Annotation of parser3/src/main/pa_string.C, revision 1.220
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:
1.220 ! misha 8: static const char * const IDENT_STRING_C="$Date: 2009-04-10 11:33:16 $";
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:
1.220 ! 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:
1.210 misha 291: size_t String::length(Charset& charset) const {
292: if(charset.isUTF8()){
1.220 ! misha 293: size_t len=0;
! 294: body.for_each<size_t *>(NULL, CORD_batched_len, &len);
! 295: return len;
1.210 misha 296: } else
297: return body.length();
298: }
299:
1.174 paf 300: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
301: String& String::mid(size_t substr_begin, size_t substr_end) const {
302: String& result=*new String;
303:
304: size_t self_length=length();
305: substr_begin=min(substr_begin, self_length);
306: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 307: size_t substr_length=substr_end-substr_begin;
308: if(!substr_length)
1.107 parser 309: return result;
1.53 paf 310:
1.176 paf 311: // first: their langs
312: result.langs.append(result.body, langs, substr_begin, substr_length);
313: // next: letters themselves
314: result.body=body.mid(substr_begin, substr_length);
1.174 paf 315:
316: ASSERT_STRING_INVARIANT(result);
1.53 paf 317: return result;
1.54 paf 318: }
319:
1.211 misha 320: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
321: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 322: String& result=*new String;
323:
1.211 misha 324: size_t self_length=(helper_length)?helper_length:length(charset);
325:
326: if(!self_length)
327: return result;
328:
329: from=min(min(to, from), self_length);
1.210 misha 330: to=min(max(to, from), self_length);
1.211 misha 331:
1.210 misha 332: size_t substr_length=to-from;
1.211 misha 333:
1.210 misha 334: if(!substr_length)
335: return result;
336:
337: if(charset.isUTF8()){
338: const XMLByte* srcPtr=(const XMLByte*)cstrm();
1.211 misha 339: const XMLByte* srcEnd=srcPtr+body.length();
1.210 misha 340:
1.212 misha 341: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.210 misha 342: from=getUTF8BytePos(srcPtr, srcEnd, from);
343: substr_length=getUTF8BytePos(srcPtr+from, srcEnd, substr_length);
344: if(!substr_length)
345: return result;
346: }
347:
348: // first: their langs
349: result.langs.append(result.body, langs, from, substr_length);
350: // next: letters themselves
351: result.body=body.mid(from, substr_length);
352:
353: ASSERT_STRING_INVARIANT(result);
354: return result;
355: }
356:
1.176 paf 357: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 358: size_t substr_length=substr.length();
359: while(true) {
360: size_t substr_begin=body.pos(substr, this_offset);
361:
362: if(substr_begin==CORD_NOT_FOUND)
363: return STRING_NOT_FOUND;
1.174 paf 364:
1.183 paf 365: if(langs.check_lang(lang, substr_begin, substr_length))
366: return substr_begin;
367:
368: this_offset=substr_begin+substr_length;
369: }
1.58 paf 370: }
371:
1.174 paf 372: size_t String::pos(const String& substr,
373: size_t this_offset, Language lang) const {
374: return pos(substr.body, this_offset, lang);
1.60 paf 375: }
376:
1.210 misha 377: size_t String::pos(Charset& charset, const String& substr,
378: size_t this_offset, Language lang) const {
379:
1.212 misha 380: size_t result;
1.210 misha 381: if(charset.isUTF8()){
382: const XMLByte* srcPtr=(const XMLByte*)cstrm();
1.212 misha 383: const XMLByte* srcEnd=srcPtr+body.length();
384:
385: // convert 'this_offset' from 'characters' to 'bytes'
386: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
387:
388: result=pos(substr.body, this_offset, lang);
389: if(result==CORD_NOT_FOUND)
390: return STRING_NOT_FOUND;
391:
392: // convert 'result' from 'bytes' to 'characters'
393: result=getUTF8CharPos(srcPtr, srcEnd, result);
394: } else {
395: result=pos(substr.body, this_offset, lang);
396: if(result==CORD_NOT_FOUND)
397: return STRING_NOT_FOUND;
1.210 misha 398: }
399:
400: return result;
401: }
402:
1.174 paf 403: void String::split(ArrayString& result,
404: size_t& pos_after,
405: const char* delim,
406: Language lang, int limit) const {
407: size_t self_length=length();
408: if(size_t delim_length=strlen(delim)) {
1.186 paf 409: size_t pos_before;
1.60 paf 410: // while we have 'delim'...
1.174 paf 411: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 412: result+=&mid(pos_after, pos_before);
1.174 paf 413: pos_after=pos_before+delim_length;
1.60 paf 414: }
415: // last piece
1.174 paf 416: if(pos_after<self_length && limit) {
417: result+=&mid(pos_after, self_length);
418: pos_after=self_length;
1.60 paf 419: }
420: } else { // empty delim
421: result+=this;
1.174 paf 422: pos_after+=self_length;
1.60 paf 423: }
424: }
425:
1.174 paf 426: void String::split(ArrayString& result,
427: size_t& pos_after,
428: const String& delim, Language lang,
429: int limit) const {
1.140 paf 430: if(!delim.is_empty()) {
1.186 paf 431: size_t pos_before;
1.60 paf 432: // while we have 'delim'...
1.174 paf 433: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 434: result+=&mid(pos_after, pos_before);
1.174 paf 435: pos_after=pos_before+delim.length();
1.60 paf 436: }
437: // last piece
1.174 paf 438: if(pos_after<length() && limit) {
439: result+=&mid(pos_after, length());
440: pos_after=length();
1.60 paf 441: }
442: } else { // empty delim
443: result+=this;
1.174 paf 444: pos_after+=length();
1.60 paf 445: }
1.61 paf 446: }
447:
1.209 misha 448: enum Match_feature {
449: MF_NEED_PRE_POST_MATCH = 0x01,
450: MF_JUST_COUNT_MATCHES = 0x02
451: };
452:
453: static void regex_options(const String* options, int* result, int* match_features){
1.63 paf 454: struct Regex_option {
1.216 misha 455: const char* key;
456: const char* keyAlt;
1.209 misha 457: int clear;
458: int set;
1.63 paf 459: int *result;
1.209 misha 460: int flag;
1.63 paf 461: } regex_option[]={
1.189 paf 462: {"i", "I", 0, PCRE_CASELESS, result, 0}, // a=A
1.216 misha 463: {"s", "S", 0, PCRE_DOTALL, result, 0}, // ^\n\n$ [default]
1.189 paf 464: {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result, 0}, // ^aaa\n$^bbb\n$
1.216 misha 465: {"x", 0, 0, PCRE_EXTENDED, result, 0}, // whitespace in regex ignored
466: {"U", 0, 0, PCRE_UNGREEDY, result, 0}, // ungreedy patterns (greedy by default)
1.189 paf 467: {"g", "G", 0, 1, result+1, 0}, // many rows
1.209 misha 468: {"'", 0, 0, 0, 0, MF_NEED_PRE_POST_MATCH},
1.216 misha 469: {"n", 0, 0, 0, 0, MF_JUST_COUNT_MATCHES},
1.189 paf 470: {0, 0, 0, 0, 0, 0}
1.63 paf 471: };
1.171 paf 472: result[0]=PCRE_EXTRA | PCRE_DOTALL | PCRE_DOLLAR_ENDONLY;
1.63 paf 473: result[1]=0;
474:
1.174 paf 475: if(options && !options->is_empty())
1.216 misha 476: for(Regex_option *o=regex_option; o->key; o++)
1.209 misha 477: if(
1.216 misha 478: options->pos(o->key)!=STRING_NOT_FOUND
479: || (o->keyAlt && options->pos(o->keyAlt)!=STRING_NOT_FOUND)
1.209 misha 480: ){
481: if(o->flag){
482: (*match_features) |= o->flag;
483: } else {
1.154 paf 484: *o->result &= ~o->clear;
485: *o->result |= o->set;
486: }
1.63 paf 487: }
488: }
489:
1.174 paf 490: Table* String::match(Charset& source_charset,
491: const String& regexp,
492: const String* options,
493: Row_action row_action, void *info,
1.209 misha 494: int& matches_count) const {
1.140 paf 495: if(regexp.is_empty())
1.218 misha 496: throw Exception(PARSER_RUNTIME,
1.174 paf 497: 0,
1.73 paf 498: "regexp is empty");
1.154 paf 499:
1.205 paf 500: const char* pattern=regexp.cstr(String::L_UNSPECIFIED); // fix any tainted with L_REGEX
1.174 paf 501: const char* errptr;
1.62 paf 502: int erroffset;
1.209 misha 503: int option_bits[2]={0};
504: int match_features=0;
505: regex_options(options, option_bits, &match_features);
506: bool need_pre_post_match=(match_features & MF_NEED_PRE_POST_MATCH) != 0;
507: bool just_count_matches=(match_features & MF_JUST_COUNT_MATCHES) != 0;
1.174 paf 508: bool global=option_bits[1]!=0;
1.215 misha 509:
1.219 misha 510: if(source_charset.isUTF8()){
511: // @todo (for UTF-8): check string & pattern and use PCRE_NO_UTF8_CHECK option
512: option_bits[0]|=PCRE_UTF8;
513: }
1.215 misha 514:
515: pcre *code=pcre_compile(pattern, option_bits[0],
1.62 paf 516: &errptr, &erroffset,
1.174 paf 517: source_charset.pcre_tables);
1.62 paf 518:
1.67 paf 519: if(!code)
1.218 misha 520: throw Exception(PCRE_EXCEPTION_TYPE,
1.174 paf 521: ®exp.mid(erroffset, regexp.length()),
1.74 paf 522: "regular expression syntax error - %s", errptr);
1.62 paf 523:
1.215 misha 524: int subpatterns=pcre_info(code, 0, 0);
1.174 paf 525: if(subpatterns<0) {
1.215 misha 526: pcre_free(code);
1.218 misha 527: throw Exception(PCRE_EXCEPTION_TYPE,
1.174 paf 528: ®exp,
1.76 paf 529: "pcre_info error (%d)",
1.174 paf 530: subpatterns);
1.63 paf 531: }
532:
1.174 paf 533: const char* subject=cstr();
534: size_t subject_length=strlen(subject);
535: const int oveclength=(1/*match*/+MAX_MATCH_GROUPS)*3;
536: int ovector[oveclength];
1.155 paf 537:
538: // create table
1.173 paf 539: Table::Action_options table_options;
1.174 paf 540: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 541:
1.64 paf 542: int exec_option_bits=0;
1.154 paf 543: int prestart=0;
544: int poststart=0;
1.174 paf 545: int postfinish=length();
1.63 paf 546: while(true) {
1.215 misha 547: int exec_substrings=pcre_exec(code, 0,
1.174 paf 548: subject, subject_length, prestart,
549: exec_option_bits, ovector, oveclength);
1.63 paf 550:
551: if(exec_substrings==PCRE_ERROR_NOMATCH) {
1.215 misha 552: pcre_free(code);
1.174 paf 553: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
1.209 misha 554: return just_count_matches ? 0 : &table;
1.63 paf 555: }
556:
557: if(exec_substrings<0) {
1.215 misha 558: pcre_free(code);
1.218 misha 559: throw Exception(PCRE_EXCEPTION_TYPE,
1.174 paf 560: ®exp,
1.217 misha 561: print_pcre_exec_error_text(exec_substrings),
1.63 paf 562: exec_substrings);
563: }
564:
1.154 paf 565: int prefinish=ovector[0];
566: poststart=ovector[1];
1.174 paf 567: ArrayString* row=new ArrayString;
568: if(need_pre_post_match) {
569: *row+=&mid(0, prefinish); // .prematch column value
570: *row+=&mid(prefinish, poststart); // .match
571: *row+=&mid(poststart, postfinish); // .postmatch
572: } else {
1.185 paf 573: *row+=&Empty; // .prematch column value
574: *row+=&Empty; // .match
575: *row+=&Empty; // .postmatch
1.174 paf 576: }
1.63 paf 577:
578: for(int i=1; i<exec_substrings; i++) {
1.69 paf 579: // -1:-1 case handled peacefully by mid() itself
1.174 paf 580: *row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value
1.63 paf 581: }
582:
1.209 misha 583: matches_count++;
1.174 paf 584: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 585:
1.219 misha 586: if(!global || prestart==poststart) { // last step
1.215 misha 587: pcre_free(code);
1.174 paf 588: row_action(table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info);
1.209 misha 589: return just_count_matches ? 0 : &table;
1.63 paf 590: }
1.154 paf 591: prestart=poststart;
1.63 paf 592:
593: /*
594: if(option_bits[0] & PCRE_MULTILINE)
1.64 paf 595: exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL
1.63 paf 596: */
597: }
1.82 parser 598: }
599:
1.174 paf 600: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
601: String& result=*new String();
602: if(is_empty())
603: return result;
604:
605: char* new_cstr=cstrm();
1.216 misha 606:
1.181 paf 607: if(source_charset.isUTF8()) {
1.216 misha 608: size_t new_cstr_len=length();
1.181 paf 609: switch(kind) {
610: case CC_UPPER:
1.192 paf 611: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 612: break;
613: case CC_LOWER:
1.192 paf 614: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 615: break;
616: default:
617: assert(!"unknown change case kind");
618: break; // never
619: }
620:
621: } else {
622: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 623:
1.181 paf 624: const unsigned char *a;
625: const unsigned char *b;
626: switch(kind) {
627: case CC_UPPER:
628: a=tables+lcc_offset;
629: b=tables+fcc_offset;
630: break;
631: case CC_LOWER:
632: a=tables+lcc_offset;
633: b=0;
634: break;
635: default:
636: assert(!"unknown change case kind");
637: a=b=0; // calm, compiler
638: break; // never
639: }
640:
1.192 paf 641: char *dest=new_cstr;
1.181 paf 642: unsigned char index;
1.190 paf 643: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 644: unsigned char c=a[index];
645: if(b)
646: c=b[c];
647:
648: *dest++=(char)c;
649: }
1.174 paf 650: }
1.176 paf 651: result.langs=langs;
1.174 paf 652: result.body=new_cstr;
1.89 parser 653:
1.101 parser 654: return result;
655: }
656:
1.213 misha 657: const String& String::escape(Charset& source_charset) const {
658: if(is_empty())
659: return *this;
660:
661: return Charset::escape(*this, source_charset);
662: }
663:
1.174 paf 664: const String& String::replace(const Dictionary& dict) const {
665: String& result=*new String();
666: const char* old_cstr=cstr();
667: const char* prematch_begin=old_cstr;
668:
669: const char* current=old_cstr;
670: while(*current) {
1.184 paf 671: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
1.174 paf 672: // prematch
673: if(size_t prematch_length=current-prematch_begin) {
1.179 paf 674: result.langs.append(result.body, langs, prematch_begin-old_cstr, prematch_length);
1.174 paf 675: result.body.append_strdup_know_length(prematch_begin, prematch_length);
1.101 parser 676: }
677:
1.174 paf 678: // match
679: // skip 'a' in 'current'; move prematch_begin
1.184 paf 680: current+=subst.from_length; prematch_begin=current;
1.174 paf 681:
1.184 paf 682: if(const String* b=subst.to) // are there any b?
1.174 paf 683: result<<*b;
684: } else // simply advance
685: current++;
686: }
1.156 paf 687:
1.174 paf 688: // postmatch
689: if(size_t postmatch_length=current-prematch_begin) {
1.179 paf 690: result.langs.append(result.body, langs, prematch_begin-old_cstr, postmatch_length);
1.174 paf 691: result.body.append_strdup_know_length(prematch_begin, postmatch_length);
692: }
1.156 paf 693:
1.174 paf 694: ASSERT_STRING_INVARIANT(result);
1.82 parser 695: return result;
1.61 paf 696: }
1.113 parser 697:
1.180 paf 698: static int serialize_body_char(char c, char** cur) {
699: *((*cur)++)=c;
700: return 0; // 0=continue
701: };
1.174 paf 702: static int serialize_body_piece(const char* s, char** cur) {
703: size_t length=strlen(s);
704: memcpy(*cur, s, length); *cur+=length;
1.178 paf 705: return 0; // 0=continue
1.174 paf 706: };
1.178 paf 707: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
708: // lang
1.191 paf 709: **cur=alang; (*cur)++;
710: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 711: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
712:
713: return 0; // 0=continue
714: }
1.174 paf 715: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 716: size_t fragments_count=langs.count();
1.202 paf 717: size_t body_length=body.length();
1.174 paf 718: size_t buf_length=
1.178 paf 719: prolog_length //1
720: +sizeof(size_t) //2
1.202 paf 721: +body_length //3
722: +1 // 4 for zero terminator used in deserialize
723: +sizeof(size_t) //5
724: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
725:
1.174 paf 726: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
727:
728: // 1: prolog
729: char *cur=result.str+prolog_length;
1.202 paf 730: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
731: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
732: // 3: letters
733: body.for_each(serialize_body_char, serialize_body_piece, &cur);
734: // 4: zero terminator
735: *cur++=0;
736: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 737: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 738: // 6: lang info
1.178 paf 739: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 740:
1.174 paf 741: return result;
1.113 parser 742: }
1.202 paf 743: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
744: size_t in_buf=buf_size;
745: if(in_buf<=prolog_size)
1.148 paf 746: return false;
1.202 paf 747: in_buf-=prolog_size;
1.135 paf 748:
1.174 paf 749: // 1: prolog
1.202 paf 750: const char* cur=(const char* )buf+prolog_size;
1.113 parser 751:
1.207 paf 752: // 2: chars.count
1.202 paf 753: size_t body_length;
754: if(in_buf<sizeof(body_length)) // body.length don't fit?
755: return false;
756: // [WARNING: not cast, addresses must be %4=0 on sparc]
757: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
758: in_buf-=sizeof(body_length);
759:
760: if(in_buf<body_length+1) // letters+terminator don't fit?
761: return false;
762: // 4: zero terminator
763: if(cur[body_length] != 0) // in place?
764: return false;
765: // 3: letters
766: body=String::Body(cur, body_length);
767: cur+=body_length+1;
768: in_buf-=body_length+1;
769:
770: // 5: langs.count
1.191 paf 771: size_t fragments_count;
1.202 paf 772: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 773: return false;
1.191 paf 774: // [WARNING: not cast, addresses must be %4=0 on sparc]
775: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 776: in_buf-=sizeof(fragments_count);
1.174 paf 777:
778: if(fragments_count) {
1.202 paf 779: // 6: lang info
1.174 paf 780: size_t total_length=0;
781: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 782: char lang;
783: size_t fragment_length;
784: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 785: if(in_buf<piece_length) // lang+length
1.174 paf 786: return false;
787:
1.191 paf 788: // lang
789: lang=*cur++;
790: // length [WARNING: not cast, addresses must be %4=0 on sparc]
791: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
792:
1.206 paf 793: size_t combined_length=total_length+fragment_length;
794: if(combined_length>body_length)
795: return false; // file curruption
1.191 paf 796: // uchar needed to prevent propagating 0x80 bit to upper bytes
797: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 798: total_length=combined_length;
1.202 paf 799: in_buf-=piece_length;
1.174 paf 800: }
1.128 paf 801:
1.202 paf 802: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 803: return false;
1.174 paf 804: }
1.202 paf 805: if(in_buf!=0) // some strange extra bytes
806: return false;
1.113 parser 807:
1.174 paf 808: ASSERT_STRING_INVARIANT(*this);
1.148 paf 809: return true;
1.176 paf 810: }
811:
812: const char* String::Body::v() const {
813: return CORD_to_const_char_star(body);
814: }
1.201 paf 815: void String::Body::dump() const {
816: CORD_dump(body);
817: }
818:
1.176 paf 819: const char* String::Languages::v() const {
1.177 paf 820: if(opt.is_not_just_lang)
1.176 paf 821: return CORD_to_const_char_star(langs);
822: else
823: return (const char*)&langs;
824: }
1.201 paf 825: void String::Languages::dump() const {
826: if(opt.is_not_just_lang)
827: CORD_dump(langs);
828: else
829: puts((const char*)&langs);
830: }
1.176 paf 831: const char* String::v() const {
1.198 paf 832: const uint LIMIT_VIEW=20;
1.176 paf 833: char* buf=(char*)malloc(MAX_STRING);
834: const char*body_view=body.v();
835: const char*langs_view=langs.v();
836: snprintf(buf, MAX_STRING,
1.178 paf 837: "%d:%.*s%s} "
1.176 paf 838: "{%d:%s",
1.178 paf 839: langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"",
1.176 paf 840: strlen(body_view), body_view
841: );
842:
843: return buf;
1.113 parser 844: }
1.201 paf 845: void String::dump() const {
846: body.dump();
847: langs.dump();
848: }
1.195 paf 849: const String& String::trim(String::Trim_kind kind, const char* chars) const {
850: if(!length())
851: return *this;
852:
853: size_t substr_begin, substr_length;
854: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length);
855: if(new_body==body) // we received unchanged pointer, do likewise
856: return *this;
857: // new_body differs from body, adjust langs along
858:
859: String& result=*new String;
860: if(!new_body) // body.trim produced empty result
861: return result;
862: // body.trim produced nonempty result
863:
864: // first: their langs
865: result.langs.append(result.body, langs, substr_begin, substr_length);
866: // next: letters themselves
867: result.body=new_body;
868:
869: ASSERT_STRING_INVARIANT(result);
870: return result;
1.198 paf 871: }
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