Annotation of parser3/src/main/pa_common.C, revision 1.140
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)
1.111 paf 6: */
1.16 paf 7:
1.140 ! paf 8: static const char* IDENT_COMMON_C="$Date: 2002/12/24 09:55:33 $";
1.1 paf 9:
10: #include "pa_common.h"
1.4 paf 11: #include "pa_exception.h"
1.14 paf 12: #include "pa_globals.h"
1.126 paf 13: #include "pa_hash.h"
14: #include "pa_vstring.h"
1.135 paf 15: #include "pa_vdate.h"
1.1 paf 16:
1.98 paf 17: #ifdef WIN32
18: # include <windows.h>
1.126 paf 19: #else
20: # define closesocket close
1.98 paf 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:
1.138 paf 32: #ifdef HAVE_FTRUNCATE
33: # define PA_O_TRUNC 0
34: #else
35: # ifdef _O_TRUNC
36: # define PA_O_TRUNC _O_TRUNC
37: # else
38: # error you must have either ftruncate function or _O_TRUNC bit declared
39: # endif
40: #endif
41:
1.93 paf 42: // locking constants
43:
1.99 paf 44: #ifdef HAVE_FLOCK
45:
46: static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); }
47: static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); }
48: static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); }
49: static int unlock(int fd) { return flock(fd, LOCK_UN); }
50:
1.98 paf 51: #else
1.99 paf 52: #ifdef HAVE__LOCKING
1.98 paf 53:
1.126 paf 54: #define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1)
1.99 paf 55: static int lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); }
56: static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); }
57: static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); }
58: static int unlock(int fd) { FLOCK(_LK_UNLCK); }
1.93 paf 59:
1.99 paf 60: #else
61: #ifdef HAVE_FCNTL
1.93 paf 62:
1.126 paf 63: #define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls)
1.99 paf 64: static int lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); }
65: static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); }
66: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); }
67: static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); }
1.93 paf 68:
69: #else
70: #ifdef HAVE_LOCKF
1.99 paf 71:
1.126 paf 72: #define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1)
1.99 paf 73: static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking
74: static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); }
75: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); }
76: static int unlock(int fd) { FLOCK(F_TLOCK); }
77:
1.93 paf 78: #else
1.99 paf 79:
80: #error unable to find file locking func
81:
82: #endif
1.93 paf 83: #endif
84: #endif
85: #endif
86:
1.127 paf 87: #define DEFAULT_USER_AGENT "parser3"
88:
89:
1.126 paf 90: void fix_line_breaks(char* buf, size_t& size) {
1.139 paf 91: if(size==0)
92: return;
93:
1.87 paf 94: //_asm int 3;
1.126 paf 95: const char* const eob=buf+size;
96: char* dest=buf;
1.72 parser 97: // fix DOS: \r\n -> \n
98: // fix Macintosh: \r -> \n
1.126 paf 99: char* bol=buf;
1.137 paf 100: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 101: size_t len=eol-bol;
102: if(dest!=bol)
1.126 paf 103: memcpy(dest, bol, len);
1.72 parser 104: dest+=len;
1.126 paf 105: *dest++='\n';
1.72 parser 106:
1.126 paf 107: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 108: bol=eol+2;
1.126 paf 109: size--;
110: } else // \r, not \n = Macintosh
1.72 parser 111: bol=eol+1;
112: }
113: // last piece without \r, including terminating 0
114: if(dest!=bol)
1.126 paf 115: memcpy(dest, bol, eob-bol);
1.72 parser 116: }
1.18 paf 117:
1.126 paf 118: char* file_read_text(Pool& pool, const String& file_spec,
119: bool fail_on_read_problem,
120: Hash *params, Hash** out_fields) {
1.72 parser 121: void *result; size_t size;
1.126 paf 122: return file_read(pool, file_spec, result, size, true, params, out_fields, fail_on_read_problem)?(char *)result:0;
123: }
124:
125: //http request stuff
126: /* ************************ http stuff *********************** */
127:
128: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
129: memset(addr, 0, sizeof(*addr));
130: addr->sin_family=AF_INET;
131: addr->sin_port=htons(port);
132: if(host) {
133: if(struct hostent *hostIP=gethostbyname(host))
134: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
135: else
136: return false;
137: } else
138: addr->sin_addr.s_addr=INADDR_ANY;
139: return true;
140: }
141:
142: static void http_read_response(String& response, int sock){
1.130 paf 143: bool check_status=true;
144: ssize_t EOLat=0;
1.126 paf 145: while(true) {
146: char *buf=(char *)response.pool().malloc(MAX_STRING);
147: ssize_t size=recv(sock, buf, MAX_STRING, 0);
148: if(size<=0)
149: break;
1.130 paf 150: response.APPEND_TAINTED(buf, size, "remote HTTP server response", 0);
151: if(check_status && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response
152: check_status=false;
153:
154: const String& status_line=response.mid(0, (size_t)EOLat);
155: Array astatus(response.pool());
156: size_t pos_after_ref=0; status_line.split(astatus, &pos_after_ref, " ", 1);
157: const String* status_code=astatus.get_string(1);
158:
159: if(!status_code || *status_code!="200")
160: throw Exception("http.status",
161: status_code,
162: "invalid HTTP response status");
163: }
1.126 paf 164: }
1.130 paf 165: if(check_status)
166: throw Exception("http.response",
167: 0,
168: "bad response from host - no status found (size=%lu)", response.size());
1.126 paf 169: }
170:
171: /* ********************** request *************************** */
172:
173: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
174: # define WE_CAN_USE_ALARM
175: #endif
176:
177: #ifdef WE_CAN_USE_ALARM
178: static sigjmp_buf timeout_env;
179: static void timeout_handler(int sig){
180: siglongjmp(timeout_env, 1);
181: }
182: #endif
183:
184: static void http_request(String& response,
185: const String *origin_string,
186: const char* host, int port,
187: const char* request,
188: int timeout){
189: if(!host)
190: throw Exception("http.host",
191: origin_string,
192: "zero hostname"); //never
193:
194: #ifdef WE_CAN_USE_ALARM
195: signal(SIGALRM, timeout_handler);
196: #endif
197: int sock=-1;
198: try {
199: #ifdef WE_CAN_USE_ALARM
200: if(sigsetjmp(timeout_env, 1))
201: throw Exception("http.timeout",
202: origin_string,
203: "timeout occured while retrieving document");
204: else {
205: alarm(timeout);
206: #endif
207: struct sockaddr_in dest;
208:
209: if(!set_addr(&dest, host, port))
210: throw Exception("http.host",
211: origin_string,
1.127 paf 212: "can not resolve hostname \"%s\"", host);
1.126 paf 213:
214: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
215: throw Exception("http.connect",
216: origin_string,
1.127 paf 217: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 218: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
219: throw Exception("http.connect",
220: origin_string,
1.127 paf 221: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 222: size_t request_size=strlen(request);
223: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
224: throw Exception("http.connect",
225: origin_string,
1.127 paf 226: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 227:
228: http_read_response(response, sock);
229: if(sock>=0)
230: closesocket(sock);
231: #ifdef WE_CAN_USE_ALARM
232: }
233: #endif
234: } catch(...) {
235: if(sock>=0)
236: closesocket(sock);
237: #ifdef WE_CAN_USE_ALARM
238: alarm(0);
239: #endif
240: /*re*/throw;
241: }
242: }
243:
1.127 paf 244: #ifndef DOXYGEN
245: struct Http_pass_header_info {
246: String* request;
247: bool user_agent_specified;
248: };
249: #endif
250: static void http_pass_header(const Hash::Key& key, Hash::Val *value, void *info)
1.126 paf 251: {
1.127 paf 252: Http_pass_header_info& i=*static_cast<Http_pass_header_info *>(info);
253: Pool& pool=i.request->pool();
254:
1.135 paf 255: *(i.request)<<key<<": "
1.136 paf 256: << attributed_meaning_to_string(*static_cast<Value *>(value), String::UL_HTTP_HEADER, false)
1.135 paf 257: <<"\n";
258:
1.127 paf 259: if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT")
260: i.user_agent_specified=true;
1.126 paf 261: }
262: static void file_read_http(Pool& pool, const String& file_spec,
263: void*& data, size_t& data_size,
1.127 paf 264: Hash *options=0, Hash** out_fields=0) {
1.126 paf 265: char host[MAX_STRING];
1.129 paf 266: const char* uri;
1.126 paf 267: int port;
268: const char* method="GET";
269: int timeout=2;
1.127 paf 270: Value *vheaders=0;
1.126 paf 271:
1.133 paf 272: String connect_string(pool);
273: // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one
274: connect_string.append(file_spec, String::UL_URI); // tainted pieces -> URI pieces
275:
276: char* connect_string_cstr=connect_string.cstr(String::UL_UNSPECIFIED);
277: if(strncmp(connect_string_cstr, "http://", 7)!=0)
1.126 paf 278: throw Exception(0,
1.133 paf 279: &connect_string,
1.126 paf 280: "does not start with http://"); //never
1.133 paf 281: connect_string_cstr+=7;
1.126 paf 282:
1.133 paf 283: strncpy(host, connect_string_cstr, sizeof(host)-1); host[sizeof(host)-1]=0;
1.126 paf 284: char* host_uri=lsplit(host, '/');
1.133 paf 285: uri=host_uri?connect_string_cstr+(host_uri-1-host):"/";
1.126 paf 286: char* port_cstr=lsplit(host, ':');
287: char* error_pos=0;
288: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
289:
1.127 paf 290: if(options) {
291: int valid_options=0;
292: if(Value *vmethod=static_cast<Value *>(options->get(*http_method_name))) {
293: valid_options++;
294: method=vmethod->as_string().cstr();
295: }
296: if(Value *vtimeout=static_cast<Value *>(options->get(*http_timeout_name))) {
297: valid_options++;
298: timeout=vtimeout->as_int();
299: }
300: if(vheaders=static_cast<Value *>(options->get(*http_headers_name))) {
301: valid_options++;
302: }
303: if(valid_options!=options->size())
304: throw Exception("parser.runtime",
305: 0,
306: "invalid option passed");
1.133 paf 307: }
1.126 paf 308:
309: //making request
1.127 paf 310: String request(pool);
1.126 paf 311: request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";
1.127 paf 312: bool user_agent_specified=false;
313: if(vheaders && !vheaders->is_string()) { // allow empty
1.133 paf 314: if(Hash *headers=vheaders->get_hash(&connect_string)) {
1.127 paf 315: Http_pass_header_info info={&request};
316: headers->for_each(http_pass_header, &info);
317: user_agent_specified=info.user_agent_specified;
318: } else
319: throw Exception("parser.runtime",
1.133 paf 320: &connect_string,
1.127 paf 321: "headers param must be hash");
322: };
323: if(!user_agent_specified) // defaulting
324: request << "user-agent: " DEFAULT_USER_AGENT "\n";
1.126 paf 325: request<<"\n";
326:
327: //sending request
328: String response(pool);
329: http_request(response,
1.133 paf 330: &connect_string, host, port, request.cstr(String::UL_UNSPECIFIED), timeout);
1.126 paf 331:
332: //processing results
333: int pos=response.pos("\r\n\r\n", 4);
334: if(pos<1){
335: throw Exception("http.response",
1.133 paf 336: &connect_string,
1.126 paf 337: "bad response from host - no headers found");
338: }
339: String header_block=response.mid(0, pos);
340: String body=response.mid(pos+4, response.size());
341:
342: Array aheaders(pool);
343: Hash& headers=*new(pool) Hash(pool);
344: size_t pos_after_ref=0;
345: header_block.split(aheaders, &pos_after_ref, "\r\n", 2);
346:
347: //processing status code
348: const String *status_line=aheaders.get_string(0);
349: if(!status_line){
350: throw Exception("http.response",
1.133 paf 351: &connect_string,
1.126 paf 352: "bad response from host - no status line ");
353: }
354: //processing headers
355: for(int i=1;i<aheaders.size();i++) {
356: if(const String *line=aheaders.get_string(i)) {
357: pos=line->pos(": ", 2);
358: if(pos<1)
359: throw Exception("http.response",
1.133 paf 360: &connect_string,
1.126 paf 361: "bad response from host - bad header \"%s\"", line->cstr());
362:
1.128 paf 363: headers.put(
364: line->mid(0, pos).change_case(pool, String::CC_UPPER),
365: new(pool) VString(line->mid(pos+2, line->size())));
1.126 paf 366: } else
367: throw Exception("http.response",
1.133 paf 368: &connect_string,
1.126 paf 369: "bad response from host - bad headers \"%s\"", header_block.cstr());
370: }
371:
372: // output response
373: data=body.cstr(); data_size=body.size();
1.130 paf 374: if(out_fields)
1.126 paf 375: *out_fields=&headers;
1.34 paf 376: }
1.123 paf 377:
378: #ifndef DOXYGEN
379: struct File_read_action_info {
380: void **data; size_t *data_size;
1.126 paf 381: };
1.123 paf 382: #endif
1.126 paf 383: static void file_read_action(Pool& pool,
384: struct stat& finfo,
1.123 paf 385: int f,
1.126 paf 386: const String& file_spec, const char* fname, bool as_text,
1.123 paf 387: void *context) {
1.126 paf 388: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 389: if(size_t to_read_size=(size_t)finfo.st_size) {
1.126 paf 390: *info.data=pool.malloc(to_read_size+(as_text?1:0), 3);
391: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 392:
393: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 394: throw Exception(0,
1.123 paf 395: &file_spec,
396: "read failed: actually read %lu bytes count not in [0..%lu] valid range",
1.126 paf 397: *info.data_size, to_read_size);
1.123 paf 398: } else { // empty file
399: if(as_text) {
1.126 paf 400: *info.data=pool.malloc(1);
1.123 paf 401: *(char*)(*info.data)=0;
402: } else
403: *info.data=0;
404: *info.data_size=0;
405: return;
406: }
1.126 paf 407: }
408: bool file_read(Pool& pool, const String& file_spec,
409: void*& data, size_t& data_size,
410: bool as_text, Hash *params, Hash** out_fields,
411: bool fail_on_read_problem) {
412: bool result;
413: if(file_spec.starts_with("http://", 7)) {
414: // fail on read problem
415: file_read_http(pool, file_spec, data, data_size, params, out_fields);
416: result=true;
417: } else {
418: File_read_action_info info={&data, &data_size};
419: result=file_read_action_under_lock(pool, file_spec,
420: "read", file_read_action, &info,
421: as_text, fail_on_read_problem);
422: }
1.123 paf 423:
1.126 paf 424: if(result && as_text) {
1.131 paf 425: // UTF-8 signature: EF BB BF
426: if(data_size>=3) {
427: char *in=(char *)data;
428: if((in[0] == '\xEF') && (in[1] == '\xBB') &&
429: (in[2] == '\xBF')) {
430: data=in+3; data_size-=3;// skip prefix
431: }
432: }
433:
1.126 paf 434: fix_line_breaks((char *)(data), data_size);
1.123 paf 435: // note: after fixing
1.126 paf 436: ((char*&)(data))[data_size]=0;
1.123 paf 437: }
1.126 paf 438:
439: return result;
1.123 paf 440: }
441:
442: bool file_read_action_under_lock(Pool& pool, const String& file_spec,
1.126 paf 443: const char* action_name, File_read_action action, void *context,
444: bool as_text,
1.123 paf 445: bool fail_on_read_problem) {
1.126 paf 446: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.33 paf 447: int f;
448:
449: // first open, next stat:
1.45 paf 450: // directory update of NTFS hard links performed on open.
1.33 paf 451: // ex:
452: // a.html:^test[] and b.html hardlink to a.html
453: // user inserts ! before ^test in a.html
1.126 paf 454: // directory entry of b.html in NTFS not updated at once,
1.35 paf 455: // they delay update till open, so we would receive "!^test[" string
456: // if would do stat, next open.
1.123 paf 457: // later: it seems, even this does not help sometimes
1.98 paf 458: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 459: try {
460: if(lock_shared_blocking(f)!=0)
1.126 paf 461: throw Exception("file.lock",
1.123 paf 462: &file_spec,
463: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 464: strerror(errno), errno, fname);
1.123 paf 465:
1.124 paf 466: struct stat finfo;
467: if(stat(fname, &finfo)!=0)
468: throw Exception("file.missing", // hardly possible: we just opened it OK
469: &file_spec,
470: "stat failed: %s (%d), actual filename '%s'",
1.126 paf 471: strerror(errno), errno, fname);
1.124 paf 472:
1.140 ! paf 473: #ifdef PA_SAFE_MODE
! 474: if(finfo.st_uid/*foreign?*/!=geteuid()
! 475: && finfo.st_gid/*foreign?*/!=getegid())
1.126 paf 476: throw Exception("parser.runtime",
477: &file_spec,
1.140 ! paf 478: "parser is in safe mode: reading files of foreign group and user disabled [recompile parser with --disable-safe-mode configure option], actual filename '%s'",
1.126 paf 479: fname);
1.105 paf 480: #endif
1.32 paf 481:
1.126 paf 482: action(pool, finfo, f, file_spec, fname, as_text, context);
1.123 paf 483: } catch(...) {
1.126 paf 484: unlock(f);close(f);
1.123 paf 485: if(fail_on_read_problem)
486: /*re*/throw;
487: return false;
488: }
1.87 paf 489:
1.126 paf 490: unlock(f);close(f);
1.72 parser 491: return true;
1.118 paf 492: } else {
493: if(fail_on_read_problem)
1.126 paf 494: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 495: &file_spec,
1.123 paf 496: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 497: action_name, strerror(errno), errno, fname);
1.118 paf 498: return false;
499: }
1.8 paf 500: }
501:
1.63 parser 502: static void create_dir_for_file(const String& file_spec) {
503: size_t pos_after=1;
504: int pos_before;
505: while((pos_before=file_spec.pos("/", 1, pos_after))>=0) {
1.126 paf 506: mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775);
1.63 parser 507: pos_after=pos_before+1;
508: }
509: }
510:
1.98 paf 511: bool file_write_action_under_lock(
1.28 paf 512: const String& file_spec,
1.126 paf 513: const char* action_name, File_write_action action, void *context,
514: bool as_text,
515: bool do_append,
516: bool do_block,
1.110 paf 517: bool fail_on_lock_problem) {
1.126 paf 518: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.28 paf 519: int f;
1.80 paf 520: if(access(fname, W_OK)!=0) // no
1.126 paf 521: create_dir_for_file(file_spec);
1.50 paf 522:
1.80 paf 523: if((f=open(fname,
524: O_CREAT|O_RDWR
525: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 526: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.99 paf 527: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.126 paf 528: Exception e("file.lock",
1.110 paf 529: &file_spec,
530: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 531: strerror(errno), errno, fname);
532: close(f);
1.110 paf 533: if(fail_on_lock_problem)
534: throw e;
1.98 paf 535: return false;
536: }
1.96 paf 537:
538: try {
1.126 paf 539: action(f, context);
1.96 paf 540: } catch(...) {
1.138 paf 541: #ifdef HAVE_FTRUNCATE
1.104 paf 542: if(!do_append)
1.125 paf 543: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 544: #endif
1.126 paf 545: unlock(f);close(f);
1.96 paf 546: /*re*/throw;
547: }
1.80 paf 548:
1.138 paf 549: #ifdef HAVE_FTRUNCATE
1.104 paf 550: if(!do_append)
1.125 paf 551: 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 552: #endif
1.126 paf 553: unlock(f);close(f);
1.98 paf 554: return true;
1.80 paf 555: } else
1.126 paf 556: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 557: &file_spec,
1.96 paf 558: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 559: action_name, strerror(errno), errno, fname);
1.96 paf 560: // here should be nothing, see rethrow above
561: }
562:
563: #ifndef DOXYGEN
564: struct File_write_action_info {
565: const void *data; size_t size;
1.126 paf 566: };
1.96 paf 567: #endif
568: static void file_write_action(int f, void *context) {
1.126 paf 569: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.113 paf 570: if(info.size) {
1.126 paf 571: int written=write(f, info.data, info.size);
1.116 paf 572: if(written<0)
1.126 paf 573: throw Exception(0,
574: 0,
575: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 576: }
1.96 paf 577: }
578: void file_write(
579: const String& file_spec,
580: const void *data, size_t size,
1.126 paf 581: bool as_text,
1.96 paf 582: bool do_append) {
1.126 paf 583: File_write_action_info info={data, size};
1.98 paf 584: file_write_action_under_lock(
1.96 paf 585: file_spec,
1.126 paf 586: "write", file_write_action, &info,
587: as_text,
588: do_append);
1.30 paf 589: }
590:
1.63 parser 591: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 592: static void rmdir(const String& file_spec, size_t pos_after) {
593: int pos_before;
594: if((pos_before=file_spec.pos("/", 1, pos_after))>=0)
1.126 paf 595: rmdir(file_spec, pos_before+1);
1.50 paf 596:
1.126 paf 597: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_FILE_SPEC));
1.50 paf 598: }
1.95 paf 599: bool file_delete(const String& file_spec, bool fail_on_read_problem) {
1.126 paf 600: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.54 parser 601: if(unlink(fname)!=0)
1.93 paf 602: if(fail_on_read_problem)
1.126 paf 603: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 604: &file_spec,
605: "unlink failed: %s (%d), actual filename '%s'",
1.126 paf 606: strerror(errno), errno, fname);
1.93 paf 607: else
608: return false;
1.50 paf 609:
1.126 paf 610: rmdir(file_spec, 1);
1.93 paf 611: return true;
1.60 parser 612: }
1.95 paf 613: void file_move(const String& old_spec, const String& new_spec) {
1.126 paf 614: const char* old_spec_cstr=old_spec.cstr(String::UL_FILE_SPEC);
615: const char* new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC);
1.63 parser 616:
1.126 paf 617: create_dir_for_file(new_spec);
1.63 parser 618:
1.60 parser 619: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 620: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 621: &old_spec,
622: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.126 paf 623: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 624:
1.126 paf 625: rmdir(old_spec, 1);
1.31 paf 626: }
627:
1.51 paf 628:
1.126 paf 629: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 630: struct stat lfinfo;
631: bool result=stat(fname, &lfinfo)==0;
632: if(afinfo)
633: *afinfo=lfinfo;
634: return result;
1.119 paf 635: }
636:
637: bool entry_exists(const String& file_spec) {
1.126 paf 638: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
639: return entry_exists(fname, 0);
1.118 paf 640: }
641:
1.51 paf 642: static bool entry_readable(const String& file_spec, bool need_dir) {
1.126 paf 643: char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.120 paf 644: if(need_dir) {
1.126 paf 645: size_t size=strlen(fname);
1.120 paf 646: while(size) {
1.126 paf 647: char c=fname[size-1];
1.120 paf 648: if(c=='/' || c=='\\')
649: fname[--size]=0;
650: else
651: break;
652: }
653: }
1.51 paf 654: struct stat finfo;
1.118 paf 655: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 656: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 657: return is_dir==need_dir;
658: }
659: return false;
660: }
1.31 paf 661: bool file_readable(const String& file_spec) {
1.126 paf 662: return entry_readable(file_spec, false);
1.51 paf 663: }
664: bool dir_readable(const String& file_spec) {
1.126 paf 665: return entry_readable(file_spec, true);
1.65 parser 666: }
667: String *file_readable(const String& path, const String& name) {
1.126 paf 668: String *result=new(path.pool()) String(path);
669: *result << "/";
1.65 parser 670: *result << name;
671: return file_readable(*result)?result:0;
1.43 paf 672: }
673: bool file_executable(const String& file_spec) {
1.64 parser 674: return access(file_spec.cstr(String::UL_FILE_SPEC), X_OK)==0;
1.44 paf 675: }
676:
1.64 parser 677: bool file_stat(const String& file_spec,
1.58 parser 678: size_t& rsize,
1.126 paf 679: time_t& ratime,
680: time_t& rmtime,
681: time_t& rctime,
1.64 parser 682: bool fail_on_read_problem) {
1.126 paf 683: Pool& pool=file_spec.pool();
684: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.44 paf 685: struct stat finfo;
686: if(stat(fname, &finfo)!=0)
1.64 parser 687: if(fail_on_read_problem)
1.126 paf 688: throw Exception("file.missing",
1.67 parser 689: &file_spec,
690: "getting file size failed: %s (%d), real filename '%s'",
1.126 paf 691: strerror(errno), errno, fname);
1.64 parser 692: else
693: return false;
1.58 parser 694: rsize=finfo.st_size;
695: ratime=finfo.st_atime;
696: rmtime=finfo.st_mtime;
697: rctime=finfo.st_ctime;
1.64 parser 698: return true;
1.18 paf 699: }
700:
1.126 paf 701: char* getrow(char* *row_ref, char delim) {
702: char* result=*row_ref;
1.8 paf 703: if(result) {
1.126 paf 704: *row_ref=strchr(result, delim);
1.8 paf 705: if(*row_ref)
706: *((*row_ref)++)=0;
707: else if(!*result)
708: return 0;
709: }
710: return result;
711: }
712:
1.126 paf 713: char* lsplit(char* string, char delim) {
1.23 paf 714: if(string) {
1.126 paf 715: char* v=strchr(string, delim);
1.8 paf 716: if(v) {
717: *v=0;
718: return v+1;
719: }
720: }
721: return 0;
722: }
723:
1.126 paf 724: char* lsplit(char* *string_ref, char delim) {
725: char* result=*string_ref;
726: char* next=lsplit(*string_ref, delim);
1.8 paf 727: *string_ref=next;
728: return result;
1.9 paf 729: }
730:
1.126 paf 731: char* rsplit(char* string, char delim) {
1.18 paf 732: if(string) {
1.126 paf 733: char* v=strrchr(string, delim);
1.18 paf 734: if(v) {
1.9 paf 735: *v=0;
736: return v+1;
737: }
738: }
739: return NULL;
1.10 paf 740: }
741:
1.37 paf 742: /// @todo less stupid type detection
1.126 paf 743: char* format(Pool& pool, double value, char* fmt) {
744: char local_buf[MAX_NUMBER];
1.108 paf 745: size_t size;
746:
1.10 paf 747: if(fmt)
748: if(strpbrk(fmt, "diouxX"))
749: if(strpbrk(fmt, "ouxX"))
1.126 paf 750: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 751: else
1.126 paf 752: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 753: else
1.126 paf 754: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 755: else
1.126 paf 756: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 757:
1.126 paf 758: char* pool_buf=(char *)pool.malloc(size+1, 4);
759: memcpy(pool_buf, local_buf, size+1);
1.108 paf 760: return pool_buf;
1.12 paf 761: }
762:
1.36 paf 763: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 764: #ifdef WIN32
765: do{
1.126 paf 766: int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout);
1.12 paf 767: if(chunk_written<=0)
768: break;
769: size-=chunk_written;
1.36 paf 770: buf=((const char*)buf)+chunk_written;
1.126 paf 771: } while(size>0);
1.12 paf 772:
773: return size;
774: #else
1.126 paf 775: return fwrite(buf, 1, size, stdout);
1.12 paf 776: #endif
1.2 paf 777: }
1.14 paf 778:
1.126 paf 779: char* unescape_chars(Pool& pool, const char* cp, int len) {
780: char* s=(char *)pool.malloc(len + 1, 5);
1.14 paf 781: enum EscapeState {
1.33 paf 782: EscapeRest,
783: EscapeFirst,
1.14 paf 784: EscapeSecond
785: } escapeState=EscapeRest;
786: int escapedValue=0;
787: int srcPos=0;
788: int dstPos=0;
789: while(srcPos < len) {
1.126 paf 790: int ch=cp[srcPos];
1.14 paf 791: switch(escapeState) {
792: case EscapeRest:
793: if(ch=='%') {
794: escapeState=EscapeFirst;
795: } else if(ch=='+') {
1.126 paf 796: s[dstPos++]=' ';
1.14 paf 797: } else {
798: s[dstPos++]=ch;
799: }
800: break;
801: case EscapeFirst:
802: escapedValue=hex_value[ch] << 4;
803: escapeState=EscapeSecond;
804: break;
805: case EscapeSecond:
1.126 paf 806: escapedValue +=hex_value[ch];
1.14 paf 807: s[dstPos++]=escapedValue;
808: escapeState=EscapeRest;
809: break;
810: }
1.126 paf 811: srcPos++;
1.14 paf 812: }
813: s[dstPos]=0;
814: return s;
1.24 paf 815: }
816:
817: #ifdef WIN32
1.126 paf 818: void back_slashes_to_slashes(char* s) {
1.24 paf 819: if(s)
820: for(; *s; s++)
821: if(*s=='\\')
1.126 paf 822: *s='/';
1.24 paf 823: }
1.42 paf 824: /*
1.126 paf 825: void slashes_to_back_slashes(char* s) {
1.42 paf 826: if(s)
827: for(; *s; s++)
828: if(*s=='/')
1.126 paf 829: *s='\\';
1.42 paf 830: }
831: */
1.24 paf 832: #endif
1.41 paf 833:
1.126 paf 834: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 835: while(true) {
836: if(!(*s1)) {
837: if(!(*s2))
838: return true;
839: else
840: return !strict;
841: } else if(!(*s2))
842: return !strict;
843: if(isalpha(*s1)) {
844: if(tolower(*s1) !=tolower(*s2))
845: return false;
846: } else if((*s1) !=(*s2))
847: return false;
1.126 paf 848: s1++;
849: s2++;
1.41 paf 850: }
1.57 parser 851: }
852:
1.84 paf 853: static bool isLeap(int year) {
1.57 parser 854: return !(
855: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 856: );
1.57 parser 857: }
858:
859: int getMonthDays(int year, int month) {
860: int monthDays[]={
1.126 paf 861: 31,
862: isLeap(year) ? 29 : 28,
863: 31,
864: 30,
865: 31,
866: 30,
867: 31,
868: 31,
869: 30,
870: 31,
871: 30,
1.57 parser 872: 31
1.126 paf 873: };
874: return monthDays[month];
1.41 paf 875: }
1.69 parser 876:
1.126 paf 877: void remove_crlf(char* start, char* end) {
878: for(char* p=start; p<end; p++)
1.69 parser 879: switch(*p) {
1.126 paf 880: case '\n': *p='|'; break;
881: case '\r': *p=' '; break;
1.69 parser 882: }
1.91 paf 883: }
884:
885:
886: /// must be last in this file
887: #undef vsnprintf
1.126 paf 888: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 889: if(!s)
890: return 0;
891:
892: int r;
893: // note: on win32& maybe somewhere else
894: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
895: --s;
896: #if _MSC_VER
897: /*
898: win32:
899: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
900:
901: if the number of bytes to write exceeds buffer, then count bytes are written and –1 is returned
902: */
1.126 paf 903: r=_vsnprintf(b, s, f, l);
1.91 paf 904: if(r<0)
905: r=s;
906: #else
1.126 paf 907: r=vsnprintf(b, s, f, l);
1.91 paf 908: /*
909: solaris:
910: man vsnprintf
911:
912: The snprintf() function returns the number of characters
913: formatted, that is, the number of characters that would have
914: been written to the buffer if it were large enough. If the
915: value of n is 0 on a call to snprintf(), an unspecified
916: value less than 1 is returned.
917: */
918:
919: if(r<0)
920: r=0;
921: else if(r>s)
922: r=s;
923: #endif
924: b[r]=0;
925: return r;
926: }
927:
1.126 paf 928: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 929: va_list l;
1.126 paf 930: va_start(l, f);
931: int r=__vsnprintf(b, s, f, l);
932: va_end(l);
1.91 paf 933: return r;
1.98 paf 934: }
935:
936: int pa_sleep(unsigned long secs, unsigned long usecs) {
1.126 paf 937: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
1.98 paf 938:
939: #ifdef WIN32
1.126 paf 940: Sleep(secs * 1000 + usecs / 1000);
1.98 paf 941: return 0;
942: #else
943: struct timeval t;
944: t.tv_sec = secs;
945: t.tv_usec = usecs;
1.126 paf 946: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
1.98 paf 947: #endif
1.135 paf 948: }
949:
950:
951: // attributed meaning
952:
953: static size_t date_attribute(const VDate& vdate, char *buf, size_t buf_size) {
954: const char month_names[12][4]={
955: "Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec"};
956: const char days[7][4]={
957: "Sun","Mon","Tue","Wed","Thu","Fri","Sat"};
958:
959: time_t when=vdate.get_time();
960: struct tm *tms=gmtime(&when);
961: if(!tms)
962: throw Exception(0,
963: 0,
964: "bad time in attribute value (seconds from epoch=%ld)", when);
965: return snprintf(buf, MAX_STRING, "%s, %.2d-%s-%.4d %.2d:%.2d:%.2d GMT",
966: days[tms->tm_wday],
967: tms->tm_mday,month_names[tms->tm_mon],tms->tm_year+1900,
968: tms->tm_hour,tms->tm_min,tms->tm_sec);
969: }
970: static void append_attribute_meaning(String& result,
1.136 paf 971: Value& value, String::Untaint_lang lang, bool forced) {
1.135 paf 972: if(const String *string=value.get_string())
1.136 paf 973: result.append(string->join_chains(result.pool(), 0), lang, forced);
1.135 paf 974: else
975: if(Value *vdate=value.as(VDATE_TYPE, false)) {
976: char *buf=(char *)result.malloc(MAX_STRING);
977: size_t size=date_attribute(*static_cast<VDate *>(vdate),
978: buf, MAX_STRING);
979:
980: result.APPEND_CLEAN(buf, size, "converted from date", 0);
981: } else
982: throw Exception("parser.runtime",
983: &result,
984: "trying to append here neither string nor date (%s)",
985: value.type());
986: }
987: #ifndef DOXYGEN
988: struct Attributed_meaning_info {
989: String *header; // header line being constructed
990: String::Untaint_lang lang; // language in which to append to that line
1.136 paf 991: bool forced; // do they force that lang?
1.135 paf 992: };
993: #endif
994: static void append_attribute_subattribute(const Hash::Key& akey, Hash::Val *avalue,
995: void *info) {
996: if(akey==VALUE_NAME)
997: return;
998:
999: Attributed_meaning_info& ami=*static_cast<Attributed_meaning_info *>(info);
1000:
1001: // ...; charset=windows1251
1002: *ami.header << "; ";
1.136 paf 1003: ami.header->append(akey, ami.lang, ami.forced);
1.135 paf 1004: *ami.header << "=";
1.136 paf 1005: append_attribute_meaning(*ami.header, *static_cast<Value *>(avalue), ami.lang, ami.forced);
1.135 paf 1006: }
1007: const String& attributed_meaning_to_string(Value& meaning,
1.136 paf 1008: String::Untaint_lang lang, bool forced) {
1.135 paf 1009: String &result=*new(meaning.pool()) String(meaning.pool());
1010: if(Hash *hash=meaning.get_hash(0)) {
1011: // $value(value) $subattribute(subattribute value)
1012: if(Value *value=static_cast<Value *>(hash->get(*value_name)))
1.136 paf 1013: append_attribute_meaning(result, *value, lang, forced);
1.135 paf 1014:
1015: Attributed_meaning_info attributed_meaning_info={&result, lang};
1016: hash->for_each(append_attribute_subattribute, &attributed_meaning_info);
1017: } else // result value
1.136 paf 1018: append_attribute_meaning(result, meaning, lang, forced);
1.135 paf 1019:
1020: return result;
1.74 parser 1021: }
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