Annotation of parser3/src/main/pa_common.C, revision 1.311
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)
1.306 moko 6: */
1.210 paf 7:
1.1 paf 8: #include "pa_common.h"
1.4 paf 9: #include "pa_exception.h"
1.154 paf 10: #include "pa_hash.h"
1.14 paf 11: #include "pa_globals.h"
1.154 paf 12: #include "pa_charsets.h"
1.214 paf 13: #include "pa_http.h"
1.223 misha 14: #include "pa_request_charsets.h"
1.237 misha 15: #include "pcre.h"
1.241 misha 16: #include "pa_request.h"
1.98 paf 17:
1.283 moko 18: #include "pa_idna.h"
19: #include "pa_convert_utf.h"
20:
1.273 moko 21: #ifdef _MSC_VER
1.276 moko 22: #include <windows.h>
1.273 moko 23: #include <direct.h>
24: #endif
25:
1.277 moko 26: #ifdef _MSC_VER
27: #define pa_mkdir(path, mode) _mkdir(path)
28: #else
29: #define pa_mkdir(path, mode) mkdir(path, mode)
30: #endif
31:
1.311 ! moko 32: volatile const char * IDENT_PA_COMMON_C="$Id: pa_common.C,v 1.310 2019/11/23 23:48:41 moko Exp $" IDENT_PA_COMMON_H IDENT_PA_HASH_H IDENT_PA_ARRAY_H IDENT_PA_STACK_H;
1.267 moko 33:
1.93 paf 34: // some maybe-undefined constants
35:
1.82 paf 36: #ifndef _O_TEXT
37: # define _O_TEXT 0
38: #endif
39: #ifndef _O_BINARY
40: # define _O_BINARY 0
1.47 paf 41: #endif
1.80 paf 42:
1.138 paf 43: #ifdef HAVE_FTRUNCATE
44: # define PA_O_TRUNC 0
45: #else
46: # ifdef _O_TRUNC
47: # define PA_O_TRUNC _O_TRUNC
48: # else
49: # error you must have either ftruncate function or _O_TRUNC bit declared
50: # endif
1.154 paf 51: #endif
1.176 paf 52:
1.154 paf 53: // defines for globals
54:
55: #define FILE_STATUS_NAME "status"
56:
57: // globals
58:
59: const String file_status_name(FILE_STATUS_NAME);
60:
1.305 moko 61: String sql_bind_name(SQL_BIND_NAME);
62: String sql_limit_name(PA_SQL_LIMIT_NAME);
63: String sql_offset_name(PA_SQL_OFFSET_NAME);
64: String sql_default_name(SQL_DEFAULT_NAME);
65: String sql_distinct_name(SQL_DISTINCT_NAME);
66: String sql_value_type_name(SQL_VALUE_TYPE_NAME);
67:
1.301 moko 68: // forwards
69:
70: const UTF16* pa_utf16_encode(const char* in, Charset& source_charset);
71:
1.154 paf 72: // functions
1.127 paf 73:
1.301 moko 74: #ifdef _MSC_VER
75:
76: int pa_stat(const char *pathname, struct stat *buffer){
77: const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source());
78: return _wstat64((const wchar_t *)utf16name, buffer);
79: }
80:
81: int pa_open(const char *pathname, int flags, int mode){
82: const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source());
83: return _wopen((const wchar_t *)utf16name, flags, mode);
84: }
85:
86: FILE *pa_fopen(const char *pathname, const char *mode){
87: const UTF16* utf16name=pa_utf16_encode(pathname, pa_thread_request().charsets.source());
88: const UTF16* utf16mode=pa_utf16_encode(mode, pa_thread_request().charsets.source());
89: return _wfopen((const wchar_t *)utf16name, (const wchar_t *)utf16mode);
90: }
91:
92: #endif
93:
1.206 paf 94: /// these options were handled but not checked elsewhere, now check them
1.239 misha 95: int pa_get_valid_file_options_count(HashStringValue& options) {
1.206 paf 96: int result=0;
97: if(options.get(PA_SQL_LIMIT_NAME))
98: result++;
99: if(options.get(PA_SQL_OFFSET_NAME))
100: result++;
101: if(options.get(PA_COLUMN_SEPARATOR_NAME))
102: result++;
103: if(options.get(PA_COLUMN_ENCLOSER_NAME))
104: result++;
1.223 misha 105: if(options.get(PA_CHARSET_NAME))
106: result++;
1.206 paf 107: return result;
108: }
109:
1.123 paf 110: #ifndef DOXYGEN
111: struct File_read_action_info {
1.154 paf 112: char **data; size_t *data_size;
1.310 moko 113: char* buf; size_t offset; size_t limit;
1.126 paf 114: };
1.123 paf 115: #endif
1.271 moko 116:
1.289 moko 117: static void file_read_action(struct stat& finfo, int f, const String& file_spec, void *context) {
1.310 moko 118: File_read_action_info& info = *static_cast<File_read_action_info *>(context);
119: size_t to_read_size = info.limit;
1.188 paf 120: if(!to_read_size)
1.310 moko 121: to_read_size = check_file_size(finfo.st_size, file_spec);
1.271 moko 122: if(to_read_size) {
1.188 paf 123: if(info.offset)
124: lseek(f, info.offset, SEEK_SET);
1.310 moko 125: *info.data = info.buf ? info.buf : (char *)pa_malloc_atomic(to_read_size+1);
126: ssize_t result = read(f, *info.data, to_read_size);
1.271 moko 127: if(result<0)
128: throw Exception("file.read", &file_spec, "read failed: %s (%d)", strerror(errno), errno);
1.310 moko 129: *info.data_size = result;
1.123 paf 130: } else { // empty file
1.209 paf 131: // for both, text and binary: for text we need that terminator, for binary we need nonzero pointer to be able to save such files
1.310 moko 132: *info.data = (char *)pa_malloc_atomic(1);
133: *(char*)(*info.data) = 0;
134: *info.data_size = 0;
1.123 paf 135: return;
136: }
1.126 paf 137: }
1.241 misha 138:
1.310 moko 139: File_read_result file_read_binary(const String& file_spec, bool fail_on_read_problem, char* buf, size_t offset, size_t limit) {
140: File_read_result result = {false, 0, 0, 0};
141: File_read_action_info info = {&result.str, &result.length, buf, offset, limit};
1.309 moko 142:
1.310 moko 143: result.success = file_read_action_under_lock(file_spec, "read", file_read_action, &info, 0, fail_on_read_problem);
1.309 moko 144: return result;
145: }
146:
1.154 paf 147: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
1.310 moko 148: bool as_text, HashStringValue *options,
1.229 misha 149: bool fail_on_read_problem,
1.310 moko 150: size_t offset = 0, size_t limit = 0, bool transcode_text_result = true) {
151: File_read_result result = {false, 0, 0, 0};
152: if(options){
153: int valid_options = pa_get_valid_file_options_count(*options);
154: if(valid_options != options->count())
1.262 misha 155: throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION);
1.241 misha 156: }
1.203 paf 157:
1.310 moko 158: File_read_action_info info = {&result.str, &result.length, 0, offset, limit};
1.161 paf 159:
1.310 moko 160: result.success = file_read_action_under_lock(file_spec, "read", file_read_action, &info, as_text, fail_on_read_problem);
1.223 misha 161:
1.241 misha 162: if(as_text){
163: if(result.success){
1.310 moko 164: Charset* asked_charset = 0;
165: if(options)
166: if(Value* vcharset_name = options->get(PA_CHARSET_NAME))
167: asked_charset = &pa_charsets.get(vcharset_name->as_string());
1.263 misha 168:
1.310 moko 169: asked_charset = pa_charsets.checkBOM(result.str, result.length, asked_charset);
1.287 moko 170:
1.263 misha 171: 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
1.310 moko 172: String::C body = String::C(result.str, result.length);
1.263 misha 173: body=Charset::transcode(body, *asked_charset, charsets.source());
1.236 misha 174:
1.310 moko 175: result.str = const_cast<char*>(body.str); // hacking a little
176: result.length = body.length;
1.131 paf 177: }
178: }
1.241 misha 179: if(result.length)
180: fix_line_breaks(result.str, result.length);
1.123 paf 181: }
1.241 misha 182:
183: return result;
184: }
185:
186: File_read_result file_load(Request& r, const String& file_spec,
1.310 moko 187: bool as_text, HashStringValue *options,
1.241 misha 188: bool fail_on_read_problem,
1.310 moko 189: bool transcode_text_result) {
190:
191: size_t offset = 0;
192: size_t limit = 0;
193:
194: if(options){
195: if(Value *voffset = (Value *)options->get(sql_offset_name))
196: offset = r.process(*voffset).as_int();
197: if(Value *vlimit = (Value *)options->get(sql_limit_name))
198: limit = r.process(*vlimit).as_int();
199: // no check on options count here
200: }
1.241 misha 201:
202: if(file_spec.starts_with("http://")) {
1.310 moko 203: if(offset || limit)
1.289 moko 204: throw Exception(PARSER_RUNTIME, 0, "offset and load options are not supported for HTTP:// file load");
1.241 misha 205:
206: // fail on read problem
1.310 moko 207: File_read_http_result http = pa_internal_file_read_http(r, file_spec, as_text, options, transcode_text_result);
208:
209: File_read_result result = {true, http.str, http.length, http.headers};
210: return result;
1.241 misha 211: } else
1.310 moko 212: return file_read(r.charsets, file_spec, as_text, options, fail_on_read_problem, offset, limit, transcode_text_result);
213: }
1.126 paf 214:
1.310 moko 215: char* file_read_text(Request_charsets& charsets, const String& file_spec, bool fail_on_read_problem) {
216: File_read_result file = file_read(charsets, file_spec, true, 0, fail_on_read_problem);
217: return file.success ? file.str : 0;
1.123 paf 218: }
219:
1.310 moko 220: char* file_load_text(Request& r, const String& file_spec, bool fail_on_read_problem, HashStringValue* options, bool transcode_result) {
221: File_read_result file = file_load(r, file_spec, true, options, fail_on_read_problem, transcode_result);
222: return file.success ? file.str : 0;
223: }
1.257 pretende 224:
1.154 paf 225: #ifdef PA_SAFE_MODE
1.259 misha 226: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
1.154 paf 227: if(finfo.st_uid/*foreign?*/!=geteuid()
228: && finfo.st_gid/*foreign?*/!=getegid())
1.289 moko 229: throw Exception(PARSER_RUNTIME,
230: &file_spec,
231: "parser is in safe mode: reading files of foreign group and user disabled "
232: "[recompile parser with --disable-safe-mode configure option], "
233: "actual filename '%s', fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
234: fname, finfo.st_uid, geteuid(), finfo.st_gid, getegid()
235: );
1.259 misha 236: }
237: #else
238: void check_safe_mode(struct stat, const String&, const char*) {
239: }
1.257 pretende 240: #endif
1.259 misha 241:
1.257 pretende 242:
1.149 paf 243:
1.154 paf 244: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 245: const char* action_name, File_read_action action, void *context,
246: bool as_text,
1.123 paf 247: bool fail_on_read_problem) {
1.247 misha 248: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.33 paf 249: int f;
250:
251: // first open, next stat:
1.45 paf 252: // directory update of NTFS hard links performed on open.
1.33 paf 253: // ex:
254: // a.html:^test[] and b.html hardlink to a.html
255: // user inserts ! before ^test in a.html
1.126 paf 256: // directory entry of b.html in NTFS not updated at once,
1.35 paf 257: // they delay update till open, so we would receive "!^test[" string
258: // if would do stat, next open.
1.123 paf 259: // later: it seems, even this does not help sometimes
1.301 moko 260: if((f=pa_open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 261: try {
1.162 paf 262: if(pa_lock_shared_blocking(f)!=0)
1.289 moko 263: throw Exception("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.123 paf 264:
1.124 paf 265: struct stat finfo;
1.298 moko 266: if(pa_fstat(f, &finfo)!=0)
1.124 paf 267: throw Exception("file.missing", // hardly possible: we just opened it OK
1.289 moko 268: &file_spec, "stat failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.124 paf 269:
1.149 paf 270: check_safe_mode(finfo, file_spec, fname);
1.32 paf 271:
1.289 moko 272: action(finfo, f, file_spec, context);
1.123 paf 273: } catch(...) {
1.162 paf 274: pa_unlock(f);close(f);
1.123 paf 275: if(fail_on_read_problem)
1.154 paf 276: rethrow;
1.289 moko 277: return false;
1.123 paf 278: }
1.87 paf 279:
1.162 paf 280: pa_unlock(f);close(f);
1.72 parser 281: return true;
1.229 misha 282: } else {
1.118 paf 283: if(fail_on_read_problem)
1.289 moko 284: throw Exception(errno==EACCES ? "file.access" : (errno==ENOENT || errno==ENOTDIR || errno==ENODEV) ? "file.missing" : 0,
285: &file_spec, "%s failed: %s (%d), actual filename '%s'", action_name, strerror(errno), errno, fname);
1.118 paf 286: return false;
287: }
1.8 paf 288: }
289:
1.202 paf 290: void create_dir_for_file(const String& file_spec) {
1.63 parser 291: size_t pos_after=1;
1.154 paf 292: size_t pos_before;
293: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
1.277 moko 294: pa_mkdir(file_spec.mid(0, pos_before).taint_cstr(String::L_FILE_SPEC), 0775);
1.63 parser 295: pos_after=pos_before+1;
296: }
297: }
298:
1.98 paf 299: bool file_write_action_under_lock(
1.28 paf 300: const String& file_spec,
1.225 misha 301: const char* action_name,
302: File_write_action action,
303: void *context,
1.126 paf 304: bool as_text,
305: bool do_append,
306: bool do_block,
1.110 paf 307: bool fail_on_lock_problem) {
1.247 misha 308: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.28 paf 309: int f;
1.80 paf 310: if(access(fname, W_OK)!=0) // no
1.126 paf 311: create_dir_for_file(file_spec);
1.50 paf 312:
1.301 moko 313: if((f=pa_open(fname,
1.80 paf 314: O_CREAT|O_RDWR
315: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 316: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 317: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.289 moko 318: Exception e("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.126 paf 319: close(f);
1.110 paf 320: if(fail_on_lock_problem)
321: throw e;
1.98 paf 322: return false;
323: }
1.96 paf 324:
1.158 paf 325: try {
1.254 misha 326: #if (defined(HAVE_FCHMOD) && defined(PA_SAFE_MODE))
327: struct stat finfo;
1.298 moko 328: if(pa_fstat(f, &finfo)==0 && finfo.st_mode & 0111)
1.254 misha 329: fchmod(f, finfo.st_mode & 0666/*clear executable bits*/); // backward: ignore errors if any
330: #endif
331: action(f, context);
1.158 paf 332: } catch(...) {
1.138 paf 333: #ifdef HAVE_FTRUNCATE
1.104 paf 334: if(!do_append)
1.125 paf 335: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 336: #endif
1.162 paf 337: pa_unlock(f);close(f);
1.154 paf 338: rethrow;
1.158 paf 339: }
1.80 paf 340:
1.138 paf 341: #ifdef HAVE_FTRUNCATE
1.104 paf 342: if(!do_append)
1.125 paf 343: 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 344: #endif
1.162 paf 345: pa_unlock(f);close(f);
1.98 paf 346: return true;
1.80 paf 347: } else
1.289 moko 348: 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 349: // here should be nothing, see rethrow above
350: }
351:
352: #ifndef DOXYGEN
353: struct File_write_action_info {
1.250 misha 354: const char* str;
355: size_t length;
1.126 paf 356: };
1.96 paf 357: #endif
1.271 moko 358:
1.96 paf 359: static void file_write_action(int f, void *context) {
1.126 paf 360: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 361: if(info.length) {
1.271 moko 362: ssize_t written=write(f, info.str, info.length);
1.116 paf 363: if(written<0)
1.271 moko 364: throw Exception("file.write", 0, "write failed: %s (%d)", strerror(errno), errno);
1.275 moko 365: if((size_t)written!=info.length)
1.271 moko 366: throw Exception("file.write", 0, "write failed: %u of %u bytes written", written, info.length);
1.113 paf 367: }
1.96 paf 368: }
1.271 moko 369:
1.96 paf 370: void file_write(
1.250 misha 371: Request_charsets& charsets,
372: const String& file_spec,
373: const char* data,
374: size_t size,
1.126 paf 375: bool as_text,
1.250 misha 376: bool do_append,
377: Charset* asked_charset) {
378:
379: if(as_text && asked_charset){
380: String::C body=String::C(data, size);
381: body=Charset::transcode(body, charsets.source(), *asked_charset);
382: data=body.str;
383: size=body.length;
384: };
385:
1.126 paf 386: File_write_action_info info={data, size};
1.225 misha 387:
1.98 paf 388: file_write_action_under_lock(
1.154 paf 389: file_spec,
1.225 misha 390: "write",
391: file_write_action,
392: &info,
1.154 paf 393: as_text,
394: do_append);
1.30 paf 395: }
396:
1.261 misha 397: static size_t get_dir(char* fname, size_t helper_length){
398: bool dir=false;
399: size_t pos=0;
400: for(pos=helper_length; pos; pos--){
401: char c=fname[pos-1];
402: if(c=='/' || c=='\\'){
403: fname[pos-1]=0;
404: dir=true;
405: } else if(dir) break;
406: }
407: return pos;
408: }
409:
1.311 ! moko 410: static bool entry_exists(const String& file_spec, bool need_dir) {
! 411: char *fname=file_spec.taint_cstrm(String::L_FILE_SPEC);
1.261 misha 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;
1.311 ! moko 424: if(entry_exists(fname, &finfo)) {
1.261 misha 425: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
426: return is_dir==need_dir;
427: }
428: return false;
429: }
430:
1.63 parser 431: // throws nothing! [this is required in file_move & file_delete]
1.277 moko 432: static void rmdir(const String& file_spec, size_t pos_after) {
1.261 misha 433: char* dir_spec=file_spec.taint_cstrm(String::L_FILE_SPEC);
434: size_t length=strlen(dir_spec);
435: while( (length=get_dir(dir_spec, length)) && (length > pos_after) ){
1.274 moko 436: #ifdef _MSC_VER
1.311 ! moko 437: if(!entry_exists(dir_spec, true))
1.261 misha 438: break;
439: DWORD attrs=GetFileAttributes(dir_spec);
440: if(
441: (attrs==INVALID_FILE_ATTRIBUTES)
442: || !(attrs & FILE_ATTRIBUTE_DIRECTORY)
443: || (attrs & FILE_ATTRIBUTE_REPARSE_POINT)
444: )
445: break;
446: #endif
447: if( rmdir(dir_spec) )
448: break;
449: };
1.50 paf 450: }
1.239 misha 451:
1.269 misha 452: bool file_delete(const String& file_spec, bool fail_on_problem, bool keep_empty_dirs) {
1.247 misha 453: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.282 moko 454: if(unlink(fname)!=0) {
1.164 paf 455: if(fail_on_problem)
1.289 moko 456: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
457: &file_spec, "unlink failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.93 paf 458: else
459: return false;
1.282 moko 460: }
1.50 paf 461:
1.269 misha 462: if(!keep_empty_dirs)
463: rmdir(file_spec, 1);
464:
1.93 paf 465: return true;
1.60 parser 466: }
1.239 misha 467:
1.269 misha 468: void file_move(const String& old_spec, const String& new_spec, bool keep_empty_dirs) {
1.247 misha 469: const char* old_spec_cstr=old_spec.taint_cstr(String::L_FILE_SPEC);
470: const char* new_spec_cstr=new_spec.taint_cstr(String::L_FILE_SPEC);
1.63 parser 471:
1.126 paf 472: create_dir_for_file(new_spec);
1.63 parser 473:
1.60 parser 474: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.289 moko 475: throw Exception(errno==EACCES ? "file.access" : errno==ENOENT ? "file.missing" : 0,
476: &old_spec, "rename failed: %s (%d), actual filename '%s' to '%s'", strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 477:
1.269 misha 478: if(!keep_empty_dirs)
479: rmdir(old_spec, 1);
1.31 paf 480: }
481:
1.51 paf 482:
1.126 paf 483: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 484: struct stat lfinfo;
1.298 moko 485: bool result=pa_stat(fname, &lfinfo)==0;
1.118 paf 486: if(afinfo)
487: *afinfo=lfinfo;
488: return result;
1.119 paf 489: }
490:
491: bool entry_exists(const String& file_spec) {
1.247 misha 492: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.126 paf 493: return entry_exists(fname, 0);
1.118 paf 494: }
495:
1.215 paf 496: bool file_exist(const String& file_spec) {
1.311 ! moko 497: return entry_exists(file_spec, false);
1.51 paf 498: }
1.239 misha 499:
1.215 paf 500: bool dir_exists(const String& file_spec) {
1.311 ! moko 501: return entry_exists(file_spec, true);
1.65 parser 502: }
1.239 misha 503:
1.215 paf 504: const String* file_exist(const String& path, const String& name) {
1.154 paf 505: String& result=*new String(path);
1.270 moko 506: if(path.last_char() != '/')
507: result << "/";
1.154 paf 508: result << name;
1.215 paf 509: return file_exist(result)?&result:0;
1.43 paf 510: }
1.239 misha 511:
1.43 paf 512: bool file_executable(const String& file_spec) {
1.247 misha 513: return access(file_spec.taint_cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 514: }
515:
1.296 moko 516: 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 517: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.154 paf 518: struct stat finfo;
1.298 moko 519: if(pa_stat(fname, &finfo)!=0) {
1.64 parser 520: if(fail_on_read_problem)
1.289 moko 521: throw Exception("file.missing", &file_spec, "getting file size failed: %s (%d), real filename '%s'", strerror(errno), errno, fname);
1.64 parser 522: else
523: return false;
1.282 moko 524: }
1.58 parser 525: rsize=finfo.st_size;
1.299 moko 526: ratime=(time_t)finfo.st_atime;
527: rmtime=(time_t)finfo.st_mtime;
528: rctime=(time_t)finfo.st_ctime;
1.64 parser 529: return true;
1.18 paf 530: }
531:
1.300 moko 532: size_t check_file_size(uint64_t size, const String& file_spec){
533: if(size > pa_file_size_limit)
534: throw Exception(PARSER_RUNTIME, &file_spec, "content size of %.15g bytes exceeds the limit (%.15g bytes)", (double)size, (double)pa_file_size_limit);
535: return (size_t)size;
536: }
537:
1.278 moko 538: /**
539: String related functions
540: */
541:
542: bool capitalized(const char* s){
543: bool upper=true;
544: for(const char* c=s; *c; c++){
545: if(*c != (upper ? toupper((unsigned char)*c) : tolower((unsigned char)*c)))
546: return false;
547: upper=strchr("-_ ", *c) != 0;
548: }
549: return true;
550: }
551:
552: const char* capitalize(const char* s){
553: if(!s || capitalized(s))
554: return s;
555:
556: char* result=pa_strdup(s);
557: if(result){
558: bool upper=true;
559: for(char* c=result; *c; c++){
560: *c=upper ? (char)toupper((unsigned char)*c) : (char)tolower((unsigned char)*c);
561: upper=strchr("-_ ", *c) != 0;
562: }
563: }
564: return (const char*)result;
565: }
566:
1.290 moko 567: char *str_lower(const char *s, size_t helper_length){
568: char *result=pa_strdup(s, helper_length);
569: for(char* c=result; *c; c++)
570: *c=(char)tolower((unsigned char)*c);
571: return result;
572: }
573:
574: char *str_upper(const char *s, size_t helper_length){
575: char *result=pa_strdup(s, helper_length);
576: for(char* c=result; *c; c++)
577: *c=(char)toupper((unsigned char)*c);
578: return result;
579: }
580:
1.278 moko 581: void fix_line_breaks(char *str, size_t& length) {
582: //_asm int 3;
583: const char* const eob=str+length;
584: char* dest=str;
585: // fix DOS: \r\n -> \n
586: // fix Macintosh: \r -> \n
587: char* bol=str;
588: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
589: size_t len=eol-bol;
590: if(dest!=bol)
591: memmove(dest, bol, len);
592: dest+=len;
593: *dest++='\n';
594:
595: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
596: bol=eol+2;
597: length--;
598: } else // \r, not \n = Macintosh
599: bol=eol+1;
600: }
601: // last piece without \r
602: if(dest!=bol)
603: memmove(dest, bol, eob-bol);
604: str[length]=0; // terminating
605: }
606:
607: /**
608: scans for @a delim[default \n] in @a *row_ref,
609: @return piece of line before it or end of string, if no @a delim found
610: assigns @a *row_ref to point right after delimiter if there were one
611: or to zero if no @a delim were found.
612: */
613:
1.126 paf 614: char* getrow(char* *row_ref, char delim) {
1.229 misha 615: char* result=*row_ref;
616: if(result) {
1.126 paf 617: *row_ref=strchr(result, delim);
1.8 paf 618: if(*row_ref)
619: *((*row_ref)++)=0;
620: else if(!*result)
621: return 0;
1.229 misha 622: }
623: return result;
1.8 paf 624: }
625:
1.126 paf 626: char* lsplit(char* string, char delim) {
1.229 misha 627: if(string) {
1.126 paf 628: char* v=strchr(string, delim);
1.8 paf 629: if(v) {
630: *v=0;
631: return v+1;
632: }
1.229 misha 633: }
634: return 0;
1.8 paf 635: }
636:
1.126 paf 637: char* lsplit(char* *string_ref, char delim) {
1.229 misha 638: char* result=*string_ref;
1.126 paf 639: char* next=lsplit(*string_ref, delim);
1.229 misha 640: *string_ref=next;
641: return result;
1.9 paf 642: }
643:
1.126 paf 644: char* rsplit(char* string, char delim) {
1.229 misha 645: if(string) {
1.126 paf 646: char* v=strrchr(string, delim);
1.18 paf 647: if(v) {
1.9 paf 648: *v=0;
649: return v+1;
650: }
1.229 misha 651: }
652: return NULL;
1.10 paf 653: }
654:
1.229 misha 655:
656: // format: %[flags][width][.precision]type http://msdn.microsoft.com/ru-ru/library/56e442dc(en-us,VS.80).aspx
657: // flags: '-', '+', ' ', '#', '0' http://msdn.microsoft.com/ru-ru/library/8aky45ct(en-us,VS.80).aspx
658: // width, precision: non negative decimal number
659: enum FormatType {
660: FormatInvalid,
661: FormatInt,
662: FormatUInt,
663: FormatDouble
664: };
1.272 moko 665: FormatType format_type(const char* fmt){
1.229 misha 666: enum FormatState {
667: Percent,
668: Flags,
669: Width,
670: Precision,
671: Done
672: } state=Percent;
673:
674: FormatType result=FormatInvalid;
675:
1.272 moko 676: const char* pos=fmt;
1.229 misha 677: while(char c=*(pos++)){
678: switch(state){
679: case Percent:
680: if(c=='%'){
681: state=Flags;
682: } else {
683: return FormatInvalid; // 1st char must be '%' only
684: }
685: break;
686: case Flags:
687: if(strchr("-+ #0", c)!=0){
688: break;
689: }
690: // go to the next step
691: case Width:
692: if(c=='.'){
693: state=Precision;
694: break;
695: }
696: // go to the next step
697: case Precision:
698: if(c>='0' && c<='9'){
699: if(state == Flags) state=Width; // no more flags
700: break;
701: } else if(c=='d' || c=='i'){
702: result=FormatInt;
703: } else if(strchr("feEgG", c)!=0){
704: result=FormatDouble;
705: } else if(strchr("uoxX", c)!=0){
706: result=FormatUInt;
707: } else {
708: return FormatInvalid; // invalid char
709: }
710: state=Done;
711: break;
712: case Done:
713: return FormatInvalid; // no chars allowed after 'type'
714: }
715: }
716: return result;
717: }
718:
719:
1.272 moko 720: const char* format(double value, const char* fmt) {
1.229 misha 721: char local_buf[MAX_NUMBER];
1.235 misha 722: int size=-1;
1.229 misha 723:
724: if(fmt && strlen(fmt)){
725: switch(format_type(fmt)){
726: case FormatDouble:
727: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
728: break;
729: case FormatInt:
730: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
731: break;
732: case FormatUInt:
1.126 paf 733: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.229 misha 734: break;
735: case FormatInvalid:
1.289 moko 736: throw Exception(PARSER_RUNTIME, 0, "Incorrect format string '%s' was specified.", fmt);
1.229 misha 737: }
738: } else
739: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
740:
741: if(size < 0 || size >= MAX_NUMBER-1){ // on win32 we manually reduce max size while printing
1.304 moko 742: throw Exception(PARSER_RUNTIME, 0, "Error occurred white executing snprintf with format string '%s'.", fmt);
1.229 misha 743: }
744:
1.235 misha 745: return pa_strdup(local_buf, (size_t)size);
1.12 paf 746: }
747:
1.36 paf 748: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 749: #ifdef WIN32
1.187 paf 750: size_t to_write = size;
1.12 paf 751: do{
1.154 paf 752: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 753: if(chunk_written<=0)
754: break;
755: size-=chunk_written;
1.36 paf 756: buf=((const char*)buf)+chunk_written;
1.126 paf 757: } while(size>0);
1.12 paf 758:
1.187 paf 759: return to_write-size;
1.12 paf 760: #else
1.126 paf 761: return fwrite(buf, 1, size, stdout);
1.12 paf 762: #endif
1.2 paf 763: }
1.14 paf 764:
1.229 misha 765: enum EscapeState {
766: EscapeRest,
767: EscapeFirst,
768: EscapeSecond,
769: EscapeUnicode
770: };
771:
1.236 misha 772: // @todo prescan for reduce required size (unescaped sting in 1 byte charset requires less memory usually)
1.258 misha 773: char* unescape_chars(const char* cp, int len, Charset* charset, bool js){
1.236 misha 774: char* s=new(PointerFreeGC) char[len+1]; // must be enough (%uXXXX==6 bytes, max utf-8 char length==6 bytes)
1.230 misha 775: char* dst=s;
1.229 misha 776: EscapeState escapeState=EscapeRest;
777: uint escapedValue=0;
778: int srcPos=0;
1.230 misha 779: short int jsCnt=0;
1.236 misha 780: while(srcPos<len){
1.229 misha 781: uchar c=(uchar)cp[srcPos];
1.258 misha 782: if(c=='%' || (c=='\\' && js)){
1.229 misha 783: escapeState=EscapeFirst;
784: } else {
785: switch(escapeState) {
786: case EscapeRest:
1.286 moko 787: if(c=='+' && !js){
1.230 misha 788: *dst++=' ';
1.229 misha 789: } else {
1.230 misha 790: *dst++=c;
1.229 misha 791: }
792: break;
793: case EscapeFirst:
1.232 misha 794: if(charset && c=='u'){
1.229 misha 795: // escaped unicode value: %u0430
796: jsCnt=0;
797: escapedValue=0;
798: escapeState=EscapeUnicode;
799: } else {
1.231 misha 800: if(isxdigit(c)){
1.229 misha 801: escapedValue=hex_value[c] << 4;
802: escapeState=EscapeSecond;
803: } else {
1.230 misha 804: *dst++=c;
1.229 misha 805: escapeState=EscapeRest;
806: }
807: }
808: break;
809: case EscapeSecond:
1.231 misha 810: if(isxdigit(c)){
1.229 misha 811: escapedValue+=hex_value[c];
1.230 misha 812: *dst++=(char)escapedValue;
1.229 misha 813: }
814: escapeState=EscapeRest;
815: break;
816: case EscapeUnicode:
1.231 misha 817: if(isxdigit(c)){
1.229 misha 818: escapedValue=(escapedValue << 4) + hex_value[c];
819: if(++jsCnt==4){
1.230 misha 820: // transcode utf8 char to client charset (we can lost some chars here)
1.232 misha 821: charset->store_Char((XMLByte*&)dst, (XMLCh)escapedValue, '?');
1.229 misha 822: escapeState=EscapeRest;
823: }
824: } else {
825: // not full unicode value
826: escapeState=EscapeRest;
827: }
828: break;
829: }
830: }
831:
832: srcPos++;
833: }
834:
1.230 misha 835: *dst=0; // zero-termination
1.229 misha 836: return s;
837: }
1.24 paf 838:
1.268 misha 839: char *search_stop(char*& current, char cstop_at) {
840: // sanity check
841: if(!current)
842: return 0;
843:
844: // skip leading WS
845: while(*current==' ' || *current=='\t')
846: current++;
847: if(!*current)
848: return current=0;
849:
850: char *result=current;
851: if(char *pstop_at=strchr(current, cstop_at)) {
852: *pstop_at=0;
853: current=pstop_at+1;
854: } else
855: current=0;
856: return result;
857: }
858:
1.24 paf 859: #ifdef WIN32
1.126 paf 860: void back_slashes_to_slashes(char* s) {
1.24 paf 861: if(s)
862: for(; *s; s++)
863: if(*s=='\\')
1.126 paf 864: *s='/';
1.24 paf 865: }
866: #endif
1.41 paf 867:
1.232 misha 868: size_t strpos(const char *str, const char *substr) {
869: const char *p = strstr(str, substr);
870: return (p==0)?STRING_NOT_FOUND:p-str;
871: }
872:
1.226 misha 873: int remove_crlf(char* start, char* end) {
874: char* from=start;
875: char* to=start;
876: bool skip=false;
877: while(from < end){
878: switch(*from){
879: case '\n':
880: case '\r':
881: case '\t':
882: case ' ':
883: if(!skip){
884: *to=' ';
885: to++;
886: skip=true;
887: }
888: break;
889: default:
890: if(from != to)
891: *to=*from;
892: to++;
893: skip=false;
1.69 parser 894: }
1.226 misha 895: from++;
896: }
897: return to-start;
1.91 paf 898: }
899:
1.279 moko 900: const char* hex_digits="0123456789ABCDEF";
901:
1.278 moko 902: const char* hex_string(unsigned char* bytes, size_t size, bool upcase) {
903: char *bytes_hex=new(PointerFreeGC) char [size*2/*byte->hh*/+1/*for zero-teminator*/];
904: unsigned char *src=bytes;
905: unsigned char *end=bytes+size;
906: char *dest=bytes_hex;
907:
1.279 moko 908: const char *hex=upcase? hex_digits : "0123456789abcdef";
1.278 moko 909:
910: for(; src<end; src++) {
911: *dest++=hex[*src/0x10];
912: *dest++=hex[*src%0x10];
913: }
914: *dest=0;
915:
916: return bytes_hex;
917: }
1.91 paf 918:
1.221 misha 919: int file_block_read(const int f, unsigned char* buffer, const size_t size){
920: int nCount = read(f, buffer, size);
921: if (nCount < 0)
1.289 moko 922: throw Exception("file.read", 0, "read failed: %s (%d)", strerror(errno), errno);
1.221 misha 923: return nCount;
924: }
925:
1.278 moko 926: static unsigned long crc32Table[256];
927: static void InitCrc32Table()
928: {
929: if(crc32Table[1] == 0){
930: // This is the official polynomial used by CRC32 in PKZip.
931: // Often times the polynomial shown reversed as 0x04C11DB7.
932: static const unsigned long dwPolynomial = 0xEDB88320;
933:
934: for(int i = 0; i < 256; i++)
935: {
936: unsigned long dwCrc = i;
937: for(int j = 8; j > 0; j--)
938: {
939: if(dwCrc & 1)
940: dwCrc = (dwCrc >> 1) ^ dwPolynomial;
941: else
942: dwCrc >>= 1;
943: }
944: crc32Table[i] = dwCrc;
945: }
946: }
947: }
948:
949: inline void CalcCrc32(const unsigned char byte, unsigned long &crc32)
950: {
951: crc32 = ((crc32) >> 8) ^ crc32Table[(byte) ^ ((crc32) & 0x000000FF)];
952: }
953:
954:
1.297 moko 955: unsigned long pa_crc32(const char *in, size_t in_size){
1.218 misha 956: unsigned long crc32=0xFFFFFFFF;
1.220 misha 957:
1.240 misha 958: InitCrc32Table();
1.239 misha 959: for(size_t i = 0; i<in_size; i++)
960: CalcCrc32(in[i], crc32);
1.220 misha 961:
1.218 misha 962: return ~crc32;
963: }
964:
1.289 moko 965: static void file_crc32_file_action(struct stat& finfo, int f, const String&, void *context) {
1.218 misha 966: unsigned long& crc32=*static_cast<unsigned long *>(context);
967: if(finfo.st_size) {
968: InitCrc32Table();
1.220 misha 969: int nCount=0;
1.218 misha 970: do {
1.221 misha 971: unsigned char buffer[FILE_BUFFER_SIZE];
972: nCount = file_block_read(f, buffer, sizeof(buffer));
1.220 misha 973: for(int i = 0; i < nCount; i++) CalcCrc32(buffer[i], crc32);
974: } while(nCount > 0);
1.218 misha 975: }
976: }
977:
1.297 moko 978: unsigned long pa_crc32(const String& file_spec){
1.278 moko 979: unsigned long crc32=0xFFFFFFFF;
980: file_read_action_under_lock(file_spec, "crc32", file_crc32_file_action, &crc32);
981: return ~crc32;
982: }
983:
984: // content-type: xxx; charset=WE-NEED-THIS
985: // content-type: xxx; charset="WE-NEED-THIS"
986: // content-type: xxx; charset="WE-NEED-THIS";
987: Charset* detect_charset(const char* content_type){
988: if(content_type){
1.291 moko 989: char* CONTENT_TYPE=str_upper(content_type);
1.278 moko 990:
991: if(const char* begin=strstr(CONTENT_TYPE, "CHARSET=")){
992: begin+=8; // skip "CHARSET="
993: char* end=0;
994: if(*begin && (*begin=='"' || *begin =='\'')){
995: char quote=*begin;
996: begin++;
997: end=(char*)strchr(begin, quote);
998: }
999: if(!end)
1000: end=(char*)strchr(begin, ';');
1001:
1002: if(end)
1003: *end=0; // terminator
1004:
1.295 moko 1005: return *begin ? &pa_charsets.get_direct(begin) : 0;
1.278 moko 1006: }
1007: }
1008: return 0;
1009: }
1010:
1.301 moko 1011: const UTF16* pa_utf16_encode(const char* in, Charset& source_charset){
1.302 moko 1012: if(!in)
1013: return 0;
1.301 moko 1014:
1.302 moko 1015: String::C sIn(in,strlen(in));
1.301 moko 1016:
1.302 moko 1017: UTF16* utf16=(UTF16*)pa_malloc_atomic(sIn.length*2+2);
1018: UTF16* utf16_end=utf16;
1.301 moko 1019:
1.302 moko 1020: if(!source_charset.isUTF8())
1021: sIn=Charset::transcode(sIn, source_charset, pa_UTF8_charset);
1.301 moko 1022:
1.302 moko 1023: int status=pa_convertUTF8toUTF16((const UTF8**)&sIn.str, (const UTF8*)(sIn.str+sIn.length), &utf16_end, utf16+sIn.length, strictConversion);
1024: if(status != conversionOK)
1025: throw Exception("utf-16 encode", new String(in), "utf-16 conversion failed (%d)", status);
1.301 moko 1026:
1.302 moko 1027: *utf16_end=0;
1.301 moko 1028:
1.302 moko 1029: return utf16;
1.301 moko 1030: }
1031:
1032: const char* pa_utf16_decode(const UTF16* in, Charset& asked_charset){
1033: if(!in)
1034: return 0;
1035:
1036: const UTF16* utf16_start=in;
1037: const UTF16* utf16_end;
1038:
1039: for(utf16_end=in; *utf16_end; utf16_end++);
1040:
1041: char *result = (char *)pa_malloc_atomic((utf16_end-in)*6+1);
1042: char *result_end = result;
1043:
1044: int status=pa_convertUTF16toUTF8(&utf16_start, utf16_end, (UTF8**)&result_end, (UTF8*)(result+(utf16_end-in)*6), strictConversion);
1045:
1046: if(status != conversionOK)
1047: throw Exception("utf-16 decode", 0, "utf conversion failed (%d)", status);
1048:
1049: *result_end='\0';
1050:
1.302 moko 1051: if(asked_charset.isUTF8())
1052: return result;
1.301 moko 1053:
1.302 moko 1054: return Charset::transcode(result, pa_UTF8_charset, asked_charset).cstr();
1.301 moko 1055: }
1.278 moko 1056:
1.283 moko 1057: static bool is_latin(const char *in){
1058: for(; *in; in++){
1059: if ((unsigned char)(*in) > 0x7F)
1060: return false;
1061: }
1062: return true;
1063: }
1064:
1065: #define MAX_IDNA_LENGTH 256
1066:
1.301 moko 1067: const char *pa_idna_encode(const char *in, Charset& source_charset){
1.283 moko 1068: if(!in || is_latin(in))
1069: return in;
1070:
1071: uint32_t utf32[MAX_IDNA_LENGTH];
1072: uint32_t *utf32_end=utf32;
1073:
1074: String::C sIn(in,strlen(in));
1075:
1076: if(!source_charset.isUTF8())
1.295 moko 1077: sIn=Charset::transcode(sIn, source_charset, pa_UTF8_charset);
1.283 moko 1078:
1079: int status=pa_convertUTF8toUTF32((const UTF8**)&sIn.str, (const UTF8*)(sIn.str+sIn.length), &utf32_end, utf32+MAX_IDNA_LENGTH-1, strictConversion);
1080: if(status != conversionOK)
1081: throw Exception("idna encode", new String(in), "utf conversion failed (%d)", status);
1082:
1083: *utf32_end=0;
1084:
1085: char *result = (char *)pa_malloc(MAX_IDNA_LENGTH);
1086: status=pa_idna_to_ascii_4z(utf32, result, MAX_IDNA_LENGTH, 0);
1087: if(status != IDNA_SUCCESS)
1088: throw Exception("idna encode", new String(in), "encode failed: %s", pa_idna_strerror(status));
1089:
1090: return result;
1091: }
1092:
1093: const char *pa_idna_decode(const char *in, Charset &asked_charset){
1094: if(!in || !(*in))
1095: return in;
1096:
1097: uint32_t utf32[MAX_IDNA_LENGTH];
1098: const uint32_t *utf32_start=utf32;
1099: uint32_t *utf32_end;
1100:
1101: int status=pa_idna_to_unicode_4z(in, utf32, MAX_IDNA_LENGTH, 0);
1102: if(status != IDNA_SUCCESS)
1103: throw Exception("idna decode", new String(in), "decode failed: %s", pa_idna_strerror(status));
1104:
1105: for(utf32_end=utf32; *utf32_end; utf32_end++);
1106:
1107: char *result = (char *)pa_malloc(MAX_IDNA_LENGTH);
1108: char *result_end = result;
1109:
1110: status=pa_convertUTF32toUTF8(&utf32_start, utf32_end, (UTF8**)&result_end, (UTF8*)(result+MAX_IDNA_LENGTH-1), strictConversion);
1111:
1112: if(status != conversionOK)
1113: throw Exception("idna decode", new String(in), "utf conversion failed (%d)", status);
1114:
1115: *result_end='\0';
1116:
1117: if(!asked_charset.isUTF8())
1.295 moko 1118: result = (char *)Charset::transcode(result, pa_UTF8_charset, asked_charset).cstr();
1.283 moko 1119:
1120: return result;
1121: }
1.292 moko 1122: /// must be last in this file
1123: #undef vsnprintf
1124: int pa_vsnprintf(char* b, size_t s, const char* f, va_list l) {
1125: if(!s)
1126: return 0;
1127:
1128: int r;
1129: // note: on win32 & maybe somewhere else
1130: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1131: // http://stackoverflow.com/questions/2915672/snprintf-and-visual-studio-2010
1132: --s;
1133:
1134: // clients do not check for negative 's', feature: ignore such prints
1135: if((ssize_t)s<0)
1136: return 0;
1137:
1138: #ifdef _MSC_VER
1139: // win32: if the number of bytes to write exceeds buffer, then count bytes are written and -1 is returned
1140: r=_vsnprintf(b, s, f, l);
1141: if(r<0)
1142: r=s;
1143: #else
1144: r=vsnprintf(b, s, f, l);
1145: /*
1146: solaris: man vsnprintf
1147:
1148: The snprintf() function returns the number of characters
1149: formatted, that is, the number of characters that would have
1150: been written to the buffer if it were large enough. If the
1151: value of n is 0 on a call to snprintf(), an unspecified
1152: value less than 1 is returned.
1153: */
1154:
1155: if(r<0)
1156: r=0;
1157: else if((size_t)r>s)
1158: r=s;
1159: #endif
1160: b[r]=0;
1161: return r;
1162: }
1163:
1164: int pa_snprintf(char* b, size_t s, const char* f, ...) {
1165: va_list l;
1166: va_start(l, f);
1167: int r=pa_vsnprintf(b, s, f, l);
1168: va_end(l);
1169: return r;
1170: }
1171:
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