Annotation of parser3/src/main/pa_common.C, revision 1.293
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.288 moko 4: Copyright (c) 2000-2015 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:
1.283 moko 39: #include "pa_idna.h"
40: #include "pa_convert_utf.h"
41:
1.273 moko 42: #ifdef _MSC_VER
1.276 moko 43: #include <windows.h>
1.273 moko 44: #include <direct.h>
45: #endif
46:
1.277 moko 47: #ifdef _MSC_VER
48: #define pa_mkdir(path, mode) _mkdir(path)
49: #else
50: #define pa_mkdir(path, mode) mkdir(path, mode)
51: #endif
52:
1.293 ! moko 53: volatile const char * IDENT_PA_COMMON_C="$Id: pa_common.C,v 1.292 2016/09/21 14:01:45 moko Exp $" IDENT_PA_COMMON_H IDENT_PA_HASH_H IDENT_PA_ARRAY_H IDENT_PA_STACK_H;
1.267 moko 54:
1.93 paf 55: // some maybe-undefined constants
56:
1.82 paf 57: #ifndef _O_TEXT
58: # define _O_TEXT 0
59: #endif
60: #ifndef _O_BINARY
61: # define _O_BINARY 0
1.47 paf 62: #endif
1.80 paf 63:
1.138 paf 64: #ifdef HAVE_FTRUNCATE
65: # define PA_O_TRUNC 0
66: #else
67: # ifdef _O_TRUNC
68: # define PA_O_TRUNC _O_TRUNC
69: # else
70: # error you must have either ftruncate function or _O_TRUNC bit declared
71: # endif
1.154 paf 72: #endif
1.176 paf 73:
1.154 paf 74: // defines for globals
75:
76: #define FILE_STATUS_NAME "status"
77:
1.293 ! moko 78: // defines for vs2015 linking with old libs
! 79: #if _MSC_VER >= 1900
! 80: int (WINAPIV * __vsnprintf)(char *, size_t, const char*, va_list) = _vsnprintf;
! 81: #endif
! 82:
1.154 paf 83: // globals
84:
85: const String file_status_name(FILE_STATUS_NAME);
86:
87: // functions
1.127 paf 88:
1.271 moko 89: char* file_read_text(Request_charsets& charsets, const String& file_spec, bool fail_on_read_problem, HashStringValue* params, bool transcode_result) {
90: File_read_result file=file_read(charsets, file_spec, true, params, fail_on_read_problem, 0, 0, 0, transcode_result);
1.154 paf 91: return file.success?file.str:0;
1.126 paf 92: }
93:
1.271 moko 94: char* file_load_text(Request& r, const String& file_spec, bool fail_on_read_problem, HashStringValue* params, bool transcode_result) {
95: File_read_result file=file_load(r, file_spec, true, params, fail_on_read_problem, 0, 0, 0, transcode_result);
1.241 misha 96: return file.success?file.str:0;
97: }
98:
1.206 paf 99: /// these options were handled but not checked elsewhere, now check them
1.239 misha 100: int pa_get_valid_file_options_count(HashStringValue& options) {
1.206 paf 101: int result=0;
102: if(options.get(PA_SQL_LIMIT_NAME))
103: result++;
104: if(options.get(PA_SQL_OFFSET_NAME))
105: result++;
106: if(options.get(PA_COLUMN_SEPARATOR_NAME))
107: result++;
108: if(options.get(PA_COLUMN_ENCLOSER_NAME))
109: result++;
1.223 misha 110: if(options.get(PA_CHARSET_NAME))
111: result++;
1.206 paf 112: return result;
113: }
114:
1.123 paf 115: #ifndef DOXYGEN
116: struct File_read_action_info {
1.154 paf 117: char **data; size_t *data_size;
1.188 paf 118: char* buf; size_t offset; size_t count;
1.126 paf 119: };
1.123 paf 120: #endif
1.271 moko 121:
1.289 moko 122: static void file_read_action(struct stat& finfo, int f, const String& file_spec, void *context) {
1.126 paf 123: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.188 paf 124: size_t to_read_size=info.count;
125: if(!to_read_size)
126: to_read_size=(size_t)finfo.st_size;
1.271 moko 127: if(to_read_size) {
1.188 paf 128: if(info.offset)
129: lseek(f, info.offset, SEEK_SET);
1.271 moko 130: *info.data=info.buf ? info.buf : (char *)pa_malloc_atomic(to_read_size+1);
131: ssize_t result=read(f, *info.data, to_read_size);
132: if(result<0)
133: throw Exception("file.read", &file_spec, "read failed: %s (%d)", strerror(errno), errno);
134: *info.data_size=result;
1.123 paf 135: } else { // empty file
1.209 paf 136: // 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 137: *info.data=(char *)pa_malloc_atomic(1);
1.209 paf 138: *(char*)(*info.data)=0;
1.123 paf 139: *info.data_size=0;
140: return;
141: }
1.126 paf 142: }
1.241 misha 143:
1.154 paf 144: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
1.229 misha 145: bool as_text, HashStringValue *params,
146: bool fail_on_read_problem,
1.234 misha 147: char* buf, size_t offset, size_t count, bool transcode_text_result) {
1.167 paf 148: File_read_result result={false, 0, 0, 0};
1.241 misha 149: if(params){
150: int valid_options=pa_get_valid_file_options_count(*params);
151: if(valid_options!=params->count())
1.262 misha 152: throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION);
1.241 misha 153: }
1.203 paf 154:
1.241 misha 155: File_read_action_info info={&result.str, &result.length, buf, offset, count};
1.161 paf 156:
1.289 moko 157: result.success=file_read_action_under_lock(file_spec, "read", file_read_action, &info, as_text, fail_on_read_problem);
1.223 misha 158:
1.241 misha 159: if(as_text){
160: if(result.success){
1.263 misha 161: Charset* asked_charset=0;
162: if(params)
163: if(Value* vcharset_name=params->get(PA_CHARSET_NAME))
1.291 moko 164: asked_charset=&::charsets.get(vcharset_name->as_string());
1.263 misha 165:
1.287 moko 166: asked_charset=::charsets.checkBOM(result.str, result.length, asked_charset);
167:
1.263 misha 168: 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
169: String::C body=String::C(result.str, result.length);
170: body=Charset::transcode(body, *asked_charset, charsets.source());
1.236 misha 171:
1.263 misha 172: result.str=const_cast<char*>(body.str); // hacking a little
173: result.length=body.length;
1.131 paf 174: }
175: }
1.241 misha 176: if(result.length)
177: fix_line_breaks(result.str, result.length);
1.123 paf 178: }
1.241 misha 179:
180: return result;
181: }
182:
183: File_read_result file_load(Request& r, const String& file_spec,
184: bool as_text, HashStringValue *params,
185: bool fail_on_read_problem,
186: char* buf, size_t offset, size_t count, bool transcode_text_result) {
187:
188: File_read_result result={false, 0, 0, 0};
189: if(file_spec.starts_with("http://")) {
190: if(offset || count)
1.289 moko 191: throw Exception(PARSER_RUNTIME, 0, "offset and load options are not supported for HTTP:// file load");
1.241 misha 192:
193: // fail on read problem
194: File_read_http_result http=pa_internal_file_read_http(r, file_spec, as_text, params, transcode_text_result);
195: result.success=true;
196: result.str=http.str;
197: result.length=http.length;
198: result.headers=http.headers;
199: } else
200: result=
201: file_read(r.charsets, file_spec, as_text, params, fail_on_read_problem, buf, offset, count, transcode_text_result);
1.126 paf 202:
203: return result;
1.123 paf 204: }
205:
1.257 pretende 206:
1.154 paf 207: #ifdef PA_SAFE_MODE
1.259 misha 208: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
1.154 paf 209: if(finfo.st_uid/*foreign?*/!=geteuid()
210: && finfo.st_gid/*foreign?*/!=getegid())
1.289 moko 211: throw Exception(PARSER_RUNTIME,
212: &file_spec,
213: "parser is in safe mode: reading files of foreign group and user disabled "
214: "[recompile parser with --disable-safe-mode configure option], "
215: "actual filename '%s', fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
216: fname, finfo.st_uid, geteuid(), finfo.st_gid, getegid()
217: );
1.259 misha 218: }
219: #else
220: void check_safe_mode(struct stat, const String&, const char*) {
221: }
1.257 pretende 222: #endif
1.259 misha 223:
1.257 pretende 224:
1.149 paf 225:
1.154 paf 226: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 227: const char* action_name, File_read_action action, void *context,
228: bool as_text,
1.123 paf 229: bool fail_on_read_problem) {
1.247 misha 230: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.33 paf 231: int f;
232:
233: // first open, next stat:
1.45 paf 234: // directory update of NTFS hard links performed on open.
1.33 paf 235: // ex:
236: // a.html:^test[] and b.html hardlink to a.html
237: // user inserts ! before ^test in a.html
1.126 paf 238: // directory entry of b.html in NTFS not updated at once,
1.35 paf 239: // they delay update till open, so we would receive "!^test[" string
240: // if would do stat, next open.
1.123 paf 241: // later: it seems, even this does not help sometimes
1.229 misha 242: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 243: try {
1.162 paf 244: if(pa_lock_shared_blocking(f)!=0)
1.289 moko 245: throw Exception("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.123 paf 246:
1.124 paf 247: struct stat finfo;
1.254 misha 248: if(fstat(f, &finfo)!=0)
1.124 paf 249: throw Exception("file.missing", // hardly possible: we just opened it OK
1.289 moko 250: &file_spec, "stat failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.124 paf 251:
1.149 paf 252: check_safe_mode(finfo, file_spec, fname);
1.32 paf 253:
1.289 moko 254: action(finfo, f, file_spec, context);
1.123 paf 255: } catch(...) {
1.162 paf 256: pa_unlock(f);close(f);
1.123 paf 257: if(fail_on_read_problem)
1.154 paf 258: rethrow;
1.289 moko 259: return false;
1.123 paf 260: }
1.87 paf 261:
1.162 paf 262: pa_unlock(f);close(f);
1.72 parser 263: return true;
1.229 misha 264: } else {
1.118 paf 265: if(fail_on_read_problem)
1.289 moko 266: throw Exception(errno==EACCES ? "file.access" : (errno==ENOENT || errno==ENOTDIR || errno==ENODEV) ? "file.missing" : 0,
267: &file_spec, "%s failed: %s (%d), actual filename '%s'", action_name, strerror(errno), errno, fname);
1.118 paf 268: return false;
269: }
1.8 paf 270: }
271:
1.202 paf 272: void create_dir_for_file(const String& file_spec) {
1.63 parser 273: size_t pos_after=1;
1.154 paf 274: size_t pos_before;
275: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
1.277 moko 276: pa_mkdir(file_spec.mid(0, pos_before).taint_cstr(String::L_FILE_SPEC), 0775);
1.63 parser 277: pos_after=pos_before+1;
278: }
279: }
280:
1.98 paf 281: bool file_write_action_under_lock(
1.28 paf 282: const String& file_spec,
1.225 misha 283: const char* action_name,
284: File_write_action action,
285: void *context,
1.126 paf 286: bool as_text,
287: bool do_append,
288: bool do_block,
1.110 paf 289: bool fail_on_lock_problem) {
1.247 misha 290: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.28 paf 291: int f;
1.80 paf 292: if(access(fname, W_OK)!=0) // no
1.126 paf 293: create_dir_for_file(file_spec);
1.50 paf 294:
1.80 paf 295: if((f=open(fname,
296: O_CREAT|O_RDWR
297: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 298: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 299: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.289 moko 300: Exception e("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.126 paf 301: close(f);
1.110 paf 302: if(fail_on_lock_problem)
303: throw e;
1.98 paf 304: return false;
305: }
1.96 paf 306:
1.158 paf 307: try {
1.254 misha 308: #if (defined(HAVE_FCHMOD) && defined(PA_SAFE_MODE))
309: struct stat finfo;
310: if(fstat(f, &finfo)==0 && finfo.st_mode & 0111)
311: fchmod(f, finfo.st_mode & 0666/*clear executable bits*/); // backward: ignore errors if any
312: #endif
313: action(f, context);
1.158 paf 314: } catch(...) {
1.138 paf 315: #ifdef HAVE_FTRUNCATE
1.104 paf 316: if(!do_append)
1.125 paf 317: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 318: #endif
1.162 paf 319: pa_unlock(f);close(f);
1.154 paf 320: rethrow;
1.158 paf 321: }
1.80 paf 322:
1.138 paf 323: #ifdef HAVE_FTRUNCATE
1.104 paf 324: if(!do_append)
1.125 paf 325: 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 326: #endif
1.162 paf 327: pa_unlock(f);close(f);
1.98 paf 328: return true;
1.80 paf 329: } else
1.289 moko 330: throw Exception(errno==EACCES ? "file.access" : 0, &file_spec, "%s failed: %s (%d), actual filename '%s'", action_name, strerror(errno), errno, fname);
1.96 paf 331: // here should be nothing, see rethrow above
332: }
333:
334: #ifndef DOXYGEN
335: struct File_write_action_info {
1.250 misha 336: const char* str;
337: size_t length;
1.126 paf 338: };
1.96 paf 339: #endif
1.271 moko 340:
1.96 paf 341: static void file_write_action(int f, void *context) {
1.126 paf 342: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 343: if(info.length) {
1.271 moko 344: ssize_t written=write(f, info.str, info.length);
1.116 paf 345: if(written<0)
1.271 moko 346: throw Exception("file.write", 0, "write failed: %s (%d)", strerror(errno), errno);
1.275 moko 347: if((size_t)written!=info.length)
1.271 moko 348: throw Exception("file.write", 0, "write failed: %u of %u bytes written", written, info.length);
1.113 paf 349: }
1.96 paf 350: }
1.271 moko 351:
1.96 paf 352: void file_write(
1.250 misha 353: Request_charsets& charsets,
354: const String& file_spec,
355: const char* data,
356: size_t size,
1.126 paf 357: bool as_text,
1.250 misha 358: bool do_append,
359: Charset* asked_charset) {
360:
361: if(as_text && asked_charset){
362: String::C body=String::C(data, size);
363: body=Charset::transcode(body, charsets.source(), *asked_charset);
364: data=body.str;
365: size=body.length;
366: };
367:
1.126 paf 368: File_write_action_info info={data, size};
1.225 misha 369:
1.98 paf 370: file_write_action_under_lock(
1.154 paf 371: file_spec,
1.225 misha 372: "write",
373: file_write_action,
374: &info,
1.154 paf 375: as_text,
376: do_append);
1.30 paf 377: }
378:
1.261 misha 379: static size_t get_dir(char* fname, size_t helper_length){
380: bool dir=false;
381: size_t pos=0;
382: for(pos=helper_length; pos; pos--){
383: char c=fname[pos-1];
384: if(c=='/' || c=='\\'){
385: fname[pos-1]=0;
386: dir=true;
387: } else if(dir) break;
388: }
389: return pos;
390: }
391:
392: static bool entry_readable(char* fname, bool need_dir) {
393: if(need_dir){
394: size_t size=strlen(fname);
395: while(size) {
396: char c=fname[size-1];
397: if(c=='/' || c=='\\')
398: fname[--size]=0;
399: else
400: break;
401: }
402: }
403:
404: struct stat finfo;
405: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
406: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
407: return is_dir==need_dir;
408: }
409: return false;
410: }
411:
412: static bool entry_readable(const String& file_spec, bool need_dir) {
413: return entry_readable(file_spec.taint_cstrm(String::L_FILE_SPEC), need_dir);
414: }
415:
1.63 parser 416: // throws nothing! [this is required in file_move & file_delete]
1.277 moko 417: static void rmdir(const String& file_spec, size_t pos_after) {
1.261 misha 418: char* dir_spec=file_spec.taint_cstrm(String::L_FILE_SPEC);
419: size_t length=strlen(dir_spec);
420: while( (length=get_dir(dir_spec, length)) && (length > pos_after) ){
1.274 moko 421: #ifdef _MSC_VER
1.261 misha 422: if(!entry_readable(dir_spec, true))
423: break;
424: DWORD attrs=GetFileAttributes(dir_spec);
425: if(
426: (attrs==INVALID_FILE_ATTRIBUTES)
427: || !(attrs & FILE_ATTRIBUTE_DIRECTORY)
428: || (attrs & FILE_ATTRIBUTE_REPARSE_POINT)
429: )
430: break;
431: #endif
432: if( rmdir(dir_spec) )
433: break;
434: };
1.50 paf 435: }
1.239 misha 436:
1.269 misha 437: bool file_delete(const String& file_spec, bool fail_on_problem, bool keep_empty_dirs) {
1.247 misha 438: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.282 moko 439: if(unlink(fname)!=0) {
1.164 paf 440: if(fail_on_problem)
1.289 moko 441: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
442: &file_spec, "unlink failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.93 paf 443: else
444: return false;
1.282 moko 445: }
1.50 paf 446:
1.269 misha 447: if(!keep_empty_dirs)
448: rmdir(file_spec, 1);
449:
1.93 paf 450: return true;
1.60 parser 451: }
1.239 misha 452:
1.269 misha 453: void file_move(const String& old_spec, const String& new_spec, bool keep_empty_dirs) {
1.247 misha 454: const char* old_spec_cstr=old_spec.taint_cstr(String::L_FILE_SPEC);
455: const char* new_spec_cstr=new_spec.taint_cstr(String::L_FILE_SPEC);
1.63 parser 456:
1.126 paf 457: create_dir_for_file(new_spec);
1.63 parser 458:
1.60 parser 459: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.289 moko 460: throw Exception(errno==EACCES ? "file.access" : errno==ENOENT ? "file.missing" : 0,
461: &old_spec, "rename failed: %s (%d), actual filename '%s' to '%s'", strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 462:
1.269 misha 463: if(!keep_empty_dirs)
464: rmdir(old_spec, 1);
1.31 paf 465: }
466:
1.51 paf 467:
1.126 paf 468: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 469: struct stat lfinfo;
470: bool result=stat(fname, &lfinfo)==0;
471: if(afinfo)
472: *afinfo=lfinfo;
473: return result;
1.119 paf 474: }
475:
476: bool entry_exists(const String& file_spec) {
1.247 misha 477: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.126 paf 478: return entry_exists(fname, 0);
1.118 paf 479: }
480:
1.215 paf 481: bool file_exist(const String& file_spec) {
1.126 paf 482: return entry_readable(file_spec, false);
1.51 paf 483: }
1.239 misha 484:
1.215 paf 485: bool dir_exists(const String& file_spec) {
1.126 paf 486: return entry_readable(file_spec, true);
1.65 parser 487: }
1.239 misha 488:
1.215 paf 489: const String* file_exist(const String& path, const String& name) {
1.154 paf 490: String& result=*new String(path);
1.270 moko 491: if(path.last_char() != '/')
492: result << "/";
1.154 paf 493: result << name;
1.215 paf 494: return file_exist(result)?&result:0;
1.43 paf 495: }
1.239 misha 496:
1.43 paf 497: bool file_executable(const String& file_spec) {
1.247 misha 498: return access(file_spec.taint_cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 499: }
500:
1.64 parser 501: bool file_stat(const String& file_spec,
1.229 misha 502: size_t& rsize,
503: time_t& ratime,
504: time_t& rmtime,
505: time_t& rctime,
506: bool fail_on_read_problem) {
1.247 misha 507: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.154 paf 508: struct stat finfo;
1.282 moko 509: if(stat(fname, &finfo)!=0) {
1.64 parser 510: if(fail_on_read_problem)
1.289 moko 511: throw Exception("file.missing", &file_spec, "getting file size failed: %s (%d), real filename '%s'", strerror(errno), errno, fname);
1.64 parser 512: else
513: return false;
1.282 moko 514: }
1.58 parser 515: rsize=finfo.st_size;
516: ratime=finfo.st_atime;
517: rmtime=finfo.st_mtime;
518: rctime=finfo.st_ctime;
1.64 parser 519: return true;
1.18 paf 520: }
521:
1.278 moko 522: /**
523: String related functions
524: */
525:
526: bool capitalized(const char* s){
527: bool upper=true;
528: for(const char* c=s; *c; c++){
529: if(*c != (upper ? toupper((unsigned char)*c) : tolower((unsigned char)*c)))
530: return false;
531: upper=strchr("-_ ", *c) != 0;
532: }
533: return true;
534: }
535:
536: const char* capitalize(const char* s){
537: if(!s || capitalized(s))
538: return s;
539:
540: char* result=pa_strdup(s);
541: if(result){
542: bool upper=true;
543: for(char* c=result; *c; c++){
544: *c=upper ? (char)toupper((unsigned char)*c) : (char)tolower((unsigned char)*c);
545: upper=strchr("-_ ", *c) != 0;
546: }
547: }
548: return (const char*)result;
549: }
550:
1.290 moko 551: char *str_lower(const char *s, size_t helper_length){
552: char *result=pa_strdup(s, helper_length);
553: for(char* c=result; *c; c++)
554: *c=(char)tolower((unsigned char)*c);
555: return result;
556: }
557:
558: char *str_upper(const char *s, size_t helper_length){
559: char *result=pa_strdup(s, helper_length);
560: for(char* c=result; *c; c++)
561: *c=(char)toupper((unsigned char)*c);
562: return result;
563: }
564:
1.278 moko 565: void fix_line_breaks(char *str, size_t& length) {
566: //_asm int 3;
567: const char* const eob=str+length;
568: char* dest=str;
569: // fix DOS: \r\n -> \n
570: // fix Macintosh: \r -> \n
571: char* bol=str;
572: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
573: size_t len=eol-bol;
574: if(dest!=bol)
575: memmove(dest, bol, len);
576: dest+=len;
577: *dest++='\n';
578:
579: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
580: bol=eol+2;
581: length--;
582: } else // \r, not \n = Macintosh
583: bol=eol+1;
584: }
585: // last piece without \r
586: if(dest!=bol)
587: memmove(dest, bol, eob-bol);
588: str[length]=0; // terminating
589: }
590:
591: /**
592: scans for @a delim[default \n] in @a *row_ref,
593: @return piece of line before it or end of string, if no @a delim found
594: assigns @a *row_ref to point right after delimiter if there were one
595: or to zero if no @a delim were found.
596: */
597:
1.126 paf 598: char* getrow(char* *row_ref, char delim) {
1.229 misha 599: char* result=*row_ref;
600: if(result) {
1.126 paf 601: *row_ref=strchr(result, delim);
1.8 paf 602: if(*row_ref)
603: *((*row_ref)++)=0;
604: else if(!*result)
605: return 0;
1.229 misha 606: }
607: return result;
1.8 paf 608: }
609:
1.126 paf 610: char* lsplit(char* string, char delim) {
1.229 misha 611: if(string) {
1.126 paf 612: char* v=strchr(string, delim);
1.8 paf 613: if(v) {
614: *v=0;
615: return v+1;
616: }
1.229 misha 617: }
618: return 0;
1.8 paf 619: }
620:
1.126 paf 621: char* lsplit(char* *string_ref, char delim) {
1.229 misha 622: char* result=*string_ref;
1.126 paf 623: char* next=lsplit(*string_ref, delim);
1.229 misha 624: *string_ref=next;
625: return result;
1.9 paf 626: }
627:
1.126 paf 628: char* rsplit(char* string, char delim) {
1.229 misha 629: if(string) {
1.126 paf 630: char* v=strrchr(string, delim);
1.18 paf 631: if(v) {
1.9 paf 632: *v=0;
633: return v+1;
634: }
1.229 misha 635: }
636: return NULL;
1.10 paf 637: }
638:
1.229 misha 639:
640: // format: %[flags][width][.precision]type http://msdn.microsoft.com/ru-ru/library/56e442dc(en-us,VS.80).aspx
641: // flags: '-', '+', ' ', '#', '0' http://msdn.microsoft.com/ru-ru/library/8aky45ct(en-us,VS.80).aspx
642: // width, precision: non negative decimal number
643: enum FormatType {
644: FormatInvalid,
645: FormatInt,
646: FormatUInt,
647: FormatDouble
648: };
1.272 moko 649: FormatType format_type(const char* fmt){
1.229 misha 650: enum FormatState {
651: Percent,
652: Flags,
653: Width,
654: Precision,
655: Done
656: } state=Percent;
657:
658: FormatType result=FormatInvalid;
659:
1.272 moko 660: const char* pos=fmt;
1.229 misha 661: while(char c=*(pos++)){
662: switch(state){
663: case Percent:
664: if(c=='%'){
665: state=Flags;
666: } else {
667: return FormatInvalid; // 1st char must be '%' only
668: }
669: break;
670: case Flags:
671: if(strchr("-+ #0", c)!=0){
672: break;
673: }
674: // go to the next step
675: case Width:
676: if(c=='.'){
677: state=Precision;
678: break;
679: }
680: // go to the next step
681: case Precision:
682: if(c>='0' && c<='9'){
683: if(state == Flags) state=Width; // no more flags
684: break;
685: } else if(c=='d' || c=='i'){
686: result=FormatInt;
687: } else if(strchr("feEgG", c)!=0){
688: result=FormatDouble;
689: } else if(strchr("uoxX", c)!=0){
690: result=FormatUInt;
691: } else {
692: return FormatInvalid; // invalid char
693: }
694: state=Done;
695: break;
696: case Done:
697: return FormatInvalid; // no chars allowed after 'type'
698: }
699: }
700: return result;
701: }
702:
703:
1.272 moko 704: const char* format(double value, const char* fmt) {
1.229 misha 705: char local_buf[MAX_NUMBER];
1.235 misha 706: int size=-1;
1.229 misha 707:
708: if(fmt && strlen(fmt)){
709: switch(format_type(fmt)){
710: case FormatDouble:
711: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
712: break;
713: case FormatInt:
714: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
715: break;
716: case FormatUInt:
1.126 paf 717: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.229 misha 718: break;
719: case FormatInvalid:
1.289 moko 720: throw Exception(PARSER_RUNTIME, 0, "Incorrect format string '%s' was specified.", fmt);
1.229 misha 721: }
722: } else
723: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
724:
725: if(size < 0 || size >= MAX_NUMBER-1){ // on win32 we manually reduce max size while printing
1.289 moko 726: throw Exception(PARSER_RUNTIME, 0, "Error occure white executing snprintf with format string '%s'.", fmt);
1.229 misha 727: }
728:
1.235 misha 729: return pa_strdup(local_buf, (size_t)size);
1.12 paf 730: }
731:
1.36 paf 732: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 733: #ifdef WIN32
1.187 paf 734: size_t to_write = size;
1.12 paf 735: do{
1.154 paf 736: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 737: if(chunk_written<=0)
738: break;
739: size-=chunk_written;
1.36 paf 740: buf=((const char*)buf)+chunk_written;
1.126 paf 741: } while(size>0);
1.12 paf 742:
1.187 paf 743: return to_write-size;
1.12 paf 744: #else
1.126 paf 745: return fwrite(buf, 1, size, stdout);
1.12 paf 746: #endif
1.2 paf 747: }
1.14 paf 748:
1.229 misha 749: enum EscapeState {
750: EscapeRest,
751: EscapeFirst,
752: EscapeSecond,
753: EscapeUnicode
754: };
755:
1.236 misha 756: // @todo prescan for reduce required size (unescaped sting in 1 byte charset requires less memory usually)
1.258 misha 757: char* unescape_chars(const char* cp, int len, Charset* charset, bool js){
1.236 misha 758: char* s=new(PointerFreeGC) char[len+1]; // must be enough (%uXXXX==6 bytes, max utf-8 char length==6 bytes)
1.230 misha 759: char* dst=s;
1.229 misha 760: EscapeState escapeState=EscapeRest;
761: uint escapedValue=0;
762: int srcPos=0;
1.230 misha 763: short int jsCnt=0;
1.236 misha 764: while(srcPos<len){
1.229 misha 765: uchar c=(uchar)cp[srcPos];
1.258 misha 766: if(c=='%' || (c=='\\' && js)){
1.229 misha 767: escapeState=EscapeFirst;
768: } else {
769: switch(escapeState) {
770: case EscapeRest:
1.286 moko 771: if(c=='+' && !js){
1.230 misha 772: *dst++=' ';
1.229 misha 773: } else {
1.230 misha 774: *dst++=c;
1.229 misha 775: }
776: break;
777: case EscapeFirst:
1.232 misha 778: if(charset && c=='u'){
1.229 misha 779: // escaped unicode value: %u0430
780: jsCnt=0;
781: escapedValue=0;
782: escapeState=EscapeUnicode;
783: } else {
1.231 misha 784: if(isxdigit(c)){
1.229 misha 785: escapedValue=hex_value[c] << 4;
786: escapeState=EscapeSecond;
787: } else {
1.230 misha 788: *dst++=c;
1.229 misha 789: escapeState=EscapeRest;
790: }
791: }
792: break;
793: case EscapeSecond:
1.231 misha 794: if(isxdigit(c)){
1.229 misha 795: escapedValue+=hex_value[c];
1.230 misha 796: *dst++=(char)escapedValue;
1.229 misha 797: }
798: escapeState=EscapeRest;
799: break;
800: case EscapeUnicode:
1.231 misha 801: if(isxdigit(c)){
1.229 misha 802: escapedValue=(escapedValue << 4) + hex_value[c];
803: if(++jsCnt==4){
1.230 misha 804: // transcode utf8 char to client charset (we can lost some chars here)
1.232 misha 805: charset->store_Char((XMLByte*&)dst, (XMLCh)escapedValue, '?');
1.229 misha 806: escapeState=EscapeRest;
807: }
808: } else {
809: // not full unicode value
810: escapeState=EscapeRest;
811: }
812: break;
813: }
814: }
815:
816: srcPos++;
817: }
818:
1.230 misha 819: *dst=0; // zero-termination
1.229 misha 820: return s;
821: }
1.24 paf 822:
1.268 misha 823: char *search_stop(char*& current, char cstop_at) {
824: // sanity check
825: if(!current)
826: return 0;
827:
828: // skip leading WS
829: while(*current==' ' || *current=='\t')
830: current++;
831: if(!*current)
832: return current=0;
833:
834: char *result=current;
835: if(char *pstop_at=strchr(current, cstop_at)) {
836: *pstop_at=0;
837: current=pstop_at+1;
838: } else
839: current=0;
840: return result;
841: }
842:
1.24 paf 843: #ifdef WIN32
1.126 paf 844: void back_slashes_to_slashes(char* s) {
1.24 paf 845: if(s)
846: for(; *s; s++)
847: if(*s=='\\')
1.126 paf 848: *s='/';
1.24 paf 849: }
850: #endif
1.41 paf 851:
1.232 misha 852: size_t strpos(const char *str, const char *substr) {
853: const char *p = strstr(str, substr);
854: return (p==0)?STRING_NOT_FOUND:p-str;
855: }
856:
1.226 misha 857: int remove_crlf(char* start, char* end) {
858: char* from=start;
859: char* to=start;
860: bool skip=false;
861: while(from < end){
862: switch(*from){
863: case '\n':
864: case '\r':
865: case '\t':
866: case ' ':
867: if(!skip){
868: *to=' ';
869: to++;
870: skip=true;
871: }
872: break;
873: default:
874: if(from != to)
875: *to=*from;
876: to++;
877: skip=false;
1.69 parser 878: }
1.226 misha 879: from++;
880: }
881: return to-start;
1.91 paf 882: }
883:
1.279 moko 884: const char* hex_digits="0123456789ABCDEF";
885:
1.278 moko 886: const char* hex_string(unsigned char* bytes, size_t size, bool upcase) {
887: char *bytes_hex=new(PointerFreeGC) char [size*2/*byte->hh*/+1/*for zero-teminator*/];
888: unsigned char *src=bytes;
889: unsigned char *end=bytes+size;
890: char *dest=bytes_hex;
891:
1.279 moko 892: const char *hex=upcase? hex_digits : "0123456789abcdef";
1.278 moko 893:
894: for(; src<end; src++) {
895: *dest++=hex[*src/0x10];
896: *dest++=hex[*src%0x10];
897: }
898: *dest=0;
899:
900: return bytes_hex;
901: }
1.91 paf 902:
1.178 paf 903: /* mime64 functions are from libgmime[http://spruce.sourceforge.net/gmime/] lib */
904: /*
905: * Authors: Michael Zucchi <notzed@helixcode.com>
906: * Jeffrey Stedfast <fejj@helixcode.com>
907: *
908: * Copyright 2000 Helix Code, Inc. (www.helixcode.com)
909: *
910: * This program is free software; you can redistribute it and/or modify
911: * it under the terms of the GNU General Public License as published by
912: * the Free Software Foundation; either version 2 of the License, or
913: * (at your option) any later version.
914: *
915: * This program is distributed in the hope that it will be useful,
916: * but WITHOUT ANY WARRANTY; without even the implied warranty of
917: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
918: * GNU General Public License for more details.
919: *
920: * You should have received a copy of the GNU General Public License
921: * along with this program; if not, write to the Free Software
922: * Foundation, Inc., 59 Temple Street #330, Boston, MA 02111-1307, USA.
923: *
924: */
1.271 moko 925: static const char *base64_alphabet =
1.178 paf 926: "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
927:
928: /**
929: * g_mime_utils_base64_encode_step:
930: * @in: input stream
931: * @inlen: length of the input
932: * @out: output string
933: * @state: holds the number of bits that are stored in @save
934: * @save: leftover bits that have not yet been encoded
935: *
936: * Base64 encodes a chunk of data. Performs an 'encode step', only
937: * encodes blocks of 3 characters to the output at a time, saves
938: * left-over state in state and save (initialise to 0 on first
939: * invocation).
940: *
941: * Returns the number of bytes encoded.
942: **/
1.252 misha 943:
944: #define BASE64_GROUPS_IN_LINE 19
945:
1.178 paf 946: static size_t
947: g_mime_utils_base64_encode_step (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
948: {
1.186 paf 949: register const unsigned char *inptr;
1.178 paf 950: register unsigned char *outptr;
951:
952: if (inlen <= 0)
953: return 0;
954:
955: inptr = in;
956: outptr = out;
957:
958: if (inlen + ((unsigned char *)save)[0] > 2) {
959: const unsigned char *inend = in + inlen - 2;
960: register int c1 = 0, c2 = 0, c3 = 0;
961: register int already;
962:
963: already = *state;
964:
965: switch (((char *)save)[0]) {
966: case 1: c1 = ((unsigned char *)save)[1]; goto skip1;
967: case 2: c1 = ((unsigned char *)save)[1];
968: c2 = ((unsigned char *)save)[2]; goto skip2;
969: }
970:
971: /* yes, we jump into the loop, no i'm not going to change it, its beautiful! */
972: while (inptr < inend) {
973: c1 = *inptr++;
974: skip1:
975: c2 = *inptr++;
976: skip2:
977: c3 = *inptr++;
978: *outptr++ = base64_alphabet [c1 >> 2];
979: *outptr++ = base64_alphabet [(c2 >> 4) | ((c1 & 0x3) << 4)];
980: *outptr++ = base64_alphabet [((c2 & 0x0f) << 2) | (c3 >> 6)];
981: *outptr++ = base64_alphabet [c3 & 0x3f];
982: /* this is a bit ugly ... */
1.252 misha 983: if ((++already) >= BASE64_GROUPS_IN_LINE) {
1.178 paf 984: *outptr++ = '\n';
985: already = 0;
986: }
987: }
988:
989: ((unsigned char *)save)[0] = 0;
990: inlen = 2 - (inptr - inend);
991: *state = already;
992: }
993:
994: //d(printf ("state = %d, inlen = %d\n", (int)((char *)save)[0], inlen));
995:
996: if (inlen > 0) {
997: register char *saveout;
998:
999: /* points to the slot for the next char to save */
1000: saveout = & (((char *)save)[1]) + ((char *)save)[0];
1001:
1002: /* inlen can only be 0 1 or 2 */
1003: switch (inlen) {
1004: case 2: *saveout++ = *inptr++;
1005: case 1: *saveout++ = *inptr++;
1006: }
1.216 paf 1007: *(char *)save = *(char *)save+(char)inlen;
1.178 paf 1008: }
1009:
1010: /*d(printf ("mode = %d\nc1 = %c\nc2 = %c\n",
1011: (int)((char *)save)[0],
1012: (int)((char *)save)[1],
1013: (int)((char *)save)[2]));*/
1014:
1015: return (outptr - out);
1016: }
1017:
1018: /**
1019: * g_mime_utils_base64_encode_close:
1020: * @in: input stream
1021: * @inlen: length of the input
1022: * @out: output string
1023: * @state: holds the number of bits that are stored in @save
1024: * @save: leftover bits that have not yet been encoded
1025: *
1026: * Base64 encodes the input stream to the output stream. Call this
1027: * when finished encoding data with g_mime_utils_base64_encode_step to
1028: * flush off the last little bit.
1029: *
1030: * Returns the number of bytes encoded.
1031: **/
1032: static size_t
1033: g_mime_utils_base64_encode_close (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1034: {
1035: unsigned char *outptr = out;
1036: int c1, c2;
1037:
1038: if (inlen > 0)
1039: outptr += g_mime_utils_base64_encode_step (in, inlen, outptr, state, save);
1040:
1041: c1 = ((unsigned char *)save)[1];
1042: c2 = ((unsigned char *)save)[2];
1043:
1044: switch (((unsigned char *)save)[0]) {
1045: case 2:
1046: outptr[2] = base64_alphabet [(c2 & 0x0f) << 2];
1047: goto skip;
1048: case 1:
1049: outptr[2] = '=';
1050: skip:
1051: outptr[0] = base64_alphabet [c1 >> 2];
1052: outptr[1] = base64_alphabet [c2 >> 4 | ((c1 & 0x3) << 4)];
1053: outptr[3] = '=';
1054: outptr += 4;
1055: break;
1056: }
1057:
1058: *outptr++ = 0;
1059:
1060: *save = 0;
1061: *state = 0;
1062:
1063: return (outptr - out);
1064: }
1065:
1.210 paf 1066: static unsigned char gmime_base64_rank[256] = {
1.266 misha 1067: 255,255,255,255,255,255,255,255,255,254,254,255,255,254,255,255,
1.210 paf 1068: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1.266 misha 1069: 254,255,255,255,255,255,255,255,255,255,255, 62,255,255,255, 63,
1.210 paf 1070: 52, 53, 54, 55, 56, 57, 58, 59, 60, 61,255,255,255, 0,255,255,
1071: 255, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,
1072: 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25,255,255,255,255,255,
1073: 255, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40,
1074: 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51,255,255,255,255,255,
1075: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1076: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1077: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1078: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1079: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1080: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1081: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1082: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1083: };
1084:
1085: /**
1086: * g_mime_utils_base64_decode_step:
1087: * @in: input stream
1088: * @inlen: max length of data to decode
1089: * @out: output stream
1090: * @state: holds the number of bits that are stored in @save
1091: * @save: leftover bits that have not yet been decoded
1.266 misha 1092: * @strict: only base64 and whitespace chars are allowed
1.210 paf 1093: *
1094: * Decodes a chunk of base64 encoded data.
1095: *
1096: * Returns the number of bytes decoded (which have been dumped in @out).
1097: **/
1098: size_t
1.266 misha 1099: 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 1100: {
1.213 paf 1101: const unsigned char *inptr;
1102: unsigned char *outptr;
1.210 paf 1103: const unsigned char *inend;
1.213 paf 1104: int saved;
1.210 paf 1105: unsigned char c;
1106: int i;
1107:
1108: inend = in + inlen;
1109: outptr = out;
1110:
1111: /* convert 4 base64 bytes to 3 normal bytes */
1112: saved = *save;
1113: i = *state;
1114: inptr = in;
1115: while (inptr < inend) {
1116: c = gmime_base64_rank[*inptr++];
1.266 misha 1117: switch(c) {
1118: case 0xff: // non-base64 and non-whitespace chars. not allowed in strict mode
1119: if(strict)
1120: throw Exception(BASE64_FORMAT, 0, "Invalid base64 char on position %d is detected", inptr-in-1);
1121: case 0xfe: // whitespace chars 0x09, 0x0A, 0x0D, 0x20 are allowed in any mode
1122: break;
1123: default:
1124: saved = (saved << 6) | c;
1125: i++;
1126: if (i == 4) {
1127: *outptr++ = (unsigned char)(saved >> 16);
1128: *outptr++ = (unsigned char)(saved >> 8);
1129: *outptr++ = (unsigned char)(saved);
1130: i = 0;
1131: }
1.210 paf 1132: }
1133: }
1134:
1135: *save = saved;
1136: *state = i;
1137:
1138: /* quick scan back for '=' on the end somewhere */
1139: /* fortunately we can drop 1 output char for each trailing = (upto 2) */
1140: i = 2;
1141: while (inptr > in && i) {
1142: inptr--;
1.266 misha 1143: if (gmime_base64_rank[*inptr] <= 0xfe) {
1.210 paf 1144: if (*inptr == '=' && outptr > out)
1145: outptr--;
1146: i--;
1147: }
1148: }
1149:
1150: /* if i != 0 then there is a truncation error! */
1151: return (outptr - out);
1152: }
1153:
1154:
1.239 misha 1155: char* pa_base64_encode(const char *in, size_t in_size){
1.252 misha 1156: size_t new_size = ((in_size / 3 + 1) * 4);
1157: new_size += new_size / (BASE64_GROUPS_IN_LINE * 4)/*new lines*/ + 1/*zero terminator*/;
1158: char* result = new(PointerFreeGC) char[new_size];
1.178 paf 1159: int state=0;
1160: int save=0;
1.183 paf 1161: #ifndef NDEBUG
1162: size_t filled=
1163: #endif
1.251 misha 1164: g_mime_utils_base64_encode_close ((const unsigned char*)in, in_size, (unsigned char*)result, &state, &save);
1165:
1166: //throw Exception(PARSER_RUNTIME, 0, "%d %d %d", in_size, new_size, filled);
1167: assert(filled <= new_size);
1.178 paf 1168:
1169: return result;
1.98 paf 1170: }
1.210 paf 1171:
1.278 moko 1172: struct File_base64_action_info {
1173: unsigned char** base64;
1174: };
1.222 misha 1175:
1.289 moko 1176: static void file_base64_file_action(struct stat& finfo, int f, const String&, void *context) {
1.222 misha 1177:
1178: if(finfo.st_size) {
1179: File_base64_action_info& info=*static_cast<File_base64_action_info *>(context);
1180: *info.base64=new(PointerFreeGC) unsigned char[finfo.st_size * 2 + 6];
1181: unsigned char* base64 = *info.base64;
1182: int state=0;
1183: int save=0;
1184: int nCount;
1185: do {
1186: unsigned char buffer[FILE_BUFFER_SIZE];
1187: nCount = file_block_read(f, buffer, sizeof(buffer));
1188: if( nCount ){
1189: size_t filled=g_mime_utils_base64_encode_step ((const unsigned char*)buffer, nCount, base64, &state, &save);
1190: base64+=filled;
1191: }
1192: } while(nCount > 0);
1193: g_mime_utils_base64_encode_close (0, 0, base64, &state, &save);
1194: }
1195: }
1196:
1.278 moko 1197: char* pa_base64_encode(const String& file_spec){
1198: unsigned char* base64=0;
1199: File_base64_action_info info={&base64};
1200:
1.289 moko 1201: file_read_action_under_lock(file_spec, "pa_base64_encode", file_base64_file_action, &info);
1.278 moko 1202:
1203: return (char*)base64;
1204: }
1205:
1.265 misha 1206: void pa_base64_decode(const char *in, size_t in_size, char*& result, size_t& result_size, bool strict) {
1.264 misha 1207: // every 4 base64 bytes are converted into 3 normal bytes
1208: // not full set (tail) of 4-bytes set is ignored
1209: size_t new_size=in_size/4*3;
1210: result=new(PointerFreeGC) char[new_size+1/*terminator*/];
1211:
1.210 paf 1212: int state=0;
1213: int save=0;
1.289 moko 1214: result_size=g_mime_utils_base64_decode_step ((const unsigned char*)in, in_size, (unsigned char*)result, &state, &save, strict);
1.264 misha 1215: assert(result_size <= new_size);
1.211 paf 1216: result[result_size]=0; // for text files
1.265 misha 1217:
1218: if(strict && state!=0)
1.266 misha 1219: throw Exception(BASE64_FORMAT, 0, "Unexpected end of chars");
1.210 paf 1220: }
1.218 misha 1221:
1222:
1.221 misha 1223: int file_block_read(const int f, unsigned char* buffer, const size_t size){
1224: int nCount = read(f, buffer, size);
1225: if (nCount < 0)
1.289 moko 1226: throw Exception("file.read", 0, "read failed: %s (%d)", strerror(errno), errno);
1.221 misha 1227: return nCount;
1228: }
1229:
1.278 moko 1230: static unsigned long crc32Table[256];
1231: static void InitCrc32Table()
1232: {
1233: if(crc32Table[1] == 0){
1234: // This is the official polynomial used by CRC32 in PKZip.
1235: // Often times the polynomial shown reversed as 0x04C11DB7.
1236: static const unsigned long dwPolynomial = 0xEDB88320;
1237:
1238: for(int i = 0; i < 256; i++)
1239: {
1240: unsigned long dwCrc = i;
1241: for(int j = 8; j > 0; j--)
1242: {
1243: if(dwCrc & 1)
1244: dwCrc = (dwCrc >> 1) ^ dwPolynomial;
1245: else
1246: dwCrc >>= 1;
1247: }
1248: crc32Table[i] = dwCrc;
1249: }
1250: }
1251: }
1252:
1253: inline void CalcCrc32(const unsigned char byte, unsigned long &crc32)
1254: {
1255: crc32 = ((crc32) >> 8) ^ crc32Table[(byte) ^ ((crc32) & 0x000000FF)];
1256: }
1257:
1258:
1.239 misha 1259: const unsigned long pa_crc32(const char *in, size_t in_size){
1.218 misha 1260: unsigned long crc32=0xFFFFFFFF;
1.220 misha 1261:
1.240 misha 1262: InitCrc32Table();
1.239 misha 1263: for(size_t i = 0; i<in_size; i++)
1264: CalcCrc32(in[i], crc32);
1.220 misha 1265:
1.218 misha 1266: return ~crc32;
1267: }
1268:
1.289 moko 1269: static void file_crc32_file_action(struct stat& finfo, int f, const String&, void *context) {
1.218 misha 1270: unsigned long& crc32=*static_cast<unsigned long *>(context);
1271: if(finfo.st_size) {
1272: InitCrc32Table();
1.220 misha 1273: int nCount=0;
1.218 misha 1274: do {
1.221 misha 1275: unsigned char buffer[FILE_BUFFER_SIZE];
1276: nCount = file_block_read(f, buffer, sizeof(buffer));
1.220 misha 1277: for(int i = 0; i < nCount; i++) CalcCrc32(buffer[i], crc32);
1278: } while(nCount > 0);
1.218 misha 1279: }
1280: }
1281:
1.278 moko 1282: const unsigned long pa_crc32(const String& file_spec){
1283: unsigned long crc32=0xFFFFFFFF;
1284: file_read_action_under_lock(file_spec, "crc32", file_crc32_file_action, &crc32);
1285: return ~crc32;
1286: }
1287:
1288: // content-type: xxx; charset=WE-NEED-THIS
1289: // content-type: xxx; charset="WE-NEED-THIS"
1290: // content-type: xxx; charset="WE-NEED-THIS";
1291: Charset* detect_charset(const char* content_type){
1292: if(content_type){
1.291 moko 1293: char* CONTENT_TYPE=str_upper(content_type);
1.278 moko 1294:
1295: if(const char* begin=strstr(CONTENT_TYPE, "CHARSET=")){
1296: begin+=8; // skip "CHARSET="
1297: char* end=0;
1298: if(*begin && (*begin=='"' || *begin =='\'')){
1299: char quote=*begin;
1300: begin++;
1301: end=(char*)strchr(begin, quote);
1302: }
1303: if(!end)
1304: end=(char*)strchr(begin, ';');
1305:
1306: if(end)
1307: *end=0; // terminator
1308:
1.291 moko 1309: return *begin?&charsets.get_direct(begin):0;
1.278 moko 1310: }
1311: }
1312: return 0;
1313: }
1314:
1315:
1.283 moko 1316: static bool is_latin(const char *in){
1317: for(; *in; in++){
1318: if ((unsigned char)(*in) > 0x7F)
1319: return false;
1320: }
1321: return true;
1322: }
1323:
1324: #define MAX_IDNA_LENGTH 256
1325:
1326: const char *pa_idna_encode(const char *in, Charset &source_charset){
1327: if(!in || is_latin(in))
1328: return in;
1329:
1330: uint32_t utf32[MAX_IDNA_LENGTH];
1331: uint32_t *utf32_end=utf32;
1332:
1333: String::C sIn(in,strlen(in));
1334:
1335: if(!source_charset.isUTF8())
1336: sIn=Charset::transcode(sIn, source_charset, UTF8_charset);
1337:
1338: int status=pa_convertUTF8toUTF32((const UTF8**)&sIn.str, (const UTF8*)(sIn.str+sIn.length), &utf32_end, utf32+MAX_IDNA_LENGTH-1, strictConversion);
1339: if(status != conversionOK)
1340: throw Exception("idna encode", new String(in), "utf conversion failed (%d)", status);
1341:
1342: *utf32_end=0;
1343:
1344: char *result = (char *)pa_malloc(MAX_IDNA_LENGTH);
1345: status=pa_idna_to_ascii_4z(utf32, result, MAX_IDNA_LENGTH, 0);
1346: if(status != IDNA_SUCCESS)
1347: throw Exception("idna encode", new String(in), "encode failed: %s", pa_idna_strerror(status));
1348:
1349: return result;
1350: }
1351:
1352: const char *pa_idna_decode(const char *in, Charset &asked_charset){
1353: if(!in || !(*in))
1354: return in;
1355:
1356: uint32_t utf32[MAX_IDNA_LENGTH];
1357: const uint32_t *utf32_start=utf32;
1358: uint32_t *utf32_end;
1359:
1360: int status=pa_idna_to_unicode_4z(in, utf32, MAX_IDNA_LENGTH, 0);
1361: if(status != IDNA_SUCCESS)
1362: throw Exception("idna decode", new String(in), "decode failed: %s", pa_idna_strerror(status));
1363:
1364: for(utf32_end=utf32; *utf32_end; utf32_end++);
1365:
1366: char *result = (char *)pa_malloc(MAX_IDNA_LENGTH);
1367: char *result_end = result;
1368:
1369: status=pa_convertUTF32toUTF8(&utf32_start, utf32_end, (UTF8**)&result_end, (UTF8*)(result+MAX_IDNA_LENGTH-1), strictConversion);
1370:
1371: if(status != conversionOK)
1372: throw Exception("idna decode", new String(in), "utf conversion failed (%d)", status);
1373:
1374: *result_end='\0';
1375:
1376: if(!asked_charset.isUTF8())
1377: result = (char *)Charset::transcode(result, UTF8_charset, asked_charset).cstr();
1378:
1379: return result;
1380: }
1.292 moko 1381: /// must be last in this file
1382: #undef vsnprintf
1383: int pa_vsnprintf(char* b, size_t s, const char* f, va_list l) {
1384: if(!s)
1385: return 0;
1386:
1387: int r;
1388: // note: on win32 & maybe somewhere else
1389: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1390: // http://stackoverflow.com/questions/2915672/snprintf-and-visual-studio-2010
1391: --s;
1392:
1393: // clients do not check for negative 's', feature: ignore such prints
1394: if((ssize_t)s<0)
1395: return 0;
1396:
1397: #ifdef _MSC_VER
1398: // win32: if the number of bytes to write exceeds buffer, then count bytes are written and -1 is returned
1399: r=_vsnprintf(b, s, f, l);
1400: if(r<0)
1401: r=s;
1402: #else
1403: r=vsnprintf(b, s, f, l);
1404: /*
1405: solaris: man vsnprintf
1406:
1407: The snprintf() function returns the number of characters
1408: formatted, that is, the number of characters that would have
1409: been written to the buffer if it were large enough. If the
1410: value of n is 0 on a call to snprintf(), an unspecified
1411: value less than 1 is returned.
1412: */
1413:
1414: if(r<0)
1415: r=0;
1416: else if((size_t)r>s)
1417: r=s;
1418: #endif
1419: b[r]=0;
1420: return r;
1421: }
1422:
1423: int pa_snprintf(char* b, size_t s, const char* f, ...) {
1424: va_list l;
1425: va_start(l, f);
1426: int r=pa_vsnprintf(b, s, f, l);
1427: va_end(l);
1428: return r;
1429: }
1430:
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