Annotation of parser3/src/main/pa_common.C, revision 1.303
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.303 ! moko 4: Copyright (c) 2000-2017 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:
1.303 ! moko 53: volatile const char * IDENT_PA_COMMON_C="$Id: pa_common.C,v 1.302 2017/02/06 16:33:33 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:
78: // globals
79:
80: const String file_status_name(FILE_STATUS_NAME);
81:
1.301 moko 82: // forwards
83:
84: const UTF16* pa_utf16_encode(const char* in, Charset& source_charset);
85:
1.154 paf 86: // functions
1.127 paf 87:
1.301 moko 88: #ifdef _MSC_VER
89:
90: int pa_stat(const char *pathname, struct stat *buffer){
91: const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source());
92: return _wstat64((const wchar_t *)utf16name, buffer);
93: }
94:
95: int pa_open(const char *pathname, int flags, int mode){
96: const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source());
97: return _wopen((const wchar_t *)utf16name, flags, mode);
98: }
99:
100: FILE *pa_fopen(const char *pathname, const char *mode){
101: const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source());
102: const UTF16* utf16mode=pa_utf16_encode(mode, pa_thread_request().charsets.source());
103: return _wfopen((const wchar_t *)utf16name, (const wchar_t *)utf16mode);
104: }
105:
106: #endif
107:
1.271 moko 108: char* file_read_text(Request_charsets& charsets, const String& file_spec, bool fail_on_read_problem, HashStringValue* params, bool transcode_result) {
109: File_read_result file=file_read(charsets, file_spec, true, params, fail_on_read_problem, 0, 0, 0, transcode_result);
1.154 paf 110: return file.success?file.str:0;
1.126 paf 111: }
112:
1.271 moko 113: char* file_load_text(Request& r, const String& file_spec, bool fail_on_read_problem, HashStringValue* params, bool transcode_result) {
114: File_read_result file=file_load(r, file_spec, true, params, fail_on_read_problem, 0, 0, 0, transcode_result);
1.241 misha 115: return file.success?file.str:0;
116: }
117:
1.206 paf 118: /// these options were handled but not checked elsewhere, now check them
1.239 misha 119: int pa_get_valid_file_options_count(HashStringValue& options) {
1.206 paf 120: int result=0;
121: if(options.get(PA_SQL_LIMIT_NAME))
122: result++;
123: if(options.get(PA_SQL_OFFSET_NAME))
124: result++;
125: if(options.get(PA_COLUMN_SEPARATOR_NAME))
126: result++;
127: if(options.get(PA_COLUMN_ENCLOSER_NAME))
128: result++;
1.223 misha 129: if(options.get(PA_CHARSET_NAME))
130: result++;
1.206 paf 131: return result;
132: }
133:
1.123 paf 134: #ifndef DOXYGEN
135: struct File_read_action_info {
1.154 paf 136: char **data; size_t *data_size;
1.188 paf 137: char* buf; size_t offset; size_t count;
1.126 paf 138: };
1.123 paf 139: #endif
1.271 moko 140:
1.289 moko 141: static void file_read_action(struct stat& finfo, int f, const String& file_spec, void *context) {
1.126 paf 142: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.188 paf 143: size_t to_read_size=info.count;
144: if(!to_read_size)
1.300 moko 145: to_read_size=check_file_size(finfo.st_size, file_spec);
1.271 moko 146: if(to_read_size) {
1.188 paf 147: if(info.offset)
148: lseek(f, info.offset, SEEK_SET);
1.271 moko 149: *info.data=info.buf ? info.buf : (char *)pa_malloc_atomic(to_read_size+1);
150: ssize_t result=read(f, *info.data, to_read_size);
151: if(result<0)
152: throw Exception("file.read", &file_spec, "read failed: %s (%d)", strerror(errno), errno);
153: *info.data_size=result;
1.123 paf 154: } else { // empty file
1.209 paf 155: // 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 156: *info.data=(char *)pa_malloc_atomic(1);
1.209 paf 157: *(char*)(*info.data)=0;
1.123 paf 158: *info.data_size=0;
159: return;
160: }
1.126 paf 161: }
1.241 misha 162:
1.154 paf 163: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
1.229 misha 164: bool as_text, HashStringValue *params,
165: bool fail_on_read_problem,
1.234 misha 166: char* buf, size_t offset, size_t count, bool transcode_text_result) {
1.167 paf 167: File_read_result result={false, 0, 0, 0};
1.241 misha 168: if(params){
169: int valid_options=pa_get_valid_file_options_count(*params);
170: if(valid_options!=params->count())
1.262 misha 171: throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION);
1.241 misha 172: }
1.203 paf 173:
1.241 misha 174: File_read_action_info info={&result.str, &result.length, buf, offset, count};
1.161 paf 175:
1.289 moko 176: result.success=file_read_action_under_lock(file_spec, "read", file_read_action, &info, as_text, fail_on_read_problem);
1.223 misha 177:
1.241 misha 178: if(as_text){
179: if(result.success){
1.263 misha 180: Charset* asked_charset=0;
181: if(params)
182: if(Value* vcharset_name=params->get(PA_CHARSET_NAME))
1.295 moko 183: asked_charset=&pa_charsets.get(vcharset_name->as_string());
1.263 misha 184:
1.295 moko 185: asked_charset=pa_charsets.checkBOM(result.str, result.length, asked_charset);
1.287 moko 186:
1.263 misha 187: 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
188: String::C body=String::C(result.str, result.length);
189: body=Charset::transcode(body, *asked_charset, charsets.source());
1.236 misha 190:
1.263 misha 191: result.str=const_cast<char*>(body.str); // hacking a little
192: result.length=body.length;
1.131 paf 193: }
194: }
1.241 misha 195: if(result.length)
196: fix_line_breaks(result.str, result.length);
1.123 paf 197: }
1.241 misha 198:
199: return result;
200: }
201:
202: File_read_result file_load(Request& r, const String& file_spec,
203: bool as_text, HashStringValue *params,
204: bool fail_on_read_problem,
205: char* buf, size_t offset, size_t count, bool transcode_text_result) {
206:
207: File_read_result result={false, 0, 0, 0};
208: if(file_spec.starts_with("http://")) {
209: if(offset || count)
1.289 moko 210: throw Exception(PARSER_RUNTIME, 0, "offset and load options are not supported for HTTP:// file load");
1.241 misha 211:
212: // fail on read problem
213: File_read_http_result http=pa_internal_file_read_http(r, file_spec, as_text, params, transcode_text_result);
214: result.success=true;
215: result.str=http.str;
216: result.length=http.length;
217: result.headers=http.headers;
218: } else
219: result=
220: file_read(r.charsets, file_spec, as_text, params, fail_on_read_problem, buf, offset, count, transcode_text_result);
1.126 paf 221:
222: return result;
1.123 paf 223: }
224:
1.257 pretende 225:
1.154 paf 226: #ifdef PA_SAFE_MODE
1.259 misha 227: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
1.154 paf 228: if(finfo.st_uid/*foreign?*/!=geteuid()
229: && finfo.st_gid/*foreign?*/!=getegid())
1.289 moko 230: throw Exception(PARSER_RUNTIME,
231: &file_spec,
232: "parser is in safe mode: reading files of foreign group and user disabled "
233: "[recompile parser with --disable-safe-mode configure option], "
234: "actual filename '%s', fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
235: fname, finfo.st_uid, geteuid(), finfo.st_gid, getegid()
236: );
1.259 misha 237: }
238: #else
239: void check_safe_mode(struct stat, const String&, const char*) {
240: }
1.257 pretende 241: #endif
1.259 misha 242:
1.257 pretende 243:
1.149 paf 244:
1.154 paf 245: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 246: const char* action_name, File_read_action action, void *context,
247: bool as_text,
1.123 paf 248: bool fail_on_read_problem) {
1.247 misha 249: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.33 paf 250: int f;
251:
252: // first open, next stat:
1.45 paf 253: // directory update of NTFS hard links performed on open.
1.33 paf 254: // ex:
255: // a.html:^test[] and b.html hardlink to a.html
256: // user inserts ! before ^test in a.html
1.126 paf 257: // directory entry of b.html in NTFS not updated at once,
1.35 paf 258: // they delay update till open, so we would receive "!^test[" string
259: // if would do stat, next open.
1.123 paf 260: // later: it seems, even this does not help sometimes
1.301 moko 261: if((f=pa_open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 262: try {
1.162 paf 263: if(pa_lock_shared_blocking(f)!=0)
1.289 moko 264: throw Exception("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.123 paf 265:
1.124 paf 266: struct stat finfo;
1.298 moko 267: if(pa_fstat(f, &finfo)!=0)
1.124 paf 268: throw Exception("file.missing", // hardly possible: we just opened it OK
1.289 moko 269: &file_spec, "stat failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.124 paf 270:
1.149 paf 271: check_safe_mode(finfo, file_spec, fname);
1.32 paf 272:
1.289 moko 273: action(finfo, f, file_spec, context);
1.123 paf 274: } catch(...) {
1.162 paf 275: pa_unlock(f);close(f);
1.123 paf 276: if(fail_on_read_problem)
1.154 paf 277: rethrow;
1.289 moko 278: return false;
1.123 paf 279: }
1.87 paf 280:
1.162 paf 281: pa_unlock(f);close(f);
1.72 parser 282: return true;
1.229 misha 283: } else {
1.118 paf 284: if(fail_on_read_problem)
1.289 moko 285: throw Exception(errno==EACCES ? "file.access" : (errno==ENOENT || errno==ENOTDIR || errno==ENODEV) ? "file.missing" : 0,
286: &file_spec, "%s failed: %s (%d), actual filename '%s'", action_name, strerror(errno), errno, fname);
1.118 paf 287: return false;
288: }
1.8 paf 289: }
290:
1.202 paf 291: void create_dir_for_file(const String& file_spec) {
1.63 parser 292: size_t pos_after=1;
1.154 paf 293: size_t pos_before;
294: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
1.277 moko 295: pa_mkdir(file_spec.mid(0, pos_before).taint_cstr(String::L_FILE_SPEC), 0775);
1.63 parser 296: pos_after=pos_before+1;
297: }
298: }
299:
1.98 paf 300: bool file_write_action_under_lock(
1.28 paf 301: const String& file_spec,
1.225 misha 302: const char* action_name,
303: File_write_action action,
304: void *context,
1.126 paf 305: bool as_text,
306: bool do_append,
307: bool do_block,
1.110 paf 308: bool fail_on_lock_problem) {
1.247 misha 309: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.28 paf 310: int f;
1.80 paf 311: if(access(fname, W_OK)!=0) // no
1.126 paf 312: create_dir_for_file(file_spec);
1.50 paf 313:
1.301 moko 314: if((f=pa_open(fname,
1.80 paf 315: O_CREAT|O_RDWR
316: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 317: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 318: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.289 moko 319: Exception e("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.126 paf 320: close(f);
1.110 paf 321: if(fail_on_lock_problem)
322: throw e;
1.98 paf 323: return false;
324: }
1.96 paf 325:
1.158 paf 326: try {
1.254 misha 327: #if (defined(HAVE_FCHMOD) && defined(PA_SAFE_MODE))
328: struct stat finfo;
1.298 moko 329: if(pa_fstat(f, &finfo)==0 && finfo.st_mode & 0111)
1.254 misha 330: fchmod(f, finfo.st_mode & 0666/*clear executable bits*/); // backward: ignore errors if any
331: #endif
332: action(f, context);
1.158 paf 333: } catch(...) {
1.138 paf 334: #ifdef HAVE_FTRUNCATE
1.104 paf 335: if(!do_append)
1.125 paf 336: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 337: #endif
1.162 paf 338: pa_unlock(f);close(f);
1.154 paf 339: rethrow;
1.158 paf 340: }
1.80 paf 341:
1.138 paf 342: #ifdef HAVE_FTRUNCATE
1.104 paf 343: if(!do_append)
1.125 paf 344: 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 345: #endif
1.162 paf 346: pa_unlock(f);close(f);
1.98 paf 347: return true;
1.80 paf 348: } else
1.289 moko 349: 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 350: // here should be nothing, see rethrow above
351: }
352:
353: #ifndef DOXYGEN
354: struct File_write_action_info {
1.250 misha 355: const char* str;
356: size_t length;
1.126 paf 357: };
1.96 paf 358: #endif
1.271 moko 359:
1.96 paf 360: static void file_write_action(int f, void *context) {
1.126 paf 361: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 362: if(info.length) {
1.271 moko 363: ssize_t written=write(f, info.str, info.length);
1.116 paf 364: if(written<0)
1.271 moko 365: throw Exception("file.write", 0, "write failed: %s (%d)", strerror(errno), errno);
1.275 moko 366: if((size_t)written!=info.length)
1.271 moko 367: throw Exception("file.write", 0, "write failed: %u of %u bytes written", written, info.length);
1.113 paf 368: }
1.96 paf 369: }
1.271 moko 370:
1.96 paf 371: void file_write(
1.250 misha 372: Request_charsets& charsets,
373: const String& file_spec,
374: const char* data,
375: size_t size,
1.126 paf 376: bool as_text,
1.250 misha 377: bool do_append,
378: Charset* asked_charset) {
379:
380: if(as_text && asked_charset){
381: String::C body=String::C(data, size);
382: body=Charset::transcode(body, charsets.source(), *asked_charset);
383: data=body.str;
384: size=body.length;
385: };
386:
1.126 paf 387: File_write_action_info info={data, size};
1.225 misha 388:
1.98 paf 389: file_write_action_under_lock(
1.154 paf 390: file_spec,
1.225 misha 391: "write",
392: file_write_action,
393: &info,
1.154 paf 394: as_text,
395: do_append);
1.30 paf 396: }
397:
1.261 misha 398: static size_t get_dir(char* fname, size_t helper_length){
399: bool dir=false;
400: size_t pos=0;
401: for(pos=helper_length; pos; pos--){
402: char c=fname[pos-1];
403: if(c=='/' || c=='\\'){
404: fname[pos-1]=0;
405: dir=true;
406: } else if(dir) break;
407: }
408: return pos;
409: }
410:
411: static bool entry_readable(char* fname, bool need_dir) {
412: if(need_dir){
413: size_t size=strlen(fname);
414: while(size) {
415: char c=fname[size-1];
416: if(c=='/' || c=='\\')
417: fname[--size]=0;
418: else
419: break;
420: }
421: }
422:
423: struct stat finfo;
424: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
425: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
426: return is_dir==need_dir;
427: }
428: return false;
429: }
430:
431: static bool entry_readable(const String& file_spec, bool need_dir) {
432: return entry_readable(file_spec.taint_cstrm(String::L_FILE_SPEC), need_dir);
433: }
434:
1.63 parser 435: // throws nothing! [this is required in file_move & file_delete]
1.277 moko 436: static void rmdir(const String& file_spec, size_t pos_after) {
1.261 misha 437: char* dir_spec=file_spec.taint_cstrm(String::L_FILE_SPEC);
438: size_t length=strlen(dir_spec);
439: while( (length=get_dir(dir_spec, length)) && (length > pos_after) ){
1.274 moko 440: #ifdef _MSC_VER
1.261 misha 441: if(!entry_readable(dir_spec, true))
442: break;
443: DWORD attrs=GetFileAttributes(dir_spec);
444: if(
445: (attrs==INVALID_FILE_ATTRIBUTES)
446: || !(attrs & FILE_ATTRIBUTE_DIRECTORY)
447: || (attrs & FILE_ATTRIBUTE_REPARSE_POINT)
448: )
449: break;
450: #endif
451: if( rmdir(dir_spec) )
452: break;
453: };
1.50 paf 454: }
1.239 misha 455:
1.269 misha 456: bool file_delete(const String& file_spec, bool fail_on_problem, bool keep_empty_dirs) {
1.247 misha 457: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.282 moko 458: if(unlink(fname)!=0) {
1.164 paf 459: if(fail_on_problem)
1.289 moko 460: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
461: &file_spec, "unlink failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.93 paf 462: else
463: return false;
1.282 moko 464: }
1.50 paf 465:
1.269 misha 466: if(!keep_empty_dirs)
467: rmdir(file_spec, 1);
468:
1.93 paf 469: return true;
1.60 parser 470: }
1.239 misha 471:
1.269 misha 472: void file_move(const String& old_spec, const String& new_spec, bool keep_empty_dirs) {
1.247 misha 473: const char* old_spec_cstr=old_spec.taint_cstr(String::L_FILE_SPEC);
474: const char* new_spec_cstr=new_spec.taint_cstr(String::L_FILE_SPEC);
1.63 parser 475:
1.126 paf 476: create_dir_for_file(new_spec);
1.63 parser 477:
1.60 parser 478: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.289 moko 479: throw Exception(errno==EACCES ? "file.access" : errno==ENOENT ? "file.missing" : 0,
480: &old_spec, "rename failed: %s (%d), actual filename '%s' to '%s'", strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 481:
1.269 misha 482: if(!keep_empty_dirs)
483: rmdir(old_spec, 1);
1.31 paf 484: }
485:
1.51 paf 486:
1.126 paf 487: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 488: struct stat lfinfo;
1.298 moko 489: bool result=pa_stat(fname, &lfinfo)==0;
1.118 paf 490: if(afinfo)
491: *afinfo=lfinfo;
492: return result;
1.119 paf 493: }
494:
495: bool entry_exists(const String& file_spec) {
1.247 misha 496: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.126 paf 497: return entry_exists(fname, 0);
1.118 paf 498: }
499:
1.215 paf 500: bool file_exist(const String& file_spec) {
1.126 paf 501: return entry_readable(file_spec, false);
1.51 paf 502: }
1.239 misha 503:
1.215 paf 504: bool dir_exists(const String& file_spec) {
1.126 paf 505: return entry_readable(file_spec, true);
1.65 parser 506: }
1.239 misha 507:
1.215 paf 508: const String* file_exist(const String& path, const String& name) {
1.154 paf 509: String& result=*new String(path);
1.270 moko 510: if(path.last_char() != '/')
511: result << "/";
1.154 paf 512: result << name;
1.215 paf 513: return file_exist(result)?&result:0;
1.43 paf 514: }
1.239 misha 515:
1.43 paf 516: bool file_executable(const String& file_spec) {
1.247 misha 517: return access(file_spec.taint_cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 518: }
519:
1.296 moko 520: bool file_stat(const String& file_spec, uint64_t& rsize, time_t& ratime, time_t& rmtime, time_t& rctime, bool fail_on_read_problem) {
1.247 misha 521: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.154 paf 522: struct stat finfo;
1.298 moko 523: if(pa_stat(fname, &finfo)!=0) {
1.64 parser 524: if(fail_on_read_problem)
1.289 moko 525: throw Exception("file.missing", &file_spec, "getting file size failed: %s (%d), real filename '%s'", strerror(errno), errno, fname);
1.64 parser 526: else
527: return false;
1.282 moko 528: }
1.58 parser 529: rsize=finfo.st_size;
1.299 moko 530: ratime=(time_t)finfo.st_atime;
531: rmtime=(time_t)finfo.st_mtime;
532: rctime=(time_t)finfo.st_ctime;
1.64 parser 533: return true;
1.18 paf 534: }
535:
1.300 moko 536: size_t check_file_size(uint64_t size, const String& file_spec){
537: if(size > pa_file_size_limit)
538: throw Exception(PARSER_RUNTIME, &file_spec, "content size of %.15g bytes exceeds the limit (%.15g bytes)", (double)size, (double)pa_file_size_limit);
539: return (size_t)size;
540: }
541:
1.278 moko 542: /**
543: String related functions
544: */
545:
546: bool capitalized(const char* s){
547: bool upper=true;
548: for(const char* c=s; *c; c++){
549: if(*c != (upper ? toupper((unsigned char)*c) : tolower((unsigned char)*c)))
550: return false;
551: upper=strchr("-_ ", *c) != 0;
552: }
553: return true;
554: }
555:
556: const char* capitalize(const char* s){
557: if(!s || capitalized(s))
558: return s;
559:
560: char* result=pa_strdup(s);
561: if(result){
562: bool upper=true;
563: for(char* c=result; *c; c++){
564: *c=upper ? (char)toupper((unsigned char)*c) : (char)tolower((unsigned char)*c);
565: upper=strchr("-_ ", *c) != 0;
566: }
567: }
568: return (const char*)result;
569: }
570:
1.290 moko 571: char *str_lower(const char *s, size_t helper_length){
572: char *result=pa_strdup(s, helper_length);
573: for(char* c=result; *c; c++)
574: *c=(char)tolower((unsigned char)*c);
575: return result;
576: }
577:
578: char *str_upper(const char *s, size_t helper_length){
579: char *result=pa_strdup(s, helper_length);
580: for(char* c=result; *c; c++)
581: *c=(char)toupper((unsigned char)*c);
582: return result;
583: }
584:
1.278 moko 585: void fix_line_breaks(char *str, size_t& length) {
586: //_asm int 3;
587: const char* const eob=str+length;
588: char* dest=str;
589: // fix DOS: \r\n -> \n
590: // fix Macintosh: \r -> \n
591: char* bol=str;
592: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
593: size_t len=eol-bol;
594: if(dest!=bol)
595: memmove(dest, bol, len);
596: dest+=len;
597: *dest++='\n';
598:
599: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
600: bol=eol+2;
601: length--;
602: } else // \r, not \n = Macintosh
603: bol=eol+1;
604: }
605: // last piece without \r
606: if(dest!=bol)
607: memmove(dest, bol, eob-bol);
608: str[length]=0; // terminating
609: }
610:
611: /**
612: scans for @a delim[default \n] in @a *row_ref,
613: @return piece of line before it or end of string, if no @a delim found
614: assigns @a *row_ref to point right after delimiter if there were one
615: or to zero if no @a delim were found.
616: */
617:
1.126 paf 618: char* getrow(char* *row_ref, char delim) {
1.229 misha 619: char* result=*row_ref;
620: if(result) {
1.126 paf 621: *row_ref=strchr(result, delim);
1.8 paf 622: if(*row_ref)
623: *((*row_ref)++)=0;
624: else if(!*result)
625: return 0;
1.229 misha 626: }
627: return result;
1.8 paf 628: }
629:
1.126 paf 630: char* lsplit(char* string, char delim) {
1.229 misha 631: if(string) {
1.126 paf 632: char* v=strchr(string, delim);
1.8 paf 633: if(v) {
634: *v=0;
635: return v+1;
636: }
1.229 misha 637: }
638: return 0;
1.8 paf 639: }
640:
1.126 paf 641: char* lsplit(char* *string_ref, char delim) {
1.229 misha 642: char* result=*string_ref;
1.126 paf 643: char* next=lsplit(*string_ref, delim);
1.229 misha 644: *string_ref=next;
645: return result;
1.9 paf 646: }
647:
1.126 paf 648: char* rsplit(char* string, char delim) {
1.229 misha 649: if(string) {
1.126 paf 650: char* v=strrchr(string, delim);
1.18 paf 651: if(v) {
1.9 paf 652: *v=0;
653: return v+1;
654: }
1.229 misha 655: }
656: return NULL;
1.10 paf 657: }
658:
1.229 misha 659:
660: // format: %[flags][width][.precision]type http://msdn.microsoft.com/ru-ru/library/56e442dc(en-us,VS.80).aspx
661: // flags: '-', '+', ' ', '#', '0' http://msdn.microsoft.com/ru-ru/library/8aky45ct(en-us,VS.80).aspx
662: // width, precision: non negative decimal number
663: enum FormatType {
664: FormatInvalid,
665: FormatInt,
666: FormatUInt,
667: FormatDouble
668: };
1.272 moko 669: FormatType format_type(const char* fmt){
1.229 misha 670: enum FormatState {
671: Percent,
672: Flags,
673: Width,
674: Precision,
675: Done
676: } state=Percent;
677:
678: FormatType result=FormatInvalid;
679:
1.272 moko 680: const char* pos=fmt;
1.229 misha 681: while(char c=*(pos++)){
682: switch(state){
683: case Percent:
684: if(c=='%'){
685: state=Flags;
686: } else {
687: return FormatInvalid; // 1st char must be '%' only
688: }
689: break;
690: case Flags:
691: if(strchr("-+ #0", c)!=0){
692: break;
693: }
694: // go to the next step
695: case Width:
696: if(c=='.'){
697: state=Precision;
698: break;
699: }
700: // go to the next step
701: case Precision:
702: if(c>='0' && c<='9'){
703: if(state == Flags) state=Width; // no more flags
704: break;
705: } else if(c=='d' || c=='i'){
706: result=FormatInt;
707: } else if(strchr("feEgG", c)!=0){
708: result=FormatDouble;
709: } else if(strchr("uoxX", c)!=0){
710: result=FormatUInt;
711: } else {
712: return FormatInvalid; // invalid char
713: }
714: state=Done;
715: break;
716: case Done:
717: return FormatInvalid; // no chars allowed after 'type'
718: }
719: }
720: return result;
721: }
722:
723:
1.272 moko 724: const char* format(double value, const char* fmt) {
1.229 misha 725: char local_buf[MAX_NUMBER];
1.235 misha 726: int size=-1;
1.229 misha 727:
728: if(fmt && strlen(fmt)){
729: switch(format_type(fmt)){
730: case FormatDouble:
731: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
732: break;
733: case FormatInt:
734: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
735: break;
736: case FormatUInt:
1.126 paf 737: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.229 misha 738: break;
739: case FormatInvalid:
1.289 moko 740: throw Exception(PARSER_RUNTIME, 0, "Incorrect format string '%s' was specified.", fmt);
1.229 misha 741: }
742: } else
743: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
744:
745: if(size < 0 || size >= MAX_NUMBER-1){ // on win32 we manually reduce max size while printing
1.289 moko 746: throw Exception(PARSER_RUNTIME, 0, "Error occure white executing snprintf with format string '%s'.", fmt);
1.229 misha 747: }
748:
1.235 misha 749: return pa_strdup(local_buf, (size_t)size);
1.12 paf 750: }
751:
1.36 paf 752: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 753: #ifdef WIN32
1.187 paf 754: size_t to_write = size;
1.12 paf 755: do{
1.154 paf 756: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 757: if(chunk_written<=0)
758: break;
759: size-=chunk_written;
1.36 paf 760: buf=((const char*)buf)+chunk_written;
1.126 paf 761: } while(size>0);
1.12 paf 762:
1.187 paf 763: return to_write-size;
1.12 paf 764: #else
1.126 paf 765: return fwrite(buf, 1, size, stdout);
1.12 paf 766: #endif
1.2 paf 767: }
1.14 paf 768:
1.229 misha 769: enum EscapeState {
770: EscapeRest,
771: EscapeFirst,
772: EscapeSecond,
773: EscapeUnicode
774: };
775:
1.236 misha 776: // @todo prescan for reduce required size (unescaped sting in 1 byte charset requires less memory usually)
1.258 misha 777: char* unescape_chars(const char* cp, int len, Charset* charset, bool js){
1.236 misha 778: char* s=new(PointerFreeGC) char[len+1]; // must be enough (%uXXXX==6 bytes, max utf-8 char length==6 bytes)
1.230 misha 779: char* dst=s;
1.229 misha 780: EscapeState escapeState=EscapeRest;
781: uint escapedValue=0;
782: int srcPos=0;
1.230 misha 783: short int jsCnt=0;
1.236 misha 784: while(srcPos<len){
1.229 misha 785: uchar c=(uchar)cp[srcPos];
1.258 misha 786: if(c=='%' || (c=='\\' && js)){
1.229 misha 787: escapeState=EscapeFirst;
788: } else {
789: switch(escapeState) {
790: case EscapeRest:
1.286 moko 791: if(c=='+' && !js){
1.230 misha 792: *dst++=' ';
1.229 misha 793: } else {
1.230 misha 794: *dst++=c;
1.229 misha 795: }
796: break;
797: case EscapeFirst:
1.232 misha 798: if(charset && c=='u'){
1.229 misha 799: // escaped unicode value: %u0430
800: jsCnt=0;
801: escapedValue=0;
802: escapeState=EscapeUnicode;
803: } else {
1.231 misha 804: if(isxdigit(c)){
1.229 misha 805: escapedValue=hex_value[c] << 4;
806: escapeState=EscapeSecond;
807: } else {
1.230 misha 808: *dst++=c;
1.229 misha 809: escapeState=EscapeRest;
810: }
811: }
812: break;
813: case EscapeSecond:
1.231 misha 814: if(isxdigit(c)){
1.229 misha 815: escapedValue+=hex_value[c];
1.230 misha 816: *dst++=(char)escapedValue;
1.229 misha 817: }
818: escapeState=EscapeRest;
819: break;
820: case EscapeUnicode:
1.231 misha 821: if(isxdigit(c)){
1.229 misha 822: escapedValue=(escapedValue << 4) + hex_value[c];
823: if(++jsCnt==4){
1.230 misha 824: // transcode utf8 char to client charset (we can lost some chars here)
1.232 misha 825: charset->store_Char((XMLByte*&)dst, (XMLCh)escapedValue, '?');
1.229 misha 826: escapeState=EscapeRest;
827: }
828: } else {
829: // not full unicode value
830: escapeState=EscapeRest;
831: }
832: break;
833: }
834: }
835:
836: srcPos++;
837: }
838:
1.230 misha 839: *dst=0; // zero-termination
1.229 misha 840: return s;
841: }
1.24 paf 842:
1.268 misha 843: char *search_stop(char*& current, char cstop_at) {
844: // sanity check
845: if(!current)
846: return 0;
847:
848: // skip leading WS
849: while(*current==' ' || *current=='\t')
850: current++;
851: if(!*current)
852: return current=0;
853:
854: char *result=current;
855: if(char *pstop_at=strchr(current, cstop_at)) {
856: *pstop_at=0;
857: current=pstop_at+1;
858: } else
859: current=0;
860: return result;
861: }
862:
1.24 paf 863: #ifdef WIN32
1.126 paf 864: void back_slashes_to_slashes(char* s) {
1.24 paf 865: if(s)
866: for(; *s; s++)
867: if(*s=='\\')
1.126 paf 868: *s='/';
1.24 paf 869: }
870: #endif
1.41 paf 871:
1.232 misha 872: size_t strpos(const char *str, const char *substr) {
873: const char *p = strstr(str, substr);
874: return (p==0)?STRING_NOT_FOUND:p-str;
875: }
876:
1.226 misha 877: int remove_crlf(char* start, char* end) {
878: char* from=start;
879: char* to=start;
880: bool skip=false;
881: while(from < end){
882: switch(*from){
883: case '\n':
884: case '\r':
885: case '\t':
886: case ' ':
887: if(!skip){
888: *to=' ';
889: to++;
890: skip=true;
891: }
892: break;
893: default:
894: if(from != to)
895: *to=*from;
896: to++;
897: skip=false;
1.69 parser 898: }
1.226 misha 899: from++;
900: }
901: return to-start;
1.91 paf 902: }
903:
1.279 moko 904: const char* hex_digits="0123456789ABCDEF";
905:
1.278 moko 906: const char* hex_string(unsigned char* bytes, size_t size, bool upcase) {
907: char *bytes_hex=new(PointerFreeGC) char [size*2/*byte->hh*/+1/*for zero-teminator*/];
908: unsigned char *src=bytes;
909: unsigned char *end=bytes+size;
910: char *dest=bytes_hex;
911:
1.279 moko 912: const char *hex=upcase? hex_digits : "0123456789abcdef";
1.278 moko 913:
914: for(; src<end; src++) {
915: *dest++=hex[*src/0x10];
916: *dest++=hex[*src%0x10];
917: }
918: *dest=0;
919:
920: return bytes_hex;
921: }
1.91 paf 922:
1.178 paf 923: /* mime64 functions are from libgmime[http://spruce.sourceforge.net/gmime/] lib */
924: /*
925: * Authors: Michael Zucchi <notzed@helixcode.com>
926: * Jeffrey Stedfast <fejj@helixcode.com>
927: *
928: * Copyright 2000 Helix Code, Inc. (www.helixcode.com)
929: *
930: * This program is free software; you can redistribute it and/or modify
931: * it under the terms of the GNU General Public License as published by
932: * the Free Software Foundation; either version 2 of the License, or
933: * (at your option) any later version.
934: *
935: * This program is distributed in the hope that it will be useful,
936: * but WITHOUT ANY WARRANTY; without even the implied warranty of
937: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
938: * GNU General Public License for more details.
939: *
940: * You should have received a copy of the GNU General Public License
941: * along with this program; if not, write to the Free Software
942: * Foundation, Inc., 59 Temple Street #330, Boston, MA 02111-1307, USA.
943: *
944: */
1.271 moko 945: static const char *base64_alphabet =
1.178 paf 946: "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
947:
948: /**
949: * g_mime_utils_base64_encode_step:
950: * @in: input stream
951: * @inlen: length of the input
952: * @out: output string
953: * @state: holds the number of bits that are stored in @save
954: * @save: leftover bits that have not yet been encoded
955: *
956: * Base64 encodes a chunk of data. Performs an 'encode step', only
957: * encodes blocks of 3 characters to the output at a time, saves
958: * left-over state in state and save (initialise to 0 on first
959: * invocation).
960: *
961: * Returns the number of bytes encoded.
962: **/
1.252 misha 963:
964: #define BASE64_GROUPS_IN_LINE 19
965:
1.178 paf 966: static size_t
967: g_mime_utils_base64_encode_step (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
968: {
1.186 paf 969: register const unsigned char *inptr;
1.178 paf 970: register unsigned char *outptr;
971:
972: if (inlen <= 0)
973: return 0;
974:
975: inptr = in;
976: outptr = out;
977:
978: if (inlen + ((unsigned char *)save)[0] > 2) {
979: const unsigned char *inend = in + inlen - 2;
980: register int c1 = 0, c2 = 0, c3 = 0;
981: register int already;
982:
983: already = *state;
984:
985: switch (((char *)save)[0]) {
986: case 1: c1 = ((unsigned char *)save)[1]; goto skip1;
987: case 2: c1 = ((unsigned char *)save)[1];
988: c2 = ((unsigned char *)save)[2]; goto skip2;
989: }
990:
991: /* yes, we jump into the loop, no i'm not going to change it, its beautiful! */
992: while (inptr < inend) {
993: c1 = *inptr++;
994: skip1:
995: c2 = *inptr++;
996: skip2:
997: c3 = *inptr++;
998: *outptr++ = base64_alphabet [c1 >> 2];
999: *outptr++ = base64_alphabet [(c2 >> 4) | ((c1 & 0x3) << 4)];
1000: *outptr++ = base64_alphabet [((c2 & 0x0f) << 2) | (c3 >> 6)];
1001: *outptr++ = base64_alphabet [c3 & 0x3f];
1002: /* this is a bit ugly ... */
1.252 misha 1003: if ((++already) >= BASE64_GROUPS_IN_LINE) {
1.178 paf 1004: *outptr++ = '\n';
1005: already = 0;
1006: }
1007: }
1008:
1009: ((unsigned char *)save)[0] = 0;
1010: inlen = 2 - (inptr - inend);
1011: *state = already;
1012: }
1013:
1014: //d(printf ("state = %d, inlen = %d\n", (int)((char *)save)[0], inlen));
1015:
1016: if (inlen > 0) {
1017: register char *saveout;
1018:
1019: /* points to the slot for the next char to save */
1020: saveout = & (((char *)save)[1]) + ((char *)save)[0];
1021:
1022: /* inlen can only be 0 1 or 2 */
1023: switch (inlen) {
1024: case 2: *saveout++ = *inptr++;
1025: case 1: *saveout++ = *inptr++;
1026: }
1.216 paf 1027: *(char *)save = *(char *)save+(char)inlen;
1.178 paf 1028: }
1029:
1030: /*d(printf ("mode = %d\nc1 = %c\nc2 = %c\n",
1031: (int)((char *)save)[0],
1032: (int)((char *)save)[1],
1033: (int)((char *)save)[2]));*/
1034:
1035: return (outptr - out);
1036: }
1037:
1038: /**
1039: * g_mime_utils_base64_encode_close:
1040: * @in: input stream
1041: * @inlen: length of the input
1042: * @out: output string
1043: * @state: holds the number of bits that are stored in @save
1044: * @save: leftover bits that have not yet been encoded
1045: *
1046: * Base64 encodes the input stream to the output stream. Call this
1047: * when finished encoding data with g_mime_utils_base64_encode_step to
1048: * flush off the last little bit.
1049: *
1050: * Returns the number of bytes encoded.
1051: **/
1052: static size_t
1053: g_mime_utils_base64_encode_close (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1054: {
1055: unsigned char *outptr = out;
1056: int c1, c2;
1057:
1058: if (inlen > 0)
1059: outptr += g_mime_utils_base64_encode_step (in, inlen, outptr, state, save);
1060:
1061: c1 = ((unsigned char *)save)[1];
1062: c2 = ((unsigned char *)save)[2];
1063:
1064: switch (((unsigned char *)save)[0]) {
1065: case 2:
1066: outptr[2] = base64_alphabet [(c2 & 0x0f) << 2];
1067: goto skip;
1068: case 1:
1069: outptr[2] = '=';
1070: skip:
1071: outptr[0] = base64_alphabet [c1 >> 2];
1072: outptr[1] = base64_alphabet [c2 >> 4 | ((c1 & 0x3) << 4)];
1073: outptr[3] = '=';
1074: outptr += 4;
1075: break;
1076: }
1077:
1078: *outptr++ = 0;
1079:
1080: *save = 0;
1081: *state = 0;
1082:
1083: return (outptr - out);
1084: }
1085:
1.210 paf 1086: static unsigned char gmime_base64_rank[256] = {
1.266 misha 1087: 255,255,255,255,255,255,255,255,255,254,254,255,255,254,255,255,
1.210 paf 1088: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1.266 misha 1089: 254,255,255,255,255,255,255,255,255,255,255, 62,255,255,255, 63,
1.210 paf 1090: 52, 53, 54, 55, 56, 57, 58, 59, 60, 61,255,255,255, 0,255,255,
1091: 255, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,
1092: 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25,255,255,255,255,255,
1093: 255, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40,
1094: 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51,255,255,255,255,255,
1095: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1096: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1097: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1098: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1099: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1100: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1101: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1102: 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
1103: };
1104:
1105: /**
1106: * g_mime_utils_base64_decode_step:
1107: * @in: input stream
1108: * @inlen: max length of data to decode
1109: * @out: output stream
1110: * @state: holds the number of bits that are stored in @save
1111: * @save: leftover bits that have not yet been decoded
1.266 misha 1112: * @strict: only base64 and whitespace chars are allowed
1.210 paf 1113: *
1114: * Decodes a chunk of base64 encoded data.
1115: *
1116: * Returns the number of bytes decoded (which have been dumped in @out).
1117: **/
1118: size_t
1.266 misha 1119: 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 1120: {
1.213 paf 1121: const unsigned char *inptr;
1122: unsigned char *outptr;
1.210 paf 1123: const unsigned char *inend;
1.213 paf 1124: int saved;
1.210 paf 1125: unsigned char c;
1126: int i;
1127:
1128: inend = in + inlen;
1129: outptr = out;
1130:
1131: /* convert 4 base64 bytes to 3 normal bytes */
1132: saved = *save;
1133: i = *state;
1134: inptr = in;
1135: while (inptr < inend) {
1136: c = gmime_base64_rank[*inptr++];
1.266 misha 1137: switch(c) {
1138: case 0xff: // non-base64 and non-whitespace chars. not allowed in strict mode
1139: if(strict)
1140: throw Exception(BASE64_FORMAT, 0, "Invalid base64 char on position %d is detected", inptr-in-1);
1141: case 0xfe: // whitespace chars 0x09, 0x0A, 0x0D, 0x20 are allowed in any mode
1142: break;
1143: default:
1144: saved = (saved << 6) | c;
1145: i++;
1146: if (i == 4) {
1147: *outptr++ = (unsigned char)(saved >> 16);
1148: *outptr++ = (unsigned char)(saved >> 8);
1149: *outptr++ = (unsigned char)(saved);
1150: i = 0;
1151: }
1.210 paf 1152: }
1153: }
1154:
1155: *save = saved;
1156: *state = i;
1157:
1158: /* quick scan back for '=' on the end somewhere */
1159: /* fortunately we can drop 1 output char for each trailing = (upto 2) */
1160: i = 2;
1161: while (inptr > in && i) {
1162: inptr--;
1.266 misha 1163: if (gmime_base64_rank[*inptr] <= 0xfe) {
1.210 paf 1164: if (*inptr == '=' && outptr > out)
1165: outptr--;
1166: i--;
1167: }
1168: }
1169:
1170: /* if i != 0 then there is a truncation error! */
1171: return (outptr - out);
1172: }
1173:
1174:
1.239 misha 1175: char* pa_base64_encode(const char *in, size_t in_size){
1.252 misha 1176: size_t new_size = ((in_size / 3 + 1) * 4);
1177: new_size += new_size / (BASE64_GROUPS_IN_LINE * 4)/*new lines*/ + 1/*zero terminator*/;
1178: char* result = new(PointerFreeGC) char[new_size];
1.178 paf 1179: int state=0;
1180: int save=0;
1.183 paf 1181: #ifndef NDEBUG
1182: size_t filled=
1183: #endif
1.251 misha 1184: g_mime_utils_base64_encode_close ((const unsigned char*)in, in_size, (unsigned char*)result, &state, &save);
1185:
1186: //throw Exception(PARSER_RUNTIME, 0, "%d %d %d", in_size, new_size, filled);
1187: assert(filled <= new_size);
1.178 paf 1188:
1189: return result;
1.98 paf 1190: }
1.210 paf 1191:
1.278 moko 1192: struct File_base64_action_info {
1193: unsigned char** base64;
1194: };
1.222 misha 1195:
1.300 moko 1196: static void file_base64_file_action(struct stat& finfo, int f, const String& file_spec, void *context) {
1.222 misha 1197:
1198: if(finfo.st_size) {
1199: File_base64_action_info& info=*static_cast<File_base64_action_info *>(context);
1.300 moko 1200: *info.base64=new(PointerFreeGC) unsigned char[check_file_size(finfo.st_size, file_spec) * 2 + 6];
1.222 misha 1201: unsigned char* base64 = *info.base64;
1202: int state=0;
1203: int save=0;
1204: int nCount;
1205: do {
1206: unsigned char buffer[FILE_BUFFER_SIZE];
1207: nCount = file_block_read(f, buffer, sizeof(buffer));
1208: if( nCount ){
1209: size_t filled=g_mime_utils_base64_encode_step ((const unsigned char*)buffer, nCount, base64, &state, &save);
1210: base64+=filled;
1211: }
1212: } while(nCount > 0);
1213: g_mime_utils_base64_encode_close (0, 0, base64, &state, &save);
1214: }
1215: }
1216:
1.278 moko 1217: char* pa_base64_encode(const String& file_spec){
1218: unsigned char* base64=0;
1219: File_base64_action_info info={&base64};
1220:
1.289 moko 1221: file_read_action_under_lock(file_spec, "pa_base64_encode", file_base64_file_action, &info);
1.278 moko 1222:
1223: return (char*)base64;
1224: }
1225:
1.265 misha 1226: void pa_base64_decode(const char *in, size_t in_size, char*& result, size_t& result_size, bool strict) {
1.264 misha 1227: // every 4 base64 bytes are converted into 3 normal bytes
1228: // not full set (tail) of 4-bytes set is ignored
1229: size_t new_size=in_size/4*3;
1230: result=new(PointerFreeGC) char[new_size+1/*terminator*/];
1231:
1.210 paf 1232: int state=0;
1233: int save=0;
1.289 moko 1234: result_size=g_mime_utils_base64_decode_step ((const unsigned char*)in, in_size, (unsigned char*)result, &state, &save, strict);
1.264 misha 1235: assert(result_size <= new_size);
1.211 paf 1236: result[result_size]=0; // for text files
1.265 misha 1237:
1238: if(strict && state!=0)
1.266 misha 1239: throw Exception(BASE64_FORMAT, 0, "Unexpected end of chars");
1.210 paf 1240: }
1.218 misha 1241:
1242:
1.221 misha 1243: int file_block_read(const int f, unsigned char* buffer, const size_t size){
1244: int nCount = read(f, buffer, size);
1245: if (nCount < 0)
1.289 moko 1246: throw Exception("file.read", 0, "read failed: %s (%d)", strerror(errno), errno);
1.221 misha 1247: return nCount;
1248: }
1249:
1.278 moko 1250: static unsigned long crc32Table[256];
1251: static void InitCrc32Table()
1252: {
1253: if(crc32Table[1] == 0){
1254: // This is the official polynomial used by CRC32 in PKZip.
1255: // Often times the polynomial shown reversed as 0x04C11DB7.
1256: static const unsigned long dwPolynomial = 0xEDB88320;
1257:
1258: for(int i = 0; i < 256; i++)
1259: {
1260: unsigned long dwCrc = i;
1261: for(int j = 8; j > 0; j--)
1262: {
1263: if(dwCrc & 1)
1264: dwCrc = (dwCrc >> 1) ^ dwPolynomial;
1265: else
1266: dwCrc >>= 1;
1267: }
1268: crc32Table[i] = dwCrc;
1269: }
1270: }
1271: }
1272:
1273: inline void CalcCrc32(const unsigned char byte, unsigned long &crc32)
1274: {
1275: crc32 = ((crc32) >> 8) ^ crc32Table[(byte) ^ ((crc32) & 0x000000FF)];
1276: }
1277:
1278:
1.297 moko 1279: unsigned long pa_crc32(const char *in, size_t in_size){
1.218 misha 1280: unsigned long crc32=0xFFFFFFFF;
1.220 misha 1281:
1.240 misha 1282: InitCrc32Table();
1.239 misha 1283: for(size_t i = 0; i<in_size; i++)
1284: CalcCrc32(in[i], crc32);
1.220 misha 1285:
1.218 misha 1286: return ~crc32;
1287: }
1288:
1.289 moko 1289: static void file_crc32_file_action(struct stat& finfo, int f, const String&, void *context) {
1.218 misha 1290: unsigned long& crc32=*static_cast<unsigned long *>(context);
1291: if(finfo.st_size) {
1292: InitCrc32Table();
1.220 misha 1293: int nCount=0;
1.218 misha 1294: do {
1.221 misha 1295: unsigned char buffer[FILE_BUFFER_SIZE];
1296: nCount = file_block_read(f, buffer, sizeof(buffer));
1.220 misha 1297: for(int i = 0; i < nCount; i++) CalcCrc32(buffer[i], crc32);
1298: } while(nCount > 0);
1.218 misha 1299: }
1300: }
1301:
1.297 moko 1302: unsigned long pa_crc32(const String& file_spec){
1.278 moko 1303: unsigned long crc32=0xFFFFFFFF;
1304: file_read_action_under_lock(file_spec, "crc32", file_crc32_file_action, &crc32);
1305: return ~crc32;
1306: }
1307:
1308: // content-type: xxx; charset=WE-NEED-THIS
1309: // content-type: xxx; charset="WE-NEED-THIS"
1310: // content-type: xxx; charset="WE-NEED-THIS";
1311: Charset* detect_charset(const char* content_type){
1312: if(content_type){
1.291 moko 1313: char* CONTENT_TYPE=str_upper(content_type);
1.278 moko 1314:
1315: if(const char* begin=strstr(CONTENT_TYPE, "CHARSET=")){
1316: begin+=8; // skip "CHARSET="
1317: char* end=0;
1318: if(*begin && (*begin=='"' || *begin =='\'')){
1319: char quote=*begin;
1320: begin++;
1321: end=(char*)strchr(begin, quote);
1322: }
1323: if(!end)
1324: end=(char*)strchr(begin, ';');
1325:
1326: if(end)
1327: *end=0; // terminator
1328:
1.295 moko 1329: return *begin ? &pa_charsets.get_direct(begin) : 0;
1.278 moko 1330: }
1331: }
1332: return 0;
1333: }
1334:
1.301 moko 1335: const UTF16* pa_utf16_encode(const char* in, Charset& source_charset){
1.302 moko 1336: if(!in)
1337: return 0;
1.301 moko 1338:
1.302 moko 1339: String::C sIn(in,strlen(in));
1.301 moko 1340:
1.302 moko 1341: UTF16* utf16=(UTF16*)pa_malloc_atomic(sIn.length*2+2);
1342: UTF16* utf16_end=utf16;
1.301 moko 1343:
1.302 moko 1344: if(!source_charset.isUTF8())
1345: sIn=Charset::transcode(sIn, source_charset, pa_UTF8_charset);
1.301 moko 1346:
1.302 moko 1347: int status=pa_convertUTF8toUTF16((const UTF8**)&sIn.str, (const UTF8*)(sIn.str+sIn.length), &utf16_end, utf16+sIn.length, strictConversion);
1348: if(status != conversionOK)
1349: throw Exception("utf-16 encode", new String(in), "utf-16 conversion failed (%d)", status);
1.301 moko 1350:
1.302 moko 1351: *utf16_end=0;
1.301 moko 1352:
1.302 moko 1353: return utf16;
1.301 moko 1354: }
1355:
1356: const char* pa_utf16_decode(const UTF16* in, Charset& asked_charset){
1357: if(!in)
1358: return 0;
1359:
1360: const UTF16* utf16_start=in;
1361: const UTF16* utf16_end;
1362:
1363: for(utf16_end=in; *utf16_end; utf16_end++);
1364:
1365: char *result = (char *)pa_malloc_atomic((utf16_end-in)*6+1);
1366: char *result_end = result;
1367:
1368: int status=pa_convertUTF16toUTF8(&utf16_start, utf16_end, (UTF8**)&result_end, (UTF8*)(result+(utf16_end-in)*6), strictConversion);
1369:
1370: if(status != conversionOK)
1371: throw Exception("utf-16 decode", 0, "utf conversion failed (%d)", status);
1372:
1373: *result_end='\0';
1374:
1.302 moko 1375: if(asked_charset.isUTF8())
1376: return result;
1.301 moko 1377:
1.302 moko 1378: return Charset::transcode(result, pa_UTF8_charset, asked_charset).cstr();
1.301 moko 1379: }
1.278 moko 1380:
1.283 moko 1381: static bool is_latin(const char *in){
1382: for(; *in; in++){
1383: if ((unsigned char)(*in) > 0x7F)
1384: return false;
1385: }
1386: return true;
1387: }
1388:
1389: #define MAX_IDNA_LENGTH 256
1390:
1.301 moko 1391: const char *pa_idna_encode(const char *in, Charset& source_charset){
1.283 moko 1392: if(!in || is_latin(in))
1393: return in;
1394:
1395: uint32_t utf32[MAX_IDNA_LENGTH];
1396: uint32_t *utf32_end=utf32;
1397:
1398: String::C sIn(in,strlen(in));
1399:
1400: if(!source_charset.isUTF8())
1.295 moko 1401: sIn=Charset::transcode(sIn, source_charset, pa_UTF8_charset);
1.283 moko 1402:
1403: int status=pa_convertUTF8toUTF32((const UTF8**)&sIn.str, (const UTF8*)(sIn.str+sIn.length), &utf32_end, utf32+MAX_IDNA_LENGTH-1, strictConversion);
1404: if(status != conversionOK)
1405: throw Exception("idna encode", new String(in), "utf conversion failed (%d)", status);
1406:
1407: *utf32_end=0;
1408:
1409: char *result = (char *)pa_malloc(MAX_IDNA_LENGTH);
1410: status=pa_idna_to_ascii_4z(utf32, result, MAX_IDNA_LENGTH, 0);
1411: if(status != IDNA_SUCCESS)
1412: throw Exception("idna encode", new String(in), "encode failed: %s", pa_idna_strerror(status));
1413:
1414: return result;
1415: }
1416:
1417: const char *pa_idna_decode(const char *in, Charset &asked_charset){
1418: if(!in || !(*in))
1419: return in;
1420:
1421: uint32_t utf32[MAX_IDNA_LENGTH];
1422: const uint32_t *utf32_start=utf32;
1423: uint32_t *utf32_end;
1424:
1425: int status=pa_idna_to_unicode_4z(in, utf32, MAX_IDNA_LENGTH, 0);
1426: if(status != IDNA_SUCCESS)
1427: throw Exception("idna decode", new String(in), "decode failed: %s", pa_idna_strerror(status));
1428:
1429: for(utf32_end=utf32; *utf32_end; utf32_end++);
1430:
1431: char *result = (char *)pa_malloc(MAX_IDNA_LENGTH);
1432: char *result_end = result;
1433:
1434: status=pa_convertUTF32toUTF8(&utf32_start, utf32_end, (UTF8**)&result_end, (UTF8*)(result+MAX_IDNA_LENGTH-1), strictConversion);
1435:
1436: if(status != conversionOK)
1437: throw Exception("idna decode", new String(in), "utf conversion failed (%d)", status);
1438:
1439: *result_end='\0';
1440:
1441: if(!asked_charset.isUTF8())
1.295 moko 1442: result = (char *)Charset::transcode(result, pa_UTF8_charset, asked_charset).cstr();
1.283 moko 1443:
1444: return result;
1445: }
1.292 moko 1446: /// must be last in this file
1447: #undef vsnprintf
1448: int pa_vsnprintf(char* b, size_t s, const char* f, va_list l) {
1449: if(!s)
1450: return 0;
1451:
1452: int r;
1453: // note: on win32 & maybe somewhere else
1454: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1455: // http://stackoverflow.com/questions/2915672/snprintf-and-visual-studio-2010
1456: --s;
1457:
1458: // clients do not check for negative 's', feature: ignore such prints
1459: if((ssize_t)s<0)
1460: return 0;
1461:
1462: #ifdef _MSC_VER
1463: // win32: if the number of bytes to write exceeds buffer, then count bytes are written and -1 is returned
1464: r=_vsnprintf(b, s, f, l);
1465: if(r<0)
1466: r=s;
1467: #else
1468: r=vsnprintf(b, s, f, l);
1469: /*
1470: solaris: man vsnprintf
1471:
1472: The snprintf() function returns the number of characters
1473: formatted, that is, the number of characters that would have
1474: been written to the buffer if it were large enough. If the
1475: value of n is 0 on a call to snprintf(), an unspecified
1476: value less than 1 is returned.
1477: */
1478:
1479: if(r<0)
1480: r=0;
1481: else if((size_t)r>s)
1482: r=s;
1483: #endif
1484: b[r]=0;
1485: return r;
1486: }
1487:
1488: int pa_snprintf(char* b, size_t s, const char* f, ...) {
1489: va_list l;
1490: va_start(l, f);
1491: int r=pa_vsnprintf(b, s, f, l);
1492: va_end(l);
1493: return r;
1494: }
1495:
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