Annotation of parser3/src/main/pa_common.C, revision 1.103
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.102 paf 4: Copyright(c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com)
1.101 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.16 paf 6:
1.103 ! paf 7: $Id: pa_common.C,v 1.102 2002/02/08 08:32:34 paf Exp $
1.1 paf 8: */
9:
10: #include "pa_common.h"
1.2 paf 11: #include "pa_types.h"
1.4 paf 12: #include "pa_exception.h"
1.14 paf 13: #include "pa_pool.h"
14: #include "pa_globals.h"
15: #include "pa_value.h"
16: #include "pa_hash.h"
17: #include "pa_string.h"
1.1 paf 18:
1.98 paf 19: #ifdef WIN32
20: # include <windows.h>
21: #endif
22:
1.93 paf 23: // some maybe-undefined constants
24:
1.82 paf 25: #ifndef _O_TEXT
26: # define _O_TEXT 0
27: #endif
28: #ifndef _O_BINARY
29: # define _O_BINARY 0
1.47 paf 30: #endif
1.80 paf 31: #ifndef O_TRUNC
32: # define O_TRUNC 0
33: #endif
34:
1.93 paf 35: // locking constants
36:
1.99 paf 37: #ifdef HAVE_FLOCK
38:
39: static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); }
40: static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); }
41: static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); }
42: static int unlock(int fd) { return flock(fd, LOCK_UN); }
43:
1.98 paf 44: #else
1.99 paf 45: #ifdef HAVE__LOCKING
1.98 paf 46:
1.99 paf 47: #define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1)
48: static int lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); }
49: static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); }
50: static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); }
51: static int unlock(int fd) { FLOCK(_LK_UNLCK); }
1.93 paf 52:
1.99 paf 53: #else
54: #ifdef HAVE_FCNTL
1.93 paf 55:
1.99 paf 56: #define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls)
57: static int lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); }
58: static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); }
59: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); }
60: static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); }
1.93 paf 61:
62: #else
63: #ifdef HAVE_LOCKF
1.99 paf 64:
65: #define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1)
66: static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking
67: static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); }
68: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); }
69: static int unlock(int fd) { FLOCK(F_TLOCK); }
70:
1.93 paf 71: #else
1.99 paf 72:
73: #error unable to find file locking func
74:
75: #endif
1.93 paf 76: #endif
77: #endif
78: #endif
79:
1.86 paf 80: static char *strnchr(char *buf, size_t size, char c) {
1.103 ! paf 81: // sanity check
! 82: if(!buf)
! 83: return 0;
! 84:
1.86 paf 85: for(; size-->0; buf++) {
86: if(*buf==c)
87: return buf;
88: }
89:
90: return 0;
91: }
92:
93: void fix_line_breaks(char *buf, size_t& size) {
1.87 paf 94: //_asm int 3;
1.86 paf 95: const char * const eob=buf+size;
1.87 paf 96: char *dest=buf;
1.72 parser 97: // fix DOS: \r\n -> \n
98: // fix Macintosh: \r -> \n
1.87 paf 99: char *bol=buf;
1.86 paf 100: while(char *eol=strnchr(bol, eob -bol, '\r')) {
1.72 parser 101: size_t len=eol-bol;
102: if(dest!=bol)
103: memcpy(dest, bol, len);
104: dest+=len;
105: *dest++='\n';
106:
1.87 paf 107: if(&eol[1]<eob && eol[1]=='\n') { // \r,\n = DOS
1.72 parser 108: bol=eol+2;
109: size--;
110: } else // \r,not \n = Macintosh
111: bol=eol+1;
112: }
113: // last piece without \r, including terminating 0
114: if(dest!=bol)
1.87 paf 115: memcpy(dest, bol, eob-bol);
1.72 parser 116: }
1.18 paf 117:
1.28 paf 118: char *file_read_text(Pool& pool, const String& file_spec, bool fail_on_read_problem) {
1.72 parser 119: void *result; size_t size;
120: return file_read(pool, file_spec, result, size, true, fail_on_read_problem)?(char *)result:0;
1.34 paf 121: }
122: bool file_read(Pool& pool, const String& file_spec,
1.87 paf 123: void*& data, size_t& data_size, bool as_text,
1.44 paf 124: bool fail_on_read_problem,
125: size_t offset, size_t limit) {
1.64 parser 126: const char *fname=file_spec.cstr(String::UL_FILE_SPEC);
1.62 parser 127: //printf("file_read(%s)\n", fname);
1.33 paf 128: int f;
1.2 paf 129: struct stat finfo;
1.33 paf 130:
131: // first open, next stat:
1.45 paf 132: // directory update of NTFS hard links performed on open.
1.33 paf 133: // ex:
134: // a.html:^test[] and b.html hardlink to a.html
135: // user inserts ! before ^test in a.html
136: // directory entry of b.html in NTFS not updated at once,
1.35 paf 137: // they delay update till open, so we would receive "!^test[" string
138: // if would do stat, next open.
1.98 paf 139: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.99 paf 140: lock_shared_blocking(f);
1.98 paf 141: if(stat(fname, &finfo)!=0) {
142: Exception e(0, 0,
143: &file_spec,
144: "stat failed: %s (%d), actual filename '%s'",
145: strerror(errno), errno, fname);
1.99 paf 146: unlock(f);
1.98 paf 147: close(f);
148: if(fail_on_read_problem)
149: throw e;
150: return false;
151: }
152: size_t max_size=limit?min(offset+limit, (size_t)finfo.st_size)-offset:finfo.st_size;
1.47 paf 153: if(!max_size) { // eof
1.83 paf 154: if(as_text) {
155: data=pool.malloc(1);
156: *(char*)data=0;
157: } else
158: data=0;
1.98 paf 159: data_size=0;
1.59 parser 160: } else {
1.78 paf 161: data=pool.malloc(max_size+(as_text?1:0), 3);
1.59 parser 162: if(offset)
163: lseek(f, offset, SEEK_SET);
1.98 paf 164: data_size=read(f, data, max_size);
1.47 paf 165: }
1.99 paf 166: unlock(f);
1.2 paf 167: close(f);
1.59 parser 168: if(!max_size) // eof
169: return true;
1.32 paf 170:
1.98 paf 171: if(int(data_size)<0 || data_size>max_size)
1.73 parser 172: throw Exception(0, 0,
1.33 paf 173: &file_spec,
1.44 paf 174: "read failed: actually read %d bytes count not in [0..%lu] valid range",
1.98 paf 175: data_size, (unsigned long)max_size); //never
1.87 paf 176:
177: if(as_text) {
178: fix_line_breaks((char *)data, data_size);
179: // note: after fixing
180: ((char*&)data)[data_size]=0;
181: }
1.72 parser 182: return true;
1.2 paf 183: }
1.4 paf 184: if(fail_on_read_problem)
1.73 parser 185: throw Exception(0, 0,
1.33 paf 186: &file_spec,
1.54 parser 187: "read failed: %s (%d), actual filename '%s'",
188: strerror(errno), errno, fname);
1.34 paf 189: return false;
1.8 paf 190: }
191:
1.63 parser 192: static void create_dir_for_file(const String& file_spec) {
193: size_t pos_after=1;
194: int pos_before;
195: while((pos_before=file_spec.pos("/", 1, pos_after))>=0) {
1.64 parser 196: mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775);
1.63 parser 197: pos_after=pos_before+1;
198: }
199: }
200:
1.98 paf 201: bool file_write_action_under_lock(
1.28 paf 202: const String& file_spec,
1.96 paf 203: const char *action_name, void (*action)(int, void *), void *context,
1.80 paf 204: bool as_text,
1.98 paf 205: bool do_append,
206: bool do_block) {
1.64 parser 207: const char *fname=file_spec.cstr(String::UL_FILE_SPEC);
1.28 paf 208: int f;
1.80 paf 209: if(access(fname, W_OK)!=0) // no
1.63 parser 210: create_dir_for_file(file_spec);
1.50 paf 211:
1.80 paf 212: if((f=open(fname,
213: O_CREAT|O_RDWR
214: |(as_text?_O_TEXT:_O_BINARY)
1.98 paf 215: |(do_append?O_APPEND:0), 0664))>=0) {
1.99 paf 216: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.98 paf 217: close(f);
218: return false;
219: }
1.96 paf 220:
221: try {
222: action(f, context);
223: } catch(...) {
1.98 paf 224: #if O_TRUNC==0
225: ftruncate(f, tell(f));
226: #endif
1.99 paf 227: unlock(f);
1.96 paf 228: close(f);
229: /*re*/throw;
230: }
1.80 paf 231:
1.98 paf 232: #if O_TRUNC==0
233: ftruncate(f, tell(f));
234: #endif
1.99 paf 235: unlock(f);
1.80 paf 236: close(f);
1.98 paf 237: return true;
1.80 paf 238: } else
239: throw Exception(0, 0,
240: &file_spec,
1.96 paf 241: "%s failed: %s (%d), actual filename '%s'",
242: action_name, strerror(errno), errno, fname);
243: // here should be nothing, see rethrow above
244: }
245:
246: #ifndef DOXYGEN
247: struct File_write_action_info {
248: const void *data; size_t size;
249: };
250: #endif
251: static void file_write_action(int f, void *context) {
252: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
253: if(info.size)
254: write(f, info.data, info.size);
255: }
256: void file_write(
257: const String& file_spec,
258: const void *data, size_t size,
259: bool as_text,
260: bool do_append) {
261: File_write_action_info info={data, size};
1.98 paf 262: file_write_action_under_lock(
1.96 paf 263: file_spec,
264: "write", file_write_action, &info,
265: as_text,
266: do_append);
1.30 paf 267: }
268:
1.63 parser 269: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 270: static void rmdir(const String& file_spec, size_t pos_after) {
271: int pos_before;
272: if((pos_before=file_spec.pos("/", 1, pos_after))>=0)
273: rmdir(file_spec, pos_before+1);
274:
1.64 parser 275: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_FILE_SPEC));
1.50 paf 276: }
1.95 paf 277: bool file_delete(const String& file_spec, bool fail_on_read_problem) {
1.64 parser 278: const char *fname=file_spec.cstr(String::UL_FILE_SPEC);
1.54 parser 279: if(unlink(fname)!=0)
1.93 paf 280: if(fail_on_read_problem)
281: throw Exception(0, 0,
282: &file_spec,
283: "unlink failed: %s (%d), actual filename '%s'",
284: strerror(errno), errno, fname);
285: else
286: return false;
1.50 paf 287:
288: rmdir(file_spec, 1);
1.93 paf 289: return true;
1.60 parser 290: }
1.95 paf 291: void file_move(const String& old_spec, const String& new_spec) {
1.64 parser 292: const char *old_spec_cstr=old_spec.cstr(String::UL_FILE_SPEC);
293: const char *new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC);
1.63 parser 294:
295: create_dir_for_file(new_spec);
296:
1.60 parser 297: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.73 parser 298: throw Exception(0, 0,
1.60 parser 299: &old_spec,
300: "rename failed: %s (%d), actual filename '%s' to '%s'",
301: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 302:
303: rmdir(old_spec, 1);
1.31 paf 304: }
305:
1.51 paf 306:
307: static bool entry_readable(const String& file_spec, bool need_dir) {
1.64 parser 308: const char *fname=file_spec.cstr(String::UL_FILE_SPEC);
1.51 paf 309: struct stat finfo;
310: if(access(fname, R_OK)==0 && stat(fname, &finfo)==0) {
1.73 parser 311: bool is_dir=finfo.st_mode&S_IFDIR != 0;
1.51 paf 312: return is_dir==need_dir;
313: }
314: return false;
315: }
1.31 paf 316: bool file_readable(const String& file_spec) {
1.51 paf 317: return entry_readable(file_spec, false);
318: }
319: bool dir_readable(const String& file_spec) {
320: return entry_readable(file_spec, true);
1.65 parser 321: }
322: String *file_readable(const String& path, const String& name) {
323: String *result=new(path.pool()) String(path);
324: *result << "/";
325: *result << name;
326: return file_readable(*result)?result:0;
1.43 paf 327: }
328: bool file_executable(const String& file_spec) {
1.64 parser 329: return access(file_spec.cstr(String::UL_FILE_SPEC), X_OK)==0;
1.44 paf 330: }
331:
1.64 parser 332: bool file_stat(const String& file_spec,
1.58 parser 333: size_t& rsize,
334: time_t& ratime,
335: time_t& rmtime,
1.64 parser 336: time_t& rctime,
337: bool fail_on_read_problem) {
1.44 paf 338: Pool& pool=file_spec.pool();
1.64 parser 339: const char *fname=file_spec.cstr(String::UL_FILE_SPEC);
1.44 paf 340: struct stat finfo;
341: if(stat(fname, &finfo)!=0)
1.64 parser 342: if(fail_on_read_problem)
1.73 parser 343: throw Exception(0, 0,
1.67 parser 344: &file_spec,
345: "getting file size failed: %s (%d), real filename '%s'",
346: strerror(errno), errno, fname);
1.64 parser 347: else
348: return false;
1.58 parser 349: rsize=finfo.st_size;
350: ratime=finfo.st_atime;
351: rmtime=finfo.st_mtime;
352: rctime=finfo.st_ctime;
1.64 parser 353: return true;
1.18 paf 354: }
355:
1.8 paf 356: char *getrow(char **row_ref, char delim) {
357: char *result=*row_ref;
358: if(result) {
359: *row_ref=strchr(result, delim);
360: if(*row_ref)
361: *((*row_ref)++)=0;
362: else if(!*result)
363: return 0;
364: }
365: return result;
366: }
367:
1.23 paf 368: char *lsplit(char *string, char delim) {
369: if(string) {
370: char *v=strchr(string, delim);
1.8 paf 371: if(v) {
372: *v=0;
373: return v+1;
374: }
375: }
376: return 0;
377: }
378:
379: char *lsplit(char **string_ref, char delim) {
380: char *result=*string_ref;
381: char *next=lsplit(*string_ref, delim);
382: *string_ref=next;
383: return result;
1.9 paf 384: }
385:
386: char *rsplit(char *string, char delim) {
1.18 paf 387: if(string) {
1.9 paf 388: char *v=strrchr(string, delim);
1.18 paf 389: if(v) {
1.9 paf 390: *v=0;
391: return v+1;
392: }
393: }
394: return NULL;
1.10 paf 395: }
396:
1.37 paf 397: /// @todo less stupid type detection
1.10 paf 398: char *format(Pool& pool, double value, char *fmt) {
1.78 paf 399: char *result=(char *)pool.malloc(MAX_NUMBER, 4);
1.10 paf 400: if(fmt)
401: if(strpbrk(fmt, "diouxX"))
402: if(strpbrk(fmt, "ouxX"))
1.33 paf 403: snprintf(result, MAX_NUMBER, fmt, (uint)value );
1.10 paf 404: else
1.33 paf 405: snprintf(result, MAX_NUMBER, fmt, (int)value );
1.10 paf 406: else
407: snprintf(result, MAX_NUMBER, fmt, value);
408: else
1.33 paf 409: snprintf(result, MAX_NUMBER, "%d", (int)value);
1.10 paf 410:
411: return result;
1.12 paf 412: }
413:
1.36 paf 414: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 415: #ifdef WIN32
416: do{
417: int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout);
418: if(chunk_written<=0)
419: break;
420: size-=chunk_written;
1.36 paf 421: buf=((const char*)buf)+chunk_written;
1.12 paf 422: } while(size>0);
423:
424: return size;
425: #else
1.13 paf 426: return fwrite(buf, 1, size, stdout);
1.12 paf 427: #endif
1.2 paf 428: }
1.14 paf 429:
1.77 parser 430: char *unescape_chars(Pool& pool, const char *cp, int len) {
1.78 paf 431: char *s=(char *)pool.malloc(len + 1, 5);
1.14 paf 432: enum EscapeState {
1.33 paf 433: EscapeRest,
434: EscapeFirst,
1.14 paf 435: EscapeSecond
436: } escapeState=EscapeRest;
437: int escapedValue=0;
438: int srcPos=0;
439: int dstPos=0;
440: while(srcPos < len) {
441: int ch=cp[srcPos];
442: switch(escapeState) {
443: case EscapeRest:
444: if(ch=='%') {
445: escapeState=EscapeFirst;
446: } else if(ch=='+') {
447: s[dstPos++]=' ';
448: } else {
449: s[dstPos++]=ch;
450: }
451: break;
452: case EscapeFirst:
453: escapedValue=hex_value[ch] << 4;
454: escapeState=EscapeSecond;
455: break;
456: case EscapeSecond:
457: escapedValue +=hex_value[ch];
458: s[dstPos++]=escapedValue;
459: escapeState=EscapeRest;
460: break;
461: }
462: srcPos++;
463: }
464: s[dstPos]=0;
465: return s;
466: }
467:
1.52 paf 468: /// used by attributed_meaning_to_string / append_attribute_subattribute
1.40 paf 469: struct Attributed_meaning_info {
1.52 paf 470: String *header; // header line being constructed
471: String::Untaint_lang lang; // language in which to append to that line
1.40 paf 472: };
1.17 paf 473: static void append_attribute_subattribute(const Hash::Key& akey, Hash::Val *avalue,
1.14 paf 474: void *info) {
475: if(akey==VALUE_NAME)
476: return;
477:
1.40 paf 478: Attributed_meaning_info& ami=*static_cast<Attributed_meaning_info *>(info);
479:
1.14 paf 480: // ...; charset=windows1251
1.48 paf 481: *ami.header << "; ";
1.70 parser 482: ami.header->append(akey, ami.lang);
1.48 paf 483: *ami.header << "=";
1.70 parser 484: ami.header->append(static_cast<Value *>(avalue)->as_string(), ami.lang);
1.14 paf 485: }
1.49 paf 486: const String& attributed_meaning_to_string(Value& meaning,
487: String::Untaint_lang lang) {
1.20 paf 488: String &result=*new(meaning.pool()) String(meaning.pool());
1.75 parser 489: if(Hash *hash=meaning.get_hash(0)) {
1.14 paf 490: // $value(value) $subattribute(subattribute value)
491: if(Value *value=static_cast<Value *>(hash->get(*value_name)))
1.40 paf 492: result.append(value->as_string(), lang, true);
1.14 paf 493:
1.40 paf 494: Attributed_meaning_info attributed_meaning_info={
1.52 paf 495: &result,
496: lang
1.40 paf 497: };
498: hash->for_each(append_attribute_subattribute, &attributed_meaning_info);
1.14 paf 499: } else // result value
1.40 paf 500: result.append(meaning.as_string(), lang, true);
1.14 paf 501:
502: return result;
1.24 paf 503: }
504:
505: #ifdef WIN32
506: void back_slashes_to_slashes(char *s) {
507: if(s)
508: for(; *s; s++)
509: if(*s=='\\')
510: *s='/';
511: }
1.42 paf 512: /*
513: void slashes_to_back_slashes(char *s) {
514: if(s)
515: for(; *s; s++)
516: if(*s=='/')
517: *s='\\';
518: }
519: */
1.24 paf 520: #endif
1.41 paf 521:
522: bool StrEqNc(const char *s1, const char *s2, bool strict) {
523: while(true) {
524: if(!(*s1)) {
525: if(!(*s2))
526: return true;
527: else
528: return !strict;
529: } else if(!(*s2))
530: return !strict;
531: if(isalpha(*s1)) {
532: if(tolower(*s1) !=tolower(*s2))
533: return false;
534: } else if((*s1) !=(*s2))
535: return false;
536: s1++;
537: s2++;
538: }
1.57 parser 539: }
540:
1.84 paf 541: static bool isLeap(int year) {
1.57 parser 542: return !(
543: (year % 4) || ((year % 400) && !(year % 100))
544: );
545: }
546:
547: int getMonthDays(int year, int month) {
548: int monthDays[]={
549: 31,
550: isLeap(year) ? 29 : 28,
551: 31,
552: 30,
553: 31,
554: 30,
555: 31,
556: 31,
557: 30,
558: 31,
559: 30,
560: 31
561: };
562: return monthDays[month];
1.41 paf 563: }
1.69 parser 564:
565: void remove_crlf(char *start, char *end) {
566: for(char *p=start; p<end; p++)
567: switch(*p) {
568: case '\n': *p='|'; break;
569: case '\r': *p=' '; break;
570: }
1.91 paf 571: }
572:
573:
574: /// must be last in this file
575: #undef vsnprintf
576: int __vsnprintf(char *b, size_t s, const char *f, va_list l) {
577: if(!s)
578: return 0;
579:
580: int r;
581: // note: on win32& maybe somewhere else
582: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
583: --s;
584: #if _MSC_VER
585: /*
586: win32:
587: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
588:
589: if the number of bytes to write exceeds buffer, then count bytes are written and –1 is returned
590: */
591: r=_vsnprintf(b, s, f, l);
592: if(r<0)
593: r=s;
594: #else
595: r=vsnprintf(b, s, f, l);
596: /*
597: solaris:
598: man vsnprintf
599:
600: The snprintf() function returns the number of characters
601: formatted, that is, the number of characters that would have
602: been written to the buffer if it were large enough. If the
603: value of n is 0 on a call to snprintf(), an unspecified
604: value less than 1 is returned.
605: */
606:
607: if(r<0)
608: r=0;
609: else if(r>s)
610: r=s;
611: #endif
612: b[r]=0;
613: return r;
614: }
615:
616: int __snprintf(char *b, size_t s, const char *f, ...) {
617: va_list l;
618: va_start(l, f);
619: int r=__vsnprintf(b, s, f, l);
620: va_end(l);
621: return r;
1.98 paf 622: }
623:
624: int pa_sleep(unsigned long secs, unsigned long usecs) {
625: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
626:
627: #ifdef WIN32
628: Sleep(secs * 1000 + usecs / 1000);
629: return 0;
630: #else
631: struct timeval t;
632: t.tv_sec = secs;
633: t.tv_usec = usecs;
634: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
635: #endif
1.74 parser 636: }
E-mail: