Annotation of parser3/src/main/pa_common.C, revision 1.271
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.267 moko 4: Copyright (c) 2000-2012 Art. Lebedev Studio (http://www.artlebedev.com)
1.101 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.16 paf 6:
1.210 paf 7: * BASE64 part
8: * Authors: Michael Zucchi <notzed@ximian.com>
9: * Jeffrey Stedfast <fejj@ximian.com>
10: *
11: * Copyright 2000-2004 Ximian, Inc. (www.ximian.com)
12: *
13: * This program is free software; you can redistribute it and/or modify
14: * it under the terms of the GNU General Public License as published by
15: * the Free Software Foundation; either version 2 of the License, or
16: * (at your option) any later version.
17: *
18: * This program is distributed in the hope that it will be useful,
19: * but WITHOUT ANY WARRANTY; without even the implied warranty of
20: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
21: * GNU General Public License for more details.
22: *
23: * You should have received a copy of the GNU General Public License
24: * along with this program; if not, write to the Free Software
25: * Foundation, Inc., 59 Temple Street #330, Boston, MA 02111-1307, USA.
26: *
27: */
28:
1.1 paf 29: #include "pa_common.h"
1.4 paf 30: #include "pa_exception.h"
1.154 paf 31: #include "pa_hash.h"
1.14 paf 32: #include "pa_globals.h"
1.154 paf 33: #include "pa_charsets.h"
1.214 paf 34: #include "pa_http.h"
1.223 misha 35: #include "pa_request_charsets.h"
1.237 misha 36: #include "pcre.h"
1.241 misha 37: #include "pa_request.h"
1.98 paf 38:
1.271 ! moko 39: volatile const char * IDENT_PA_COMMON_C="$Id: pa_common.C,v 1.270 2013-04-21 21:59:07 moko Exp $" IDENT_PA_COMMON_H IDENT_PA_HASH_H IDENT_PA_ARRAY_H IDENT_PA_STACK_H;
1.267 moko 40:
1.93 paf 41: // some maybe-undefined constants
42:
1.82 paf 43: #ifndef _O_TEXT
44: # define _O_TEXT 0
45: #endif
46: #ifndef _O_BINARY
47: # define _O_BINARY 0
1.47 paf 48: #endif
1.80 paf 49:
1.138 paf 50: #ifdef HAVE_FTRUNCATE
51: # define PA_O_TRUNC 0
52: #else
53: # ifdef _O_TRUNC
54: # define PA_O_TRUNC _O_TRUNC
55: # else
56: # error you must have either ftruncate function or _O_TRUNC bit declared
57: # endif
1.154 paf 58: #endif
1.176 paf 59:
1.154 paf 60: // defines for globals
61:
62: #define FILE_STATUS_NAME "status"
63:
64: // globals
65:
66: const String file_status_name(FILE_STATUS_NAME);
67:
68: // functions
1.127 paf 69:
1.255 misha 70: bool capitalized(const char* s){
71: bool upper=true;
72: for(const char* c=s; *c; c++){
73: if(*c != (upper ? toupper((unsigned char)*c) : tolower((unsigned char)*c)))
74: return false;
75: upper=strchr("-_ ", *c) != 0;
76: }
77: return true;
78: }
79:
1.249 misha 80: const char* capitalize(const char* s){
1.255 misha 81: if(!s || capitalized(s))
82: return s;
83:
1.249 misha 84: char* result=pa_strdup(s);
85: if(result){
86: bool upper=true;
87: for(char* c=result; *c; c++){
88: *c=upper ? (char)toupper((unsigned char)*c) : (char)tolower((unsigned char)*c);
89: upper=strchr("-_ ", *c) != 0;
90: }
91: }
92: return (const char*)result;
93: }
94:
1.154 paf 95: void fix_line_breaks(char *str, size_t& length) {
1.87 paf 96: //_asm int 3;
1.154 paf 97: const char* const eob=str+length;
98: char* dest=str;
1.72 parser 99: // fix DOS: \r\n -> \n
100: // fix Macintosh: \r -> \n
1.154 paf 101: char* bol=str;
1.137 paf 102: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 103: size_t len=eol-bol;
104: if(dest!=bol)
1.260 misha 105: memmove(dest, bol, len);
1.72 parser 106: dest+=len;
1.126 paf 107: *dest++='\n';
1.72 parser 108:
1.126 paf 109: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 110: bol=eol+2;
1.154 paf 111: length--;
1.126 paf 112: } else // \r, not \n = Macintosh
1.72 parser 113: bol=eol+1;
114: }
1.154 paf 115: // last piece without \r
1.72 parser 116: if(dest!=bol)
1.260 misha 117: memmove(dest, bol, eob-bol);
1.154 paf 118: str[length]=0; // terminating
1.72 parser 119: }
1.18 paf 120:
1.271 ! moko 121: char* file_read_text(Request_charsets& charsets, const String& file_spec, bool fail_on_read_problem, HashStringValue* params, bool transcode_result) {
! 122: File_read_result file=file_read(charsets, file_spec, true, params, fail_on_read_problem, 0, 0, 0, transcode_result);
1.154 paf 123: return file.success?file.str:0;
1.126 paf 124: }
125:
1.271 ! moko 126: char* file_load_text(Request& r, const String& file_spec, bool fail_on_read_problem, HashStringValue* params, bool transcode_result) {
! 127: File_read_result file=file_load(r, file_spec, true, params, fail_on_read_problem, 0, 0, 0, transcode_result);
1.241 misha 128: return file.success?file.str:0;
129: }
130:
1.206 paf 131: /// these options were handled but not checked elsewhere, now check them
1.239 misha 132: int pa_get_valid_file_options_count(HashStringValue& options) {
1.206 paf 133: int result=0;
134: if(options.get(PA_SQL_LIMIT_NAME))
135: result++;
136: if(options.get(PA_SQL_OFFSET_NAME))
137: result++;
138: if(options.get(PA_COLUMN_SEPARATOR_NAME))
139: result++;
140: if(options.get(PA_COLUMN_ENCLOSER_NAME))
141: result++;
1.223 misha 142: if(options.get(PA_CHARSET_NAME))
143: result++;
1.206 paf 144: return result;
145: }
146:
1.123 paf 147: #ifndef DOXYGEN
148: struct File_read_action_info {
1.154 paf 149: char **data; size_t *data_size;
1.188 paf 150: char* buf; size_t offset; size_t count;
1.126 paf 151: };
1.123 paf 152: #endif
1.271 ! moko 153:
! 154: static void file_read_action(struct stat& finfo, int f, const String& file_spec, const char* /*fname*/, bool as_text, void *context) {
1.126 paf 155: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.188 paf 156: size_t to_read_size=info.count;
157: if(!to_read_size)
158: to_read_size=(size_t)finfo.st_size;
159: assert( !(info.buf && as_text) );
1.271 ! moko 160: if(to_read_size) {
1.188 paf 161: if(info.offset)
162: lseek(f, info.offset, SEEK_SET);
1.271 ! moko 163: *info.data=info.buf ? info.buf : (char *)pa_malloc_atomic(to_read_size+1);
! 164: ssize_t result=read(f, *info.data, to_read_size);
! 165: if(result<0)
! 166: throw Exception("file.read", &file_spec, "read failed: %s (%d)", strerror(errno), errno);
! 167: *info.data_size=result;
1.123 paf 168: } else { // empty file
1.209 paf 169: // for both, text and binary: for text we need that terminator, for binary we need nonzero pointer to be able to save such files
1.253 misha 170: *info.data=(char *)pa_malloc_atomic(1);
1.209 paf 171: *(char*)(*info.data)=0;
1.123 paf 172: *info.data_size=0;
173: return;
174: }
1.126 paf 175: }
1.241 misha 176:
1.154 paf 177: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
1.229 misha 178: bool as_text, HashStringValue *params,
179: bool fail_on_read_problem,
1.234 misha 180: char* buf, size_t offset, size_t count, bool transcode_text_result) {
1.167 paf 181: File_read_result result={false, 0, 0, 0};
1.241 misha 182: if(params){
183: int valid_options=pa_get_valid_file_options_count(*params);
184: if(valid_options!=params->count())
1.262 misha 185: throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION);
1.241 misha 186: }
1.203 paf 187:
1.241 misha 188: File_read_action_info info={&result.str, &result.length, buf, offset, count};
1.161 paf 189:
1.241 misha 190: result.success=file_read_action_under_lock(file_spec,
191: "read", file_read_action, &info,
192: as_text, fail_on_read_problem);
1.223 misha 193:
1.241 misha 194: if(as_text){
195: if(result.success){
1.263 misha 196: Charset* asked_charset=0;
1.236 misha 197: if(result.length>=3 && strncmp(result.str, "\xEF\xBB\xBF", 3)==0){
1.240 misha 198: // skip UTF-8 signature (BOM code)
1.236 misha 199: result.str+=3;
200: result.length-=3;
1.263 misha 201: asked_charset=&UTF8_charset;
1.236 misha 202: }
203:
1.263 misha 204: if(params)
205: if(Value* vcharset_name=params->get(PA_CHARSET_NAME))
206: asked_charset=&::charsets.get(vcharset_name->as_string().change_case(charsets.source(), String::CC_UPPER));
207:
208: if(result.length && transcode_text_result && asked_charset){ // length must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below
209: String::C body=String::C(result.str, result.length);
210: body=Charset::transcode(body, *asked_charset, charsets.source());
1.236 misha 211:
1.263 misha 212: result.str=const_cast<char*>(body.str); // hacking a little
213: result.length=body.length;
1.131 paf 214: }
215: }
1.241 misha 216: if(result.length)
217: fix_line_breaks(result.str, result.length);
1.123 paf 218: }
1.241 misha 219:
220: return result;
221: }
222:
223: File_read_result file_load(Request& r, const String& file_spec,
224: bool as_text, HashStringValue *params,
225: bool fail_on_read_problem,
226: char* buf, size_t offset, size_t count, bool transcode_text_result) {
227:
228: File_read_result result={false, 0, 0, 0};
229: if(file_spec.starts_with("http://")) {
230: if(offset || count)
231: throw Exception(PARSER_RUNTIME,
232: 0,
233: "offset and load options are not supported for HTTP:// file load");
234:
235: // fail on read problem
236: File_read_http_result http=pa_internal_file_read_http(r, file_spec, as_text, params, transcode_text_result);
237: result.success=true;
238: result.str=http.str;
239: result.length=http.length;
240: result.headers=http.headers;
241: } else
242: result=
243: file_read(r.charsets, file_spec, as_text, params, fail_on_read_problem, buf, offset, count, transcode_text_result);
1.126 paf 244:
245: return result;
1.123 paf 246: }
247:
1.257 pretende 248:
1.154 paf 249: #ifdef PA_SAFE_MODE
1.259 misha 250: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
1.154 paf 251: if(finfo.st_uid/*foreign?*/!=geteuid()
252: && finfo.st_gid/*foreign?*/!=getegid())
1.224 misha 253: throw Exception(PARSER_RUNTIME,
1.154 paf 254: &file_spec,
255: "parser is in safe mode: "
256: "reading files of foreign group and user disabled "
257: "[recompile parser with --disable-safe-mode configure option], "
258: "actual filename '%s', "
259: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
260: fname,
261: finfo.st_uid, geteuid(),
1.259 misha 262: finfo.st_gid, getegid());
263: }
264: #else
265: void check_safe_mode(struct stat, const String&, const char*) {
266: }
1.257 pretende 267: #endif
1.259 misha 268:
1.257 pretende 269:
1.149 paf 270:
1.154 paf 271: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 272: const char* action_name, File_read_action action, void *context,
273: bool as_text,
1.123 paf 274: bool fail_on_read_problem) {
1.247 misha 275: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.33 paf 276: int f;
277:
278: // first open, next stat:
1.45 paf 279: // directory update of NTFS hard links performed on open.
1.33 paf 280: // ex:
281: // a.html:^test[] and b.html hardlink to a.html
282: // user inserts ! before ^test in a.html
1.126 paf 283: // directory entry of b.html in NTFS not updated at once,
1.35 paf 284: // they delay update till open, so we would receive "!^test[" string
285: // if would do stat, next open.
1.123 paf 286: // later: it seems, even this does not help sometimes
1.229 misha 287: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 288: try {
1.162 paf 289: if(pa_lock_shared_blocking(f)!=0)
1.126 paf 290: throw Exception("file.lock",
1.123 paf 291: &file_spec,
292: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 293: strerror(errno), errno, fname);
1.123 paf 294:
1.124 paf 295: struct stat finfo;
1.254 misha 296: if(fstat(f, &finfo)!=0)
1.124 paf 297: throw Exception("file.missing", // hardly possible: we just opened it OK
298: &file_spec,
299: "stat failed: %s (%d), actual filename '%s'",
1.154 paf 300: strerror(errno), errno, fname);
1.124 paf 301:
1.149 paf 302: check_safe_mode(finfo, file_spec, fname);
1.32 paf 303:
1.154 paf 304: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 305: } catch(...) {
1.162 paf 306: pa_unlock(f);close(f);
1.123 paf 307: if(fail_on_read_problem)
1.154 paf 308: rethrow;
1.123 paf 309: return false;
310: }
1.87 paf 311:
1.162 paf 312: pa_unlock(f);close(f);
1.72 parser 313: return true;
1.229 misha 314: } else {
1.118 paf 315: if(fail_on_read_problem)
1.256 misha 316: throw Exception(errno==EACCES?"file.access"
317: :(errno==ENOENT || errno==ENOTDIR || errno==ENODEV)?"file.missing":0,
1.118 paf 318: &file_spec,
1.123 paf 319: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 320: action_name, strerror(errno), errno, fname);
1.118 paf 321: return false;
322: }
1.8 paf 323: }
324:
1.202 paf 325: void create_dir_for_file(const String& file_spec) {
1.63 parser 326: size_t pos_after=1;
1.154 paf 327: size_t pos_before;
328: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
1.247 misha 329: mkdir(file_spec.mid(0, pos_before).taint_cstr(String::L_FILE_SPEC), 0775);
1.63 parser 330: pos_after=pos_before+1;
331: }
332: }
333:
1.98 paf 334: bool file_write_action_under_lock(
1.28 paf 335: const String& file_spec,
1.225 misha 336: const char* action_name,
337: File_write_action action,
338: void *context,
1.126 paf 339: bool as_text,
340: bool do_append,
341: bool do_block,
1.110 paf 342: bool fail_on_lock_problem) {
1.247 misha 343: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.28 paf 344: int f;
1.80 paf 345: if(access(fname, W_OK)!=0) // no
1.126 paf 346: create_dir_for_file(file_spec);
1.50 paf 347:
1.80 paf 348: if((f=open(fname,
349: O_CREAT|O_RDWR
350: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 351: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 352: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.126 paf 353: Exception e("file.lock",
1.110 paf 354: &file_spec,
355: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 356: strerror(errno), errno, fname);
1.126 paf 357: close(f);
1.110 paf 358: if(fail_on_lock_problem)
359: throw e;
1.98 paf 360: return false;
361: }
1.96 paf 362:
1.158 paf 363: try {
1.254 misha 364: #if (defined(HAVE_FCHMOD) && defined(PA_SAFE_MODE))
365: struct stat finfo;
366: if(fstat(f, &finfo)==0 && finfo.st_mode & 0111)
367: fchmod(f, finfo.st_mode & 0666/*clear executable bits*/); // backward: ignore errors if any
368: #endif
369: action(f, context);
1.158 paf 370: } catch(...) {
1.138 paf 371: #ifdef HAVE_FTRUNCATE
1.104 paf 372: if(!do_append)
1.125 paf 373: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 374: #endif
1.162 paf 375: pa_unlock(f);close(f);
1.154 paf 376: rethrow;
1.158 paf 377: }
1.80 paf 378:
1.138 paf 379: #ifdef HAVE_FTRUNCATE
1.104 paf 380: if(!do_append)
1.125 paf 381: ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov <virtan@rotabanner.com> for discovering]
1.138 paf 382: #endif
1.162 paf 383: pa_unlock(f);close(f);
1.98 paf 384: return true;
1.80 paf 385: } else
1.126 paf 386: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 387: &file_spec,
1.96 paf 388: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 389: action_name, strerror(errno), errno, fname);
1.96 paf 390: // here should be nothing, see rethrow above
391: }
392:
393: #ifndef DOXYGEN
394: struct File_write_action_info {
1.250 misha 395: const char* str;
396: size_t length;
1.126 paf 397: };
1.96 paf 398: #endif
1.271 ! moko 399:
1.96 paf 400: static void file_write_action(int f, void *context) {
1.126 paf 401: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 402: if(info.length) {
1.271 ! moko 403: ssize_t written=write(f, info.str, info.length);
1.116 paf 404: if(written<0)
1.271 ! moko 405: throw Exception("file.write", 0, "write failed: %s (%d)", strerror(errno), errno);
! 406: if(written!=info.length)
! 407: throw Exception("file.write", 0, "write failed: %u of %u bytes written", written, info.length);
1.113 paf 408: }
1.96 paf 409: }
1.271 ! moko 410:
1.96 paf 411: void file_write(
1.250 misha 412: Request_charsets& charsets,
413: const String& file_spec,
414: const char* data,
415: size_t size,
1.126 paf 416: bool as_text,
1.250 misha 417: bool do_append,
418: Charset* asked_charset) {
419:
420: if(as_text && asked_charset){
421: String::C body=String::C(data, size);
422: body=Charset::transcode(body, charsets.source(), *asked_charset);
423: data=body.str;
424: size=body.length;
425: };
426:
1.126 paf 427: File_write_action_info info={data, size};
1.225 misha 428:
1.98 paf 429: file_write_action_under_lock(
1.154 paf 430: file_spec,
1.225 misha 431: "write",
432: file_write_action,
433: &info,
1.154 paf 434: as_text,
435: do_append);
1.30 paf 436: }
437:
1.261 misha 438: static size_t get_dir(char* fname, size_t helper_length){
439: bool dir=false;
440: size_t pos=0;
441: for(pos=helper_length; pos; pos--){
442: char c=fname[pos-1];
443: if(c=='/' || c=='\\'){
444: fname[pos-1]=0;
445: dir=true;
446: } else if(dir) break;
447: }
448: return pos;
449: }
450:
451: static bool entry_readable(char* fname, bool need_dir) {
452: if(need_dir){
453: size_t size=strlen(fname);
454: while(size) {
455: char c=fname[size-1];
456: if(c=='/' || c=='\\')
457: fname[--size]=0;
458: else
459: break;
460: }
461: }
462:
463: struct stat finfo;
464: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
465: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
466: return is_dir==need_dir;
467: }
468: return false;
469: }
470:
471: static bool entry_readable(const String& file_spec, bool need_dir) {
472: return entry_readable(file_spec.taint_cstrm(String::L_FILE_SPEC), need_dir);
473: }
474:
1.63 parser 475: // throws nothing! [this is required in file_move & file_delete]
1.261 misha 476: static void rmdir(const String& file_spec, size_t pos_after=0) {
477: char* dir_spec=file_spec.taint_cstrm(String::L_FILE_SPEC);
478: size_t length=strlen(dir_spec);
479: while( (length=get_dir(dir_spec, length)) && (length > pos_after) ){
480: #ifdef WIN32
481: if(!entry_readable(dir_spec, true))
482: break;
483: DWORD attrs=GetFileAttributes(dir_spec);
484: if(
485: (attrs==INVALID_FILE_ATTRIBUTES)
486: || !(attrs & FILE_ATTRIBUTE_DIRECTORY)
487: || (attrs & FILE_ATTRIBUTE_REPARSE_POINT)
488: )
489: break;
490: #endif
491: if( rmdir(dir_spec) )
492: break;
493: };
1.50 paf 494: }
1.239 misha 495:
1.269 misha 496: bool file_delete(const String& file_spec, bool fail_on_problem, bool keep_empty_dirs) {
1.247 misha 497: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.54 parser 498: if(unlink(fname)!=0)
1.164 paf 499: if(fail_on_problem)
1.126 paf 500: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 501: &file_spec,
502: "unlink failed: %s (%d), actual filename '%s'",
1.154 paf 503: strerror(errno), errno, fname);
1.93 paf 504: else
505: return false;
1.50 paf 506:
1.269 misha 507: if(!keep_empty_dirs)
508: rmdir(file_spec, 1);
509:
1.93 paf 510: return true;
1.60 parser 511: }
1.239 misha 512:
1.269 misha 513: void file_move(const String& old_spec, const String& new_spec, bool keep_empty_dirs) {
1.247 misha 514: const char* old_spec_cstr=old_spec.taint_cstr(String::L_FILE_SPEC);
515: const char* new_spec_cstr=new_spec.taint_cstr(String::L_FILE_SPEC);
1.63 parser 516:
1.126 paf 517: create_dir_for_file(new_spec);
1.63 parser 518:
1.60 parser 519: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 520: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 521: &old_spec,
522: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 paf 523: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 524:
1.269 misha 525: if(!keep_empty_dirs)
526: rmdir(old_spec, 1);
1.31 paf 527: }
528:
1.51 paf 529:
1.126 paf 530: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 531: struct stat lfinfo;
532: bool result=stat(fname, &lfinfo)==0;
533: if(afinfo)
534: *afinfo=lfinfo;
535: return result;
1.119 paf 536: }
537:
538: bool entry_exists(const String& file_spec) {
1.247 misha 539: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.126 paf 540: return entry_exists(fname, 0);
1.118 paf 541: }
542:
1.215 paf 543: bool file_exist(const String& file_spec) {
1.126 paf 544: return entry_readable(file_spec, false);
1.51 paf 545: }
1.239 misha 546:
1.215 paf 547: bool dir_exists(const String& file_spec) {
1.126 paf 548: return entry_readable(file_spec, true);
1.65 parser 549: }
1.239 misha 550:
1.215 paf 551: const String* file_exist(const String& path, const String& name) {
1.154 paf 552: String& result=*new String(path);
1.270 moko 553: if(path.last_char() != '/')
554: result << "/";
1.154 paf 555: result << name;
1.215 paf 556: return file_exist(result)?&result:0;
1.43 paf 557: }
1.239 misha 558:
1.43 paf 559: bool file_executable(const String& file_spec) {
1.247 misha 560: return access(file_spec.taint_cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 561: }
562:
1.64 parser 563: bool file_stat(const String& file_spec,
1.229 misha 564: size_t& rsize,
565: time_t& ratime,
566: time_t& rmtime,
567: time_t& rctime,
568: bool fail_on_read_problem) {
1.247 misha 569: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.154 paf 570: struct stat finfo;
1.44 paf 571: if(stat(fname, &finfo)!=0)
1.64 parser 572: if(fail_on_read_problem)
1.126 paf 573: throw Exception("file.missing",
1.67 parser 574: &file_spec,
575: "getting file size failed: %s (%d), real filename '%s'",
1.154 paf 576: strerror(errno), errno, fname);
1.64 parser 577: else
578: return false;
1.58 parser 579: rsize=finfo.st_size;
580: ratime=finfo.st_atime;
581: rmtime=finfo.st_mtime;
582: rctime=finfo.st_ctime;
1.64 parser 583: return true;
1.18 paf 584: }
585:
1.126 paf 586: char* getrow(char* *row_ref, char delim) {
1.229 misha 587: char* result=*row_ref;
588: if(result) {
1.126 paf 589: *row_ref=strchr(result, delim);
1.8 paf 590: if(*row_ref)
591: *((*row_ref)++)=0;
592: else if(!*result)
593: return 0;
1.229 misha 594: }
595: return result;
1.8 paf 596: }
597:
1.126 paf 598: char* lsplit(char* string, char delim) {
1.229 misha 599: if(string) {
1.126 paf 600: char* v=strchr(string, delim);
1.8 paf 601: if(v) {
602: *v=0;
603: return v+1;
604: }
1.229 misha 605: }
606: return 0;
1.8 paf 607: }
608:
1.126 paf 609: char* lsplit(char* *string_ref, char delim) {
1.229 misha 610: char* result=*string_ref;
1.126 paf 611: char* next=lsplit(*string_ref, delim);
1.229 misha 612: *string_ref=next;
613: return result;
1.9 paf 614: }
615:
1.126 paf 616: char* rsplit(char* string, char delim) {
1.229 misha 617: if(string) {
1.126 paf 618: char* v=strrchr(string, delim);
1.18 paf 619: if(v) {
1.9 paf 620: *v=0;
621: return v+1;
622: }
1.229 misha 623: }
624: return NULL;
1.10 paf 625: }
626:
1.229 misha 627:
628: // format: %[flags][width][.precision]type http://msdn.microsoft.com/ru-ru/library/56e442dc(en-us,VS.80).aspx
629: // flags: '-', '+', ' ', '#', '0' http://msdn.microsoft.com/ru-ru/library/8aky45ct(en-us,VS.80).aspx
630: // width, precision: non negative decimal number
631: enum FormatType {
632: FormatInvalid,
633: FormatInt,
634: FormatUInt,
635: FormatDouble
636: };
637: FormatType format_type(char* fmt){
638: enum FormatState {
639: Percent,
640: Flags,
641: Width,
642: Precision,
643: Done
644: } state=Percent;
645:
646: FormatType result=FormatInvalid;
647:
648: char* pos=fmt;
649: while(char c=*(pos++)){
650: switch(state){
651: case Percent:
652: if(c=='%'){
653: state=Flags;
654: } else {
655: return FormatInvalid; // 1st char must be '%' only
656: }
657: break;
658: case Flags:
659: if(strchr("-+ #0", c)!=0){
660: break;
661: }
662: // go to the next step
663: case Width:
664: if(c=='.'){
665: state=Precision;
666: break;
667: }
668: // go to the next step
669: case Precision:
670: if(c>='0' && c<='9'){
671: if(state == Flags) state=Width; // no more flags
672: break;
673: } else if(c=='d' || c=='i'){
674: result=FormatInt;
675: } else if(strchr("feEgG", c)!=0){
676: result=FormatDouble;
677: } else if(strchr("uoxX", c)!=0){
678: result=FormatUInt;
679: } else {
680: return FormatInvalid; // invalid char
681: }
682: state=Done;
683: break;
684: case Done:
685: return FormatInvalid; // no chars allowed after 'type'
686: }
687: }
688: return result;
689: }
690:
691:
1.154 paf 692: const char* format(double value, char* fmt) {
1.229 misha 693: char local_buf[MAX_NUMBER];
1.235 misha 694: int size=-1;
1.229 misha 695:
696: if(fmt && strlen(fmt)){
697: switch(format_type(fmt)){
698: case FormatDouble:
699: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
700: break;
701: case FormatInt:
702: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
703: break;
704: case FormatUInt:
1.126 paf 705: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.229 misha 706: break;
707: case FormatInvalid:
708: throw Exception(PARSER_RUNTIME,
709: 0,
710: "Incorrect format string '%s' was specified.", fmt);
711: }
712: } else
713: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
714:
715: if(size < 0 || size >= MAX_NUMBER-1){ // on win32 we manually reduce max size while printing
716: throw Exception(PARSER_RUNTIME,
717: 0,
718: "Error occure white executing snprintf with format string '%s'.", fmt);
719: }
720:
1.235 misha 721: return pa_strdup(local_buf, (size_t)size);
1.12 paf 722: }
723:
1.36 paf 724: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 725: #ifdef WIN32
1.187 paf 726: size_t to_write = size;
1.12 paf 727: do{
1.154 paf 728: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 729: if(chunk_written<=0)
730: break;
731: size-=chunk_written;
1.36 paf 732: buf=((const char*)buf)+chunk_written;
1.126 paf 733: } while(size>0);
1.12 paf 734:
1.187 paf 735: return to_write-size;
1.12 paf 736: #else
1.126 paf 737: return fwrite(buf, 1, size, stdout);
1.12 paf 738: #endif
1.2 paf 739: }
1.14 paf 740:
1.229 misha 741: enum EscapeState {
742: EscapeRest,
743: EscapeFirst,
744: EscapeSecond,
745: EscapeUnicode
746: };
747:
1.236 misha 748: // @todo prescan for reduce required size (unescaped sting in 1 byte charset requires less memory usually)
1.258 misha 749: char* unescape_chars(const char* cp, int len, Charset* charset, bool js){
1.236 misha 750: char* s=new(PointerFreeGC) char[len+1]; // must be enough (%uXXXX==6 bytes, max utf-8 char length==6 bytes)
1.230 misha 751: char* dst=s;
1.229 misha 752: EscapeState escapeState=EscapeRest;
753: uint escapedValue=0;
754: int srcPos=0;
1.230 misha 755: short int jsCnt=0;
1.236 misha 756: while(srcPos<len){
1.229 misha 757: uchar c=(uchar)cp[srcPos];
1.258 misha 758: if(c=='%' || (c=='\\' && js)){
1.229 misha 759: escapeState=EscapeFirst;
760: } else {
761: switch(escapeState) {
762: case EscapeRest:
1.258 misha 763: if(!js && c=='+'){
1.230 misha 764: *dst++=' ';
1.229 misha 765: } else {
1.230 misha 766: *dst++=c;
1.229 misha 767: }
768: break;
769: case EscapeFirst:
1.232 misha 770: if(charset && c=='u'){
1.229 misha 771: // escaped unicode value: %u0430
772: jsCnt=0;
773: escapedValue=0;
774: escapeState=EscapeUnicode;
775: } else {
1.231 misha 776: if(isxdigit(c)){
1.229 misha 777: escapedValue=hex_value[c] << 4;
778: escapeState=EscapeSecond;
779: } else {
1.230 misha 780: *dst++=c;
1.229 misha 781: escapeState=EscapeRest;
782: }
783: }
784: break;
785: case EscapeSecond:
1.231 misha 786: if(isxdigit(c)){
1.229 misha 787: escapedValue+=hex_value[c];
1.230 misha 788: *dst++=(char)escapedValue;
1.229 misha 789: }
790: escapeState=EscapeRest;
791: break;
792: case EscapeUnicode:
1.231 misha 793: if(isxdigit(c)){
1.229 misha 794: escapedValue=(escapedValue << 4) + hex_value[c];
795: if(++jsCnt==4){
1.230 misha 796: // transcode utf8 char to client charset (we can lost some chars here)
1.232 misha 797: charset->store_Char((XMLByte*&)dst, (XMLCh)escapedValue, '?');
1.229 misha 798: escapeState=EscapeRest;
799: }
800: } else {
801: // not full unicode value
802: escapeState=EscapeRest;
803: }
804: break;
805: }
806: }
807:
808: srcPos++;
809: }
810:
1.230 misha 811: *dst=0; // zero-termination
1.229 misha 812: return s;
813: }
1.24 paf 814:
1.268 misha 815: char *search_stop(char*& current, char cstop_at) {
816: // sanity check
817: if(!current)
818: return 0;
819:
820: // skip leading WS
821: while(*current==' ' || *current=='\t')
822: current++;
823: if(!*current)
824: return current=0;
825:
826: char *result=current;
827: if(char *pstop_at=strchr(current, cstop_at)) {
828: *pstop_at=0;
829: current=pstop_at+1;
830: } else
831: current=0;
832: return result;
833: }
834:
1.24 paf 835: #ifdef WIN32
1.126 paf 836: void back_slashes_to_slashes(char* s) {
1.24 paf 837: if(s)
838: for(; *s; s++)
839: if(*s=='\\')
1.126 paf 840: *s='/';
1.24 paf 841: }
1.42 paf 842: /*
1.126 paf 843: void slashes_to_back_slashes(char* s) {
1.42 paf 844: if(s)
845: for(; *s; s++)
846: if(*s=='/')
1.126 paf 847: *s='\\';
1.42 paf 848: }
849: */
1.24 paf 850: #endif
1.41 paf 851:
1.231 misha 852: bool StrStartFromNC(const char* str, const char* substr, bool equal){
1.41 paf 853: while(true) {
1.231 misha 854: if(!(*substr)){
855: if(!(*str))
1.41 paf 856: return true;
857: else
1.231 misha 858: return !equal;
859: }
860: if(!(*str))
861: return false;
862: if(isalpha((unsigned char)*str)) {
863: if(tolower((unsigned char)*str)!=tolower((unsigned char)*substr))
1.41 paf 864: return false;
1.231 misha 865: } else if((*str) != (*substr))
1.41 paf 866: return false;
1.231 misha 867: str++;
868: substr++;
1.41 paf 869: }
1.57 parser 870: }
871:
1.232 misha 872: size_t strpos(const char *str, const char *substr) {
873: const char *p = strstr(str, substr);
874: return (p==0)?STRING_NOT_FOUND:p-str;
875: }
876:
877: // content-type: xxx; charset=WE-NEED-THIS
878: // content-type: xxx; charset="WE-NEED-THIS"
879: // content-type: xxx; charset="WE-NEED-THIS";
1.248 misha 880: Charset* detect_charset(const char* content_type){
1.233 misha 881: if(content_type){
1.245 misha 882: char* CONTENT_TYPE=pa_strdup(content_type);
1.248 misha 883:
884: for(char *p=CONTENT_TYPE; *p; p++)
885: *p=(char)toupper((unsigned char)*p);
1.233 misha 886:
887: if(const char* begin=strstr(CONTENT_TYPE, "CHARSET=")){
888: begin+=8; // skip "CHARSET="
889: char* end=0;
890: if(*begin && (*begin=='"' || *begin =='\'')){
891: char quote=*begin;
892: begin++;
893: end=(char*)strchr(begin, quote);
894: }
895: if(!end)
896: end=(char*)strchr(begin, ';');
897:
1.244 misha 898: if(end)
1.233 misha 899: *end=0; // terminator
900:
1.245 misha 901: return *begin?&charsets.get(begin):0;
1.232 misha 902: }
903: }
904: return 0;
905: }
906:
907:
1.84 paf 908: static bool isLeap(int year) {
1.229 misha 909: return !(
910: (year % 4) || ((year % 400) && !(year % 100))
911: );
1.57 parser 912: }
913:
914: int getMonthDays(int year, int month) {
1.220 misha 915: static int monthDays[]={
1.229 misha 916: 31,
917: 28,
918: 31,
919: 30,
920: 31,
921: 30,
922: 31,
923: 31,
924: 30,
925: 31,
926: 30,
927: 31
928: };
1.228 misha 929: return (month == 1 /* january -- 0 */ && isLeap(year)) ? 29 : monthDays[month];
1.41 paf 930: }
1.69 parser 931:
1.226 misha 932: int remove_crlf(char* start, char* end) {
933: char* from=start;
934: char* to=start;
935: bool skip=false;
936: while(from < end){
937: switch(*from){
938: case '\n':
939: case '\r':
940: case '\t':
941: case ' ':
942: if(!skip){
943: *to=' ';
944: to++;
945: skip=true;
946: }
947: break;
948: default:
949: if(from != to)
950: *to=*from;
951: to++;
952: skip=false;
1.69 parser 953: }
1.226 misha 954: from++;
955: }
956: return to-start;
1.91 paf 957: }
958:
959:
960: /// must be last in this file
961: #undef vsnprintf
1.126 paf 962: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 963: if(!s)
964: return 0;
965:
966: int r;
967: // note: on win32& maybe somewhere else
968: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
969: --s;
1.172 paf 970:
971: // clients do not check for negative 's', feature: ignore such prints
972: if((ssize_t)s<0)
973: return 0;
974:
1.91 paf 975: #if _MSC_VER
976: /*
977: win32:
978: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
979:
1.154 paf 980: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 981: */
1.126 paf 982: r=_vsnprintf(b, s, f, l);
1.91 paf 983: if(r<0)
984: r=s;
985: #else
1.126 paf 986: r=vsnprintf(b, s, f, l);
1.91 paf 987: /*
988: solaris:
989: man vsnprintf
990:
991: The snprintf() function returns the number of characters
992: formatted, that is, the number of characters that would have
993: been written to the buffer if it were large enough. If the
994: value of n is 0 on a call to snprintf(), an unspecified
995: value less than 1 is returned.
996: */
997:
998: if(r<0)
999: r=0;
1.167 paf 1000: else if((size_t)r>s)
1.91 paf 1001: r=s;
1002: #endif
1003: b[r]=0;
1004: return r;
1005: }
1006:
1.126 paf 1007: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1008: va_list l;
1.241 misha 1009: va_start(l, f);
1010: int r=__vsnprintf(b, s, f, l);
1011: va_end(l);
1.91 paf 1012: return r;
1.178 paf 1013: }
1014:
1015: /* mime64 functions are from libgmime[http://spruce.sourceforge.net/gmime/] lib */
1016: /*
1017: * Authors: Michael Zucchi <notzed@helixcode.com>
1018: * Jeffrey Stedfast <fejj@helixcode.com>
1019: *
1020: * Copyright 2000 Helix Code, Inc. (www.helixcode.com)
1021: *
1022: * This program is free software; you can redistribute it and/or modify
1023: * it under the terms of the GNU General Public License as published by
1024: * the Free Software Foundation; either version 2 of the License, or
1025: * (at your option) any later version.
1026: *
1027: * This program is distributed in the hope that it will be useful,
1028: * but WITHOUT ANY WARRANTY; without even the implied warranty of
1029: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1030: * GNU General Public License for more details.
1031: *
1032: * You should have received a copy of the GNU General Public License
1033: * along with this program; if not, write to the Free Software
1034: * Foundation, Inc., 59 Temple Street #330, Boston, MA 02111-1307, USA.
1035: *
1036: */
1.271 ! moko 1037: static const char *base64_alphabet =
1.178 paf 1038: "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
1039:
1040: /**
1041: * g_mime_utils_base64_encode_step:
1042: * @in: input stream
1043: * @inlen: length of the input
1044: * @out: output string
1045: * @state: holds the number of bits that are stored in @save
1046: * @save: leftover bits that have not yet been encoded
1047: *
1048: * Base64 encodes a chunk of data. Performs an 'encode step', only
1049: * encodes blocks of 3 characters to the output at a time, saves
1050: * left-over state in state and save (initialise to 0 on first
1051: * invocation).
1052: *
1053: * Returns the number of bytes encoded.
1054: **/
1.252 misha 1055:
1056: #define BASE64_GROUPS_IN_LINE 19
1057:
1.178 paf 1058: static size_t
1059: g_mime_utils_base64_encode_step (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1060: {
1.186 paf 1061: register const unsigned char *inptr;
1.178 paf 1062: register unsigned char *outptr;
1063:
1064: if (inlen <= 0)
1065: return 0;
1066:
1067: inptr = in;
1068: outptr = out;
1069:
1070: if (inlen + ((unsigned char *)save)[0] > 2) {
1071: const unsigned char *inend = in + inlen - 2;
1072: register int c1 = 0, c2 = 0, c3 = 0;
1073: register int already;
1074:
1075: already = *state;
1076:
1077: switch (((char *)save)[0]) {
1078: case 1: c1 = ((unsigned char *)save)[1]; goto skip1;
1079: case 2: c1 = ((unsigned char *)save)[1];
1080: c2 = ((unsigned char *)save)[2]; goto skip2;
1081: }
1082:
1083: /* yes, we jump into the loop, no i'm not going to change it, its beautiful! */
1084: while (inptr < inend) {
1085: c1 = *inptr++;
1086: skip1:
1087: c2 = *inptr++;
1088: skip2:
1089: c3 = *inptr++;
1090: *outptr++ = base64_alphabet [c1 >> 2];
1091: *outptr++ = base64_alphabet [(c2 >> 4) | ((c1 & 0x3) << 4)];
1092: *outptr++ = base64_alphabet [((c2 & 0x0f) << 2) | (c3 >> 6)];
1093: *outptr++ = base64_alphabet [c3 & 0x3f];
1094: /* this is a bit ugly ... */
1.252 misha 1095: if ((++already) >= BASE64_GROUPS_IN_LINE) {
1.178 paf 1096: *outptr++ = '\n';
1097: already = 0;
1098: }
1099: }
1100:
1101: ((unsigned char *)save)[0] = 0;
1102: inlen = 2 - (inptr - inend);
1103: *state = already;
1104: }
1105:
1106: //d(printf ("state = %d, inlen = %d\n", (int)((char *)save)[0], inlen));
1107:
1108: if (inlen > 0) {
1109: register char *saveout;
1110:
1111: /* points to the slot for the next char to save */
1112: saveout = & (((char *)save)[1]) + ((char *)save)[0];
1113:
1114: /* inlen can only be 0 1 or 2 */
1115: switch (inlen) {
1116: case 2: *saveout++ = *inptr++;
1117: case 1: *saveout++ = *inptr++;
1118: }
1.216 paf 1119: *(char *)save = *(char *)save+(char)inlen;
1.178 paf 1120: }
1121:
1122: /*d(printf ("mode = %d\nc1 = %c\nc2 = %c\n",
1123: (int)((char *)save)[0],
1124: (int)((char *)save)[1],
1125: (int)((char *)save)[2]));*/
1126:
1127: return (outptr - out);
1128: }
1129:
1130: /**
1131: * g_mime_utils_base64_encode_close:
1132: * @in: input stream
1133: * @inlen: length of the input
1134: * @out: output string
1135: * @state: holds the number of bits that are stored in @save
1136: * @save: leftover bits that have not yet been encoded
1137: *
1138: * Base64 encodes the input stream to the output stream. Call this
1139: * when finished encoding data with g_mime_utils_base64_encode_step to
1140: * flush off the last little bit.
1141: *
1142: * Returns the number of bytes encoded.
1143: **/
1144: static size_t
1145: g_mime_utils_base64_encode_close (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1146: {
1147: unsigned char *outptr = out;
1148: int c1, c2;
1149:
1150: if (inlen > 0)
1151: outptr += g_mime_utils_base64_encode_step (in, inlen, outptr, state, save);
1152:
1153: c1 = ((unsigned char *)save)[1];
1154: c2 = ((unsigned char *)save)[2];
1155:
1156: switch (((unsigned char *)save)[0]) {
1157: case 2:
1158: outptr[2] = base64_alphabet [(c2 & 0x0f) << 2];
1159: goto skip;
1160: case 1:
1161: outptr[2] = '=';
1162: skip:
1163: outptr[0] = base64_alphabet [c1 >> 2];
1164: outptr[1] = base64_alphabet [c2 >> 4 | ((c1 & 0x3) << 4)];
1165: outptr[3] = '=';
1166: outptr += 4;
1167: break;
1168: }
1169:
1170: *outptr++ = 0;
1171:
1172: *save = 0;
1173: *state = 0;
1174:
1175: return (outptr - out);
1176: }
1177:
1.210 paf 1178: static unsigned char gmime_base64_rank[256] = {
1.266 misha 1179: 255,255,255,255,255,255,255,255,255,254,254,255,255,254,255,255,
1.210 paf 1180: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1.266 misha 1181: 254,255,255,255,255,255,255,255,255,255,255, 62,255,255,255, 63,
1.210 paf 1182: 52, 53, 54, 55, 56, 57, 58, 59, 60, 61,255,255,255, 0,255,255,
1183: 255, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,
1184: 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25,255,255,255,255,255,
1185: 255, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40,
1186: 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51,255,255,255,255,255,
1187: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1188: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1189: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1190: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1191: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1192: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1193: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1194: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1195: };
1196:
1197: /**
1198: * g_mime_utils_base64_decode_step:
1199: * @in: input stream
1200: * @inlen: max length of data to decode
1201: * @out: output stream
1202: * @state: holds the number of bits that are stored in @save
1203: * @save: leftover bits that have not yet been decoded
1.266 misha 1204: * @strict: only base64 and whitespace chars are allowed
1.210 paf 1205: *
1206: * Decodes a chunk of base64 encoded data.
1207: *
1208: * Returns the number of bytes decoded (which have been dumped in @out).
1209: **/
1210: size_t
1.266 misha 1211: g_mime_utils_base64_decode_step(const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save, bool strict=false)
1.210 paf 1212: {
1.213 paf 1213: const unsigned char *inptr;
1214: unsigned char *outptr;
1.210 paf 1215: const unsigned char *inend;
1.213 paf 1216: int saved;
1.210 paf 1217: unsigned char c;
1218: int i;
1219:
1220: inend = in + inlen;
1221: outptr = out;
1222:
1223: /* convert 4 base64 bytes to 3 normal bytes */
1224: saved = *save;
1225: i = *state;
1226: inptr = in;
1227: while (inptr < inend) {
1228: c = gmime_base64_rank[*inptr++];
1.266 misha 1229: switch(c) {
1230: case 0xff: // non-base64 and non-whitespace chars. not allowed in strict mode
1231: if(strict)
1232: throw Exception(BASE64_FORMAT, 0, "Invalid base64 char on position %d is detected", inptr-in-1);
1233: case 0xfe: // whitespace chars 0x09, 0x0A, 0x0D, 0x20 are allowed in any mode
1234: break;
1235: default:
1236: saved = (saved << 6) | c;
1237: i++;
1238: if (i == 4) {
1239: *outptr++ = (unsigned char)(saved >> 16);
1240: *outptr++ = (unsigned char)(saved >> 8);
1241: *outptr++ = (unsigned char)(saved);
1242: i = 0;
1243: }
1.210 paf 1244: }
1245: }
1246:
1247: *save = saved;
1248: *state = i;
1249:
1250: /* quick scan back for '=' on the end somewhere */
1251: /* fortunately we can drop 1 output char for each trailing = (upto 2) */
1252: i = 2;
1253: while (inptr > in && i) {
1254: inptr--;
1.266 misha 1255: if (gmime_base64_rank[*inptr] <= 0xfe) {
1.210 paf 1256: if (*inptr == '=' && outptr > out)
1257: outptr--;
1258: i--;
1259: }
1260: }
1261:
1262: /* if i != 0 then there is a truncation error! */
1263: return (outptr - out);
1264: }
1265:
1266:
1.239 misha 1267: char* pa_base64_encode(const char *in, size_t in_size){
1.252 misha 1268: size_t new_size = ((in_size / 3 + 1) * 4);
1269: new_size += new_size / (BASE64_GROUPS_IN_LINE * 4)/*new lines*/ + 1/*zero terminator*/;
1270: char* result = new(PointerFreeGC) char[new_size];
1.178 paf 1271: int state=0;
1272: int save=0;
1.183 paf 1273: #ifndef NDEBUG
1274: size_t filled=
1275: #endif
1.251 misha 1276: g_mime_utils_base64_encode_close ((const unsigned char*)in, in_size, (unsigned char*)result, &state, &save);
1277:
1278: //throw Exception(PARSER_RUNTIME, 0, "%d %d %d", in_size, new_size, filled);
1279: assert(filled <= new_size);
1.178 paf 1280:
1281: return result;
1.98 paf 1282: }
1.210 paf 1283:
1.222 misha 1284:
1.239 misha 1285: char* pa_base64_encode(const String& file_spec){
1.222 misha 1286: unsigned char* base64=0;
1287: File_base64_action_info info={&base64};
1288:
1289: file_read_action_under_lock(file_spec,
1290: "pa_base64_encode", file_base64_file_action, &info);
1291:
1292: return (char*)base64;
1293: }
1294:
1295:
1296: static void file_base64_file_action(
1.229 misha 1297: struct stat& finfo,
1298: int f,
1299: const String&, const char* /*fname*/, bool,
1300: void *context) {
1.222 misha 1301:
1302: if(finfo.st_size) {
1303: File_base64_action_info& info=*static_cast<File_base64_action_info *>(context);
1304: *info.base64=new(PointerFreeGC) unsigned char[finfo.st_size * 2 + 6];
1305: unsigned char* base64 = *info.base64;
1306: int state=0;
1307: int save=0;
1308: int nCount;
1309: do {
1310: unsigned char buffer[FILE_BUFFER_SIZE];
1311: nCount = file_block_read(f, buffer, sizeof(buffer));
1312: if( nCount ){
1313: size_t filled=g_mime_utils_base64_encode_step ((const unsigned char*)buffer, nCount, base64, &state, &save);
1314: base64+=filled;
1315: }
1316: } while(nCount > 0);
1317: g_mime_utils_base64_encode_close (0, 0, base64, &state, &save);
1318: }
1319: }
1320:
1.265 misha 1321: void pa_base64_decode(const char *in, size_t in_size, char*& result, size_t& result_size, bool strict) {
1.264 misha 1322: // every 4 base64 bytes are converted into 3 normal bytes
1323: // not full set (tail) of 4-bytes set is ignored
1324: size_t new_size=in_size/4*3;
1325: result=new(PointerFreeGC) char[new_size+1/*terminator*/];
1326:
1.210 paf 1327: int state=0;
1328: int save=0;
1329: result_size=
1.266 misha 1330: g_mime_utils_base64_decode_step ((const unsigned char*)in, in_size,
1331: (unsigned char*)result, &state, &save, strict);
1.264 misha 1332: assert(result_size <= new_size);
1.211 paf 1333: result[result_size]=0; // for text files
1.265 misha 1334:
1335: if(strict && state!=0)
1.266 misha 1336: throw Exception(BASE64_FORMAT, 0, "Unexpected end of chars");
1.210 paf 1337: }
1.218 misha 1338:
1339:
1.221 misha 1340: int file_block_read(const int f, unsigned char* buffer, const size_t size){
1341: int nCount = read(f, buffer, size);
1342: if (nCount < 0)
1.238 misha 1343: throw Exception("file.read",
1.221 misha 1344: 0,
1345: "read failed: %s (%d)", strerror(errno), errno);
1346: return nCount;
1347: }
1348:
1.239 misha 1349: const unsigned long pa_crc32(const char *in, size_t in_size){
1.218 misha 1350: unsigned long crc32=0xFFFFFFFF;
1.220 misha 1351:
1.240 misha 1352: InitCrc32Table();
1.239 misha 1353: for(size_t i = 0; i<in_size; i++)
1354: CalcCrc32(in[i], crc32);
1.220 misha 1355:
1.218 misha 1356: return ~crc32;
1357: }
1358:
1.239 misha 1359: const unsigned long pa_crc32(const String& file_spec){
1.218 misha 1360: unsigned long crc32=0xFFFFFFFF;
1361: file_read_action_under_lock(file_spec, "crc32", file_crc32_file_action, &crc32);
1362: return ~crc32;
1363: }
1364:
1365: static void file_crc32_file_action(
1.229 misha 1366: struct stat& finfo,
1367: int f,
1368: const String&, const char* /*fname*/, bool,
1.239 misha 1369: void *context) {
1.218 misha 1370: unsigned long& crc32=*static_cast<unsigned long *>(context);
1371: if(finfo.st_size) {
1372: InitCrc32Table();
1.220 misha 1373: int nCount=0;
1.218 misha 1374: do {
1.221 misha 1375: unsigned char buffer[FILE_BUFFER_SIZE];
1376: nCount = file_block_read(f, buffer, sizeof(buffer));
1.220 misha 1377: for(int i = 0; i < nCount; i++) CalcCrc32(buffer[i], crc32);
1378: } while(nCount > 0);
1.218 misha 1379: }
1380: }
1381:
E-mail:
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to pa_common.C CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [parser3project] / parser3 / src / main