Annotation of parser3/src/main/pa_common.C, revision 1.313
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:
1.313 ! moko 32: volatile const char * IDENT_PA_COMMON_C="$Id: pa_common.C,v 1.312 2020/02/26 12:17:30 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)
1.313 ! 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:
1.312 moko 410: bool entry_exists(const char* fname, struct stat *afinfo) {
411: struct stat lfinfo;
412: bool result=pa_stat(fname, &lfinfo)==0;
413: if(afinfo)
414: *afinfo=lfinfo;
415: return result;
416: }
417:
418: bool entry_exists(const String& file_spec) {
419: return entry_exists(file_spec.taint_cstr(String::L_FILE_SPEC), 0);
420: }
421:
422: static bool entry_ifdir(char *fname, bool need_dir) {
1.261 misha 423: if(need_dir){
424: size_t size=strlen(fname);
425: while(size) {
426: char c=fname[size-1];
427: if(c=='/' || c=='\\')
428: fname[--size]=0;
429: else
430: break;
431: }
432: }
433:
434: struct stat finfo;
1.311 moko 435: if(entry_exists(fname, &finfo)) {
1.261 misha 436: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
437: return is_dir==need_dir;
438: }
439: return false;
440: }
441:
1.312 moko 442: static bool entry_ifdir(const String& file_spec, bool need_dir) {
443: return entry_ifdir(file_spec.taint_cstrm(String::L_FILE_SPEC), need_dir);
444: }
445:
1.63 parser 446: // throws nothing! [this is required in file_move & file_delete]
1.277 moko 447: static void rmdir(const String& file_spec, size_t pos_after) {
1.261 misha 448: char* dir_spec=file_spec.taint_cstrm(String::L_FILE_SPEC);
449: size_t length=strlen(dir_spec);
450: while( (length=get_dir(dir_spec, length)) && (length > pos_after) ){
1.274 moko 451: #ifdef _MSC_VER
1.312 moko 452: if(!entry_ifdir(dir_spec, true))
1.261 misha 453: break;
454: DWORD attrs=GetFileAttributes(dir_spec);
455: if(
456: (attrs==INVALID_FILE_ATTRIBUTES)
457: || !(attrs & FILE_ATTRIBUTE_DIRECTORY)
458: || (attrs & FILE_ATTRIBUTE_REPARSE_POINT)
459: )
460: break;
461: #endif
462: if( rmdir(dir_spec) )
463: break;
464: };
1.50 paf 465: }
1.239 misha 466:
1.269 misha 467: bool file_delete(const String& file_spec, bool fail_on_problem, bool keep_empty_dirs) {
1.247 misha 468: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.282 moko 469: if(unlink(fname)!=0) {
1.164 paf 470: if(fail_on_problem)
1.289 moko 471: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
472: &file_spec, "unlink failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.93 paf 473: else
474: return false;
1.282 moko 475: }
1.50 paf 476:
1.269 misha 477: if(!keep_empty_dirs)
478: rmdir(file_spec, 1);
479:
1.93 paf 480: return true;
1.60 parser 481: }
1.239 misha 482:
1.269 misha 483: void file_move(const String& old_spec, const String& new_spec, bool keep_empty_dirs) {
1.247 misha 484: const char* old_spec_cstr=old_spec.taint_cstr(String::L_FILE_SPEC);
485: const char* new_spec_cstr=new_spec.taint_cstr(String::L_FILE_SPEC);
1.63 parser 486:
1.126 paf 487: create_dir_for_file(new_spec);
1.63 parser 488:
1.60 parser 489: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.289 moko 490: throw Exception(errno==EACCES ? "file.access" : errno==ENOENT ? "file.missing" : 0,
491: &old_spec, "rename failed: %s (%d), actual filename '%s' to '%s'", strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 492:
1.269 misha 493: if(!keep_empty_dirs)
494: rmdir(old_spec, 1);
1.31 paf 495: }
496:
1.51 paf 497:
1.215 paf 498: bool file_exist(const String& file_spec) {
1.312 moko 499: return entry_ifdir(file_spec, false);
1.51 paf 500: }
1.239 misha 501:
1.215 paf 502: bool dir_exists(const String& file_spec) {
1.312 moko 503: return entry_ifdir(file_spec, true);
1.65 parser 504: }
1.239 misha 505:
1.215 paf 506: const String* file_exist(const String& path, const String& name) {
1.154 paf 507: String& result=*new String(path);
1.270 moko 508: if(path.last_char() != '/')
509: result << "/";
1.154 paf 510: result << name;
1.215 paf 511: return file_exist(result)?&result:0;
1.43 paf 512: }
1.239 misha 513:
1.43 paf 514: bool file_executable(const String& file_spec) {
1.247 misha 515: return access(file_spec.taint_cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 516: }
517:
1.296 moko 518: 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 519: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.154 paf 520: struct stat finfo;
1.298 moko 521: if(pa_stat(fname, &finfo)!=0) {
1.64 parser 522: if(fail_on_read_problem)
1.289 moko 523: throw Exception("file.missing", &file_spec, "getting file size failed: %s (%d), real filename '%s'", strerror(errno), errno, fname);
1.64 parser 524: else
525: return false;
1.282 moko 526: }
1.58 parser 527: rsize=finfo.st_size;
1.299 moko 528: ratime=(time_t)finfo.st_atime;
529: rmtime=(time_t)finfo.st_mtime;
530: rctime=(time_t)finfo.st_ctime;
1.64 parser 531: return true;
1.18 paf 532: }
533:
1.313 ! moko 534: size_t check_file_size(uint64_t size, const String* file_spec){
1.300 moko 535: if(size > pa_file_size_limit)
1.313 ! moko 536: throw Exception(PARSER_RUNTIME, file_spec, "content size of %.15g bytes exceeds the limit (%.15g bytes)", (double)size, (double)pa_file_size_limit);
1.300 moko 537: return (size_t)size;
538: }
539:
1.278 moko 540: /**
541: String related functions
542: */
543:
544: bool capitalized(const char* s){
545: bool upper=true;
546: for(const char* c=s; *c; c++){
547: if(*c != (upper ? toupper((unsigned char)*c) : tolower((unsigned char)*c)))
548: return false;
549: upper=strchr("-_ ", *c) != 0;
550: }
551: return true;
552: }
553:
554: const char* capitalize(const char* s){
555: if(!s || capitalized(s))
556: return s;
557:
558: char* result=pa_strdup(s);
559: if(result){
560: bool upper=true;
561: for(char* c=result; *c; c++){
562: *c=upper ? (char)toupper((unsigned char)*c) : (char)tolower((unsigned char)*c);
563: upper=strchr("-_ ", *c) != 0;
564: }
565: }
566: return (const char*)result;
567: }
568:
1.290 moko 569: char *str_lower(const char *s, size_t helper_length){
570: char *result=pa_strdup(s, helper_length);
571: for(char* c=result; *c; c++)
572: *c=(char)tolower((unsigned char)*c);
573: return result;
574: }
575:
576: char *str_upper(const char *s, size_t helper_length){
577: char *result=pa_strdup(s, helper_length);
578: for(char* c=result; *c; c++)
579: *c=(char)toupper((unsigned char)*c);
580: return result;
581: }
582:
1.278 moko 583: void fix_line_breaks(char *str, size_t& length) {
584: //_asm int 3;
585: const char* const eob=str+length;
586: char* dest=str;
587: // fix DOS: \r\n -> \n
588: // fix Macintosh: \r -> \n
589: char* bol=str;
590: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
591: size_t len=eol-bol;
592: if(dest!=bol)
593: memmove(dest, bol, len);
594: dest+=len;
595: *dest++='\n';
596:
597: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
598: bol=eol+2;
599: length--;
600: } else // \r, not \n = Macintosh
601: bol=eol+1;
602: }
603: // last piece without \r
604: if(dest!=bol)
605: memmove(dest, bol, eob-bol);
606: str[length]=0; // terminating
607: }
608:
609: /**
610: scans for @a delim[default \n] in @a *row_ref,
611: @return piece of line before it or end of string, if no @a delim found
612: assigns @a *row_ref to point right after delimiter if there were one
613: or to zero if no @a delim were found.
614: */
615:
1.126 paf 616: char* getrow(char* *row_ref, char delim) {
1.229 misha 617: char* result=*row_ref;
618: if(result) {
1.126 paf 619: *row_ref=strchr(result, delim);
1.8 paf 620: if(*row_ref)
621: *((*row_ref)++)=0;
622: else if(!*result)
623: return 0;
1.229 misha 624: }
625: return result;
1.8 paf 626: }
627:
1.126 paf 628: char* lsplit(char* string, char delim) {
1.229 misha 629: if(string) {
1.126 paf 630: char* v=strchr(string, delim);
1.8 paf 631: if(v) {
632: *v=0;
633: return v+1;
634: }
1.229 misha 635: }
636: return 0;
1.8 paf 637: }
638:
1.126 paf 639: char* lsplit(char* *string_ref, char delim) {
1.229 misha 640: char* result=*string_ref;
1.126 paf 641: char* next=lsplit(*string_ref, delim);
1.229 misha 642: *string_ref=next;
643: return result;
1.9 paf 644: }
645:
1.126 paf 646: char* rsplit(char* string, char delim) {
1.229 misha 647: if(string) {
1.126 paf 648: char* v=strrchr(string, delim);
1.18 paf 649: if(v) {
1.9 paf 650: *v=0;
651: return v+1;
652: }
1.229 misha 653: }
654: return NULL;
1.10 paf 655: }
656:
1.229 misha 657:
658: // format: %[flags][width][.precision]type http://msdn.microsoft.com/ru-ru/library/56e442dc(en-us,VS.80).aspx
659: // flags: '-', '+', ' ', '#', '0' http://msdn.microsoft.com/ru-ru/library/8aky45ct(en-us,VS.80).aspx
660: // width, precision: non negative decimal number
661: enum FormatType {
662: FormatInvalid,
663: FormatInt,
664: FormatUInt,
665: FormatDouble
666: };
1.272 moko 667: FormatType format_type(const char* fmt){
1.229 misha 668: enum FormatState {
669: Percent,
670: Flags,
671: Width,
672: Precision,
673: Done
674: } state=Percent;
675:
676: FormatType result=FormatInvalid;
677:
1.272 moko 678: const char* pos=fmt;
1.229 misha 679: while(char c=*(pos++)){
680: switch(state){
681: case Percent:
682: if(c=='%'){
683: state=Flags;
684: } else {
685: return FormatInvalid; // 1st char must be '%' only
686: }
687: break;
688: case Flags:
689: if(strchr("-+ #0", c)!=0){
690: break;
691: }
692: // go to the next step
693: case Width:
694: if(c=='.'){
695: state=Precision;
696: break;
697: }
698: // go to the next step
699: case Precision:
700: if(c>='0' && c<='9'){
701: if(state == Flags) state=Width; // no more flags
702: break;
703: } else if(c=='d' || c=='i'){
704: result=FormatInt;
705: } else if(strchr("feEgG", c)!=0){
706: result=FormatDouble;
707: } else if(strchr("uoxX", c)!=0){
708: result=FormatUInt;
709: } else {
710: return FormatInvalid; // invalid char
711: }
712: state=Done;
713: break;
714: case Done:
715: return FormatInvalid; // no chars allowed after 'type'
716: }
717: }
718: return result;
719: }
720:
721:
1.272 moko 722: const char* format(double value, const char* fmt) {
1.229 misha 723: char local_buf[MAX_NUMBER];
1.235 misha 724: int size=-1;
1.229 misha 725:
726: if(fmt && strlen(fmt)){
727: switch(format_type(fmt)){
728: case FormatDouble:
729: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
730: break;
731: case FormatInt:
732: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
733: break;
734: case FormatUInt:
1.126 paf 735: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.229 misha 736: break;
737: case FormatInvalid:
1.289 moko 738: throw Exception(PARSER_RUNTIME, 0, "Incorrect format string '%s' was specified.", fmt);
1.229 misha 739: }
740: } else
741: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
742:
743: if(size < 0 || size >= MAX_NUMBER-1){ // on win32 we manually reduce max size while printing
1.304 moko 744: throw Exception(PARSER_RUNTIME, 0, "Error occurred white executing snprintf with format string '%s'.", fmt);
1.229 misha 745: }
746:
1.235 misha 747: return pa_strdup(local_buf, (size_t)size);
1.12 paf 748: }
749:
1.36 paf 750: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 751: #ifdef WIN32
1.187 paf 752: size_t to_write = size;
1.12 paf 753: do{
1.154 paf 754: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 755: if(chunk_written<=0)
756: break;
757: size-=chunk_written;
1.36 paf 758: buf=((const char*)buf)+chunk_written;
1.126 paf 759: } while(size>0);
1.12 paf 760:
1.187 paf 761: return to_write-size;
1.12 paf 762: #else
1.126 paf 763: return fwrite(buf, 1, size, stdout);
1.12 paf 764: #endif
1.2 paf 765: }
1.14 paf 766:
1.229 misha 767: enum EscapeState {
768: EscapeRest,
769: EscapeFirst,
770: EscapeSecond,
771: EscapeUnicode
772: };
773:
1.236 misha 774: // @todo prescan for reduce required size (unescaped sting in 1 byte charset requires less memory usually)
1.258 misha 775: char* unescape_chars(const char* cp, int len, Charset* charset, bool js){
1.236 misha 776: char* s=new(PointerFreeGC) char[len+1]; // must be enough (%uXXXX==6 bytes, max utf-8 char length==6 bytes)
1.230 misha 777: char* dst=s;
1.229 misha 778: EscapeState escapeState=EscapeRest;
779: uint escapedValue=0;
780: int srcPos=0;
1.230 misha 781: short int jsCnt=0;
1.236 misha 782: while(srcPos<len){
1.229 misha 783: uchar c=(uchar)cp[srcPos];
1.258 misha 784: if(c=='%' || (c=='\\' && js)){
1.229 misha 785: escapeState=EscapeFirst;
786: } else {
787: switch(escapeState) {
788: case EscapeRest:
1.286 moko 789: if(c=='+' && !js){
1.230 misha 790: *dst++=' ';
1.229 misha 791: } else {
1.230 misha 792: *dst++=c;
1.229 misha 793: }
794: break;
795: case EscapeFirst:
1.232 misha 796: if(charset && c=='u'){
1.229 misha 797: // escaped unicode value: %u0430
798: jsCnt=0;
799: escapedValue=0;
800: escapeState=EscapeUnicode;
801: } else {
1.231 misha 802: if(isxdigit(c)){
1.229 misha 803: escapedValue=hex_value[c] << 4;
804: escapeState=EscapeSecond;
805: } else {
1.230 misha 806: *dst++=c;
1.229 misha 807: escapeState=EscapeRest;
808: }
809: }
810: break;
811: case EscapeSecond:
1.231 misha 812: if(isxdigit(c)){
1.229 misha 813: escapedValue+=hex_value[c];
1.230 misha 814: *dst++=(char)escapedValue;
1.229 misha 815: }
816: escapeState=EscapeRest;
817: break;
818: case EscapeUnicode:
1.231 misha 819: if(isxdigit(c)){
1.229 misha 820: escapedValue=(escapedValue << 4) + hex_value[c];
821: if(++jsCnt==4){
1.230 misha 822: // transcode utf8 char to client charset (we can lost some chars here)
1.232 misha 823: charset->store_Char((XMLByte*&)dst, (XMLCh)escapedValue, '?');
1.229 misha 824: escapeState=EscapeRest;
825: }
826: } else {
827: // not full unicode value
828: escapeState=EscapeRest;
829: }
830: break;
831: }
832: }
833:
834: srcPos++;
835: }
836:
1.230 misha 837: *dst=0; // zero-termination
1.229 misha 838: return s;
839: }
1.24 paf 840:
1.268 misha 841: char *search_stop(char*& current, char cstop_at) {
842: // sanity check
843: if(!current)
844: return 0;
845:
846: // skip leading WS
847: while(*current==' ' || *current=='\t')
848: current++;
849: if(!*current)
850: return current=0;
851:
852: char *result=current;
853: if(char *pstop_at=strchr(current, cstop_at)) {
854: *pstop_at=0;
855: current=pstop_at+1;
856: } else
857: current=0;
858: return result;
859: }
860:
1.24 paf 861: #ifdef WIN32
1.126 paf 862: void back_slashes_to_slashes(char* s) {
1.24 paf 863: if(s)
864: for(; *s; s++)
865: if(*s=='\\')
1.126 paf 866: *s='/';
1.24 paf 867: }
868: #endif
1.41 paf 869:
1.232 misha 870: size_t strpos(const char *str, const char *substr) {
871: const char *p = strstr(str, substr);
872: return (p==0)?STRING_NOT_FOUND:p-str;
873: }
874:
1.226 misha 875: int remove_crlf(char* start, char* end) {
876: char* from=start;
877: char* to=start;
878: bool skip=false;
879: while(from < end){
880: switch(*from){
881: case '\n':
882: case '\r':
883: case '\t':
884: case ' ':
885: if(!skip){
886: *to=' ';
887: to++;
888: skip=true;
889: }
890: break;
891: default:
892: if(from != to)
893: *to=*from;
894: to++;
895: skip=false;
1.69 parser 896: }
1.226 misha 897: from++;
898: }
899: return to-start;
1.91 paf 900: }
901:
1.279 moko 902: const char* hex_digits="0123456789ABCDEF";
903:
1.278 moko 904: const char* hex_string(unsigned char* bytes, size_t size, bool upcase) {
905: char *bytes_hex=new(PointerFreeGC) char [size*2/*byte->hh*/+1/*for zero-teminator*/];
906: unsigned char *src=bytes;
907: unsigned char *end=bytes+size;
908: char *dest=bytes_hex;
909:
1.279 moko 910: const char *hex=upcase? hex_digits : "0123456789abcdef";
1.278 moko 911:
912: for(; src<end; src++) {
913: *dest++=hex[*src/0x10];
914: *dest++=hex[*src%0x10];
915: }
916: *dest=0;
917:
918: return bytes_hex;
919: }
1.91 paf 920:
1.221 misha 921: int file_block_read(const int f, unsigned char* buffer, const size_t size){
922: int nCount = read(f, buffer, size);
923: if (nCount < 0)
1.289 moko 924: throw Exception("file.read", 0, "read failed: %s (%d)", strerror(errno), errno);
1.221 misha 925: return nCount;
926: }
927:
1.278 moko 928: static unsigned long crc32Table[256];
929: static void InitCrc32Table()
930: {
931: if(crc32Table[1] == 0){
932: // This is the official polynomial used by CRC32 in PKZip.
933: // Often times the polynomial shown reversed as 0x04C11DB7.
934: static const unsigned long dwPolynomial = 0xEDB88320;
935:
936: for(int i = 0; i < 256; i++)
937: {
938: unsigned long dwCrc = i;
939: for(int j = 8; j > 0; j--)
940: {
941: if(dwCrc & 1)
942: dwCrc = (dwCrc >> 1) ^ dwPolynomial;
943: else
944: dwCrc >>= 1;
945: }
946: crc32Table[i] = dwCrc;
947: }
948: }
949: }
950:
951: inline void CalcCrc32(const unsigned char byte, unsigned long &crc32)
952: {
953: crc32 = ((crc32) >> 8) ^ crc32Table[(byte) ^ ((crc32) & 0x000000FF)];
954: }
955:
956:
1.297 moko 957: unsigned long pa_crc32(const char *in, size_t in_size){
1.218 misha 958: unsigned long crc32=0xFFFFFFFF;
1.220 misha 959:
1.240 misha 960: InitCrc32Table();
1.239 misha 961: for(size_t i = 0; i<in_size; i++)
962: CalcCrc32(in[i], crc32);
1.220 misha 963:
1.218 misha 964: return ~crc32;
965: }
966:
1.289 moko 967: static void file_crc32_file_action(struct stat& finfo, int f, const String&, void *context) {
1.218 misha 968: unsigned long& crc32=*static_cast<unsigned long *>(context);
969: if(finfo.st_size) {
970: InitCrc32Table();
1.220 misha 971: int nCount=0;
1.218 misha 972: do {
1.221 misha 973: unsigned char buffer[FILE_BUFFER_SIZE];
974: nCount = file_block_read(f, buffer, sizeof(buffer));
1.220 misha 975: for(int i = 0; i < nCount; i++) CalcCrc32(buffer[i], crc32);
976: } while(nCount > 0);
1.218 misha 977: }
978: }
979:
1.297 moko 980: unsigned long pa_crc32(const String& file_spec){
1.278 moko 981: unsigned long crc32=0xFFFFFFFF;
982: file_read_action_under_lock(file_spec, "crc32", file_crc32_file_action, &crc32);
983: return ~crc32;
984: }
985:
986: // content-type: xxx; charset=WE-NEED-THIS
987: // content-type: xxx; charset="WE-NEED-THIS"
988: // content-type: xxx; charset="WE-NEED-THIS";
989: Charset* detect_charset(const char* content_type){
990: if(content_type){
1.291 moko 991: char* CONTENT_TYPE=str_upper(content_type);
1.278 moko 992:
993: if(const char* begin=strstr(CONTENT_TYPE, "CHARSET=")){
994: begin+=8; // skip "CHARSET="
995: char* end=0;
996: if(*begin && (*begin=='"' || *begin =='\'')){
997: char quote=*begin;
998: begin++;
999: end=(char*)strchr(begin, quote);
1000: }
1001: if(!end)
1002: end=(char*)strchr(begin, ';');
1003:
1004: if(end)
1005: *end=0; // terminator
1006:
1.295 moko 1007: return *begin ? &pa_charsets.get_direct(begin) : 0;
1.278 moko 1008: }
1009: }
1010: return 0;
1011: }
1012:
1.301 moko 1013: const UTF16* pa_utf16_encode(const char* in, Charset& source_charset){
1.302 moko 1014: if(!in)
1015: return 0;
1.301 moko 1016:
1.302 moko 1017: String::C sIn(in,strlen(in));
1.301 moko 1018:
1.302 moko 1019: UTF16* utf16=(UTF16*)pa_malloc_atomic(sIn.length*2+2);
1020: UTF16* utf16_end=utf16;
1.301 moko 1021:
1.302 moko 1022: if(!source_charset.isUTF8())
1023: sIn=Charset::transcode(sIn, source_charset, pa_UTF8_charset);
1.301 moko 1024:
1.302 moko 1025: int status=pa_convertUTF8toUTF16((const UTF8**)&sIn.str, (const UTF8*)(sIn.str+sIn.length), &utf16_end, utf16+sIn.length, strictConversion);
1026: if(status != conversionOK)
1027: throw Exception("utf-16 encode", new String(in), "utf-16 conversion failed (%d)", status);
1.301 moko 1028:
1.302 moko 1029: *utf16_end=0;
1.301 moko 1030:
1.302 moko 1031: return utf16;
1.301 moko 1032: }
1033:
1034: const char* pa_utf16_decode(const UTF16* in, Charset& asked_charset){
1035: if(!in)
1036: return 0;
1037:
1038: const UTF16* utf16_start=in;
1039: const UTF16* utf16_end;
1040:
1041: for(utf16_end=in; *utf16_end; utf16_end++);
1042:
1043: char *result = (char *)pa_malloc_atomic((utf16_end-in)*6+1);
1044: char *result_end = result;
1045:
1046: int status=pa_convertUTF16toUTF8(&utf16_start, utf16_end, (UTF8**)&result_end, (UTF8*)(result+(utf16_end-in)*6), strictConversion);
1047:
1048: if(status != conversionOK)
1049: throw Exception("utf-16 decode", 0, "utf conversion failed (%d)", status);
1050:
1051: *result_end='\0';
1052:
1.302 moko 1053: if(asked_charset.isUTF8())
1054: return result;
1.301 moko 1055:
1.302 moko 1056: return Charset::transcode(result, pa_UTF8_charset, asked_charset).cstr();
1.301 moko 1057: }
1.278 moko 1058:
1.283 moko 1059: static bool is_latin(const char *in){
1060: for(; *in; in++){
1061: if ((unsigned char)(*in) > 0x7F)
1062: return false;
1063: }
1064: return true;
1065: }
1066:
1067: #define MAX_IDNA_LENGTH 256
1068:
1.301 moko 1069: const char *pa_idna_encode(const char *in, Charset& source_charset){
1.283 moko 1070: if(!in || is_latin(in))
1071: return in;
1072:
1073: uint32_t utf32[MAX_IDNA_LENGTH];
1074: uint32_t *utf32_end=utf32;
1075:
1076: String::C sIn(in,strlen(in));
1077:
1078: if(!source_charset.isUTF8())
1.295 moko 1079: sIn=Charset::transcode(sIn, source_charset, pa_UTF8_charset);
1.283 moko 1080:
1081: int status=pa_convertUTF8toUTF32((const UTF8**)&sIn.str, (const UTF8*)(sIn.str+sIn.length), &utf32_end, utf32+MAX_IDNA_LENGTH-1, strictConversion);
1082: if(status != conversionOK)
1083: throw Exception("idna encode", new String(in), "utf conversion failed (%d)", status);
1084:
1085: *utf32_end=0;
1086:
1087: char *result = (char *)pa_malloc(MAX_IDNA_LENGTH);
1088: status=pa_idna_to_ascii_4z(utf32, result, MAX_IDNA_LENGTH, 0);
1089: if(status != IDNA_SUCCESS)
1090: throw Exception("idna encode", new String(in), "encode failed: %s", pa_idna_strerror(status));
1091:
1092: return result;
1093: }
1094:
1095: const char *pa_idna_decode(const char *in, Charset &asked_charset){
1096: if(!in || !(*in))
1097: return in;
1098:
1099: uint32_t utf32[MAX_IDNA_LENGTH];
1100: const uint32_t *utf32_start=utf32;
1101: uint32_t *utf32_end;
1102:
1103: int status=pa_idna_to_unicode_4z(in, utf32, MAX_IDNA_LENGTH, 0);
1104: if(status != IDNA_SUCCESS)
1105: throw Exception("idna decode", new String(in), "decode failed: %s", pa_idna_strerror(status));
1106:
1107: for(utf32_end=utf32; *utf32_end; utf32_end++);
1108:
1109: char *result = (char *)pa_malloc(MAX_IDNA_LENGTH);
1110: char *result_end = result;
1111:
1112: status=pa_convertUTF32toUTF8(&utf32_start, utf32_end, (UTF8**)&result_end, (UTF8*)(result+MAX_IDNA_LENGTH-1), strictConversion);
1113:
1114: if(status != conversionOK)
1115: throw Exception("idna decode", new String(in), "utf conversion failed (%d)", status);
1116:
1117: *result_end='\0';
1118:
1119: if(!asked_charset.isUTF8())
1.295 moko 1120: result = (char *)Charset::transcode(result, pa_UTF8_charset, asked_charset).cstr();
1.283 moko 1121:
1122: return result;
1123: }
1.292 moko 1124: /// must be last in this file
1125: #undef vsnprintf
1126: int pa_vsnprintf(char* b, size_t s, const char* f, va_list l) {
1127: if(!s)
1128: return 0;
1129:
1130: int r;
1131: // note: on win32 & maybe somewhere else
1132: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1133: // http://stackoverflow.com/questions/2915672/snprintf-and-visual-studio-2010
1134: --s;
1135:
1136: // clients do not check for negative 's', feature: ignore such prints
1137: if((ssize_t)s<0)
1138: return 0;
1139:
1140: #ifdef _MSC_VER
1141: // win32: if the number of bytes to write exceeds buffer, then count bytes are written and -1 is returned
1142: r=_vsnprintf(b, s, f, l);
1143: if(r<0)
1144: r=s;
1145: #else
1146: r=vsnprintf(b, s, f, l);
1147: /*
1148: solaris: man vsnprintf
1149:
1150: The snprintf() function returns the number of characters
1151: formatted, that is, the number of characters that would have
1152: been written to the buffer if it were large enough. If the
1153: value of n is 0 on a call to snprintf(), an unspecified
1154: value less than 1 is returned.
1155: */
1156:
1157: if(r<0)
1158: r=0;
1159: else if((size_t)r>s)
1160: r=s;
1161: #endif
1162: b[r]=0;
1163: return r;
1164: }
1165:
1166: int pa_snprintf(char* b, size_t s, const char* f, ...) {
1167: va_list l;
1168: va_start(l, f);
1169: int r=pa_vsnprintf(b, s, f, l);
1170: va_end(l);
1171: return r;
1172: }
1173:
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