Annotation of parser3/src/main/pa_string.C, revision 1.216
1.45 paf 1: /** @file
1.174 paf 2: Parser: string class. @see untalength_t.C.
1.46 paf 3:
1.203 paf 4: Copyright (c) 2001-2005 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:
1.216 ! misha 8: static const char * const IDENT_STRING_C="$Date: 2008-08-15 15:30:12 $";
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;
232: CORD_iter5(body, 0,
233: CORD_batched_iter_fn_generic_hash_code,
234: CORD_batched_iter_fn_generic_hash_code, &result);
1.120 paf 235: return result;
1.94 parser 236: }
237:
1.174 paf 238: // String methods
239:
240: String::String(const char* cstr, size_t helper_length, bool tainted): body(CORD_EMPTY) {
241: append_help_length(cstr, helper_length, tainted?L_TAINTED:L_CLEAN);
1.115 paf 242: }
1.174 paf 243: String::String(const String::C cstr, bool tainted): body(CORD_EMPTY) {
244: append_know_length(cstr.str, cstr.length, tainted?L_TAINTED:L_CLEAN);
1.5 paf 245: }
1.28 paf 246:
1.174 paf 247: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
248: if(!known_length)
1.9 paf 249: return *this;
1.122 paf 250:
1.176 paf 251: // first: langs
252: langs.append(body, lang, known_length);
253: // next: letters themselves
1.174 paf 254: body.append_know_length(str, known_length);
1.1 paf 255:
1.174 paf 256: ASSERT_STRING_INVARIANT(*this);
1.1 paf 257: return *this;
258: }
1.174 paf 259: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
260: if(!str)
261: return *this;
262: size_t known_length=helper_length?helper_length:strlen(str);
263: if(!known_length)
264: return *this;
1.1 paf 265:
1.174 paf 266: return append_know_length(str, known_length, lang);
1.5 paf 267: }
1.174 paf 268: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
269: size_t known_length=helper_length?helper_length:strlen(str);
270: if(!known_length)
271: return *this;
1.5 paf 272:
1.176 paf 273: // first: langs
274: langs.append(body, lang, known_length);
275: // next: letters themselves
1.174 paf 276: body.append_strdup_know_length(str, known_length);
1.33 paf 277:
1.174 paf 278: ASSERT_STRING_INVARIANT(*this);
279: return *this;
1.5 paf 280: }
1.46 paf 281:
1.210 misha 282: size_t String::length(Charset& charset) const {
283: if(charset.isUTF8()){
284: const XMLByte* srcPtr=(const XMLByte*)cstrm();
285: return lengthUTF8(srcPtr, srcPtr+body.length());
286: } else
287: return body.length();
288: }
289:
1.174 paf 290: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
291: String& String::mid(size_t substr_begin, size_t substr_end) const {
292: String& result=*new String;
293:
294: size_t self_length=length();
295: substr_begin=min(substr_begin, self_length);
296: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 297: size_t substr_length=substr_end-substr_begin;
298: if(!substr_length)
1.107 parser 299: return result;
1.53 paf 300:
1.176 paf 301: // first: their langs
302: result.langs.append(result.body, langs, substr_begin, substr_length);
303: // next: letters themselves
304: result.body=body.mid(substr_begin, substr_length);
1.174 paf 305:
306: ASSERT_STRING_INVARIANT(result);
1.53 paf 307: return result;
1.54 paf 308: }
309:
1.211 misha 310: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
311: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 312: String& result=*new String;
313:
1.211 misha 314: size_t self_length=(helper_length)?helper_length:length(charset);
315:
316: if(!self_length)
317: return result;
318:
319: from=min(min(to, from), self_length);
1.210 misha 320: to=min(max(to, from), self_length);
1.211 misha 321:
1.210 misha 322: size_t substr_length=to-from;
1.211 misha 323:
1.210 misha 324: if(!substr_length)
325: return result;
326:
327: if(charset.isUTF8()){
328: const XMLByte* srcPtr=(const XMLByte*)cstrm();
1.211 misha 329: const XMLByte* srcEnd=srcPtr+body.length();
1.210 misha 330:
1.212 misha 331: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.210 misha 332: from=getUTF8BytePos(srcPtr, srcEnd, from);
333: substr_length=getUTF8BytePos(srcPtr+from, srcEnd, substr_length);
334: if(!substr_length)
335: return result;
336: }
337:
338: // first: their langs
339: result.langs.append(result.body, langs, from, substr_length);
340: // next: letters themselves
341: result.body=body.mid(from, substr_length);
342:
343: ASSERT_STRING_INVARIANT(result);
344: return result;
345: }
346:
1.176 paf 347: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 348: size_t substr_length=substr.length();
349: while(true) {
350: size_t substr_begin=body.pos(substr, this_offset);
351:
352: if(substr_begin==CORD_NOT_FOUND)
353: return STRING_NOT_FOUND;
1.174 paf 354:
1.183 paf 355: if(langs.check_lang(lang, substr_begin, substr_length))
356: return substr_begin;
357:
358: this_offset=substr_begin+substr_length;
359: }
1.58 paf 360: }
361:
1.174 paf 362: size_t String::pos(const String& substr,
363: size_t this_offset, Language lang) const {
364: return pos(substr.body, this_offset, lang);
1.60 paf 365: }
366:
1.210 misha 367: size_t String::pos(Charset& charset, const String& substr,
368: size_t this_offset, Language lang) const {
369:
1.212 misha 370: size_t result;
1.210 misha 371: if(charset.isUTF8()){
372: const XMLByte* srcPtr=(const XMLByte*)cstrm();
1.212 misha 373: const XMLByte* srcEnd=srcPtr+body.length();
374:
375: // convert 'this_offset' from 'characters' to 'bytes'
376: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
377:
378: result=pos(substr.body, this_offset, lang);
379: if(result==CORD_NOT_FOUND)
380: return STRING_NOT_FOUND;
381:
382: // convert 'result' from 'bytes' to 'characters'
383: result=getUTF8CharPos(srcPtr, srcEnd, result);
384: } else {
385: result=pos(substr.body, this_offset, lang);
386: if(result==CORD_NOT_FOUND)
387: return STRING_NOT_FOUND;
1.210 misha 388: }
389:
390: return result;
391: }
392:
1.174 paf 393: void String::split(ArrayString& result,
394: size_t& pos_after,
395: const char* delim,
396: Language lang, int limit) const {
397: size_t self_length=length();
398: if(size_t delim_length=strlen(delim)) {
1.186 paf 399: size_t pos_before;
1.60 paf 400: // while we have 'delim'...
1.174 paf 401: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 402: result+=&mid(pos_after, pos_before);
1.174 paf 403: pos_after=pos_before+delim_length;
1.60 paf 404: }
405: // last piece
1.174 paf 406: if(pos_after<self_length && limit) {
407: result+=&mid(pos_after, self_length);
408: pos_after=self_length;
1.60 paf 409: }
410: } else { // empty delim
411: result+=this;
1.174 paf 412: pos_after+=self_length;
1.60 paf 413: }
414: }
415:
1.174 paf 416: void String::split(ArrayString& result,
417: size_t& pos_after,
418: const String& delim, Language lang,
419: int limit) const {
1.140 paf 420: if(!delim.is_empty()) {
1.186 paf 421: size_t pos_before;
1.60 paf 422: // while we have 'delim'...
1.174 paf 423: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 424: result+=&mid(pos_after, pos_before);
1.174 paf 425: pos_after=pos_before+delim.length();
1.60 paf 426: }
427: // last piece
1.174 paf 428: if(pos_after<length() && limit) {
429: result+=&mid(pos_after, length());
430: pos_after=length();
1.60 paf 431: }
432: } else { // empty delim
433: result+=this;
1.174 paf 434: pos_after+=length();
1.60 paf 435: }
1.61 paf 436: }
437:
1.209 misha 438: enum Match_feature {
439: MF_NEED_PRE_POST_MATCH = 0x01,
440: MF_JUST_COUNT_MATCHES = 0x02
441: };
442:
443: static void regex_options(const String* options, int* result, int* match_features){
1.63 paf 444: struct Regex_option {
1.216 ! misha 445: const char* key;
! 446: const char* keyAlt;
1.209 misha 447: int clear;
448: int set;
1.63 paf 449: int *result;
1.209 misha 450: int flag;
1.63 paf 451: } regex_option[]={
1.189 paf 452: {"i", "I", 0, PCRE_CASELESS, result, 0}, // a=A
1.216 ! misha 453: {"s", "S", 0, PCRE_DOTALL, result, 0}, // ^\n\n$ [default]
1.189 paf 454: {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result, 0}, // ^aaa\n$^bbb\n$
1.216 ! misha 455: {"x", 0, 0, PCRE_EXTENDED, result, 0}, // whitespace in regex ignored
! 456: {"U", 0, 0, PCRE_UNGREEDY, result, 0}, // ungreedy patterns (greedy by default)
1.189 paf 457: {"g", "G", 0, 1, result+1, 0}, // many rows
1.209 misha 458: {"'", 0, 0, 0, 0, MF_NEED_PRE_POST_MATCH},
1.216 ! misha 459: {"n", 0, 0, 0, 0, MF_JUST_COUNT_MATCHES},
1.189 paf 460: {0, 0, 0, 0, 0, 0}
1.63 paf 461: };
1.171 paf 462: result[0]=PCRE_EXTRA | PCRE_DOTALL | PCRE_DOLLAR_ENDONLY;
1.63 paf 463: result[1]=0;
464:
1.174 paf 465: if(options && !options->is_empty())
1.216 ! misha 466: for(Regex_option *o=regex_option; o->key; o++)
1.209 misha 467: if(
1.216 ! misha 468: options->pos(o->key)!=STRING_NOT_FOUND
! 469: || (o->keyAlt && options->pos(o->keyAlt)!=STRING_NOT_FOUND)
1.209 misha 470: ){
471: if(o->flag){
472: (*match_features) |= o->flag;
473: } else {
1.154 paf 474: *o->result &= ~o->clear;
475: *o->result |= o->set;
476: }
1.63 paf 477: }
478: }
479:
1.174 paf 480: Table* String::match(Charset& source_charset,
481: const String& regexp,
482: const String* options,
483: Row_action row_action, void *info,
1.209 misha 484: int& matches_count) const {
1.140 paf 485: if(regexp.is_empty())
1.149 paf 486: throw Exception(0,
1.174 paf 487: 0,
1.73 paf 488: "regexp is empty");
1.154 paf 489:
1.205 paf 490: const char* pattern=regexp.cstr(String::L_UNSPECIFIED); // fix any tainted with L_REGEX
1.174 paf 491: const char* errptr;
1.62 paf 492: int erroffset;
1.209 misha 493: int option_bits[2]={0};
494: int match_features=0;
495: regex_options(options, option_bits, &match_features);
496: bool need_pre_post_match=(match_features & MF_NEED_PRE_POST_MATCH) != 0;
497: bool just_count_matches=(match_features & MF_JUST_COUNT_MATCHES) != 0;
1.174 paf 498: bool global=option_bits[1]!=0;
1.215 misha 499:
500: if(source_charset.isUTF8())
501: option_bits[0]=option_bits[0] | PCRE_UTF8;
502:
503: pcre *code=pcre_compile(pattern, option_bits[0],
1.62 paf 504: &errptr, &erroffset,
1.174 paf 505: source_charset.pcre_tables);
1.62 paf 506:
1.67 paf 507: if(!code)
1.149 paf 508: throw Exception(0,
1.174 paf 509: ®exp.mid(erroffset, regexp.length()),
1.74 paf 510: "regular expression syntax error - %s", errptr);
1.62 paf 511:
1.215 misha 512: int subpatterns=pcre_info(code, 0, 0);
1.174 paf 513: if(subpatterns<0) {
1.215 misha 514: pcre_free(code);
1.149 paf 515: throw Exception(0,
1.174 paf 516: ®exp,
1.76 paf 517: "pcre_info error (%d)",
1.174 paf 518: subpatterns);
1.63 paf 519: }
520:
1.174 paf 521: const char* subject=cstr();
522: size_t subject_length=strlen(subject);
523: const int oveclength=(1/*match*/+MAX_MATCH_GROUPS)*3;
524: int ovector[oveclength];
1.155 paf 525:
526: // create table
1.173 paf 527: Table::Action_options table_options;
1.174 paf 528: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 529:
1.64 paf 530: int exec_option_bits=0;
1.154 paf 531: int prestart=0;
532: int poststart=0;
1.174 paf 533: int postfinish=length();
1.63 paf 534: while(true) {
1.215 misha 535: int exec_substrings=pcre_exec(code, 0,
1.174 paf 536: subject, subject_length, prestart,
537: exec_option_bits, ovector, oveclength);
1.63 paf 538:
539: if(exec_substrings==PCRE_ERROR_NOMATCH) {
1.215 misha 540: pcre_free(code);
1.174 paf 541: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
1.208 misha 542: // if(global || subpatterns)
543: // return &table; // global or with subpatterns=true+result
544: // else {
545: // just_matched=false; return 0; // not global=no result
546: // }
1.209 misha 547: return just_count_matches ? 0 : &table;
1.63 paf 548: }
549:
550: if(exec_substrings<0) {
1.215 misha 551: pcre_free(code);
1.149 paf 552: throw Exception(0,
1.174 paf 553: ®exp,
1.76 paf 554: "regular expression execute error (%d)",
1.63 paf 555: exec_substrings);
556: }
557:
1.154 paf 558: int prefinish=ovector[0];
559: poststart=ovector[1];
1.174 paf 560: ArrayString* row=new ArrayString;
561: if(need_pre_post_match) {
562: *row+=&mid(0, prefinish); // .prematch column value
563: *row+=&mid(prefinish, poststart); // .match
564: *row+=&mid(poststart, postfinish); // .postmatch
565: } else {
1.185 paf 566: *row+=&Empty; // .prematch column value
567: *row+=&Empty; // .match
568: *row+=&Empty; // .postmatch
1.174 paf 569: }
1.63 paf 570:
571: for(int i=1; i<exec_substrings; i++) {
1.69 paf 572: // -1:-1 case handled peacefully by mid() itself
1.174 paf 573: *row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value
1.63 paf 574: }
575:
1.209 misha 576: matches_count++;
1.174 paf 577: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 578:
1.174 paf 579: if(!global || prestart==poststart) { // not global | going to hang
1.215 misha 580: pcre_free(code);
1.174 paf 581: row_action(table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info);
1.209 misha 582: return just_count_matches ? 0 : &table;
583: // return &table;
1.63 paf 584: }
1.154 paf 585: prestart=poststart;
1.63 paf 586:
587: /*
588: if(option_bits[0] & PCRE_MULTILINE)
1.64 paf 589: exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL
1.63 paf 590: */
591: }
1.82 parser 592: }
593:
1.174 paf 594: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
595: String& result=*new String();
596: if(is_empty())
597: return result;
598:
599: char* new_cstr=cstrm();
1.216 ! misha 600:
1.181 paf 601: if(source_charset.isUTF8()) {
1.216 ! misha 602: size_t new_cstr_len=length();
1.181 paf 603: switch(kind) {
604: case CC_UPPER:
1.192 paf 605: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 606: break;
607: case CC_LOWER:
1.192 paf 608: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 609: break;
610: default:
611: assert(!"unknown change case kind");
612: break; // never
613: }
614:
615: } else {
616: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 617:
1.181 paf 618: const unsigned char *a;
619: const unsigned char *b;
620: switch(kind) {
621: case CC_UPPER:
622: a=tables+lcc_offset;
623: b=tables+fcc_offset;
624: break;
625: case CC_LOWER:
626: a=tables+lcc_offset;
627: b=0;
628: break;
629: default:
630: assert(!"unknown change case kind");
631: a=b=0; // calm, compiler
632: break; // never
633: }
634:
1.192 paf 635: char *dest=new_cstr;
1.181 paf 636: unsigned char index;
1.190 paf 637: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 638: unsigned char c=a[index];
639: if(b)
640: c=b[c];
641:
642: *dest++=(char)c;
643: }
1.174 paf 644: }
1.176 paf 645: result.langs=langs;
1.174 paf 646: result.body=new_cstr;
1.89 parser 647:
1.101 parser 648: return result;
649: }
650:
1.213 misha 651: const String& String::escape(Charset& source_charset) const {
652: if(is_empty())
653: return *this;
654:
655: return Charset::escape(*this, source_charset);
656: }
657:
1.174 paf 658: const String& String::replace(const Dictionary& dict) const {
659: String& result=*new String();
660: const char* old_cstr=cstr();
661: const char* prematch_begin=old_cstr;
662:
663: const char* current=old_cstr;
664: while(*current) {
1.184 paf 665: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
1.174 paf 666: // prematch
667: if(size_t prematch_length=current-prematch_begin) {
1.179 paf 668: result.langs.append(result.body, langs, prematch_begin-old_cstr, prematch_length);
1.174 paf 669: result.body.append_strdup_know_length(prematch_begin, prematch_length);
1.101 parser 670: }
671:
1.174 paf 672: // match
673: // skip 'a' in 'current'; move prematch_begin
1.184 paf 674: current+=subst.from_length; prematch_begin=current;
1.174 paf 675:
1.184 paf 676: if(const String* b=subst.to) // are there any b?
1.174 paf 677: result<<*b;
678: } else // simply advance
679: current++;
680: }
1.156 paf 681:
1.174 paf 682: // postmatch
683: if(size_t postmatch_length=current-prematch_begin) {
1.179 paf 684: result.langs.append(result.body, langs, prematch_begin-old_cstr, postmatch_length);
1.174 paf 685: result.body.append_strdup_know_length(prematch_begin, postmatch_length);
686: }
1.156 paf 687:
1.174 paf 688: ASSERT_STRING_INVARIANT(result);
1.82 parser 689: return result;
1.61 paf 690: }
1.113 parser 691:
1.180 paf 692: static int serialize_body_char(char c, char** cur) {
693: *((*cur)++)=c;
694: return 0; // 0=continue
695: };
1.174 paf 696: static int serialize_body_piece(const char* s, char** cur) {
697: size_t length=strlen(s);
698: memcpy(*cur, s, length); *cur+=length;
1.178 paf 699: return 0; // 0=continue
1.174 paf 700: };
1.178 paf 701: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
702: // lang
1.191 paf 703: **cur=alang; (*cur)++;
704: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 705: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
706:
707: return 0; // 0=continue
708: }
1.174 paf 709: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 710: size_t fragments_count=langs.count();
1.202 paf 711: size_t body_length=body.length();
1.174 paf 712: size_t buf_length=
1.178 paf 713: prolog_length //1
714: +sizeof(size_t) //2
1.202 paf 715: +body_length //3
716: +1 // 4 for zero terminator used in deserialize
717: +sizeof(size_t) //5
718: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
719:
1.174 paf 720: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
721:
722: // 1: prolog
723: char *cur=result.str+prolog_length;
1.202 paf 724: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
725: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
726: // 3: letters
727: body.for_each(serialize_body_char, serialize_body_piece, &cur);
728: // 4: zero terminator
729: *cur++=0;
730: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 731: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 732: // 6: lang info
1.178 paf 733: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 734:
1.174 paf 735: return result;
1.113 parser 736: }
1.202 paf 737: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
738: size_t in_buf=buf_size;
739: if(in_buf<=prolog_size)
1.148 paf 740: return false;
1.202 paf 741: in_buf-=prolog_size;
1.135 paf 742:
1.174 paf 743: // 1: prolog
1.202 paf 744: const char* cur=(const char* )buf+prolog_size;
1.113 parser 745:
1.207 paf 746: // 2: chars.count
1.202 paf 747: size_t body_length;
748: if(in_buf<sizeof(body_length)) // body.length don't fit?
749: return false;
750: // [WARNING: not cast, addresses must be %4=0 on sparc]
751: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
752: in_buf-=sizeof(body_length);
753:
754: if(in_buf<body_length+1) // letters+terminator don't fit?
755: return false;
756: // 4: zero terminator
757: if(cur[body_length] != 0) // in place?
758: return false;
759: // 3: letters
760: body=String::Body(cur, body_length);
761: cur+=body_length+1;
762: in_buf-=body_length+1;
763:
764: // 5: langs.count
1.191 paf 765: size_t fragments_count;
1.202 paf 766: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 767: return false;
1.191 paf 768: // [WARNING: not cast, addresses must be %4=0 on sparc]
769: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 770: in_buf-=sizeof(fragments_count);
1.174 paf 771:
772: if(fragments_count) {
1.202 paf 773: // 6: lang info
1.174 paf 774: size_t total_length=0;
775: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 776: char lang;
777: size_t fragment_length;
778: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 779: if(in_buf<piece_length) // lang+length
1.174 paf 780: return false;
781:
1.191 paf 782: // lang
783: lang=*cur++;
784: // length [WARNING: not cast, addresses must be %4=0 on sparc]
785: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
786:
1.206 paf 787: size_t combined_length=total_length+fragment_length;
788: if(combined_length>body_length)
789: return false; // file curruption
1.191 paf 790: // uchar needed to prevent propagating 0x80 bit to upper bytes
791: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 792: total_length=combined_length;
1.202 paf 793: in_buf-=piece_length;
1.174 paf 794: }
1.128 paf 795:
1.202 paf 796: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 797: return false;
1.174 paf 798: }
1.202 paf 799: if(in_buf!=0) // some strange extra bytes
800: return false;
1.113 parser 801:
1.174 paf 802: ASSERT_STRING_INVARIANT(*this);
1.148 paf 803: return true;
1.176 paf 804: }
805:
806: const char* String::Body::v() const {
807: return CORD_to_const_char_star(body);
808: }
1.201 paf 809: void String::Body::dump() const {
810: CORD_dump(body);
811: }
812:
1.176 paf 813: const char* String::Languages::v() const {
1.177 paf 814: if(opt.is_not_just_lang)
1.176 paf 815: return CORD_to_const_char_star(langs);
816: else
817: return (const char*)&langs;
818: }
1.201 paf 819: void String::Languages::dump() const {
820: if(opt.is_not_just_lang)
821: CORD_dump(langs);
822: else
823: puts((const char*)&langs);
824: }
1.176 paf 825: const char* String::v() const {
1.198 paf 826: const uint LIMIT_VIEW=20;
1.176 paf 827: char* buf=(char*)malloc(MAX_STRING);
828: const char*body_view=body.v();
829: const char*langs_view=langs.v();
830: snprintf(buf, MAX_STRING,
1.178 paf 831: "%d:%.*s%s} "
1.176 paf 832: "{%d:%s",
1.178 paf 833: langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"",
1.176 paf 834: strlen(body_view), body_view
835: );
836:
837: return buf;
1.113 parser 838: }
1.201 paf 839: void String::dump() const {
840: body.dump();
841: langs.dump();
842: }
1.195 paf 843: const String& String::trim(String::Trim_kind kind, const char* chars) const {
844: if(!length())
845: return *this;
846:
847: size_t substr_begin, substr_length;
848: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length);
849: if(new_body==body) // we received unchanged pointer, do likewise
850: return *this;
851: // new_body differs from body, adjust langs along
852:
853: String& result=*new String;
854: if(!new_body) // body.trim produced empty result
855: return result;
856: // body.trim produced nonempty result
857:
858: // first: their langs
859: result.langs.append(result.body, langs, substr_begin, substr_length);
860: // next: letters themselves
861: result.body=new_body;
862:
863: ASSERT_STRING_INVARIANT(result);
864: return result;
1.198 paf 865: }
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