Annotation of parser3/src/main/pa_common.C, revision 1.137
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:
1.137 ! paf 8: static const char* IDENT_COMMON_C="$Date: 2002/12/02 10:07:39 $";
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:
1.93 paf 32: // locking constants
33:
1.99 paf 34: #ifdef HAVE_FLOCK
35:
36: static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); }
37: static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); }
38: static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); }
39: static int unlock(int fd) { return flock(fd, LOCK_UN); }
40:
1.98 paf 41: #else
1.99 paf 42: #ifdef HAVE__LOCKING
1.98 paf 43:
1.126 paf 44: #define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1)
1.99 paf 45: static int lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); }
46: static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); }
47: static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); }
48: static int unlock(int fd) { FLOCK(_LK_UNLCK); }
1.93 paf 49:
1.99 paf 50: #else
51: #ifdef HAVE_FCNTL
1.93 paf 52:
1.126 paf 53: #define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls)
1.99 paf 54: static int lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); }
55: static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); }
56: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); }
57: static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); }
1.93 paf 58:
59: #else
60: #ifdef HAVE_LOCKF
1.99 paf 61:
1.126 paf 62: #define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1)
1.99 paf 63: static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking
64: static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); }
65: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); }
66: static int unlock(int fd) { FLOCK(F_TLOCK); }
67:
1.93 paf 68: #else
1.99 paf 69:
70: #error unable to find file locking func
71:
72: #endif
1.93 paf 73: #endif
74: #endif
75: #endif
76:
1.127 paf 77: #define DEFAULT_USER_AGENT "parser3"
78:
79:
1.126 paf 80: void fix_line_breaks(char* buf, size_t& size) {
1.87 paf 81: //_asm int 3;
1.126 paf 82: const char* const eob=buf+size;
83: char* dest=buf;
1.72 parser 84: // fix DOS: \r\n -> \n
85: // fix Macintosh: \r -> \n
1.126 paf 86: char* bol=buf;
1.137 ! paf 87: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 88: size_t len=eol-bol;
89: if(dest!=bol)
1.126 paf 90: memcpy(dest, bol, len);
1.72 parser 91: dest+=len;
1.126 paf 92: *dest++='\n';
1.72 parser 93:
1.126 paf 94: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 95: bol=eol+2;
1.126 paf 96: size--;
97: } else // \r, not \n = Macintosh
1.72 parser 98: bol=eol+1;
99: }
100: // last piece without \r, including terminating 0
101: if(dest!=bol)
1.126 paf 102: memcpy(dest, bol, eob-bol);
1.72 parser 103: }
1.18 paf 104:
1.126 paf 105: char* file_read_text(Pool& pool, const String& file_spec,
106: bool fail_on_read_problem,
107: Hash *params, Hash** out_fields) {
1.72 parser 108: void *result; size_t size;
1.126 paf 109: return file_read(pool, file_spec, result, size, true, params, out_fields, fail_on_read_problem)?(char *)result:0;
110: }
111:
112: //http request stuff
113: /* ************************ http stuff *********************** */
114:
115: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
116: memset(addr, 0, sizeof(*addr));
117: addr->sin_family=AF_INET;
118: addr->sin_port=htons(port);
119: if(host) {
120: if(struct hostent *hostIP=gethostbyname(host))
121: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
122: else
123: return false;
124: } else
125: addr->sin_addr.s_addr=INADDR_ANY;
126: return true;
127: }
128:
129: static void http_read_response(String& response, int sock){
1.130 paf 130: bool check_status=true;
131: ssize_t EOLat=0;
1.126 paf 132: while(true) {
133: char *buf=(char *)response.pool().malloc(MAX_STRING);
134: ssize_t size=recv(sock, buf, MAX_STRING, 0);
135: if(size<=0)
136: break;
1.130 paf 137: response.APPEND_TAINTED(buf, size, "remote HTTP server response", 0);
138: if(check_status && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response
139: check_status=false;
140:
141: const String& status_line=response.mid(0, (size_t)EOLat);
142: Array astatus(response.pool());
143: size_t pos_after_ref=0; status_line.split(astatus, &pos_after_ref, " ", 1);
144: const String* status_code=astatus.get_string(1);
145:
146: if(!status_code || *status_code!="200")
147: throw Exception("http.status",
148: status_code,
149: "invalid HTTP response status");
150: }
1.126 paf 151: }
1.130 paf 152: if(check_status)
153: throw Exception("http.response",
154: 0,
155: "bad response from host - no status found (size=%lu)", response.size());
1.126 paf 156: }
157:
158: /* ********************** request *************************** */
159:
160: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
161: # define WE_CAN_USE_ALARM
162: #endif
163:
164: #ifdef WE_CAN_USE_ALARM
165: static sigjmp_buf timeout_env;
166: static void timeout_handler(int sig){
167: siglongjmp(timeout_env, 1);
168: }
169: #endif
170:
171: static void http_request(String& response,
172: const String *origin_string,
173: const char* host, int port,
174: const char* request,
175: int timeout){
176: if(!host)
177: throw Exception("http.host",
178: origin_string,
179: "zero hostname"); //never
180:
181: #ifdef WE_CAN_USE_ALARM
182: signal(SIGALRM, timeout_handler);
183: #endif
184: int sock=-1;
185: try {
186: #ifdef WE_CAN_USE_ALARM
187: if(sigsetjmp(timeout_env, 1))
188: throw Exception("http.timeout",
189: origin_string,
190: "timeout occured while retrieving document");
191: else {
192: alarm(timeout);
193: #endif
194: struct sockaddr_in dest;
195:
196: if(!set_addr(&dest, host, port))
197: throw Exception("http.host",
198: origin_string,
1.127 paf 199: "can not resolve hostname \"%s\"", host);
1.126 paf 200:
201: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
202: throw Exception("http.connect",
203: origin_string,
1.127 paf 204: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 205: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
206: throw Exception("http.connect",
207: origin_string,
1.127 paf 208: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 209: size_t request_size=strlen(request);
210: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
211: throw Exception("http.connect",
212: origin_string,
1.127 paf 213: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 214:
215: http_read_response(response, sock);
216: if(sock>=0)
217: closesocket(sock);
218: #ifdef WE_CAN_USE_ALARM
219: }
220: #endif
221: } catch(...) {
222: if(sock>=0)
223: closesocket(sock);
224: #ifdef WE_CAN_USE_ALARM
225: alarm(0);
226: #endif
227: /*re*/throw;
228: }
229: }
230:
1.127 paf 231: #ifndef DOXYGEN
232: struct Http_pass_header_info {
233: String* request;
234: bool user_agent_specified;
235: };
236: #endif
237: static void http_pass_header(const Hash::Key& key, Hash::Val *value, void *info)
1.126 paf 238: {
1.127 paf 239: Http_pass_header_info& i=*static_cast<Http_pass_header_info *>(info);
240: Pool& pool=i.request->pool();
241:
1.135 paf 242: *(i.request)<<key<<": "
1.136 paf 243: << attributed_meaning_to_string(*static_cast<Value *>(value), String::UL_HTTP_HEADER, false)
1.135 paf 244: <<"\n";
245:
1.127 paf 246: if(key.change_case(pool, String::CC_UPPER)=="USER-AGENT")
247: i.user_agent_specified=true;
1.126 paf 248: }
249: static void file_read_http(Pool& pool, const String& file_spec,
250: void*& data, size_t& data_size,
1.127 paf 251: Hash *options=0, Hash** out_fields=0) {
1.126 paf 252: char host[MAX_STRING];
1.129 paf 253: const char* uri;
1.126 paf 254: int port;
255: const char* method="GET";
256: int timeout=2;
1.127 paf 257: Value *vheaders=0;
1.126 paf 258:
1.133 paf 259: String connect_string(pool);
260: // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one
261: connect_string.append(file_spec, String::UL_URI); // tainted pieces -> URI pieces
262:
263: char* connect_string_cstr=connect_string.cstr(String::UL_UNSPECIFIED);
264: if(strncmp(connect_string_cstr, "http://", 7)!=0)
1.126 paf 265: throw Exception(0,
1.133 paf 266: &connect_string,
1.126 paf 267: "does not start with http://"); //never
1.133 paf 268: connect_string_cstr+=7;
1.126 paf 269:
1.133 paf 270: strncpy(host, connect_string_cstr, sizeof(host)-1); host[sizeof(host)-1]=0;
1.126 paf 271: char* host_uri=lsplit(host, '/');
1.133 paf 272: uri=host_uri?connect_string_cstr+(host_uri-1-host):"/";
1.126 paf 273: char* port_cstr=lsplit(host, ':');
274: char* error_pos=0;
275: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
276:
1.127 paf 277: if(options) {
278: int valid_options=0;
279: if(Value *vmethod=static_cast<Value *>(options->get(*http_method_name))) {
280: valid_options++;
281: method=vmethod->as_string().cstr();
282: }
283: if(Value *vtimeout=static_cast<Value *>(options->get(*http_timeout_name))) {
284: valid_options++;
285: timeout=vtimeout->as_int();
286: }
287: if(vheaders=static_cast<Value *>(options->get(*http_headers_name))) {
288: valid_options++;
289: }
290: if(valid_options!=options->size())
291: throw Exception("parser.runtime",
292: 0,
293: "invalid option passed");
1.133 paf 294: }
1.126 paf 295:
296: //making request
1.127 paf 297: String request(pool);
1.126 paf 298: request<< method <<" "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";
1.127 paf 299: bool user_agent_specified=false;
300: if(vheaders && !vheaders->is_string()) { // allow empty
1.133 paf 301: if(Hash *headers=vheaders->get_hash(&connect_string)) {
1.127 paf 302: Http_pass_header_info info={&request};
303: headers->for_each(http_pass_header, &info);
304: user_agent_specified=info.user_agent_specified;
305: } else
306: throw Exception("parser.runtime",
1.133 paf 307: &connect_string,
1.127 paf 308: "headers param must be hash");
309: };
310: if(!user_agent_specified) // defaulting
311: request << "user-agent: " DEFAULT_USER_AGENT "\n";
1.126 paf 312: request<<"\n";
313:
314: //sending request
315: String response(pool);
316: http_request(response,
1.133 paf 317: &connect_string, host, port, request.cstr(String::UL_UNSPECIFIED), timeout);
1.126 paf 318:
319: //processing results
320: int pos=response.pos("\r\n\r\n", 4);
321: if(pos<1){
322: throw Exception("http.response",
1.133 paf 323: &connect_string,
1.126 paf 324: "bad response from host - no headers found");
325: }
326: String header_block=response.mid(0, pos);
327: String body=response.mid(pos+4, response.size());
328:
329: Array aheaders(pool);
330: Hash& headers=*new(pool) Hash(pool);
331: size_t pos_after_ref=0;
332: header_block.split(aheaders, &pos_after_ref, "\r\n", 2);
333:
334: //processing status code
335: const String *status_line=aheaders.get_string(0);
336: if(!status_line){
337: throw Exception("http.response",
1.133 paf 338: &connect_string,
1.126 paf 339: "bad response from host - no status line ");
340: }
341: //processing headers
342: for(int i=1;i<aheaders.size();i++) {
343: if(const String *line=aheaders.get_string(i)) {
344: pos=line->pos(": ", 2);
345: if(pos<1)
346: throw Exception("http.response",
1.133 paf 347: &connect_string,
1.126 paf 348: "bad response from host - bad header \"%s\"", line->cstr());
349:
1.128 paf 350: headers.put(
351: line->mid(0, pos).change_case(pool, String::CC_UPPER),
352: new(pool) VString(line->mid(pos+2, line->size())));
1.126 paf 353: } else
354: throw Exception("http.response",
1.133 paf 355: &connect_string,
1.126 paf 356: "bad response from host - bad headers \"%s\"", header_block.cstr());
357: }
358:
359: // output response
360: data=body.cstr(); data_size=body.size();
1.130 paf 361: if(out_fields)
1.126 paf 362: *out_fields=&headers;
1.34 paf 363: }
1.123 paf 364:
365: #ifndef DOXYGEN
366: struct File_read_action_info {
367: void **data; size_t *data_size;
1.126 paf 368: };
1.123 paf 369: #endif
1.126 paf 370: static void file_read_action(Pool& pool,
371: struct stat& finfo,
1.123 paf 372: int f,
1.126 paf 373: const String& file_spec, const char* fname, bool as_text,
1.123 paf 374: void *context) {
1.126 paf 375: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 376: if(size_t to_read_size=(size_t)finfo.st_size) {
1.126 paf 377: *info.data=pool.malloc(to_read_size+(as_text?1:0), 3);
378: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 379:
380: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 381: throw Exception(0,
1.123 paf 382: &file_spec,
383: "read failed: actually read %lu bytes count not in [0..%lu] valid range",
1.126 paf 384: *info.data_size, to_read_size);
1.123 paf 385: } else { // empty file
386: if(as_text) {
1.126 paf 387: *info.data=pool.malloc(1);
1.123 paf 388: *(char*)(*info.data)=0;
389: } else
390: *info.data=0;
391: *info.data_size=0;
392: return;
393: }
1.126 paf 394: }
395: bool file_read(Pool& pool, const String& file_spec,
396: void*& data, size_t& data_size,
397: bool as_text, Hash *params, Hash** out_fields,
398: bool fail_on_read_problem) {
399: bool result;
400: if(file_spec.starts_with("http://", 7)) {
401: // fail on read problem
402: file_read_http(pool, file_spec, data, data_size, params, out_fields);
403: result=true;
404: } else {
405: File_read_action_info info={&data, &data_size};
406: result=file_read_action_under_lock(pool, file_spec,
407: "read", file_read_action, &info,
408: as_text, fail_on_read_problem);
409: }
1.123 paf 410:
1.126 paf 411: if(result && as_text) {
1.131 paf 412: // UTF-8 signature: EF BB BF
413: if(data_size>=3) {
414: char *in=(char *)data;
415: if((in[0] == '\xEF') && (in[1] == '\xBB') &&
416: (in[2] == '\xBF')) {
417: data=in+3; data_size-=3;// skip prefix
418: }
419: }
420:
1.126 paf 421: fix_line_breaks((char *)(data), data_size);
1.123 paf 422: // note: after fixing
1.126 paf 423: ((char*&)(data))[data_size]=0;
1.123 paf 424: }
1.126 paf 425:
426: return result;
1.123 paf 427: }
428:
429: bool file_read_action_under_lock(Pool& pool, const String& file_spec,
1.126 paf 430: const char* action_name, File_read_action action, void *context,
431: bool as_text,
1.123 paf 432: bool fail_on_read_problem) {
1.126 paf 433: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.33 paf 434: int f;
435:
436: // first open, next stat:
1.45 paf 437: // directory update of NTFS hard links performed on open.
1.33 paf 438: // ex:
439: // a.html:^test[] and b.html hardlink to a.html
440: // user inserts ! before ^test in a.html
1.126 paf 441: // directory entry of b.html in NTFS not updated at once,
1.35 paf 442: // they delay update till open, so we would receive "!^test[" string
443: // if would do stat, next open.
1.123 paf 444: // later: it seems, even this does not help sometimes
1.98 paf 445: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 446: try {
447: if(lock_shared_blocking(f)!=0)
1.126 paf 448: throw Exception("file.lock",
1.123 paf 449: &file_spec,
450: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 451: strerror(errno), errno, fname);
1.123 paf 452:
1.124 paf 453: struct stat finfo;
454: if(stat(fname, &finfo)!=0)
455: throw Exception("file.missing", // hardly possible: we just opened it OK
456: &file_spec,
457: "stat failed: %s (%d), actual filename '%s'",
1.126 paf 458: strerror(errno), errno, fname);
1.124 paf 459:
1.105 paf 460: #ifdef NO_FOREIGN_GROUP_FILES
1.132 paf 461: if(finfo.st_gid/*foreign?*/!=getegid())
1.126 paf 462: throw Exception("parser.runtime",
463: &file_spec,
1.123 paf 464: "parser reading files of foreign group disabled [recompile parser without --disable-foreign-group-files configure option], actual filename '%s'",
1.126 paf 465: fname);
1.105 paf 466: #endif
1.32 paf 467:
1.126 paf 468: action(pool, finfo, f, file_spec, fname, as_text, context);
1.123 paf 469: } catch(...) {
1.126 paf 470: unlock(f);close(f);
1.123 paf 471: if(fail_on_read_problem)
472: /*re*/throw;
473: return false;
474: }
1.87 paf 475:
1.126 paf 476: unlock(f);close(f);
1.72 parser 477: return true;
1.118 paf 478: } else {
479: if(fail_on_read_problem)
1.126 paf 480: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 481: &file_spec,
1.123 paf 482: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 483: action_name, strerror(errno), errno, fname);
1.118 paf 484: return false;
485: }
1.8 paf 486: }
487:
1.63 parser 488: static void create_dir_for_file(const String& file_spec) {
489: size_t pos_after=1;
490: int pos_before;
491: while((pos_before=file_spec.pos("/", 1, pos_after))>=0) {
1.126 paf 492: mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775);
1.63 parser 493: pos_after=pos_before+1;
494: }
495: }
496:
1.98 paf 497: bool file_write_action_under_lock(
1.28 paf 498: const String& file_spec,
1.126 paf 499: const char* action_name, File_write_action action, void *context,
500: bool as_text,
501: bool do_append,
502: bool do_block,
1.110 paf 503: bool fail_on_lock_problem) {
1.126 paf 504: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.28 paf 505: int f;
1.80 paf 506: if(access(fname, W_OK)!=0) // no
1.126 paf 507: create_dir_for_file(file_spec);
1.50 paf 508:
1.80 paf 509: if((f=open(fname,
510: O_CREAT|O_RDWR
511: |(as_text?_O_TEXT:_O_BINARY)
1.122 paf 512: |(do_append?O_APPEND:0), 0664))>=0) {
1.99 paf 513: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.126 paf 514: Exception e("file.lock",
1.110 paf 515: &file_spec,
516: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 517: strerror(errno), errno, fname);
518: close(f);
1.110 paf 519: if(fail_on_lock_problem)
520: throw e;
1.98 paf 521: return false;
522: }
1.96 paf 523:
524: try {
1.126 paf 525: action(f, context);
1.96 paf 526: } catch(...) {
1.104 paf 527: if(!do_append)
1.125 paf 528: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.126 paf 529: unlock(f);close(f);
1.96 paf 530: /*re*/throw;
531: }
1.80 paf 532:
1.104 paf 533: if(!do_append)
1.125 paf 534: ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov <virtan@rotabanner.com> for discovering]
1.126 paf 535: unlock(f);close(f);
1.98 paf 536: return true;
1.80 paf 537: } else
1.126 paf 538: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 539: &file_spec,
1.96 paf 540: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 541: action_name, strerror(errno), errno, fname);
1.96 paf 542: // here should be nothing, see rethrow above
543: }
544:
545: #ifndef DOXYGEN
546: struct File_write_action_info {
547: const void *data; size_t size;
1.126 paf 548: };
1.96 paf 549: #endif
550: static void file_write_action(int f, void *context) {
1.126 paf 551: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.113 paf 552: if(info.size) {
1.126 paf 553: int written=write(f, info.data, info.size);
1.116 paf 554: if(written<0)
1.126 paf 555: throw Exception(0,
556: 0,
557: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 558: }
1.96 paf 559: }
560: void file_write(
561: const String& file_spec,
562: const void *data, size_t size,
1.126 paf 563: bool as_text,
1.96 paf 564: bool do_append) {
1.126 paf 565: File_write_action_info info={data, size};
1.98 paf 566: file_write_action_under_lock(
1.96 paf 567: file_spec,
1.126 paf 568: "write", file_write_action, &info,
569: as_text,
570: do_append);
1.30 paf 571: }
572:
1.63 parser 573: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 574: static void rmdir(const String& file_spec, size_t pos_after) {
575: int pos_before;
576: if((pos_before=file_spec.pos("/", 1, pos_after))>=0)
1.126 paf 577: rmdir(file_spec, pos_before+1);
1.50 paf 578:
1.126 paf 579: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_FILE_SPEC));
1.50 paf 580: }
1.95 paf 581: bool file_delete(const String& file_spec, bool fail_on_read_problem) {
1.126 paf 582: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.54 parser 583: if(unlink(fname)!=0)
1.93 paf 584: if(fail_on_read_problem)
1.126 paf 585: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 586: &file_spec,
587: "unlink failed: %s (%d), actual filename '%s'",
1.126 paf 588: strerror(errno), errno, fname);
1.93 paf 589: else
590: return false;
1.50 paf 591:
1.126 paf 592: rmdir(file_spec, 1);
1.93 paf 593: return true;
1.60 parser 594: }
1.95 paf 595: void file_move(const String& old_spec, const String& new_spec) {
1.126 paf 596: const char* old_spec_cstr=old_spec.cstr(String::UL_FILE_SPEC);
597: const char* new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC);
1.63 parser 598:
1.126 paf 599: create_dir_for_file(new_spec);
1.63 parser 600:
1.60 parser 601: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 602: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 603: &old_spec,
604: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.126 paf 605: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 606:
1.126 paf 607: rmdir(old_spec, 1);
1.31 paf 608: }
609:
1.51 paf 610:
1.126 paf 611: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 612: struct stat lfinfo;
613: bool result=stat(fname, &lfinfo)==0;
614: if(afinfo)
615: *afinfo=lfinfo;
616: return result;
1.119 paf 617: }
618:
619: bool entry_exists(const String& file_spec) {
1.126 paf 620: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
621: return entry_exists(fname, 0);
1.118 paf 622: }
623:
1.51 paf 624: static bool entry_readable(const String& file_spec, bool need_dir) {
1.126 paf 625: char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.120 paf 626: if(need_dir) {
1.126 paf 627: size_t size=strlen(fname);
1.120 paf 628: while(size) {
1.126 paf 629: char c=fname[size-1];
1.120 paf 630: if(c=='/' || c=='\\')
631: fname[--size]=0;
632: else
633: break;
634: }
635: }
1.51 paf 636: struct stat finfo;
1.118 paf 637: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 638: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 639: return is_dir==need_dir;
640: }
641: return false;
642: }
1.31 paf 643: bool file_readable(const String& file_spec) {
1.126 paf 644: return entry_readable(file_spec, false);
1.51 paf 645: }
646: bool dir_readable(const String& file_spec) {
1.126 paf 647: return entry_readable(file_spec, true);
1.65 parser 648: }
649: String *file_readable(const String& path, const String& name) {
1.126 paf 650: String *result=new(path.pool()) String(path);
651: *result << "/";
1.65 parser 652: *result << name;
653: return file_readable(*result)?result:0;
1.43 paf 654: }
655: bool file_executable(const String& file_spec) {
1.64 parser 656: return access(file_spec.cstr(String::UL_FILE_SPEC), X_OK)==0;
1.44 paf 657: }
658:
1.64 parser 659: bool file_stat(const String& file_spec,
1.58 parser 660: size_t& rsize,
1.126 paf 661: time_t& ratime,
662: time_t& rmtime,
663: time_t& rctime,
1.64 parser 664: bool fail_on_read_problem) {
1.126 paf 665: Pool& pool=file_spec.pool();
666: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.44 paf 667: struct stat finfo;
668: if(stat(fname, &finfo)!=0)
1.64 parser 669: if(fail_on_read_problem)
1.126 paf 670: throw Exception("file.missing",
1.67 parser 671: &file_spec,
672: "getting file size failed: %s (%d), real filename '%s'",
1.126 paf 673: strerror(errno), errno, fname);
1.64 parser 674: else
675: return false;
1.58 parser 676: rsize=finfo.st_size;
677: ratime=finfo.st_atime;
678: rmtime=finfo.st_mtime;
679: rctime=finfo.st_ctime;
1.64 parser 680: return true;
1.18 paf 681: }
682:
1.126 paf 683: char* getrow(char* *row_ref, char delim) {
684: char* result=*row_ref;
1.8 paf 685: if(result) {
1.126 paf 686: *row_ref=strchr(result, delim);
1.8 paf 687: if(*row_ref)
688: *((*row_ref)++)=0;
689: else if(!*result)
690: return 0;
691: }
692: return result;
693: }
694:
1.126 paf 695: char* lsplit(char* string, char delim) {
1.23 paf 696: if(string) {
1.126 paf 697: char* v=strchr(string, delim);
1.8 paf 698: if(v) {
699: *v=0;
700: return v+1;
701: }
702: }
703: return 0;
704: }
705:
1.126 paf 706: char* lsplit(char* *string_ref, char delim) {
707: char* result=*string_ref;
708: char* next=lsplit(*string_ref, delim);
1.8 paf 709: *string_ref=next;
710: return result;
1.9 paf 711: }
712:
1.126 paf 713: char* rsplit(char* string, char delim) {
1.18 paf 714: if(string) {
1.126 paf 715: char* v=strrchr(string, delim);
1.18 paf 716: if(v) {
1.9 paf 717: *v=0;
718: return v+1;
719: }
720: }
721: return NULL;
1.10 paf 722: }
723:
1.37 paf 724: /// @todo less stupid type detection
1.126 paf 725: char* format(Pool& pool, double value, char* fmt) {
726: char local_buf[MAX_NUMBER];
1.108 paf 727: size_t size;
728:
1.10 paf 729: if(fmt)
730: if(strpbrk(fmt, "diouxX"))
731: if(strpbrk(fmt, "ouxX"))
1.126 paf 732: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 733: else
1.126 paf 734: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 735: else
1.126 paf 736: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 737: else
1.126 paf 738: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 739:
1.126 paf 740: char* pool_buf=(char *)pool.malloc(size+1, 4);
741: memcpy(pool_buf, local_buf, size+1);
1.108 paf 742: return pool_buf;
1.12 paf 743: }
744:
1.36 paf 745: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 746: #ifdef WIN32
747: do{
1.126 paf 748: int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout);
1.12 paf 749: if(chunk_written<=0)
750: break;
751: size-=chunk_written;
1.36 paf 752: buf=((const char*)buf)+chunk_written;
1.126 paf 753: } while(size>0);
1.12 paf 754:
755: return size;
756: #else
1.126 paf 757: return fwrite(buf, 1, size, stdout);
1.12 paf 758: #endif
1.2 paf 759: }
1.14 paf 760:
1.126 paf 761: char* unescape_chars(Pool& pool, const char* cp, int len) {
762: char* s=(char *)pool.malloc(len + 1, 5);
1.14 paf 763: enum EscapeState {
1.33 paf 764: EscapeRest,
765: EscapeFirst,
1.14 paf 766: EscapeSecond
767: } escapeState=EscapeRest;
768: int escapedValue=0;
769: int srcPos=0;
770: int dstPos=0;
771: while(srcPos < len) {
1.126 paf 772: int ch=cp[srcPos];
1.14 paf 773: switch(escapeState) {
774: case EscapeRest:
775: if(ch=='%') {
776: escapeState=EscapeFirst;
777: } else if(ch=='+') {
1.126 paf 778: s[dstPos++]=' ';
1.14 paf 779: } else {
780: s[dstPos++]=ch;
781: }
782: break;
783: case EscapeFirst:
784: escapedValue=hex_value[ch] << 4;
785: escapeState=EscapeSecond;
786: break;
787: case EscapeSecond:
1.126 paf 788: escapedValue +=hex_value[ch];
1.14 paf 789: s[dstPos++]=escapedValue;
790: escapeState=EscapeRest;
791: break;
792: }
1.126 paf 793: srcPos++;
1.14 paf 794: }
795: s[dstPos]=0;
796: return s;
1.24 paf 797: }
798:
799: #ifdef WIN32
1.126 paf 800: void back_slashes_to_slashes(char* s) {
1.24 paf 801: if(s)
802: for(; *s; s++)
803: if(*s=='\\')
1.126 paf 804: *s='/';
1.24 paf 805: }
1.42 paf 806: /*
1.126 paf 807: void slashes_to_back_slashes(char* s) {
1.42 paf 808: if(s)
809: for(; *s; s++)
810: if(*s=='/')
1.126 paf 811: *s='\\';
1.42 paf 812: }
813: */
1.24 paf 814: #endif
1.41 paf 815:
1.126 paf 816: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 817: while(true) {
818: if(!(*s1)) {
819: if(!(*s2))
820: return true;
821: else
822: return !strict;
823: } else if(!(*s2))
824: return !strict;
825: if(isalpha(*s1)) {
826: if(tolower(*s1) !=tolower(*s2))
827: return false;
828: } else if((*s1) !=(*s2))
829: return false;
1.126 paf 830: s1++;
831: s2++;
1.41 paf 832: }
1.57 parser 833: }
834:
1.84 paf 835: static bool isLeap(int year) {
1.57 parser 836: return !(
837: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 838: );
1.57 parser 839: }
840:
841: int getMonthDays(int year, int month) {
842: int monthDays[]={
1.126 paf 843: 31,
844: isLeap(year) ? 29 : 28,
845: 31,
846: 30,
847: 31,
848: 30,
849: 31,
850: 31,
851: 30,
852: 31,
853: 30,
1.57 parser 854: 31
1.126 paf 855: };
856: return monthDays[month];
1.41 paf 857: }
1.69 parser 858:
1.126 paf 859: void remove_crlf(char* start, char* end) {
860: for(char* p=start; p<end; p++)
1.69 parser 861: switch(*p) {
1.126 paf 862: case '\n': *p='|'; break;
863: case '\r': *p=' '; break;
1.69 parser 864: }
1.91 paf 865: }
866:
867:
868: /// must be last in this file
869: #undef vsnprintf
1.126 paf 870: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 871: if(!s)
872: return 0;
873:
874: int r;
875: // note: on win32& maybe somewhere else
876: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
877: --s;
878: #if _MSC_VER
879: /*
880: win32:
881: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
882:
883: if the number of bytes to write exceeds buffer, then count bytes are written and –1 is returned
884: */
1.126 paf 885: r=_vsnprintf(b, s, f, l);
1.91 paf 886: if(r<0)
887: r=s;
888: #else
1.126 paf 889: r=vsnprintf(b, s, f, l);
1.91 paf 890: /*
891: solaris:
892: man vsnprintf
893:
894: The snprintf() function returns the number of characters
895: formatted, that is, the number of characters that would have
896: been written to the buffer if it were large enough. If the
897: value of n is 0 on a call to snprintf(), an unspecified
898: value less than 1 is returned.
899: */
900:
901: if(r<0)
902: r=0;
903: else if(r>s)
904: r=s;
905: #endif
906: b[r]=0;
907: return r;
908: }
909:
1.126 paf 910: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 911: va_list l;
1.126 paf 912: va_start(l, f);
913: int r=__vsnprintf(b, s, f, l);
914: va_end(l);
1.91 paf 915: return r;
1.98 paf 916: }
917:
918: int pa_sleep(unsigned long secs, unsigned long usecs) {
1.126 paf 919: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
1.98 paf 920:
921: #ifdef WIN32
1.126 paf 922: Sleep(secs * 1000 + usecs / 1000);
1.98 paf 923: return 0;
924: #else
925: struct timeval t;
926: t.tv_sec = secs;
927: t.tv_usec = usecs;
1.126 paf 928: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
1.98 paf 929: #endif
1.135 paf 930: }
931:
932:
933: // attributed meaning
934:
935: static size_t date_attribute(const VDate& vdate, char *buf, size_t buf_size) {
936: const char month_names[12][4]={
937: "Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec"};
938: const char days[7][4]={
939: "Sun","Mon","Tue","Wed","Thu","Fri","Sat"};
940:
941: time_t when=vdate.get_time();
942: struct tm *tms=gmtime(&when);
943: if(!tms)
944: throw Exception(0,
945: 0,
946: "bad time in attribute value (seconds from epoch=%ld)", when);
947: return snprintf(buf, MAX_STRING, "%s, %.2d-%s-%.4d %.2d:%.2d:%.2d GMT",
948: days[tms->tm_wday],
949: tms->tm_mday,month_names[tms->tm_mon],tms->tm_year+1900,
950: tms->tm_hour,tms->tm_min,tms->tm_sec);
951: }
952: static void append_attribute_meaning(String& result,
1.136 paf 953: Value& value, String::Untaint_lang lang, bool forced) {
1.135 paf 954: if(const String *string=value.get_string())
1.136 paf 955: result.append(string->join_chains(result.pool(), 0), lang, forced);
1.135 paf 956: else
957: if(Value *vdate=value.as(VDATE_TYPE, false)) {
958: char *buf=(char *)result.malloc(MAX_STRING);
959: size_t size=date_attribute(*static_cast<VDate *>(vdate),
960: buf, MAX_STRING);
961:
962: result.APPEND_CLEAN(buf, size, "converted from date", 0);
963: } else
964: throw Exception("parser.runtime",
965: &result,
966: "trying to append here neither string nor date (%s)",
967: value.type());
968: }
969: #ifndef DOXYGEN
970: struct Attributed_meaning_info {
971: String *header; // header line being constructed
972: String::Untaint_lang lang; // language in which to append to that line
1.136 paf 973: bool forced; // do they force that lang?
1.135 paf 974: };
975: #endif
976: static void append_attribute_subattribute(const Hash::Key& akey, Hash::Val *avalue,
977: void *info) {
978: if(akey==VALUE_NAME)
979: return;
980:
981: Attributed_meaning_info& ami=*static_cast<Attributed_meaning_info *>(info);
982:
983: // ...; charset=windows1251
984: *ami.header << "; ";
1.136 paf 985: ami.header->append(akey, ami.lang, ami.forced);
1.135 paf 986: *ami.header << "=";
1.136 paf 987: append_attribute_meaning(*ami.header, *static_cast<Value *>(avalue), ami.lang, ami.forced);
1.135 paf 988: }
989: const String& attributed_meaning_to_string(Value& meaning,
1.136 paf 990: String::Untaint_lang lang, bool forced) {
1.135 paf 991: String &result=*new(meaning.pool()) String(meaning.pool());
992: if(Hash *hash=meaning.get_hash(0)) {
993: // $value(value) $subattribute(subattribute value)
994: if(Value *value=static_cast<Value *>(hash->get(*value_name)))
1.136 paf 995: append_attribute_meaning(result, *value, lang, forced);
1.135 paf 996:
997: Attributed_meaning_info attributed_meaning_info={&result, lang};
998: hash->for_each(append_attribute_subattribute, &attributed_meaning_info);
999: } else // result value
1.136 paf 1000: append_attribute_meaning(result, meaning, lang, forced);
1.135 paf 1001:
1002: return result;
1.74 parser 1003: }
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