Annotation of parser3/src/main/pa_common.C, revision 1.276
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:
1.273 moko 39: #ifdef _MSC_VER
1.276 ! moko 40: #include <windows.h>
1.273 moko 41: #include <direct.h>
42: #endif
43:
1.276 ! moko 44: volatile const char * IDENT_PA_COMMON_C="$Id: pa_common.C,v 1.275 2013/07/23 10:18:54 moko Exp $" IDENT_PA_COMMON_H IDENT_PA_HASH_H IDENT_PA_ARRAY_H IDENT_PA_STACK_H;
1.267 moko 45:
1.93 paf 46: // some maybe-undefined constants
47:
1.82 paf 48: #ifndef _O_TEXT
49: # define _O_TEXT 0
50: #endif
51: #ifndef _O_BINARY
52: # define _O_BINARY 0
1.47 paf 53: #endif
1.80 paf 54:
1.138 paf 55: #ifdef HAVE_FTRUNCATE
56: # define PA_O_TRUNC 0
57: #else
58: # ifdef _O_TRUNC
59: # define PA_O_TRUNC _O_TRUNC
60: # else
61: # error you must have either ftruncate function or _O_TRUNC bit declared
62: # endif
1.154 paf 63: #endif
1.176 paf 64:
1.154 paf 65: // defines for globals
66:
67: #define FILE_STATUS_NAME "status"
68:
69: // globals
70:
71: const String file_status_name(FILE_STATUS_NAME);
72:
73: // functions
1.127 paf 74:
1.255 misha 75: bool capitalized(const char* s){
76: bool upper=true;
77: for(const char* c=s; *c; c++){
78: if(*c != (upper ? toupper((unsigned char)*c) : tolower((unsigned char)*c)))
79: return false;
80: upper=strchr("-_ ", *c) != 0;
81: }
82: return true;
83: }
84:
1.249 misha 85: const char* capitalize(const char* s){
1.255 misha 86: if(!s || capitalized(s))
87: return s;
88:
1.249 misha 89: char* result=pa_strdup(s);
90: if(result){
91: bool upper=true;
92: for(char* c=result; *c; c++){
93: *c=upper ? (char)toupper((unsigned char)*c) : (char)tolower((unsigned char)*c);
94: upper=strchr("-_ ", *c) != 0;
95: }
96: }
97: return (const char*)result;
98: }
99:
1.154 paf 100: void fix_line_breaks(char *str, size_t& length) {
1.87 paf 101: //_asm int 3;
1.154 paf 102: const char* const eob=str+length;
103: char* dest=str;
1.72 parser 104: // fix DOS: \r\n -> \n
105: // fix Macintosh: \r -> \n
1.154 paf 106: char* bol=str;
1.137 paf 107: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 108: size_t len=eol-bol;
109: if(dest!=bol)
1.260 misha 110: memmove(dest, bol, len);
1.72 parser 111: dest+=len;
1.126 paf 112: *dest++='\n';
1.72 parser 113:
1.126 paf 114: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 115: bol=eol+2;
1.154 paf 116: length--;
1.126 paf 117: } else // \r, not \n = Macintosh
1.72 parser 118: bol=eol+1;
119: }
1.154 paf 120: // last piece without \r
1.72 parser 121: if(dest!=bol)
1.260 misha 122: memmove(dest, bol, eob-bol);
1.154 paf 123: str[length]=0; // terminating
1.72 parser 124: }
1.18 paf 125:
1.271 moko 126: char* file_read_text(Request_charsets& charsets, const String& file_spec, bool fail_on_read_problem, HashStringValue* params, bool transcode_result) {
127: File_read_result file=file_read(charsets, file_spec, true, params, fail_on_read_problem, 0, 0, 0, transcode_result);
1.154 paf 128: return file.success?file.str:0;
1.126 paf 129: }
130:
1.271 moko 131: char* file_load_text(Request& r, const String& file_spec, bool fail_on_read_problem, HashStringValue* params, bool transcode_result) {
132: File_read_result file=file_load(r, file_spec, true, params, fail_on_read_problem, 0, 0, 0, transcode_result);
1.241 misha 133: return file.success?file.str:0;
134: }
135:
1.206 paf 136: /// these options were handled but not checked elsewhere, now check them
1.239 misha 137: int pa_get_valid_file_options_count(HashStringValue& options) {
1.206 paf 138: int result=0;
139: if(options.get(PA_SQL_LIMIT_NAME))
140: result++;
141: if(options.get(PA_SQL_OFFSET_NAME))
142: result++;
143: if(options.get(PA_COLUMN_SEPARATOR_NAME))
144: result++;
145: if(options.get(PA_COLUMN_ENCLOSER_NAME))
146: result++;
1.223 misha 147: if(options.get(PA_CHARSET_NAME))
148: result++;
1.206 paf 149: return result;
150: }
151:
1.123 paf 152: #ifndef DOXYGEN
153: struct File_read_action_info {
1.154 paf 154: char **data; size_t *data_size;
1.188 paf 155: char* buf; size_t offset; size_t count;
1.126 paf 156: };
1.123 paf 157: #endif
1.271 moko 158:
159: 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 160: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.188 paf 161: size_t to_read_size=info.count;
162: if(!to_read_size)
163: to_read_size=(size_t)finfo.st_size;
164: assert( !(info.buf && as_text) );
1.271 moko 165: if(to_read_size) {
1.188 paf 166: if(info.offset)
167: lseek(f, info.offset, SEEK_SET);
1.271 moko 168: *info.data=info.buf ? info.buf : (char *)pa_malloc_atomic(to_read_size+1);
169: ssize_t result=read(f, *info.data, to_read_size);
170: if(result<0)
171: throw Exception("file.read", &file_spec, "read failed: %s (%d)", strerror(errno), errno);
172: *info.data_size=result;
1.123 paf 173: } else { // empty file
1.209 paf 174: // 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 175: *info.data=(char *)pa_malloc_atomic(1);
1.209 paf 176: *(char*)(*info.data)=0;
1.123 paf 177: *info.data_size=0;
178: return;
179: }
1.126 paf 180: }
1.241 misha 181:
1.154 paf 182: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
1.229 misha 183: bool as_text, HashStringValue *params,
184: bool fail_on_read_problem,
1.234 misha 185: char* buf, size_t offset, size_t count, bool transcode_text_result) {
1.167 paf 186: File_read_result result={false, 0, 0, 0};
1.241 misha 187: if(params){
188: int valid_options=pa_get_valid_file_options_count(*params);
189: if(valid_options!=params->count())
1.262 misha 190: throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION);
1.241 misha 191: }
1.203 paf 192:
1.241 misha 193: File_read_action_info info={&result.str, &result.length, buf, offset, count};
1.161 paf 194:
1.241 misha 195: result.success=file_read_action_under_lock(file_spec,
196: "read", file_read_action, &info,
197: as_text, fail_on_read_problem);
1.223 misha 198:
1.241 misha 199: if(as_text){
200: if(result.success){
1.263 misha 201: Charset* asked_charset=0;
1.236 misha 202: if(result.length>=3 && strncmp(result.str, "\xEF\xBB\xBF", 3)==0){
1.240 misha 203: // skip UTF-8 signature (BOM code)
1.236 misha 204: result.str+=3;
205: result.length-=3;
1.263 misha 206: asked_charset=&UTF8_charset;
1.236 misha 207: }
208:
1.263 misha 209: if(params)
210: if(Value* vcharset_name=params->get(PA_CHARSET_NAME))
211: asked_charset=&::charsets.get(vcharset_name->as_string().change_case(charsets.source(), String::CC_UPPER));
212:
213: 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
214: String::C body=String::C(result.str, result.length);
215: body=Charset::transcode(body, *asked_charset, charsets.source());
1.236 misha 216:
1.263 misha 217: result.str=const_cast<char*>(body.str); // hacking a little
218: result.length=body.length;
1.131 paf 219: }
220: }
1.241 misha 221: if(result.length)
222: fix_line_breaks(result.str, result.length);
1.123 paf 223: }
1.241 misha 224:
225: return result;
226: }
227:
228: File_read_result file_load(Request& r, const String& file_spec,
229: bool as_text, HashStringValue *params,
230: bool fail_on_read_problem,
231: char* buf, size_t offset, size_t count, bool transcode_text_result) {
232:
233: File_read_result result={false, 0, 0, 0};
234: if(file_spec.starts_with("http://")) {
235: if(offset || count)
236: throw Exception(PARSER_RUNTIME,
237: 0,
238: "offset and load options are not supported for HTTP:// file load");
239:
240: // fail on read problem
241: File_read_http_result http=pa_internal_file_read_http(r, file_spec, as_text, params, transcode_text_result);
242: result.success=true;
243: result.str=http.str;
244: result.length=http.length;
245: result.headers=http.headers;
246: } else
247: result=
248: file_read(r.charsets, file_spec, as_text, params, fail_on_read_problem, buf, offset, count, transcode_text_result);
1.126 paf 249:
250: return result;
1.123 paf 251: }
252:
1.257 pretende 253:
1.154 paf 254: #ifdef PA_SAFE_MODE
1.259 misha 255: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
1.154 paf 256: if(finfo.st_uid/*foreign?*/!=geteuid()
257: && finfo.st_gid/*foreign?*/!=getegid())
1.224 misha 258: throw Exception(PARSER_RUNTIME,
1.154 paf 259: &file_spec,
260: "parser is in safe mode: "
261: "reading files of foreign group and user disabled "
262: "[recompile parser with --disable-safe-mode configure option], "
263: "actual filename '%s', "
264: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
265: fname,
266: finfo.st_uid, geteuid(),
1.259 misha 267: finfo.st_gid, getegid());
268: }
269: #else
270: void check_safe_mode(struct stat, const String&, const char*) {
271: }
1.257 pretende 272: #endif
1.259 misha 273:
1.257 pretende 274:
1.149 paf 275:
1.154 paf 276: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 277: const char* action_name, File_read_action action, void *context,
278: bool as_text,
1.123 paf 279: bool fail_on_read_problem) {
1.247 misha 280: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.33 paf 281: int f;
282:
283: // first open, next stat:
1.45 paf 284: // directory update of NTFS hard links performed on open.
1.33 paf 285: // ex:
286: // a.html:^test[] and b.html hardlink to a.html
287: // user inserts ! before ^test in a.html
1.126 paf 288: // directory entry of b.html in NTFS not updated at once,
1.35 paf 289: // they delay update till open, so we would receive "!^test[" string
290: // if would do stat, next open.
1.123 paf 291: // later: it seems, even this does not help sometimes
1.229 misha 292: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 293: try {
1.162 paf 294: if(pa_lock_shared_blocking(f)!=0)
1.126 paf 295: throw Exception("file.lock",
1.123 paf 296: &file_spec,
297: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 298: strerror(errno), errno, fname);
1.123 paf 299:
1.124 paf 300: struct stat finfo;
1.254 misha 301: if(fstat(f, &finfo)!=0)
1.124 paf 302: throw Exception("file.missing", // hardly possible: we just opened it OK
303: &file_spec,
304: "stat failed: %s (%d), actual filename '%s'",
1.154 paf 305: strerror(errno), errno, fname);
1.124 paf 306:
1.149 paf 307: check_safe_mode(finfo, file_spec, fname);
1.32 paf 308:
1.154 paf 309: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 310: } catch(...) {
1.162 paf 311: pa_unlock(f);close(f);
1.123 paf 312: if(fail_on_read_problem)
1.154 paf 313: rethrow;
1.123 paf 314: return false;
315: }
1.87 paf 316:
1.162 paf 317: pa_unlock(f);close(f);
1.72 parser 318: return true;
1.229 misha 319: } else {
1.118 paf 320: if(fail_on_read_problem)
1.256 misha 321: throw Exception(errno==EACCES?"file.access"
322: :(errno==ENOENT || errno==ENOTDIR || errno==ENODEV)?"file.missing":0,
1.118 paf 323: &file_spec,
1.123 paf 324: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 325: action_name, strerror(errno), errno, fname);
1.118 paf 326: return false;
327: }
1.8 paf 328: }
329:
1.202 paf 330: void create_dir_for_file(const String& file_spec) {
1.63 parser 331: size_t pos_after=1;
1.154 paf 332: size_t pos_before;
333: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
1.247 misha 334: mkdir(file_spec.mid(0, pos_before).taint_cstr(String::L_FILE_SPEC), 0775);
1.63 parser 335: pos_after=pos_before+1;
336: }
337: }
338:
1.98 paf 339: bool file_write_action_under_lock(
1.28 paf 340: const String& file_spec,
1.225 misha 341: const char* action_name,
342: File_write_action action,
343: void *context,
1.126 paf 344: bool as_text,
345: bool do_append,
346: bool do_block,
1.110 paf 347: bool fail_on_lock_problem) {
1.247 misha 348: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.28 paf 349: int f;
1.80 paf 350: if(access(fname, W_OK)!=0) // no
1.126 paf 351: create_dir_for_file(file_spec);
1.50 paf 352:
1.80 paf 353: if((f=open(fname,
354: O_CREAT|O_RDWR
355: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 356: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 357: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.126 paf 358: Exception e("file.lock",
1.110 paf 359: &file_spec,
360: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 361: strerror(errno), errno, fname);
1.126 paf 362: close(f);
1.110 paf 363: if(fail_on_lock_problem)
364: throw e;
1.98 paf 365: return false;
366: }
1.96 paf 367:
1.158 paf 368: try {
1.254 misha 369: #if (defined(HAVE_FCHMOD) && defined(PA_SAFE_MODE))
370: struct stat finfo;
371: if(fstat(f, &finfo)==0 && finfo.st_mode & 0111)
372: fchmod(f, finfo.st_mode & 0666/*clear executable bits*/); // backward: ignore errors if any
373: #endif
374: action(f, context);
1.158 paf 375: } catch(...) {
1.138 paf 376: #ifdef HAVE_FTRUNCATE
1.104 paf 377: if(!do_append)
1.125 paf 378: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 379: #endif
1.162 paf 380: pa_unlock(f);close(f);
1.154 paf 381: rethrow;
1.158 paf 382: }
1.80 paf 383:
1.138 paf 384: #ifdef HAVE_FTRUNCATE
1.104 paf 385: if(!do_append)
1.125 paf 386: 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 387: #endif
1.162 paf 388: pa_unlock(f);close(f);
1.98 paf 389: return true;
1.80 paf 390: } else
1.126 paf 391: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 392: &file_spec,
1.96 paf 393: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 394: action_name, strerror(errno), errno, fname);
1.96 paf 395: // here should be nothing, see rethrow above
396: }
397:
398: #ifndef DOXYGEN
399: struct File_write_action_info {
1.250 misha 400: const char* str;
401: size_t length;
1.126 paf 402: };
1.96 paf 403: #endif
1.271 moko 404:
1.96 paf 405: static void file_write_action(int f, void *context) {
1.126 paf 406: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 407: if(info.length) {
1.271 moko 408: ssize_t written=write(f, info.str, info.length);
1.116 paf 409: if(written<0)
1.271 moko 410: throw Exception("file.write", 0, "write failed: %s (%d)", strerror(errno), errno);
1.275 moko 411: if((size_t)written!=info.length)
1.271 moko 412: throw Exception("file.write", 0, "write failed: %u of %u bytes written", written, info.length);
1.113 paf 413: }
1.96 paf 414: }
1.271 moko 415:
1.96 paf 416: void file_write(
1.250 misha 417: Request_charsets& charsets,
418: const String& file_spec,
419: const char* data,
420: size_t size,
1.126 paf 421: bool as_text,
1.250 misha 422: bool do_append,
423: Charset* asked_charset) {
424:
425: if(as_text && asked_charset){
426: String::C body=String::C(data, size);
427: body=Charset::transcode(body, charsets.source(), *asked_charset);
428: data=body.str;
429: size=body.length;
430: };
431:
1.126 paf 432: File_write_action_info info={data, size};
1.225 misha 433:
1.98 paf 434: file_write_action_under_lock(
1.154 paf 435: file_spec,
1.225 misha 436: "write",
437: file_write_action,
438: &info,
1.154 paf 439: as_text,
440: do_append);
1.30 paf 441: }
442:
1.261 misha 443: static size_t get_dir(char* fname, size_t helper_length){
444: bool dir=false;
445: size_t pos=0;
446: for(pos=helper_length; pos; pos--){
447: char c=fname[pos-1];
448: if(c=='/' || c=='\\'){
449: fname[pos-1]=0;
450: dir=true;
451: } else if(dir) break;
452: }
453: return pos;
454: }
455:
456: static bool entry_readable(char* fname, bool need_dir) {
457: if(need_dir){
458: size_t size=strlen(fname);
459: while(size) {
460: char c=fname[size-1];
461: if(c=='/' || c=='\\')
462: fname[--size]=0;
463: else
464: break;
465: }
466: }
467:
468: struct stat finfo;
469: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
470: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
471: return is_dir==need_dir;
472: }
473: return false;
474: }
475:
476: static bool entry_readable(const String& file_spec, bool need_dir) {
477: return entry_readable(file_spec.taint_cstrm(String::L_FILE_SPEC), need_dir);
478: }
479:
1.63 parser 480: // throws nothing! [this is required in file_move & file_delete]
1.261 misha 481: static void rmdir(const String& file_spec, size_t pos_after=0) {
482: char* dir_spec=file_spec.taint_cstrm(String::L_FILE_SPEC);
483: size_t length=strlen(dir_spec);
484: while( (length=get_dir(dir_spec, length)) && (length > pos_after) ){
1.274 moko 485: #ifdef _MSC_VER
1.261 misha 486: if(!entry_readable(dir_spec, true))
487: break;
488: DWORD attrs=GetFileAttributes(dir_spec);
489: if(
490: (attrs==INVALID_FILE_ATTRIBUTES)
491: || !(attrs & FILE_ATTRIBUTE_DIRECTORY)
492: || (attrs & FILE_ATTRIBUTE_REPARSE_POINT)
493: )
494: break;
495: #endif
496: if( rmdir(dir_spec) )
497: break;
498: };
1.50 paf 499: }
1.239 misha 500:
1.269 misha 501: bool file_delete(const String& file_spec, bool fail_on_problem, bool keep_empty_dirs) {
1.247 misha 502: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.54 parser 503: if(unlink(fname)!=0)
1.164 paf 504: if(fail_on_problem)
1.126 paf 505: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 506: &file_spec,
507: "unlink failed: %s (%d), actual filename '%s'",
1.154 paf 508: strerror(errno), errno, fname);
1.93 paf 509: else
510: return false;
1.50 paf 511:
1.269 misha 512: if(!keep_empty_dirs)
513: rmdir(file_spec, 1);
514:
1.93 paf 515: return true;
1.60 parser 516: }
1.239 misha 517:
1.269 misha 518: void file_move(const String& old_spec, const String& new_spec, bool keep_empty_dirs) {
1.247 misha 519: const char* old_spec_cstr=old_spec.taint_cstr(String::L_FILE_SPEC);
520: const char* new_spec_cstr=new_spec.taint_cstr(String::L_FILE_SPEC);
1.63 parser 521:
1.126 paf 522: create_dir_for_file(new_spec);
1.63 parser 523:
1.60 parser 524: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 525: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 526: &old_spec,
527: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 paf 528: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 529:
1.269 misha 530: if(!keep_empty_dirs)
531: rmdir(old_spec, 1);
1.31 paf 532: }
533:
1.51 paf 534:
1.126 paf 535: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 536: struct stat lfinfo;
537: bool result=stat(fname, &lfinfo)==0;
538: if(afinfo)
539: *afinfo=lfinfo;
540: return result;
1.119 paf 541: }
542:
543: bool entry_exists(const String& file_spec) {
1.247 misha 544: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.126 paf 545: return entry_exists(fname, 0);
1.118 paf 546: }
547:
1.215 paf 548: bool file_exist(const String& file_spec) {
1.126 paf 549: return entry_readable(file_spec, false);
1.51 paf 550: }
1.239 misha 551:
1.215 paf 552: bool dir_exists(const String& file_spec) {
1.126 paf 553: return entry_readable(file_spec, true);
1.65 parser 554: }
1.239 misha 555:
1.215 paf 556: const String* file_exist(const String& path, const String& name) {
1.154 paf 557: String& result=*new String(path);
1.270 moko 558: if(path.last_char() != '/')
559: result << "/";
1.154 paf 560: result << name;
1.215 paf 561: return file_exist(result)?&result:0;
1.43 paf 562: }
1.239 misha 563:
1.43 paf 564: bool file_executable(const String& file_spec) {
1.247 misha 565: return access(file_spec.taint_cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 566: }
567:
1.64 parser 568: bool file_stat(const String& file_spec,
1.229 misha 569: size_t& rsize,
570: time_t& ratime,
571: time_t& rmtime,
572: time_t& rctime,
573: bool fail_on_read_problem) {
1.247 misha 574: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.154 paf 575: struct stat finfo;
1.44 paf 576: if(stat(fname, &finfo)!=0)
1.64 parser 577: if(fail_on_read_problem)
1.126 paf 578: throw Exception("file.missing",
1.67 parser 579: &file_spec,
580: "getting file size failed: %s (%d), real filename '%s'",
1.154 paf 581: strerror(errno), errno, fname);
1.64 parser 582: else
583: return false;
1.58 parser 584: rsize=finfo.st_size;
585: ratime=finfo.st_atime;
586: rmtime=finfo.st_mtime;
587: rctime=finfo.st_ctime;
1.64 parser 588: return true;
1.18 paf 589: }
590:
1.126 paf 591: char* getrow(char* *row_ref, char delim) {
1.229 misha 592: char* result=*row_ref;
593: if(result) {
1.126 paf 594: *row_ref=strchr(result, delim);
1.8 paf 595: if(*row_ref)
596: *((*row_ref)++)=0;
597: else if(!*result)
598: return 0;
1.229 misha 599: }
600: return result;
1.8 paf 601: }
602:
1.126 paf 603: char* lsplit(char* string, char delim) {
1.229 misha 604: if(string) {
1.126 paf 605: char* v=strchr(string, delim);
1.8 paf 606: if(v) {
607: *v=0;
608: return v+1;
609: }
1.229 misha 610: }
611: return 0;
1.8 paf 612: }
613:
1.126 paf 614: char* lsplit(char* *string_ref, char delim) {
1.229 misha 615: char* result=*string_ref;
1.126 paf 616: char* next=lsplit(*string_ref, delim);
1.229 misha 617: *string_ref=next;
618: return result;
1.9 paf 619: }
620:
1.126 paf 621: char* rsplit(char* string, char delim) {
1.229 misha 622: if(string) {
1.126 paf 623: char* v=strrchr(string, delim);
1.18 paf 624: if(v) {
1.9 paf 625: *v=0;
626: return v+1;
627: }
1.229 misha 628: }
629: return NULL;
1.10 paf 630: }
631:
1.229 misha 632:
633: // format: %[flags][width][.precision]type http://msdn.microsoft.com/ru-ru/library/56e442dc(en-us,VS.80).aspx
634: // flags: '-', '+', ' ', '#', '0' http://msdn.microsoft.com/ru-ru/library/8aky45ct(en-us,VS.80).aspx
635: // width, precision: non negative decimal number
636: enum FormatType {
637: FormatInvalid,
638: FormatInt,
639: FormatUInt,
640: FormatDouble
641: };
1.272 moko 642: FormatType format_type(const char* fmt){
1.229 misha 643: enum FormatState {
644: Percent,
645: Flags,
646: Width,
647: Precision,
648: Done
649: } state=Percent;
650:
651: FormatType result=FormatInvalid;
652:
1.272 moko 653: const char* pos=fmt;
1.229 misha 654: while(char c=*(pos++)){
655: switch(state){
656: case Percent:
657: if(c=='%'){
658: state=Flags;
659: } else {
660: return FormatInvalid; // 1st char must be '%' only
661: }
662: break;
663: case Flags:
664: if(strchr("-+ #0", c)!=0){
665: break;
666: }
667: // go to the next step
668: case Width:
669: if(c=='.'){
670: state=Precision;
671: break;
672: }
673: // go to the next step
674: case Precision:
675: if(c>='0' && c<='9'){
676: if(state == Flags) state=Width; // no more flags
677: break;
678: } else if(c=='d' || c=='i'){
679: result=FormatInt;
680: } else if(strchr("feEgG", c)!=0){
681: result=FormatDouble;
682: } else if(strchr("uoxX", c)!=0){
683: result=FormatUInt;
684: } else {
685: return FormatInvalid; // invalid char
686: }
687: state=Done;
688: break;
689: case Done:
690: return FormatInvalid; // no chars allowed after 'type'
691: }
692: }
693: return result;
694: }
695:
696:
1.272 moko 697: const char* format(double value, const char* fmt) {
1.229 misha 698: char local_buf[MAX_NUMBER];
1.235 misha 699: int size=-1;
1.229 misha 700:
701: if(fmt && strlen(fmt)){
702: switch(format_type(fmt)){
703: case FormatDouble:
704: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
705: break;
706: case FormatInt:
707: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
708: break;
709: case FormatUInt:
1.126 paf 710: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.229 misha 711: break;
712: case FormatInvalid:
713: throw Exception(PARSER_RUNTIME,
714: 0,
715: "Incorrect format string '%s' was specified.", fmt);
716: }
717: } else
718: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
719:
720: if(size < 0 || size >= MAX_NUMBER-1){ // on win32 we manually reduce max size while printing
721: throw Exception(PARSER_RUNTIME,
722: 0,
723: "Error occure white executing snprintf with format string '%s'.", fmt);
724: }
725:
1.235 misha 726: return pa_strdup(local_buf, (size_t)size);
1.12 paf 727: }
728:
1.36 paf 729: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 730: #ifdef WIN32
1.187 paf 731: size_t to_write = size;
1.12 paf 732: do{
1.154 paf 733: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 734: if(chunk_written<=0)
735: break;
736: size-=chunk_written;
1.36 paf 737: buf=((const char*)buf)+chunk_written;
1.126 paf 738: } while(size>0);
1.12 paf 739:
1.187 paf 740: return to_write-size;
1.12 paf 741: #else
1.126 paf 742: return fwrite(buf, 1, size, stdout);
1.12 paf 743: #endif
1.2 paf 744: }
1.14 paf 745:
1.229 misha 746: enum EscapeState {
747: EscapeRest,
748: EscapeFirst,
749: EscapeSecond,
750: EscapeUnicode
751: };
752:
1.236 misha 753: // @todo prescan for reduce required size (unescaped sting in 1 byte charset requires less memory usually)
1.258 misha 754: char* unescape_chars(const char* cp, int len, Charset* charset, bool js){
1.236 misha 755: char* s=new(PointerFreeGC) char[len+1]; // must be enough (%uXXXX==6 bytes, max utf-8 char length==6 bytes)
1.230 misha 756: char* dst=s;
1.229 misha 757: EscapeState escapeState=EscapeRest;
758: uint escapedValue=0;
759: int srcPos=0;
1.230 misha 760: short int jsCnt=0;
1.236 misha 761: while(srcPos<len){
1.229 misha 762: uchar c=(uchar)cp[srcPos];
1.258 misha 763: if(c=='%' || (c=='\\' && js)){
1.229 misha 764: escapeState=EscapeFirst;
765: } else {
766: switch(escapeState) {
767: case EscapeRest:
1.258 misha 768: if(!js && c=='+'){
1.230 misha 769: *dst++=' ';
1.229 misha 770: } else {
1.230 misha 771: *dst++=c;
1.229 misha 772: }
773: break;
774: case EscapeFirst:
1.232 misha 775: if(charset && c=='u'){
1.229 misha 776: // escaped unicode value: %u0430
777: jsCnt=0;
778: escapedValue=0;
779: escapeState=EscapeUnicode;
780: } else {
1.231 misha 781: if(isxdigit(c)){
1.229 misha 782: escapedValue=hex_value[c] << 4;
783: escapeState=EscapeSecond;
784: } else {
1.230 misha 785: *dst++=c;
1.229 misha 786: escapeState=EscapeRest;
787: }
788: }
789: break;
790: case EscapeSecond:
1.231 misha 791: if(isxdigit(c)){
1.229 misha 792: escapedValue+=hex_value[c];
1.230 misha 793: *dst++=(char)escapedValue;
1.229 misha 794: }
795: escapeState=EscapeRest;
796: break;
797: case EscapeUnicode:
1.231 misha 798: if(isxdigit(c)){
1.229 misha 799: escapedValue=(escapedValue << 4) + hex_value[c];
800: if(++jsCnt==4){
1.230 misha 801: // transcode utf8 char to client charset (we can lost some chars here)
1.232 misha 802: charset->store_Char((XMLByte*&)dst, (XMLCh)escapedValue, '?');
1.229 misha 803: escapeState=EscapeRest;
804: }
805: } else {
806: // not full unicode value
807: escapeState=EscapeRest;
808: }
809: break;
810: }
811: }
812:
813: srcPos++;
814: }
815:
1.230 misha 816: *dst=0; // zero-termination
1.229 misha 817: return s;
818: }
1.24 paf 819:
1.268 misha 820: char *search_stop(char*& current, char cstop_at) {
821: // sanity check
822: if(!current)
823: return 0;
824:
825: // skip leading WS
826: while(*current==' ' || *current=='\t')
827: current++;
828: if(!*current)
829: return current=0;
830:
831: char *result=current;
832: if(char *pstop_at=strchr(current, cstop_at)) {
833: *pstop_at=0;
834: current=pstop_at+1;
835: } else
836: current=0;
837: return result;
838: }
839:
1.24 paf 840: #ifdef WIN32
1.126 paf 841: void back_slashes_to_slashes(char* s) {
1.24 paf 842: if(s)
843: for(; *s; s++)
844: if(*s=='\\')
1.126 paf 845: *s='/';
1.24 paf 846: }
847: #endif
1.41 paf 848:
1.231 misha 849: bool StrStartFromNC(const char* str, const char* substr, bool equal){
1.41 paf 850: while(true) {
1.231 misha 851: if(!(*substr)){
852: if(!(*str))
1.41 paf 853: return true;
854: else
1.231 misha 855: return !equal;
856: }
857: if(!(*str))
858: return false;
859: if(isalpha((unsigned char)*str)) {
860: if(tolower((unsigned char)*str)!=tolower((unsigned char)*substr))
1.41 paf 861: return false;
1.231 misha 862: } else if((*str) != (*substr))
1.41 paf 863: return false;
1.231 misha 864: str++;
865: substr++;
1.41 paf 866: }
1.57 parser 867: }
868:
1.232 misha 869: size_t strpos(const char *str, const char *substr) {
870: const char *p = strstr(str, substr);
871: return (p==0)?STRING_NOT_FOUND:p-str;
872: }
873:
874: // content-type: xxx; charset=WE-NEED-THIS
875: // content-type: xxx; charset="WE-NEED-THIS"
876: // content-type: xxx; charset="WE-NEED-THIS";
1.248 misha 877: Charset* detect_charset(const char* content_type){
1.233 misha 878: if(content_type){
1.245 misha 879: char* CONTENT_TYPE=pa_strdup(content_type);
1.248 misha 880:
881: for(char *p=CONTENT_TYPE; *p; p++)
882: *p=(char)toupper((unsigned char)*p);
1.233 misha 883:
884: if(const char* begin=strstr(CONTENT_TYPE, "CHARSET=")){
885: begin+=8; // skip "CHARSET="
886: char* end=0;
887: if(*begin && (*begin=='"' || *begin =='\'')){
888: char quote=*begin;
889: begin++;
890: end=(char*)strchr(begin, quote);
891: }
892: if(!end)
893: end=(char*)strchr(begin, ';');
894:
1.244 misha 895: if(end)
1.233 misha 896: *end=0; // terminator
897:
1.245 misha 898: return *begin?&charsets.get(begin):0;
1.232 misha 899: }
900: }
901: return 0;
902: }
903:
904:
1.84 paf 905: static bool isLeap(int year) {
1.229 misha 906: return !(
907: (year % 4) || ((year % 400) && !(year % 100))
908: );
1.57 parser 909: }
910:
911: int getMonthDays(int year, int month) {
1.220 misha 912: static int monthDays[]={
1.229 misha 913: 31,
914: 28,
915: 31,
916: 30,
917: 31,
918: 30,
919: 31,
920: 31,
921: 30,
922: 31,
923: 30,
924: 31
925: };
1.228 misha 926: return (month == 1 /* january -- 0 */ && isLeap(year)) ? 29 : monthDays[month];
1.41 paf 927: }
1.69 parser 928:
1.226 misha 929: int remove_crlf(char* start, char* end) {
930: char* from=start;
931: char* to=start;
932: bool skip=false;
933: while(from < end){
934: switch(*from){
935: case '\n':
936: case '\r':
937: case '\t':
938: case ' ':
939: if(!skip){
940: *to=' ';
941: to++;
942: skip=true;
943: }
944: break;
945: default:
946: if(from != to)
947: *to=*from;
948: to++;
949: skip=false;
1.69 parser 950: }
1.226 misha 951: from++;
952: }
953: return to-start;
1.91 paf 954: }
955:
956:
957: /// must be last in this file
958: #undef vsnprintf
1.126 paf 959: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 960: if(!s)
961: return 0;
962:
963: int r;
964: // note: on win32& maybe somewhere else
965: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
966: --s;
1.172 paf 967:
968: // clients do not check for negative 's', feature: ignore such prints
969: if((ssize_t)s<0)
970: return 0;
971:
1.273 moko 972: #ifdef _MSC_VER
1.91 paf 973: /*
974: win32:
975: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
976:
1.154 paf 977: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 978: */
1.126 paf 979: r=_vsnprintf(b, s, f, l);
1.91 paf 980: if(r<0)
981: r=s;
982: #else
1.126 paf 983: r=vsnprintf(b, s, f, l);
1.91 paf 984: /*
985: solaris:
986: man vsnprintf
987:
988: The snprintf() function returns the number of characters
989: formatted, that is, the number of characters that would have
990: been written to the buffer if it were large enough. If the
991: value of n is 0 on a call to snprintf(), an unspecified
992: value less than 1 is returned.
993: */
994:
995: if(r<0)
996: r=0;
1.167 paf 997: else if((size_t)r>s)
1.91 paf 998: r=s;
999: #endif
1000: b[r]=0;
1001: return r;
1002: }
1003:
1.126 paf 1004: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1005: va_list l;
1.241 misha 1006: va_start(l, f);
1007: int r=__vsnprintf(b, s, f, l);
1008: va_end(l);
1.91 paf 1009: return r;
1.178 paf 1010: }
1011:
1012: /* mime64 functions are from libgmime[http://spruce.sourceforge.net/gmime/] lib */
1013: /*
1014: * Authors: Michael Zucchi <notzed@helixcode.com>
1015: * Jeffrey Stedfast <fejj@helixcode.com>
1016: *
1017: * Copyright 2000 Helix Code, Inc. (www.helixcode.com)
1018: *
1019: * This program is free software; you can redistribute it and/or modify
1020: * it under the terms of the GNU General Public License as published by
1021: * the Free Software Foundation; either version 2 of the License, or
1022: * (at your option) any later version.
1023: *
1024: * This program is distributed in the hope that it will be useful,
1025: * but WITHOUT ANY WARRANTY; without even the implied warranty of
1026: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1027: * GNU General Public License for more details.
1028: *
1029: * You should have received a copy of the GNU General Public License
1030: * along with this program; if not, write to the Free Software
1031: * Foundation, Inc., 59 Temple Street #330, Boston, MA 02111-1307, USA.
1032: *
1033: */
1.271 moko 1034: static const char *base64_alphabet =
1.178 paf 1035: "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
1036:
1037: /**
1038: * g_mime_utils_base64_encode_step:
1039: * @in: input stream
1040: * @inlen: length of the input
1041: * @out: output string
1042: * @state: holds the number of bits that are stored in @save
1043: * @save: leftover bits that have not yet been encoded
1044: *
1045: * Base64 encodes a chunk of data. Performs an 'encode step', only
1046: * encodes blocks of 3 characters to the output at a time, saves
1047: * left-over state in state and save (initialise to 0 on first
1048: * invocation).
1049: *
1050: * Returns the number of bytes encoded.
1051: **/
1.252 misha 1052:
1053: #define BASE64_GROUPS_IN_LINE 19
1054:
1.178 paf 1055: static size_t
1056: g_mime_utils_base64_encode_step (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1057: {
1.186 paf 1058: register const unsigned char *inptr;
1.178 paf 1059: register unsigned char *outptr;
1060:
1061: if (inlen <= 0)
1062: return 0;
1063:
1064: inptr = in;
1065: outptr = out;
1066:
1067: if (inlen + ((unsigned char *)save)[0] > 2) {
1068: const unsigned char *inend = in + inlen - 2;
1069: register int c1 = 0, c2 = 0, c3 = 0;
1070: register int already;
1071:
1072: already = *state;
1073:
1074: switch (((char *)save)[0]) {
1075: case 1: c1 = ((unsigned char *)save)[1]; goto skip1;
1076: case 2: c1 = ((unsigned char *)save)[1];
1077: c2 = ((unsigned char *)save)[2]; goto skip2;
1078: }
1079:
1080: /* yes, we jump into the loop, no i'm not going to change it, its beautiful! */
1081: while (inptr < inend) {
1082: c1 = *inptr++;
1083: skip1:
1084: c2 = *inptr++;
1085: skip2:
1086: c3 = *inptr++;
1087: *outptr++ = base64_alphabet [c1 >> 2];
1088: *outptr++ = base64_alphabet [(c2 >> 4) | ((c1 & 0x3) << 4)];
1089: *outptr++ = base64_alphabet [((c2 & 0x0f) << 2) | (c3 >> 6)];
1090: *outptr++ = base64_alphabet [c3 & 0x3f];
1091: /* this is a bit ugly ... */
1.252 misha 1092: if ((++already) >= BASE64_GROUPS_IN_LINE) {
1.178 paf 1093: *outptr++ = '\n';
1094: already = 0;
1095: }
1096: }
1097:
1098: ((unsigned char *)save)[0] = 0;
1099: inlen = 2 - (inptr - inend);
1100: *state = already;
1101: }
1102:
1103: //d(printf ("state = %d, inlen = %d\n", (int)((char *)save)[0], inlen));
1104:
1105: if (inlen > 0) {
1106: register char *saveout;
1107:
1108: /* points to the slot for the next char to save */
1109: saveout = & (((char *)save)[1]) + ((char *)save)[0];
1110:
1111: /* inlen can only be 0 1 or 2 */
1112: switch (inlen) {
1113: case 2: *saveout++ = *inptr++;
1114: case 1: *saveout++ = *inptr++;
1115: }
1.216 paf 1116: *(char *)save = *(char *)save+(char)inlen;
1.178 paf 1117: }
1118:
1119: /*d(printf ("mode = %d\nc1 = %c\nc2 = %c\n",
1120: (int)((char *)save)[0],
1121: (int)((char *)save)[1],
1122: (int)((char *)save)[2]));*/
1123:
1124: return (outptr - out);
1125: }
1126:
1127: /**
1128: * g_mime_utils_base64_encode_close:
1129: * @in: input stream
1130: * @inlen: length of the input
1131: * @out: output string
1132: * @state: holds the number of bits that are stored in @save
1133: * @save: leftover bits that have not yet been encoded
1134: *
1135: * Base64 encodes the input stream to the output stream. Call this
1136: * when finished encoding data with g_mime_utils_base64_encode_step to
1137: * flush off the last little bit.
1138: *
1139: * Returns the number of bytes encoded.
1140: **/
1141: static size_t
1142: g_mime_utils_base64_encode_close (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1143: {
1144: unsigned char *outptr = out;
1145: int c1, c2;
1146:
1147: if (inlen > 0)
1148: outptr += g_mime_utils_base64_encode_step (in, inlen, outptr, state, save);
1149:
1150: c1 = ((unsigned char *)save)[1];
1151: c2 = ((unsigned char *)save)[2];
1152:
1153: switch (((unsigned char *)save)[0]) {
1154: case 2:
1155: outptr[2] = base64_alphabet [(c2 & 0x0f) << 2];
1156: goto skip;
1157: case 1:
1158: outptr[2] = '=';
1159: skip:
1160: outptr[0] = base64_alphabet [c1 >> 2];
1161: outptr[1] = base64_alphabet [c2 >> 4 | ((c1 & 0x3) << 4)];
1162: outptr[3] = '=';
1163: outptr += 4;
1164: break;
1165: }
1166:
1167: *outptr++ = 0;
1168:
1169: *save = 0;
1170: *state = 0;
1171:
1172: return (outptr - out);
1173: }
1174:
1.210 paf 1175: static unsigned char gmime_base64_rank[256] = {
1.266 misha 1176: 255,255,255,255,255,255,255,255,255,254,254,255,255,254,255,255,
1.210 paf 1177: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1.266 misha 1178: 254,255,255,255,255,255,255,255,255,255,255, 62,255,255,255, 63,
1.210 paf 1179: 52, 53, 54, 55, 56, 57, 58, 59, 60, 61,255,255,255, 0,255,255,
1180: 255, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,
1181: 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25,255,255,255,255,255,
1182: 255, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40,
1183: 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51,255,255,255,255,255,
1184: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1185: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1186: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
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: };
1193:
1194: /**
1195: * g_mime_utils_base64_decode_step:
1196: * @in: input stream
1197: * @inlen: max length of data to decode
1198: * @out: output stream
1199: * @state: holds the number of bits that are stored in @save
1200: * @save: leftover bits that have not yet been decoded
1.266 misha 1201: * @strict: only base64 and whitespace chars are allowed
1.210 paf 1202: *
1203: * Decodes a chunk of base64 encoded data.
1204: *
1205: * Returns the number of bytes decoded (which have been dumped in @out).
1206: **/
1207: size_t
1.266 misha 1208: 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 1209: {
1.213 paf 1210: const unsigned char *inptr;
1211: unsigned char *outptr;
1.210 paf 1212: const unsigned char *inend;
1.213 paf 1213: int saved;
1.210 paf 1214: unsigned char c;
1215: int i;
1216:
1217: inend = in + inlen;
1218: outptr = out;
1219:
1220: /* convert 4 base64 bytes to 3 normal bytes */
1221: saved = *save;
1222: i = *state;
1223: inptr = in;
1224: while (inptr < inend) {
1225: c = gmime_base64_rank[*inptr++];
1.266 misha 1226: switch(c) {
1227: case 0xff: // non-base64 and non-whitespace chars. not allowed in strict mode
1228: if(strict)
1229: throw Exception(BASE64_FORMAT, 0, "Invalid base64 char on position %d is detected", inptr-in-1);
1230: case 0xfe: // whitespace chars 0x09, 0x0A, 0x0D, 0x20 are allowed in any mode
1231: break;
1232: default:
1233: saved = (saved << 6) | c;
1234: i++;
1235: if (i == 4) {
1236: *outptr++ = (unsigned char)(saved >> 16);
1237: *outptr++ = (unsigned char)(saved >> 8);
1238: *outptr++ = (unsigned char)(saved);
1239: i = 0;
1240: }
1.210 paf 1241: }
1242: }
1243:
1244: *save = saved;
1245: *state = i;
1246:
1247: /* quick scan back for '=' on the end somewhere */
1248: /* fortunately we can drop 1 output char for each trailing = (upto 2) */
1249: i = 2;
1250: while (inptr > in && i) {
1251: inptr--;
1.266 misha 1252: if (gmime_base64_rank[*inptr] <= 0xfe) {
1.210 paf 1253: if (*inptr == '=' && outptr > out)
1254: outptr--;
1255: i--;
1256: }
1257: }
1258:
1259: /* if i != 0 then there is a truncation error! */
1260: return (outptr - out);
1261: }
1262:
1263:
1.239 misha 1264: char* pa_base64_encode(const char *in, size_t in_size){
1.252 misha 1265: size_t new_size = ((in_size / 3 + 1) * 4);
1266: new_size += new_size / (BASE64_GROUPS_IN_LINE * 4)/*new lines*/ + 1/*zero terminator*/;
1267: char* result = new(PointerFreeGC) char[new_size];
1.178 paf 1268: int state=0;
1269: int save=0;
1.183 paf 1270: #ifndef NDEBUG
1271: size_t filled=
1272: #endif
1.251 misha 1273: g_mime_utils_base64_encode_close ((const unsigned char*)in, in_size, (unsigned char*)result, &state, &save);
1274:
1275: //throw Exception(PARSER_RUNTIME, 0, "%d %d %d", in_size, new_size, filled);
1276: assert(filled <= new_size);
1.178 paf 1277:
1278: return result;
1.98 paf 1279: }
1.210 paf 1280:
1.222 misha 1281:
1.239 misha 1282: char* pa_base64_encode(const String& file_spec){
1.222 misha 1283: unsigned char* base64=0;
1284: File_base64_action_info info={&base64};
1285:
1286: file_read_action_under_lock(file_spec,
1287: "pa_base64_encode", file_base64_file_action, &info);
1288:
1289: return (char*)base64;
1290: }
1291:
1292:
1293: static void file_base64_file_action(
1.229 misha 1294: struct stat& finfo,
1295: int f,
1296: const String&, const char* /*fname*/, bool,
1297: void *context) {
1.222 misha 1298:
1299: if(finfo.st_size) {
1300: File_base64_action_info& info=*static_cast<File_base64_action_info *>(context);
1301: *info.base64=new(PointerFreeGC) unsigned char[finfo.st_size * 2 + 6];
1302: unsigned char* base64 = *info.base64;
1303: int state=0;
1304: int save=0;
1305: int nCount;
1306: do {
1307: unsigned char buffer[FILE_BUFFER_SIZE];
1308: nCount = file_block_read(f, buffer, sizeof(buffer));
1309: if( nCount ){
1310: size_t filled=g_mime_utils_base64_encode_step ((const unsigned char*)buffer, nCount, base64, &state, &save);
1311: base64+=filled;
1312: }
1313: } while(nCount > 0);
1314: g_mime_utils_base64_encode_close (0, 0, base64, &state, &save);
1315: }
1316: }
1317:
1.265 misha 1318: void pa_base64_decode(const char *in, size_t in_size, char*& result, size_t& result_size, bool strict) {
1.264 misha 1319: // every 4 base64 bytes are converted into 3 normal bytes
1320: // not full set (tail) of 4-bytes set is ignored
1321: size_t new_size=in_size/4*3;
1322: result=new(PointerFreeGC) char[new_size+1/*terminator*/];
1323:
1.210 paf 1324: int state=0;
1325: int save=0;
1326: result_size=
1.266 misha 1327: g_mime_utils_base64_decode_step ((const unsigned char*)in, in_size,
1328: (unsigned char*)result, &state, &save, strict);
1.264 misha 1329: assert(result_size <= new_size);
1.211 paf 1330: result[result_size]=0; // for text files
1.265 misha 1331:
1332: if(strict && state!=0)
1.266 misha 1333: throw Exception(BASE64_FORMAT, 0, "Unexpected end of chars");
1.210 paf 1334: }
1.218 misha 1335:
1336:
1.221 misha 1337: int file_block_read(const int f, unsigned char* buffer, const size_t size){
1338: int nCount = read(f, buffer, size);
1339: if (nCount < 0)
1.238 misha 1340: throw Exception("file.read",
1.221 misha 1341: 0,
1342: "read failed: %s (%d)", strerror(errno), errno);
1343: return nCount;
1344: }
1345:
1.239 misha 1346: const unsigned long pa_crc32(const char *in, size_t in_size){
1.218 misha 1347: unsigned long crc32=0xFFFFFFFF;
1.220 misha 1348:
1.240 misha 1349: InitCrc32Table();
1.239 misha 1350: for(size_t i = 0; i<in_size; i++)
1351: CalcCrc32(in[i], crc32);
1.220 misha 1352:
1.218 misha 1353: return ~crc32;
1354: }
1355:
1.239 misha 1356: const unsigned long pa_crc32(const String& file_spec){
1.218 misha 1357: unsigned long crc32=0xFFFFFFFF;
1358: file_read_action_under_lock(file_spec, "crc32", file_crc32_file_action, &crc32);
1359: return ~crc32;
1360: }
1361:
1362: static void file_crc32_file_action(
1.229 misha 1363: struct stat& finfo,
1364: int f,
1365: const String&, const char* /*fname*/, bool,
1.239 misha 1366: void *context) {
1.218 misha 1367: unsigned long& crc32=*static_cast<unsigned long *>(context);
1368: if(finfo.st_size) {
1369: InitCrc32Table();
1.220 misha 1370: int nCount=0;
1.218 misha 1371: do {
1.221 misha 1372: unsigned char buffer[FILE_BUFFER_SIZE];
1373: nCount = file_block_read(f, buffer, sizeof(buffer));
1.220 misha 1374: for(int i = 0; i < nCount; i++) CalcCrc32(buffer[i], crc32);
1375: } while(nCount > 0);
1.218 misha 1376: }
1377: }
1378:
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