Annotation of parser3/src/main/pa_common.C, revision 1.140.2.2
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:
1.140.2.2! paf 8: static const char* IDENT_COMMON_C="$Date: 2003/02/24 12:38:54 $";
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)
1.140.2.1 paf 174: # define PA_USE_ALARM
1.126 paf 175: #endif
176:
1.140.2.1 paf 177: #ifdef PA_USE_ALARM
1.126 paf 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,
1.140.2.2! paf 188: int timeout) {
1.126 paf 189: if(!host)
190: throw Exception("http.host",
191: origin_string,
192: "zero hostname"); //never
193:
1.140.2.1 paf 194: #ifdef PA_USE_ALARM
1.126 paf 195: signal(SIGALRM, timeout_handler);
196: #endif
197: int sock=-1;
1.140.2.1 paf 198: #ifdef PA_USE_ALARM
199: if(sigsetjmp(timeout_env, 1)) {
200: // stupid gcc [2.95.4] generated bad code
201: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
202: // rewritten simplier [though duplicating closesocket code]
203: if(sock>=0)
204: closesocket(sock);
205: throw Exception("http.timeout",
206: origin_string,
207: "timeout occured while retrieving document");
1.140.2.2! paf 208: } else
1.140.2.1 paf 209: alarm(timeout);
210: #endif
1.126 paf 211: try {
1.140.2.1 paf 212: struct sockaddr_in dest;
213:
214: if(!set_addr(&dest, host, port))
215: throw Exception("http.host",
1.126 paf 216: origin_string,
1.140.2.1 paf 217: "can not resolve hostname \"%s\"", host);
1.126 paf 218:
1.140.2.1 paf 219: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
220: throw Exception("http.connect",
221: origin_string,
222: "can not make socket: %s (%d)", strerror(errno), errno);
223: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
224: throw Exception("http.connect",
225: origin_string,
226: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
227: size_t request_size=strlen(request);
228: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
229: throw Exception("http.connect",
230: origin_string,
231: "error sending request: %s (%d)", strerror(errno), errno);
232:
233: http_read_response(response, sock);
234: closesocket(sock);
235: #ifdef PA_USE_ALARM
236: alarm(0);
1.126 paf 237: #endif
238: } catch(...) {
239: if(sock>=0)
240: closesocket(sock);
1.140.2.1 paf 241: #ifdef PA_USE_ALARM
1.126 paf 242: alarm(0);
243: #endif
244: /*re*/throw;
245: }
246: }
247:
1.127 paf 248: #ifndef DOXYGEN
249: struct Http_pass_header_info {
250: String* request;
251: bool user_agent_specified;
252: };
253: #endif
254: static void http_pass_header(const Hash::Key& key, Hash::Val *value, void *info)
1.126 paf 255: {
1.127 paf 256: Http_pass_header_info& i=*static_cast<Http_pass_header_info *>(info);
257: Pool& pool=i.request->pool();
258:
1.135 paf 259: *(i.request)<<key<<": "
1.136 paf 260: << attributed_meaning_to_string(*static_cast<Value *>(value), String::UL_HTTP_HEADER, false)
1.135 paf 261: <<"\n";
262:
1.127 paf 263: if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT")
264: i.user_agent_specified=true;
1.126 paf 265: }
266: static void file_read_http(Pool& pool, const String& file_spec,
267: void*& data, size_t& data_size,
1.127 paf 268: Hash *options=0, Hash** out_fields=0) {
1.126 paf 269: char host[MAX_STRING];
1.129 paf 270: const char* uri;
1.126 paf 271: int port;
272: const char* method="GET";
273: int timeout=2;
1.127 paf 274: Value *vheaders=0;
1.126 paf 275:
1.133 paf 276: String connect_string(pool);
277: // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one
278: connect_string.append(file_spec, String::UL_URI); // tainted pieces -> URI pieces
279:
280: char* connect_string_cstr=connect_string.cstr(String::UL_UNSPECIFIED);
281: if(strncmp(connect_string_cstr, "http://", 7)!=0)
1.126 paf 282: throw Exception(0,
1.133 paf 283: &connect_string,
1.126 paf 284: "does not start with http://"); //never
1.133 paf 285: connect_string_cstr+=7;
1.126 paf 286:
1.133 paf 287: strncpy(host, connect_string_cstr, sizeof(host)-1); host[sizeof(host)-1]=0;
1.126 paf 288: char* host_uri=lsplit(host, '/');
1.133 paf 289: uri=host_uri?connect_string_cstr+(host_uri-1-host):"/";
1.126 paf 290: char* port_cstr=lsplit(host, ':');
291: char* error_pos=0;
292: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
293:
1.127 paf 294: if(options) {
295: int valid_options=0;
296: if(Value *vmethod=static_cast<Value *>(options->get(*http_method_name))) {
297: valid_options++;
298: method=vmethod->as_string().cstr();
299: }
300: if(Value *vtimeout=static_cast<Value *>(options->get(*http_timeout_name))) {
301: valid_options++;
302: timeout=vtimeout->as_int();
303: }
304: if(vheaders=static_cast<Value *>(options->get(*http_headers_name))) {
305: valid_options++;
306: }
307: if(valid_options!=options->size())
308: throw Exception("parser.runtime",
309: 0,
310: "invalid option passed");
1.133 paf 311: }
1.126 paf 312:
313: //making request
1.127 paf 314: String request(pool);
1.126 paf 315: request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";
1.127 paf 316: bool user_agent_specified=false;
317: if(vheaders && !vheaders->is_string()) { // allow empty
1.133 paf 318: if(Hash *headers=vheaders->get_hash(&connect_string)) {
1.127 paf 319: Http_pass_header_info info={&request};
320: headers->for_each(http_pass_header, &info);
321: user_agent_specified=info.user_agent_specified;
322: } else
323: throw Exception("parser.runtime",
1.133 paf 324: &connect_string,
1.127 paf 325: "headers param must be hash");
326: };
327: if(!user_agent_specified) // defaulting
328: request << "user-agent: " DEFAULT_USER_AGENT "\n";
1.126 paf 329: request<<"\n";
330:
331: //sending request
332: String response(pool);
333: http_request(response,
1.133 paf 334: &connect_string, host, port, request.cstr(String::UL_UNSPECIFIED), timeout);
1.126 paf 335:
336: //processing results
337: int pos=response.pos("\r\n\r\n", 4);
338: if(pos<1){
339: throw Exception("http.response",
1.133 paf 340: &connect_string,
1.126 paf 341: "bad response from host - no headers found");
342: }
343: String header_block=response.mid(0, pos);
344: String body=response.mid(pos+4, response.size());
345:
346: Array aheaders(pool);
347: Hash& headers=*new(pool) Hash(pool);
348: size_t pos_after_ref=0;
349: header_block.split(aheaders, &pos_after_ref, "\r\n", 2);
350:
351: //processing status code
352: const String *status_line=aheaders.get_string(0);
353: if(!status_line){
354: throw Exception("http.response",
1.133 paf 355: &connect_string,
1.126 paf 356: "bad response from host - no status line ");
357: }
358: //processing headers
359: for(int i=1;i<aheaders.size();i++) {
360: if(const String *line=aheaders.get_string(i)) {
361: pos=line->pos(": ", 2);
362: if(pos<1)
363: throw Exception("http.response",
1.133 paf 364: &connect_string,
1.126 paf 365: "bad response from host - bad header \"%s\"", line->cstr());
366:
1.128 paf 367: headers.put(
368: line->mid(0, pos).change_case(pool, String::CC_UPPER),
369: new(pool) VString(line->mid(pos+2, line->size())));
1.126 paf 370: } else
371: throw Exception("http.response",
1.133 paf 372: &connect_string,
1.126 paf 373: "bad response from host - bad headers \"%s\"", header_block.cstr());
374: }
375:
376: // output response
377: data=body.cstr(); data_size=body.size();
1.130 paf 378: if(out_fields)
1.126 paf 379: *out_fields=&headers;
1.34 paf 380: }
1.123 paf 381:
382: #ifndef DOXYGEN
383: struct File_read_action_info {
384: void **data; size_t *data_size;
1.126 paf 385: };
1.123 paf 386: #endif
1.126 paf 387: static void file_read_action(Pool& pool,
388: struct stat& finfo,
1.123 paf 389: int f,
1.126 paf 390: const String& file_spec, const char* fname, bool as_text,
1.123 paf 391: void *context) {
1.126 paf 392: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 393: if(size_t to_read_size=(size_t)finfo.st_size) {
1.126 paf 394: *info.data=pool.malloc(to_read_size+(as_text?1:0), 3);
395: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 396:
397: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 398: throw Exception(0,
1.123 paf 399: &file_spec,
400: "read failed: actually read %lu bytes count not in [0..%lu] valid range",
1.126 paf 401: *info.data_size, to_read_size);
1.123 paf 402: } else { // empty file
403: if(as_text) {
1.126 paf 404: *info.data=pool.malloc(1);
1.123 paf 405: *(char*)(*info.data)=0;
406: } else
407: *info.data=0;
408: *info.data_size=0;
409: return;
410: }
1.126 paf 411: }
412: bool file_read(Pool& pool, const String& file_spec,
413: void*& data, size_t& data_size,
414: bool as_text, Hash *params, Hash** out_fields,
415: bool fail_on_read_problem) {
416: bool result;
417: if(file_spec.starts_with("http://", 7)) {
418: // fail on read problem
419: file_read_http(pool, file_spec, data, data_size, params, out_fields);
420: result=true;
421: } else {
422: File_read_action_info info={&data, &data_size};
423: result=file_read_action_under_lock(pool, file_spec,
424: "read", file_read_action, &info,
425: as_text, fail_on_read_problem);
426: }
1.123 paf 427:
1.126 paf 428: if(result && as_text) {
1.131 paf 429: // UTF-8 signature: EF BB BF
430: if(data_size>=3) {
431: char *in=(char *)data;
432: if((in[0] == '\xEF') && (in[1] == '\xBB') &&
433: (in[2] == '\xBF')) {
434: data=in+3; data_size-=3;// skip prefix
435: }
436: }
437:
1.126 paf 438: fix_line_breaks((char *)(data), data_size);
1.123 paf 439: // note: after fixing
1.126 paf 440: ((char*&)(data))[data_size]=0;
1.123 paf 441: }
1.126 paf 442:
443: return result;
1.123 paf 444: }
445:
446: bool file_read_action_under_lock(Pool& pool, const String& file_spec,
1.126 paf 447: const char* action_name, File_read_action action, void *context,
448: bool as_text,
1.123 paf 449: bool fail_on_read_problem) {
1.126 paf 450: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.33 paf 451: int f;
452:
453: // first open, next stat:
1.45 paf 454: // directory update of NTFS hard links performed on open.
1.33 paf 455: // ex:
456: // a.html:^test[] and b.html hardlink to a.html
457: // user inserts ! before ^test in a.html
1.126 paf 458: // directory entry of b.html in NTFS not updated at once,
1.35 paf 459: // they delay update till open, so we would receive "!^test[" string
460: // if would do stat, next open.
1.123 paf 461: // later: it seems, even this does not help sometimes
1.98 paf 462: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 463: try {
464: if(lock_shared_blocking(f)!=0)
1.126 paf 465: throw Exception("file.lock",
1.123 paf 466: &file_spec,
467: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 468: strerror(errno), errno, fname);
1.123 paf 469:
1.124 paf 470: struct stat finfo;
471: if(stat(fname, &finfo)!=0)
472: throw Exception("file.missing", // hardly possible: we just opened it OK
473: &file_spec,
474: "stat failed: %s (%d), actual filename '%s'",
1.126 paf 475: strerror(errno), errno, fname);
1.124 paf 476:
1.140 paf 477: #ifdef PA_SAFE_MODE
478: if(finfo.st_uid/*foreign?*/!=geteuid()
479: && finfo.st_gid/*foreign?*/!=getegid())
1.126 paf 480: throw Exception("parser.runtime",
481: &file_spec,
1.140 paf 482: "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 483: fname);
1.105 paf 484: #endif
1.32 paf 485:
1.126 paf 486: action(pool, finfo, f, file_spec, fname, as_text, context);
1.123 paf 487: } catch(...) {
1.126 paf 488: unlock(f);close(f);
1.123 paf 489: if(fail_on_read_problem)
490: /*re*/throw;
491: return false;
492: }
1.87 paf 493:
1.126 paf 494: unlock(f);close(f);
1.72 parser 495: return true;
1.118 paf 496: } else {
497: if(fail_on_read_problem)
1.126 paf 498: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 499: &file_spec,
1.123 paf 500: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 501: action_name, strerror(errno), errno, fname);
1.118 paf 502: return false;
503: }
1.8 paf 504: }
505:
1.63 parser 506: static void create_dir_for_file(const String& file_spec) {
507: size_t pos_after=1;
508: int pos_before;
509: while((pos_before=file_spec.pos("/", 1, pos_after))>=0) {
1.126 paf 510: mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775);
1.63 parser 511: pos_after=pos_before+1;
512: }
513: }
514:
1.98 paf 515: bool file_write_action_under_lock(
1.28 paf 516: const String& file_spec,
1.126 paf 517: const char* action_name, File_write_action action, void *context,
518: bool as_text,
519: bool do_append,
520: bool do_block,
1.110 paf 521: bool fail_on_lock_problem) {
1.126 paf 522: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.28 paf 523: int f;
1.80 paf 524: if(access(fname, W_OK)!=0) // no
1.126 paf 525: create_dir_for_file(file_spec);
1.50 paf 526:
1.80 paf 527: if((f=open(fname,
528: O_CREAT|O_RDWR
529: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 530: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.99 paf 531: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.126 paf 532: Exception e("file.lock",
1.110 paf 533: &file_spec,
534: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 535: strerror(errno), errno, fname);
536: close(f);
1.110 paf 537: if(fail_on_lock_problem)
538: throw e;
1.98 paf 539: return false;
540: }
1.96 paf 541:
542: try {
1.126 paf 543: action(f, context);
1.96 paf 544: } catch(...) {
1.138 paf 545: #ifdef HAVE_FTRUNCATE
1.104 paf 546: if(!do_append)
1.125 paf 547: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 548: #endif
1.126 paf 549: unlock(f);close(f);
1.96 paf 550: /*re*/throw;
551: }
1.80 paf 552:
1.138 paf 553: #ifdef HAVE_FTRUNCATE
1.104 paf 554: if(!do_append)
1.125 paf 555: 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 556: #endif
1.126 paf 557: unlock(f);close(f);
1.98 paf 558: return true;
1.80 paf 559: } else
1.126 paf 560: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 561: &file_spec,
1.96 paf 562: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 563: action_name, strerror(errno), errno, fname);
1.96 paf 564: // here should be nothing, see rethrow above
565: }
566:
567: #ifndef DOXYGEN
568: struct File_write_action_info {
569: const void *data; size_t size;
1.126 paf 570: };
1.96 paf 571: #endif
572: static void file_write_action(int f, void *context) {
1.126 paf 573: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.113 paf 574: if(info.size) {
1.126 paf 575: int written=write(f, info.data, info.size);
1.116 paf 576: if(written<0)
1.126 paf 577: throw Exception(0,
578: 0,
579: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 580: }
1.96 paf 581: }
582: void file_write(
583: const String& file_spec,
584: const void *data, size_t size,
1.126 paf 585: bool as_text,
1.96 paf 586: bool do_append) {
1.126 paf 587: File_write_action_info info={data, size};
1.98 paf 588: file_write_action_under_lock(
1.96 paf 589: file_spec,
1.126 paf 590: "write", file_write_action, &info,
591: as_text,
592: do_append);
1.30 paf 593: }
594:
1.63 parser 595: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 596: static void rmdir(const String& file_spec, size_t pos_after) {
597: int pos_before;
598: if((pos_before=file_spec.pos("/", 1, pos_after))>=0)
1.126 paf 599: rmdir(file_spec, pos_before+1);
1.50 paf 600:
1.126 paf 601: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_FILE_SPEC));
1.50 paf 602: }
1.95 paf 603: bool file_delete(const String& file_spec, bool fail_on_read_problem) {
1.126 paf 604: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.54 parser 605: if(unlink(fname)!=0)
1.93 paf 606: if(fail_on_read_problem)
1.126 paf 607: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 608: &file_spec,
609: "unlink failed: %s (%d), actual filename '%s'",
1.126 paf 610: strerror(errno), errno, fname);
1.93 paf 611: else
612: return false;
1.50 paf 613:
1.126 paf 614: rmdir(file_spec, 1);
1.93 paf 615: return true;
1.60 parser 616: }
1.95 paf 617: void file_move(const String& old_spec, const String& new_spec) {
1.126 paf 618: const char* old_spec_cstr=old_spec.cstr(String::UL_FILE_SPEC);
619: const char* new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC);
1.63 parser 620:
1.126 paf 621: create_dir_for_file(new_spec);
1.63 parser 622:
1.60 parser 623: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 624: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 625: &old_spec,
626: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.126 paf 627: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 628:
1.126 paf 629: rmdir(old_spec, 1);
1.31 paf 630: }
631:
1.51 paf 632:
1.126 paf 633: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 634: struct stat lfinfo;
635: bool result=stat(fname, &lfinfo)==0;
636: if(afinfo)
637: *afinfo=lfinfo;
638: return result;
1.119 paf 639: }
640:
641: bool entry_exists(const String& file_spec) {
1.126 paf 642: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
643: return entry_exists(fname, 0);
1.118 paf 644: }
645:
1.51 paf 646: static bool entry_readable(const String& file_spec, bool need_dir) {
1.126 paf 647: char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.120 paf 648: if(need_dir) {
1.126 paf 649: size_t size=strlen(fname);
1.120 paf 650: while(size) {
1.126 paf 651: char c=fname[size-1];
1.120 paf 652: if(c=='/' || c=='\\')
653: fname[--size]=0;
654: else
655: break;
656: }
657: }
1.51 paf 658: struct stat finfo;
1.118 paf 659: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 660: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 661: return is_dir==need_dir;
662: }
663: return false;
664: }
1.31 paf 665: bool file_readable(const String& file_spec) {
1.126 paf 666: return entry_readable(file_spec, false);
1.51 paf 667: }
668: bool dir_readable(const String& file_spec) {
1.126 paf 669: return entry_readable(file_spec, true);
1.65 parser 670: }
671: String *file_readable(const String& path, const String& name) {
1.126 paf 672: String *result=new(path.pool()) String(path);
673: *result << "/";
1.65 parser 674: *result << name;
675: return file_readable(*result)?result:0;
1.43 paf 676: }
677: bool file_executable(const String& file_spec) {
1.64 parser 678: return access(file_spec.cstr(String::UL_FILE_SPEC), X_OK)==0;
1.44 paf 679: }
680:
1.64 parser 681: bool file_stat(const String& file_spec,
1.58 parser 682: size_t& rsize,
1.126 paf 683: time_t& ratime,
684: time_t& rmtime,
685: time_t& rctime,
1.64 parser 686: bool fail_on_read_problem) {
1.126 paf 687: Pool& pool=file_spec.pool();
688: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.44 paf 689: struct stat finfo;
690: if(stat(fname, &finfo)!=0)
1.64 parser 691: if(fail_on_read_problem)
1.126 paf 692: throw Exception("file.missing",
1.67 parser 693: &file_spec,
694: "getting file size failed: %s (%d), real filename '%s'",
1.126 paf 695: strerror(errno), errno, fname);
1.64 parser 696: else
697: return false;
1.58 parser 698: rsize=finfo.st_size;
699: ratime=finfo.st_atime;
700: rmtime=finfo.st_mtime;
701: rctime=finfo.st_ctime;
1.64 parser 702: return true;
1.18 paf 703: }
704:
1.126 paf 705: char* getrow(char* *row_ref, char delim) {
706: char* result=*row_ref;
1.8 paf 707: if(result) {
1.126 paf 708: *row_ref=strchr(result, delim);
1.8 paf 709: if(*row_ref)
710: *((*row_ref)++)=0;
711: else if(!*result)
712: return 0;
713: }
714: return result;
715: }
716:
1.126 paf 717: char* lsplit(char* string, char delim) {
1.23 paf 718: if(string) {
1.126 paf 719: char* v=strchr(string, delim);
1.8 paf 720: if(v) {
721: *v=0;
722: return v+1;
723: }
724: }
725: return 0;
726: }
727:
1.126 paf 728: char* lsplit(char* *string_ref, char delim) {
729: char* result=*string_ref;
730: char* next=lsplit(*string_ref, delim);
1.8 paf 731: *string_ref=next;
732: return result;
1.9 paf 733: }
734:
1.126 paf 735: char* rsplit(char* string, char delim) {
1.18 paf 736: if(string) {
1.126 paf 737: char* v=strrchr(string, delim);
1.18 paf 738: if(v) {
1.9 paf 739: *v=0;
740: return v+1;
741: }
742: }
743: return NULL;
1.10 paf 744: }
745:
1.37 paf 746: /// @todo less stupid type detection
1.126 paf 747: char* format(Pool& pool, double value, char* fmt) {
748: char local_buf[MAX_NUMBER];
1.108 paf 749: size_t size;
750:
1.10 paf 751: if(fmt)
752: if(strpbrk(fmt, "diouxX"))
753: if(strpbrk(fmt, "ouxX"))
1.126 paf 754: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 755: else
1.126 paf 756: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 757: else
1.126 paf 758: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 759: else
1.126 paf 760: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 761:
1.126 paf 762: char* pool_buf=(char *)pool.malloc(size+1, 4);
763: memcpy(pool_buf, local_buf, size+1);
1.108 paf 764: return pool_buf;
1.12 paf 765: }
766:
1.36 paf 767: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 768: #ifdef WIN32
769: do{
1.126 paf 770: int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout);
1.12 paf 771: if(chunk_written<=0)
772: break;
773: size-=chunk_written;
1.36 paf 774: buf=((const char*)buf)+chunk_written;
1.126 paf 775: } while(size>0);
1.12 paf 776:
777: return size;
778: #else
1.126 paf 779: return fwrite(buf, 1, size, stdout);
1.12 paf 780: #endif
1.2 paf 781: }
1.14 paf 782:
1.126 paf 783: char* unescape_chars(Pool& pool, const char* cp, int len) {
784: char* s=(char *)pool.malloc(len + 1, 5);
1.14 paf 785: enum EscapeState {
1.33 paf 786: EscapeRest,
787: EscapeFirst,
1.14 paf 788: EscapeSecond
789: } escapeState=EscapeRest;
790: int escapedValue=0;
791: int srcPos=0;
792: int dstPos=0;
793: while(srcPos < len) {
1.126 paf 794: int ch=cp[srcPos];
1.14 paf 795: switch(escapeState) {
796: case EscapeRest:
797: if(ch=='%') {
798: escapeState=EscapeFirst;
799: } else if(ch=='+') {
1.126 paf 800: s[dstPos++]=' ';
1.14 paf 801: } else {
802: s[dstPos++]=ch;
803: }
804: break;
805: case EscapeFirst:
806: escapedValue=hex_value[ch] << 4;
807: escapeState=EscapeSecond;
808: break;
809: case EscapeSecond:
1.126 paf 810: escapedValue +=hex_value[ch];
1.14 paf 811: s[dstPos++]=escapedValue;
812: escapeState=EscapeRest;
813: break;
814: }
1.126 paf 815: srcPos++;
1.14 paf 816: }
817: s[dstPos]=0;
818: return s;
1.24 paf 819: }
820:
821: #ifdef WIN32
1.126 paf 822: void back_slashes_to_slashes(char* s) {
1.24 paf 823: if(s)
824: for(; *s; s++)
825: if(*s=='\\')
1.126 paf 826: *s='/';
1.24 paf 827: }
1.42 paf 828: /*
1.126 paf 829: void slashes_to_back_slashes(char* s) {
1.42 paf 830: if(s)
831: for(; *s; s++)
832: if(*s=='/')
1.126 paf 833: *s='\\';
1.42 paf 834: }
835: */
1.24 paf 836: #endif
1.41 paf 837:
1.126 paf 838: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 839: while(true) {
840: if(!(*s1)) {
841: if(!(*s2))
842: return true;
843: else
844: return !strict;
845: } else if(!(*s2))
846: return !strict;
847: if(isalpha(*s1)) {
848: if(tolower(*s1) !=tolower(*s2))
849: return false;
850: } else if((*s1) !=(*s2))
851: return false;
1.126 paf 852: s1++;
853: s2++;
1.41 paf 854: }
1.57 parser 855: }
856:
1.84 paf 857: static bool isLeap(int year) {
1.57 parser 858: return !(
859: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 860: );
1.57 parser 861: }
862:
863: int getMonthDays(int year, int month) {
864: int monthDays[]={
1.126 paf 865: 31,
866: isLeap(year) ? 29 : 28,
867: 31,
868: 30,
869: 31,
870: 30,
871: 31,
872: 31,
873: 30,
874: 31,
875: 30,
1.57 parser 876: 31
1.126 paf 877: };
878: return monthDays[month];
1.41 paf 879: }
1.69 parser 880:
1.126 paf 881: void remove_crlf(char* start, char* end) {
882: for(char* p=start; p<end; p++)
1.69 parser 883: switch(*p) {
1.126 paf 884: case '\n': *p='|'; break;
885: case '\r': *p=' '; break;
1.69 parser 886: }
1.91 paf 887: }
888:
889:
890: /// must be last in this file
891: #undef vsnprintf
1.126 paf 892: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 893: if(!s)
894: return 0;
895:
896: int r;
897: // note: on win32& maybe somewhere else
898: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
899: --s;
900: #if _MSC_VER
901: /*
902: win32:
903: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
904:
905: if the number of bytes to write exceeds buffer, then count bytes are written and –1 is returned
906: */
1.126 paf 907: r=_vsnprintf(b, s, f, l);
1.91 paf 908: if(r<0)
909: r=s;
910: #else
1.126 paf 911: r=vsnprintf(b, s, f, l);
1.91 paf 912: /*
913: solaris:
914: man vsnprintf
915:
916: The snprintf() function returns the number of characters
917: formatted, that is, the number of characters that would have
918: been written to the buffer if it were large enough. If the
919: value of n is 0 on a call to snprintf(), an unspecified
920: value less than 1 is returned.
921: */
922:
923: if(r<0)
924: r=0;
925: else if(r>s)
926: r=s;
927: #endif
928: b[r]=0;
929: return r;
930: }
931:
1.126 paf 932: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 933: va_list l;
1.126 paf 934: va_start(l, f);
935: int r=__vsnprintf(b, s, f, l);
936: va_end(l);
1.91 paf 937: return r;
1.98 paf 938: }
939:
940: int pa_sleep(unsigned long secs, unsigned long usecs) {
1.126 paf 941: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
1.98 paf 942:
943: #ifdef WIN32
1.126 paf 944: Sleep(secs * 1000 + usecs / 1000);
1.98 paf 945: return 0;
946: #else
947: struct timeval t;
948: t.tv_sec = secs;
949: t.tv_usec = usecs;
1.126 paf 950: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
1.98 paf 951: #endif
1.135 paf 952: }
953:
954:
955: // attributed meaning
956:
957: static size_t date_attribute(const VDate& vdate, char *buf, size_t buf_size) {
958: const char month_names[12][4]={
959: "Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec"};
960: const char days[7][4]={
961: "Sun","Mon","Tue","Wed","Thu","Fri","Sat"};
962:
963: time_t when=vdate.get_time();
964: struct tm *tms=gmtime(&when);
965: if(!tms)
966: throw Exception(0,
967: 0,
968: "bad time in attribute value (seconds from epoch=%ld)", when);
969: return snprintf(buf, MAX_STRING, "%s, %.2d-%s-%.4d %.2d:%.2d:%.2d GMT",
970: days[tms->tm_wday],
971: tms->tm_mday,month_names[tms->tm_mon],tms->tm_year+1900,
972: tms->tm_hour,tms->tm_min,tms->tm_sec);
973: }
974: static void append_attribute_meaning(String& result,
1.136 paf 975: Value& value, String::Untaint_lang lang, bool forced) {
1.135 paf 976: if(const String *string=value.get_string())
1.136 paf 977: result.append(string->join_chains(result.pool(), 0), lang, forced);
1.135 paf 978: else
979: if(Value *vdate=value.as(VDATE_TYPE, false)) {
980: char *buf=(char *)result.malloc(MAX_STRING);
981: size_t size=date_attribute(*static_cast<VDate *>(vdate),
982: buf, MAX_STRING);
983:
984: result.APPEND_CLEAN(buf, size, "converted from date", 0);
985: } else
986: throw Exception("parser.runtime",
987: &result,
988: "trying to append here neither string nor date (%s)",
989: value.type());
990: }
991: #ifndef DOXYGEN
992: struct Attributed_meaning_info {
993: String *header; // header line being constructed
994: String::Untaint_lang lang; // language in which to append to that line
1.136 paf 995: bool forced; // do they force that lang?
1.135 paf 996: };
997: #endif
998: static void append_attribute_subattribute(const Hash::Key& akey, Hash::Val *avalue,
999: void *info) {
1000: if(akey==VALUE_NAME)
1001: return;
1002:
1003: Attributed_meaning_info& ami=*static_cast<Attributed_meaning_info *>(info);
1004:
1005: // ...; charset=windows1251
1006: *ami.header << "; ";
1.136 paf 1007: ami.header->append(akey, ami.lang, ami.forced);
1.135 paf 1008: *ami.header << "=";
1.136 paf 1009: append_attribute_meaning(*ami.header, *static_cast<Value *>(avalue), ami.lang, ami.forced);
1.135 paf 1010: }
1011: const String& attributed_meaning_to_string(Value& meaning,
1.136 paf 1012: String::Untaint_lang lang, bool forced) {
1.135 paf 1013: String &result=*new(meaning.pool()) String(meaning.pool());
1014: if(Hash *hash=meaning.get_hash(0)) {
1015: // $value(value) $subattribute(subattribute value)
1016: if(Value *value=static_cast<Value *>(hash->get(*value_name)))
1.136 paf 1017: append_attribute_meaning(result, *value, lang, forced);
1.135 paf 1018:
1019: Attributed_meaning_info attributed_meaning_info={&result, lang};
1020: hash->for_each(append_attribute_subattribute, &attributed_meaning_info);
1021: } else // result value
1.136 paf 1022: append_attribute_meaning(result, meaning, lang, forced);
1.135 paf 1023:
1024: return result;
1.74 parser 1025: }
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