%s",
1.178 paf 845: langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"",
1.176 paf 846: strlen(body_view), body_view
847: );
848:
849: return buf;
1.113 parser 850: }
1.201 paf 851: void String::dump() const {
852: body.dump();
853: langs.dump();
854: }
1.195 paf 855: const String& String::trim(String::Trim_kind kind, const char* chars) const {
856: if(!length())
857: return *this;
858:
859: size_t substr_begin, substr_length;
860: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length);
861: if(new_body==body) // we received unchanged pointer, do likewise
862: return *this;
863: // new_body differs from body, adjust langs along
864:
865: String& result=*new String;
866: if(!new_body) // body.trim produced empty result
867: return result;
868: // body.trim produced nonempty result
869:
870: // first: their langs
871: result.langs.append(result.body, langs, substr_begin, substr_length);
872: // next: letters themselves
873: result.body=new_body;
874:
875: ASSERT_STRING_INVARIANT(result);
876: return result;
1.198 paf 877: }
E-mail:
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to pa_string.C CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [parser3project] / parser3 / src / main