Annotation of parser3/src/main/pa_common.C, revision 1.306
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:
1.306 ! moko 32: volatile const char * IDENT_PA_COMMON_C="$Id: pa_common.C,v 1.305 2019/09/06 10:17:07 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:
1.271 moko 94: char* file_read_text(Request_charsets& charsets, const String& file_spec, bool fail_on_read_problem, HashStringValue* params, bool transcode_result) {
95: File_read_result file=file_read(charsets, file_spec, true, params, fail_on_read_problem, 0, 0, 0, transcode_result);
1.154 paf 96: return file.success?file.str:0;
1.126 paf 97: }
98:
1.271 moko 99: char* file_load_text(Request& r, const String& file_spec, bool fail_on_read_problem, HashStringValue* params, bool transcode_result) {
100: File_read_result file=file_load(r, file_spec, true, params, fail_on_read_problem, 0, 0, 0, transcode_result);
1.241 misha 101: return file.success?file.str:0;
102: }
103:
1.206 paf 104: /// these options were handled but not checked elsewhere, now check them
1.239 misha 105: int pa_get_valid_file_options_count(HashStringValue& options) {
1.206 paf 106: int result=0;
107: if(options.get(PA_SQL_LIMIT_NAME))
108: result++;
109: if(options.get(PA_SQL_OFFSET_NAME))
110: result++;
111: if(options.get(PA_COLUMN_SEPARATOR_NAME))
112: result++;
113: if(options.get(PA_COLUMN_ENCLOSER_NAME))
114: result++;
1.223 misha 115: if(options.get(PA_CHARSET_NAME))
116: result++;
1.206 paf 117: return result;
118: }
119:
1.123 paf 120: #ifndef DOXYGEN
121: struct File_read_action_info {
1.154 paf 122: char **data; size_t *data_size;
1.188 paf 123: char* buf; size_t offset; size_t count;
1.126 paf 124: };
1.123 paf 125: #endif
1.271 moko 126:
1.289 moko 127: static void file_read_action(struct stat& finfo, int f, const String& file_spec, void *context) {
1.126 paf 128: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.188 paf 129: size_t to_read_size=info.count;
130: if(!to_read_size)
1.300 moko 131: to_read_size=check_file_size(finfo.st_size, file_spec);
1.271 moko 132: if(to_read_size) {
1.188 paf 133: if(info.offset)
134: lseek(f, info.offset, SEEK_SET);
1.271 moko 135: *info.data=info.buf ? info.buf : (char *)pa_malloc_atomic(to_read_size+1);
136: ssize_t result=read(f, *info.data, to_read_size);
137: if(result<0)
138: throw Exception("file.read", &file_spec, "read failed: %s (%d)", strerror(errno), errno);
139: *info.data_size=result;
1.123 paf 140: } else { // empty file
1.209 paf 141: // 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 142: *info.data=(char *)pa_malloc_atomic(1);
1.209 paf 143: *(char*)(*info.data)=0;
1.123 paf 144: *info.data_size=0;
145: return;
146: }
1.126 paf 147: }
1.241 misha 148:
1.154 paf 149: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
1.229 misha 150: bool as_text, HashStringValue *params,
151: bool fail_on_read_problem,
1.234 misha 152: char* buf, size_t offset, size_t count, bool transcode_text_result) {
1.167 paf 153: File_read_result result={false, 0, 0, 0};
1.241 misha 154: if(params){
155: int valid_options=pa_get_valid_file_options_count(*params);
156: if(valid_options!=params->count())
1.262 misha 157: throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION);
1.241 misha 158: }
1.203 paf 159:
1.241 misha 160: File_read_action_info info={&result.str, &result.length, buf, offset, count};
1.161 paf 161:
1.289 moko 162: result.success=file_read_action_under_lock(file_spec, "read", file_read_action, &info, as_text, fail_on_read_problem);
1.223 misha 163:
1.241 misha 164: if(as_text){
165: if(result.success){
1.263 misha 166: Charset* asked_charset=0;
167: if(params)
168: if(Value* vcharset_name=params->get(PA_CHARSET_NAME))
1.295 moko 169: asked_charset=&pa_charsets.get(vcharset_name->as_string());
1.263 misha 170:
1.295 moko 171: asked_charset=pa_charsets.checkBOM(result.str, result.length, asked_charset);
1.287 moko 172:
1.263 misha 173: 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
174: String::C body=String::C(result.str, result.length);
175: body=Charset::transcode(body, *asked_charset, charsets.source());
1.236 misha 176:
1.263 misha 177: result.str=const_cast<char*>(body.str); // hacking a little
178: result.length=body.length;
1.131 paf 179: }
180: }
1.241 misha 181: if(result.length)
182: fix_line_breaks(result.str, result.length);
1.123 paf 183: }
1.241 misha 184:
185: return result;
186: }
187:
188: File_read_result file_load(Request& r, const String& file_spec,
189: bool as_text, HashStringValue *params,
190: bool fail_on_read_problem,
191: char* buf, size_t offset, size_t count, bool transcode_text_result) {
192:
193: File_read_result result={false, 0, 0, 0};
194: if(file_spec.starts_with("http://")) {
195: if(offset || count)
1.289 moko 196: throw Exception(PARSER_RUNTIME, 0, "offset and load options are not supported for HTTP:// file load");
1.241 misha 197:
198: // fail on read problem
199: File_read_http_result http=pa_internal_file_read_http(r, file_spec, as_text, params, transcode_text_result);
200: result.success=true;
201: result.str=http.str;
202: result.length=http.length;
203: result.headers=http.headers;
204: } else
205: result=
206: file_read(r.charsets, file_spec, as_text, params, fail_on_read_problem, buf, offset, count, transcode_text_result);
1.126 paf 207:
208: return result;
1.123 paf 209: }
210:
1.257 pretende 211:
1.154 paf 212: #ifdef PA_SAFE_MODE
1.259 misha 213: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
1.154 paf 214: if(finfo.st_uid/*foreign?*/!=geteuid()
215: && finfo.st_gid/*foreign?*/!=getegid())
1.289 moko 216: throw Exception(PARSER_RUNTIME,
217: &file_spec,
218: "parser is in safe mode: reading files of foreign group and user disabled "
219: "[recompile parser with --disable-safe-mode configure option], "
220: "actual filename '%s', fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
221: fname, finfo.st_uid, geteuid(), finfo.st_gid, getegid()
222: );
1.259 misha 223: }
224: #else
225: void check_safe_mode(struct stat, const String&, const char*) {
226: }
1.257 pretende 227: #endif
1.259 misha 228:
1.257 pretende 229:
1.149 paf 230:
1.154 paf 231: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 232: const char* action_name, File_read_action action, void *context,
233: bool as_text,
1.123 paf 234: bool fail_on_read_problem) {
1.247 misha 235: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.33 paf 236: int f;
237:
238: // first open, next stat:
1.45 paf 239: // directory update of NTFS hard links performed on open.
1.33 paf 240: // ex:
241: // a.html:^test[] and b.html hardlink to a.html
242: // user inserts ! before ^test in a.html
1.126 paf 243: // directory entry of b.html in NTFS not updated at once,
1.35 paf 244: // they delay update till open, so we would receive "!^test[" string
245: // if would do stat, next open.
1.123 paf 246: // later: it seems, even this does not help sometimes
1.301 moko 247: if((f=pa_open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 248: try {
1.162 paf 249: if(pa_lock_shared_blocking(f)!=0)
1.289 moko 250: throw Exception("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.123 paf 251:
1.124 paf 252: struct stat finfo;
1.298 moko 253: if(pa_fstat(f, &finfo)!=0)
1.124 paf 254: throw Exception("file.missing", // hardly possible: we just opened it OK
1.289 moko 255: &file_spec, "stat failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.124 paf 256:
1.149 paf 257: check_safe_mode(finfo, file_spec, fname);
1.32 paf 258:
1.289 moko 259: action(finfo, f, file_spec, context);
1.123 paf 260: } catch(...) {
1.162 paf 261: pa_unlock(f);close(f);
1.123 paf 262: if(fail_on_read_problem)
1.154 paf 263: rethrow;
1.289 moko 264: return false;
1.123 paf 265: }
1.87 paf 266:
1.162 paf 267: pa_unlock(f);close(f);
1.72 parser 268: return true;
1.229 misha 269: } else {
1.118 paf 270: if(fail_on_read_problem)
1.289 moko 271: throw Exception(errno==EACCES ? "file.access" : (errno==ENOENT || errno==ENOTDIR || errno==ENODEV) ? "file.missing" : 0,
272: &file_spec, "%s failed: %s (%d), actual filename '%s'", action_name, strerror(errno), errno, fname);
1.118 paf 273: return false;
274: }
1.8 paf 275: }
276:
1.202 paf 277: void create_dir_for_file(const String& file_spec) {
1.63 parser 278: size_t pos_after=1;
1.154 paf 279: size_t pos_before;
280: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
1.277 moko 281: pa_mkdir(file_spec.mid(0, pos_before).taint_cstr(String::L_FILE_SPEC), 0775);
1.63 parser 282: pos_after=pos_before+1;
283: }
284: }
285:
1.98 paf 286: bool file_write_action_under_lock(
1.28 paf 287: const String& file_spec,
1.225 misha 288: const char* action_name,
289: File_write_action action,
290: void *context,
1.126 paf 291: bool as_text,
292: bool do_append,
293: bool do_block,
1.110 paf 294: bool fail_on_lock_problem) {
1.247 misha 295: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.28 paf 296: int f;
1.80 paf 297: if(access(fname, W_OK)!=0) // no
1.126 paf 298: create_dir_for_file(file_spec);
1.50 paf 299:
1.301 moko 300: if((f=pa_open(fname,
1.80 paf 301: O_CREAT|O_RDWR
302: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 303: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 304: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.289 moko 305: Exception e("file.lock", &file_spec, "shared lock failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.126 paf 306: close(f);
1.110 paf 307: if(fail_on_lock_problem)
308: throw e;
1.98 paf 309: return false;
310: }
1.96 paf 311:
1.158 paf 312: try {
1.254 misha 313: #if (defined(HAVE_FCHMOD) && defined(PA_SAFE_MODE))
314: struct stat finfo;
1.298 moko 315: if(pa_fstat(f, &finfo)==0 && finfo.st_mode & 0111)
1.254 misha 316: fchmod(f, finfo.st_mode & 0666/*clear executable bits*/); // backward: ignore errors if any
317: #endif
318: action(f, context);
1.158 paf 319: } catch(...) {
1.138 paf 320: #ifdef HAVE_FTRUNCATE
1.104 paf 321: if(!do_append)
1.125 paf 322: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 323: #endif
1.162 paf 324: pa_unlock(f);close(f);
1.154 paf 325: rethrow;
1.158 paf 326: }
1.80 paf 327:
1.138 paf 328: #ifdef HAVE_FTRUNCATE
1.104 paf 329: if(!do_append)
1.125 paf 330: 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 331: #endif
1.162 paf 332: pa_unlock(f);close(f);
1.98 paf 333: return true;
1.80 paf 334: } else
1.289 moko 335: 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 336: // here should be nothing, see rethrow above
337: }
338:
339: #ifndef DOXYGEN
340: struct File_write_action_info {
1.250 misha 341: const char* str;
342: size_t length;
1.126 paf 343: };
1.96 paf 344: #endif
1.271 moko 345:
1.96 paf 346: static void file_write_action(int f, void *context) {
1.126 paf 347: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 348: if(info.length) {
1.271 moko 349: ssize_t written=write(f, info.str, info.length);
1.116 paf 350: if(written<0)
1.271 moko 351: throw Exception("file.write", 0, "write failed: %s (%d)", strerror(errno), errno);
1.275 moko 352: if((size_t)written!=info.length)
1.271 moko 353: throw Exception("file.write", 0, "write failed: %u of %u bytes written", written, info.length);
1.113 paf 354: }
1.96 paf 355: }
1.271 moko 356:
1.96 paf 357: void file_write(
1.250 misha 358: Request_charsets& charsets,
359: const String& file_spec,
360: const char* data,
361: size_t size,
1.126 paf 362: bool as_text,
1.250 misha 363: bool do_append,
364: Charset* asked_charset) {
365:
366: if(as_text && asked_charset){
367: String::C body=String::C(data, size);
368: body=Charset::transcode(body, charsets.source(), *asked_charset);
369: data=body.str;
370: size=body.length;
371: };
372:
1.126 paf 373: File_write_action_info info={data, size};
1.225 misha 374:
1.98 paf 375: file_write_action_under_lock(
1.154 paf 376: file_spec,
1.225 misha 377: "write",
378: file_write_action,
379: &info,
1.154 paf 380: as_text,
381: do_append);
1.30 paf 382: }
383:
1.261 misha 384: static size_t get_dir(char* fname, size_t helper_length){
385: bool dir=false;
386: size_t pos=0;
387: for(pos=helper_length; pos; pos--){
388: char c=fname[pos-1];
389: if(c=='/' || c=='\\'){
390: fname[pos-1]=0;
391: dir=true;
392: } else if(dir) break;
393: }
394: return pos;
395: }
396:
397: static bool entry_readable(char* fname, bool need_dir) {
398: if(need_dir){
399: size_t size=strlen(fname);
400: while(size) {
401: char c=fname[size-1];
402: if(c=='/' || c=='\\')
403: fname[--size]=0;
404: else
405: break;
406: }
407: }
408:
409: struct stat finfo;
410: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
411: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
412: return is_dir==need_dir;
413: }
414: return false;
415: }
416:
417: static bool entry_readable(const String& file_spec, bool need_dir) {
418: return entry_readable(file_spec.taint_cstrm(String::L_FILE_SPEC), need_dir);
419: }
420:
1.63 parser 421: // throws nothing! [this is required in file_move & file_delete]
1.277 moko 422: static void rmdir(const String& file_spec, size_t pos_after) {
1.261 misha 423: char* dir_spec=file_spec.taint_cstrm(String::L_FILE_SPEC);
424: size_t length=strlen(dir_spec);
425: while( (length=get_dir(dir_spec, length)) && (length > pos_after) ){
1.274 moko 426: #ifdef _MSC_VER
1.261 misha 427: if(!entry_readable(dir_spec, true))
428: break;
429: DWORD attrs=GetFileAttributes(dir_spec);
430: if(
431: (attrs==INVALID_FILE_ATTRIBUTES)
432: || !(attrs & FILE_ATTRIBUTE_DIRECTORY)
433: || (attrs & FILE_ATTRIBUTE_REPARSE_POINT)
434: )
435: break;
436: #endif
437: if( rmdir(dir_spec) )
438: break;
439: };
1.50 paf 440: }
1.239 misha 441:
1.269 misha 442: bool file_delete(const String& file_spec, bool fail_on_problem, bool keep_empty_dirs) {
1.247 misha 443: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.282 moko 444: if(unlink(fname)!=0) {
1.164 paf 445: if(fail_on_problem)
1.289 moko 446: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
447: &file_spec, "unlink failed: %s (%d), actual filename '%s'", strerror(errno), errno, fname);
1.93 paf 448: else
449: return false;
1.282 moko 450: }
1.50 paf 451:
1.269 misha 452: if(!keep_empty_dirs)
453: rmdir(file_spec, 1);
454:
1.93 paf 455: return true;
1.60 parser 456: }
1.239 misha 457:
1.269 misha 458: void file_move(const String& old_spec, const String& new_spec, bool keep_empty_dirs) {
1.247 misha 459: const char* old_spec_cstr=old_spec.taint_cstr(String::L_FILE_SPEC);
460: const char* new_spec_cstr=new_spec.taint_cstr(String::L_FILE_SPEC);
1.63 parser 461:
1.126 paf 462: create_dir_for_file(new_spec);
1.63 parser 463:
1.60 parser 464: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.289 moko 465: throw Exception(errno==EACCES ? "file.access" : errno==ENOENT ? "file.missing" : 0,
466: &old_spec, "rename failed: %s (%d), actual filename '%s' to '%s'", strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 467:
1.269 misha 468: if(!keep_empty_dirs)
469: rmdir(old_spec, 1);
1.31 paf 470: }
471:
1.51 paf 472:
1.126 paf 473: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 474: struct stat lfinfo;
1.298 moko 475: bool result=pa_stat(fname, &lfinfo)==0;
1.118 paf 476: if(afinfo)
477: *afinfo=lfinfo;
478: return result;
1.119 paf 479: }
480:
481: bool entry_exists(const String& file_spec) {
1.247 misha 482: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.126 paf 483: return entry_exists(fname, 0);
1.118 paf 484: }
485:
1.215 paf 486: bool file_exist(const String& file_spec) {
1.126 paf 487: return entry_readable(file_spec, false);
1.51 paf 488: }
1.239 misha 489:
1.215 paf 490: bool dir_exists(const String& file_spec) {
1.126 paf 491: return entry_readable(file_spec, true);
1.65 parser 492: }
1.239 misha 493:
1.215 paf 494: const String* file_exist(const String& path, const String& name) {
1.154 paf 495: String& result=*new String(path);
1.270 moko 496: if(path.last_char() != '/')
497: result << "/";
1.154 paf 498: result << name;
1.215 paf 499: return file_exist(result)?&result:0;
1.43 paf 500: }
1.239 misha 501:
1.43 paf 502: bool file_executable(const String& file_spec) {
1.247 misha 503: return access(file_spec.taint_cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 504: }
505:
1.296 moko 506: 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 507: const char* fname=file_spec.taint_cstr(String::L_FILE_SPEC);
1.154 paf 508: struct stat finfo;
1.298 moko 509: if(pa_stat(fname, &finfo)!=0) {
1.64 parser 510: if(fail_on_read_problem)
1.289 moko 511: throw Exception("file.missing", &file_spec, "getting file size failed: %s (%d), real filename '%s'", strerror(errno), errno, fname);
1.64 parser 512: else
513: return false;
1.282 moko 514: }
1.58 parser 515: rsize=finfo.st_size;
1.299 moko 516: ratime=(time_t)finfo.st_atime;
517: rmtime=(time_t)finfo.st_mtime;
518: rctime=(time_t)finfo.st_ctime;
1.64 parser 519: return true;
1.18 paf 520: }
521:
1.300 moko 522: size_t check_file_size(uint64_t size, const String& file_spec){
523: if(size > pa_file_size_limit)
524: throw Exception(PARSER_RUNTIME, &file_spec, "content size of %.15g bytes exceeds the limit (%.15g bytes)", (double)size, (double)pa_file_size_limit);
525: return (size_t)size;
526: }
527:
1.278 moko 528: /**
529: String related functions
530: */
531:
532: bool capitalized(const char* s){
533: bool upper=true;
534: for(const char* c=s; *c; c++){
535: if(*c != (upper ? toupper((unsigned char)*c) : tolower((unsigned char)*c)))
536: return false;
537: upper=strchr("-_ ", *c) != 0;
538: }
539: return true;
540: }
541:
542: const char* capitalize(const char* s){
543: if(!s || capitalized(s))
544: return s;
545:
546: char* result=pa_strdup(s);
547: if(result){
548: bool upper=true;
549: for(char* c=result; *c; c++){
550: *c=upper ? (char)toupper((unsigned char)*c) : (char)tolower((unsigned char)*c);
551: upper=strchr("-_ ", *c) != 0;
552: }
553: }
554: return (const char*)result;
555: }
556:
1.290 moko 557: char *str_lower(const char *s, size_t helper_length){
558: char *result=pa_strdup(s, helper_length);
559: for(char* c=result; *c; c++)
560: *c=(char)tolower((unsigned char)*c);
561: return result;
562: }
563:
564: char *str_upper(const char *s, size_t helper_length){
565: char *result=pa_strdup(s, helper_length);
566: for(char* c=result; *c; c++)
567: *c=(char)toupper((unsigned char)*c);
568: return result;
569: }
570:
1.278 moko 571: void fix_line_breaks(char *str, size_t& length) {
572: //_asm int 3;
573: const char* const eob=str+length;
574: char* dest=str;
575: // fix DOS: \r\n -> \n
576: // fix Macintosh: \r -> \n
577: char* bol=str;
578: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
579: size_t len=eol-bol;
580: if(dest!=bol)
581: memmove(dest, bol, len);
582: dest+=len;
583: *dest++='\n';
584:
585: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
586: bol=eol+2;
587: length--;
588: } else // \r, not \n = Macintosh
589: bol=eol+1;
590: }
591: // last piece without \r
592: if(dest!=bol)
593: memmove(dest, bol, eob-bol);
594: str[length]=0; // terminating
595: }
596:
597: /**
598: scans for @a delim[default \n] in @a *row_ref,
599: @return piece of line before it or end of string, if no @a delim found
600: assigns @a *row_ref to point right after delimiter if there were one
601: or to zero if no @a delim were found.
602: */
603:
1.126 paf 604: char* getrow(char* *row_ref, char delim) {
1.229 misha 605: char* result=*row_ref;
606: if(result) {
1.126 paf 607: *row_ref=strchr(result, delim);
1.8 paf 608: if(*row_ref)
609: *((*row_ref)++)=0;
610: else if(!*result)
611: return 0;
1.229 misha 612: }
613: return result;
1.8 paf 614: }
615:
1.126 paf 616: char* lsplit(char* string, char delim) {
1.229 misha 617: if(string) {
1.126 paf 618: char* v=strchr(string, delim);
1.8 paf 619: if(v) {
620: *v=0;
621: return v+1;
622: }
1.229 misha 623: }
624: return 0;
1.8 paf 625: }
626:
1.126 paf 627: char* lsplit(char* *string_ref, char delim) {
1.229 misha 628: char* result=*string_ref;
1.126 paf 629: char* next=lsplit(*string_ref, delim);
1.229 misha 630: *string_ref=next;
631: return result;
1.9 paf 632: }
633:
1.126 paf 634: char* rsplit(char* string, char delim) {
1.229 misha 635: if(string) {
1.126 paf 636: char* v=strrchr(string, delim);
1.18 paf 637: if(v) {
1.9 paf 638: *v=0;
639: return v+1;
640: }
1.229 misha 641: }
642: return NULL;
1.10 paf 643: }
644:
1.229 misha 645:
646: // format: %[flags][width][.precision]type http://msdn.microsoft.com/ru-ru/library/56e442dc(en-us,VS.80).aspx
647: // flags: '-', '+', ' ', '#', '0' http://msdn.microsoft.com/ru-ru/library/8aky45ct(en-us,VS.80).aspx
648: // width, precision: non negative decimal number
649: enum FormatType {
650: FormatInvalid,
651: FormatInt,
652: FormatUInt,
653: FormatDouble
654: };
1.272 moko 655: FormatType format_type(const char* fmt){
1.229 misha 656: enum FormatState {
657: Percent,
658: Flags,
659: Width,
660: Precision,
661: Done
662: } state=Percent;
663:
664: FormatType result=FormatInvalid;
665:
1.272 moko 666: const char* pos=fmt;
1.229 misha 667: while(char c=*(pos++)){
668: switch(state){
669: case Percent:
670: if(c=='%'){
671: state=Flags;
672: } else {
673: return FormatInvalid; // 1st char must be '%' only
674: }
675: break;
676: case Flags:
677: if(strchr("-+ #0", c)!=0){
678: break;
679: }
680: // go to the next step
681: case Width:
682: if(c=='.'){
683: state=Precision;
684: break;
685: }
686: // go to the next step
687: case Precision:
688: if(c>='0' && c<='9'){
689: if(state == Flags) state=Width; // no more flags
690: break;
691: } else if(c=='d' || c=='i'){
692: result=FormatInt;
693: } else if(strchr("feEgG", c)!=0){
694: result=FormatDouble;
695: } else if(strchr("uoxX", c)!=0){
696: result=FormatUInt;
697: } else {
698: return FormatInvalid; // invalid char
699: }
700: state=Done;
701: break;
702: case Done:
703: return FormatInvalid; // no chars allowed after 'type'
704: }
705: }
706: return result;
707: }
708:
709:
1.272 moko 710: const char* format(double value, const char* fmt) {
1.229 misha 711: char local_buf[MAX_NUMBER];
1.235 misha 712: int size=-1;
1.229 misha 713:
714: if(fmt && strlen(fmt)){
715: switch(format_type(fmt)){
716: case FormatDouble:
717: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
718: break;
719: case FormatInt:
720: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
721: break;
722: case FormatUInt:
1.126 paf 723: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.229 misha 724: break;
725: case FormatInvalid:
1.289 moko 726: throw Exception(PARSER_RUNTIME, 0, "Incorrect format string '%s' was specified.", fmt);
1.229 misha 727: }
728: } else
729: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
730:
731: if(size < 0 || size >= MAX_NUMBER-1){ // on win32 we manually reduce max size while printing
1.304 moko 732: throw Exception(PARSER_RUNTIME, 0, "Error occurred white executing snprintf with format string '%s'.", fmt);
1.229 misha 733: }
734:
1.235 misha 735: return pa_strdup(local_buf, (size_t)size);
1.12 paf 736: }
737:
1.36 paf 738: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 739: #ifdef WIN32
1.187 paf 740: size_t to_write = size;
1.12 paf 741: do{
1.154 paf 742: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 743: if(chunk_written<=0)
744: break;
745: size-=chunk_written;
1.36 paf 746: buf=((const char*)buf)+chunk_written;
1.126 paf 747: } while(size>0);
1.12 paf 748:
1.187 paf 749: return to_write-size;
1.12 paf 750: #else
1.126 paf 751: return fwrite(buf, 1, size, stdout);
1.12 paf 752: #endif
1.2 paf 753: }
1.14 paf 754:
1.229 misha 755: enum EscapeState {
756: EscapeRest,
757: EscapeFirst,
758: EscapeSecond,
759: EscapeUnicode
760: };
761:
1.236 misha 762: // @todo prescan for reduce required size (unescaped sting in 1 byte charset requires less memory usually)
1.258 misha 763: char* unescape_chars(const char* cp, int len, Charset* charset, bool js){
1.236 misha 764: char* s=new(PointerFreeGC) char[len+1]; // must be enough (%uXXXX==6 bytes, max utf-8 char length==6 bytes)
1.230 misha 765: char* dst=s;
1.229 misha 766: EscapeState escapeState=EscapeRest;
767: uint escapedValue=0;
768: int srcPos=0;
1.230 misha 769: short int jsCnt=0;
1.236 misha 770: while(srcPos<len){
1.229 misha 771: uchar c=(uchar)cp[srcPos];
1.258 misha 772: if(c=='%' || (c=='\\' && js)){
1.229 misha 773: escapeState=EscapeFirst;
774: } else {
775: switch(escapeState) {
776: case EscapeRest:
1.286 moko 777: if(c=='+' && !js){
1.230 misha 778: *dst++=' ';
1.229 misha 779: } else {
1.230 misha 780: *dst++=c;
1.229 misha 781: }
782: break;
783: case EscapeFirst:
1.232 misha 784: if(charset && c=='u'){
1.229 misha 785: // escaped unicode value: %u0430
786: jsCnt=0;
787: escapedValue=0;
788: escapeState=EscapeUnicode;
789: } else {
1.231 misha 790: if(isxdigit(c)){
1.229 misha 791: escapedValue=hex_value[c] << 4;
792: escapeState=EscapeSecond;
793: } else {
1.230 misha 794: *dst++=c;
1.229 misha 795: escapeState=EscapeRest;
796: }
797: }
798: break;
799: case EscapeSecond:
1.231 misha 800: if(isxdigit(c)){
1.229 misha 801: escapedValue+=hex_value[c];
1.230 misha 802: *dst++=(char)escapedValue;
1.229 misha 803: }
804: escapeState=EscapeRest;
805: break;
806: case EscapeUnicode:
1.231 misha 807: if(isxdigit(c)){
1.229 misha 808: escapedValue=(escapedValue << 4) + hex_value[c];
809: if(++jsCnt==4){
1.230 misha 810: // transcode utf8 char to client charset (we can lost some chars here)
1.232 misha 811: charset->store_Char((XMLByte*&)dst, (XMLCh)escapedValue, '?');
1.229 misha 812: escapeState=EscapeRest;
813: }
814: } else {
815: // not full unicode value
816: escapeState=EscapeRest;
817: }
818: break;
819: }
820: }
821:
822: srcPos++;
823: }
824:
1.230 misha 825: *dst=0; // zero-termination
1.229 misha 826: return s;
827: }
1.24 paf 828:
1.268 misha 829: char *search_stop(char*& current, char cstop_at) {
830: // sanity check
831: if(!current)
832: return 0;
833:
834: // skip leading WS
835: while(*current==' ' || *current=='\t')
836: current++;
837: if(!*current)
838: return current=0;
839:
840: char *result=current;
841: if(char *pstop_at=strchr(current, cstop_at)) {
842: *pstop_at=0;
843: current=pstop_at+1;
844: } else
845: current=0;
846: return result;
847: }
848:
1.24 paf 849: #ifdef WIN32
1.126 paf 850: void back_slashes_to_slashes(char* s) {
1.24 paf 851: if(s)
852: for(; *s; s++)
853: if(*s=='\\')
1.126 paf 854: *s='/';
1.24 paf 855: }
856: #endif
1.41 paf 857:
1.232 misha 858: size_t strpos(const char *str, const char *substr) {
859: const char *p = strstr(str, substr);
860: return (p==0)?STRING_NOT_FOUND:p-str;
861: }
862:
1.226 misha 863: int remove_crlf(char* start, char* end) {
864: char* from=start;
865: char* to=start;
866: bool skip=false;
867: while(from < end){
868: switch(*from){
869: case '\n':
870: case '\r':
871: case '\t':
872: case ' ':
873: if(!skip){
874: *to=' ';
875: to++;
876: skip=true;
877: }
878: break;
879: default:
880: if(from != to)
881: *to=*from;
882: to++;
883: skip=false;
1.69 parser 884: }
1.226 misha 885: from++;
886: }
887: return to-start;
1.91 paf 888: }
889:
1.279 moko 890: const char* hex_digits="0123456789ABCDEF";
891:
1.278 moko 892: const char* hex_string(unsigned char* bytes, size_t size, bool upcase) {
893: char *bytes_hex=new(PointerFreeGC) char [size*2/*byte->hh*/+1/*for zero-teminator*/];
894: unsigned char *src=bytes;
895: unsigned char *end=bytes+size;
896: char *dest=bytes_hex;
897:
1.279 moko 898: const char *hex=upcase? hex_digits : "0123456789abcdef";
1.278 moko 899:
900: for(; src<end; src++) {
901: *dest++=hex[*src/0x10];
902: *dest++=hex[*src%0x10];
903: }
904: *dest=0;
905:
906: return bytes_hex;
907: }
1.91 paf 908:
1.221 misha 909: int file_block_read(const int f, unsigned char* buffer, const size_t size){
910: int nCount = read(f, buffer, size);
911: if (nCount < 0)
1.289 moko 912: throw Exception("file.read", 0, "read failed: %s (%d)", strerror(errno), errno);
1.221 misha 913: return nCount;
914: }
915:
1.278 moko 916: static unsigned long crc32Table[256];
917: static void InitCrc32Table()
918: {
919: if(crc32Table[1] == 0){
920: // This is the official polynomial used by CRC32 in PKZip.
921: // Often times the polynomial shown reversed as 0x04C11DB7.
922: static const unsigned long dwPolynomial = 0xEDB88320;
923:
924: for(int i = 0; i < 256; i++)
925: {
926: unsigned long dwCrc = i;
927: for(int j = 8; j > 0; j--)
928: {
929: if(dwCrc & 1)
930: dwCrc = (dwCrc >> 1) ^ dwPolynomial;
931: else
932: dwCrc >>= 1;
933: }
934: crc32Table[i] = dwCrc;
935: }
936: }
937: }
938:
939: inline void CalcCrc32(const unsigned char byte, unsigned long &crc32)
940: {
941: crc32 = ((crc32) >> 8) ^ crc32Table[(byte) ^ ((crc32) & 0x000000FF)];
942: }
943:
944:
1.297 moko 945: unsigned long pa_crc32(const char *in, size_t in_size){
1.218 misha 946: unsigned long crc32=0xFFFFFFFF;
1.220 misha 947:
1.240 misha 948: InitCrc32Table();
1.239 misha 949: for(size_t i = 0; i<in_size; i++)
950: CalcCrc32(in[i], crc32);
1.220 misha 951:
1.218 misha 952: return ~crc32;
953: }
954:
1.289 moko 955: static void file_crc32_file_action(struct stat& finfo, int f, const String&, void *context) {
1.218 misha 956: unsigned long& crc32=*static_cast<unsigned long *>(context);
957: if(finfo.st_size) {
958: InitCrc32Table();
1.220 misha 959: int nCount=0;
1.218 misha 960: do {
1.221 misha 961: unsigned char buffer[FILE_BUFFER_SIZE];
962: nCount = file_block_read(f, buffer, sizeof(buffer));
1.220 misha 963: for(int i = 0; i < nCount; i++) CalcCrc32(buffer[i], crc32);
964: } while(nCount > 0);
1.218 misha 965: }
966: }
967:
1.297 moko 968: unsigned long pa_crc32(const String& file_spec){
1.278 moko 969: unsigned long crc32=0xFFFFFFFF;
970: file_read_action_under_lock(file_spec, "crc32", file_crc32_file_action, &crc32);
971: return ~crc32;
972: }
973:
974: // content-type: xxx; charset=WE-NEED-THIS
975: // content-type: xxx; charset="WE-NEED-THIS"
976: // content-type: xxx; charset="WE-NEED-THIS";
977: Charset* detect_charset(const char* content_type){
978: if(content_type){
1.291 moko 979: char* CONTENT_TYPE=str_upper(content_type);
1.278 moko 980:
981: if(const char* begin=strstr(CONTENT_TYPE, "CHARSET=")){
982: begin+=8; // skip "CHARSET="
983: char* end=0;
984: if(*begin && (*begin=='"' || *begin =='\'')){
985: char quote=*begin;
986: begin++;
987: end=(char*)strchr(begin, quote);
988: }
989: if(!end)
990: end=(char*)strchr(begin, ';');
991:
992: if(end)
993: *end=0; // terminator
994:
1.295 moko 995: return *begin ? &pa_charsets.get_direct(begin) : 0;
1.278 moko 996: }
997: }
998: return 0;
999: }
1000:
1.301 moko 1001: const UTF16* pa_utf16_encode(const char* in, Charset& source_charset){
1.302 moko 1002: if(!in)
1003: return 0;
1.301 moko 1004:
1.302 moko 1005: String::C sIn(in,strlen(in));
1.301 moko 1006:
1.302 moko 1007: UTF16* utf16=(UTF16*)pa_malloc_atomic(sIn.length*2+2);
1008: UTF16* utf16_end=utf16;
1.301 moko 1009:
1.302 moko 1010: if(!source_charset.isUTF8())
1011: sIn=Charset::transcode(sIn, source_charset, pa_UTF8_charset);
1.301 moko 1012:
1.302 moko 1013: int status=pa_convertUTF8toUTF16((const UTF8**)&sIn.str, (const UTF8*)(sIn.str+sIn.length), &utf16_end, utf16+sIn.length, strictConversion);
1014: if(status != conversionOK)
1015: throw Exception("utf-16 encode", new String(in), "utf-16 conversion failed (%d)", status);
1.301 moko 1016:
1.302 moko 1017: *utf16_end=0;
1.301 moko 1018:
1.302 moko 1019: return utf16;
1.301 moko 1020: }
1021:
1022: const char* pa_utf16_decode(const UTF16* in, Charset& asked_charset){
1023: if(!in)
1024: return 0;
1025:
1026: const UTF16* utf16_start=in;
1027: const UTF16* utf16_end;
1028:
1029: for(utf16_end=in; *utf16_end; utf16_end++);
1030:
1031: char *result = (char *)pa_malloc_atomic((utf16_end-in)*6+1);
1032: char *result_end = result;
1033:
1034: int status=pa_convertUTF16toUTF8(&utf16_start, utf16_end, (UTF8**)&result_end, (UTF8*)(result+(utf16_end-in)*6), strictConversion);
1035:
1036: if(status != conversionOK)
1037: throw Exception("utf-16 decode", 0, "utf conversion failed (%d)", status);
1038:
1039: *result_end='\0';
1040:
1.302 moko 1041: if(asked_charset.isUTF8())
1042: return result;
1.301 moko 1043:
1.302 moko 1044: return Charset::transcode(result, pa_UTF8_charset, asked_charset).cstr();
1.301 moko 1045: }
1.278 moko 1046:
1.283 moko 1047: static bool is_latin(const char *in){
1048: for(; *in; in++){
1049: if ((unsigned char)(*in) > 0x7F)
1050: return false;
1051: }
1052: return true;
1053: }
1054:
1055: #define MAX_IDNA_LENGTH 256
1056:
1.301 moko 1057: const char *pa_idna_encode(const char *in, Charset& source_charset){
1.283 moko 1058: if(!in || is_latin(in))
1059: return in;
1060:
1061: uint32_t utf32[MAX_IDNA_LENGTH];
1062: uint32_t *utf32_end=utf32;
1063:
1064: String::C sIn(in,strlen(in));
1065:
1066: if(!source_charset.isUTF8())
1.295 moko 1067: sIn=Charset::transcode(sIn, source_charset, pa_UTF8_charset);
1.283 moko 1068:
1069: int status=pa_convertUTF8toUTF32((const UTF8**)&sIn.str, (const UTF8*)(sIn.str+sIn.length), &utf32_end, utf32+MAX_IDNA_LENGTH-1, strictConversion);
1070: if(status != conversionOK)
1071: throw Exception("idna encode", new String(in), "utf conversion failed (%d)", status);
1072:
1073: *utf32_end=0;
1074:
1075: char *result = (char *)pa_malloc(MAX_IDNA_LENGTH);
1076: status=pa_idna_to_ascii_4z(utf32, result, MAX_IDNA_LENGTH, 0);
1077: if(status != IDNA_SUCCESS)
1078: throw Exception("idna encode", new String(in), "encode failed: %s", pa_idna_strerror(status));
1079:
1080: return result;
1081: }
1082:
1083: const char *pa_idna_decode(const char *in, Charset &asked_charset){
1084: if(!in || !(*in))
1085: return in;
1086:
1087: uint32_t utf32[MAX_IDNA_LENGTH];
1088: const uint32_t *utf32_start=utf32;
1089: uint32_t *utf32_end;
1090:
1091: int status=pa_idna_to_unicode_4z(in, utf32, MAX_IDNA_LENGTH, 0);
1092: if(status != IDNA_SUCCESS)
1093: throw Exception("idna decode", new String(in), "decode failed: %s", pa_idna_strerror(status));
1094:
1095: for(utf32_end=utf32; *utf32_end; utf32_end++);
1096:
1097: char *result = (char *)pa_malloc(MAX_IDNA_LENGTH);
1098: char *result_end = result;
1099:
1100: status=pa_convertUTF32toUTF8(&utf32_start, utf32_end, (UTF8**)&result_end, (UTF8*)(result+MAX_IDNA_LENGTH-1), strictConversion);
1101:
1102: if(status != conversionOK)
1103: throw Exception("idna decode", new String(in), "utf conversion failed (%d)", status);
1104:
1105: *result_end='\0';
1106:
1107: if(!asked_charset.isUTF8())
1.295 moko 1108: result = (char *)Charset::transcode(result, pa_UTF8_charset, asked_charset).cstr();
1.283 moko 1109:
1110: return result;
1111: }
1.292 moko 1112: /// must be last in this file
1113: #undef vsnprintf
1114: int pa_vsnprintf(char* b, size_t s, const char* f, va_list l) {
1115: if(!s)
1116: return 0;
1117:
1118: int r;
1119: // note: on win32 & maybe somewhere else
1120: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1121: // http://stackoverflow.com/questions/2915672/snprintf-and-visual-studio-2010
1122: --s;
1123:
1124: // clients do not check for negative 's', feature: ignore such prints
1125: if((ssize_t)s<0)
1126: return 0;
1127:
1128: #ifdef _MSC_VER
1129: // win32: if the number of bytes to write exceeds buffer, then count bytes are written and -1 is returned
1130: r=_vsnprintf(b, s, f, l);
1131: if(r<0)
1132: r=s;
1133: #else
1134: r=vsnprintf(b, s, f, l);
1135: /*
1136: solaris: man vsnprintf
1137:
1138: The snprintf() function returns the number of characters
1139: formatted, that is, the number of characters that would have
1140: been written to the buffer if it were large enough. If the
1141: value of n is 0 on a call to snprintf(), an unspecified
1142: value less than 1 is returned.
1143: */
1144:
1145: if(r<0)
1146: r=0;
1147: else if((size_t)r>s)
1148: r=s;
1149: #endif
1150: b[r]=0;
1151: return r;
1152: }
1153:
1154: int pa_snprintf(char* b, size_t s, const char* f, ...) {
1155: va_list l;
1156: va_start(l, f);
1157: int r=pa_vsnprintf(b, s, f, l);
1158: va_end(l);
1159: return r;
1160: }
1161:
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