Annotation of parser3/src/main/pa_common.C, revision 1.143.2.3
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.143 paf 4: Copyright(c) 2001, 2003 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:
1.143.2.3! paf 8: static const char* IDENT_COMMON_C="$Date: 2003/01/24 08:19:09 $";
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"
1.143.2.3! paf 14: #include "pa_vstring.h"
1.1 paf 15:
1.98 paf 16: #ifdef WIN32
17: # include <windows.h>
1.126 paf 18: #else
19: # define closesocket close
1.98 paf 20: #endif
21:
1.93 paf 22: // some maybe-undefined constants
23:
1.82 paf 24: #ifndef _O_TEXT
25: # define _O_TEXT 0
26: #endif
27: #ifndef _O_BINARY
28: # define _O_BINARY 0
1.47 paf 29: #endif
1.80 paf 30:
1.138 paf 31: #ifdef HAVE_FTRUNCATE
32: # define PA_O_TRUNC 0
33: #else
34: # ifdef _O_TRUNC
35: # define PA_O_TRUNC _O_TRUNC
36: # else
37: # error you must have either ftruncate function or _O_TRUNC bit declared
38: # endif
39: #endif
40:
1.93 paf 41: // locking constants
42:
1.99 paf 43: #ifdef HAVE_FLOCK
44:
45: static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); }
46: static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); }
47: static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); }
48: static int unlock(int fd) { return flock(fd, LOCK_UN); }
49:
1.98 paf 50: #else
1.99 paf 51: #ifdef HAVE__LOCKING
1.98 paf 52:
1.126 paf 53: #define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1)
1.99 paf 54: static int lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); }
55: static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); }
56: static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); }
57: static int unlock(int fd) { FLOCK(_LK_UNLCK); }
1.93 paf 58:
1.99 paf 59: #else
60: #ifdef HAVE_FCNTL
1.93 paf 61:
1.126 paf 62: #define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls)
1.99 paf 63: static int lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); }
64: static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); }
65: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); }
66: static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); }
1.93 paf 67:
68: #else
69: #ifdef HAVE_LOCKF
1.99 paf 70:
1.126 paf 71: #define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1)
1.99 paf 72: static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking
73: static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); }
74: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); }
75: static int unlock(int fd) { FLOCK(F_TLOCK); }
76:
1.93 paf 77: #else
1.99 paf 78:
79: #error unable to find file locking func
80:
81: #endif
1.93 paf 82: #endif
83: #endif
84: #endif
85:
1.127 paf 86: #define DEFAULT_USER_AGENT "parser3"
87:
88:
1.126 paf 89: void fix_line_breaks(char* buf, size_t& size) {
1.139 paf 90: if(size==0)
91: return;
92:
1.87 paf 93: //_asm int 3;
1.126 paf 94: const char* const eob=buf+size;
95: char* dest=buf;
1.72 parser 96: // fix DOS: \r\n -> \n
97: // fix Macintosh: \r -> \n
1.126 paf 98: char* bol=buf;
1.137 paf 99: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 100: size_t len=eol-bol;
101: if(dest!=bol)
1.126 paf 102: memcpy(dest, bol, len);
1.72 parser 103: dest+=len;
1.126 paf 104: *dest++='\n';
1.72 parser 105:
1.126 paf 106: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 107: bol=eol+2;
1.126 paf 108: size--;
109: } else // \r, not \n = Macintosh
1.72 parser 110: bol=eol+1;
111: }
112: // last piece without \r, including terminating 0
113: if(dest!=bol)
1.126 paf 114: memcpy(dest, bol, eob-bol);
1.72 parser 115: }
1.18 paf 116:
1.143.2.3! paf 117: char* file_read_text(Pool& pool, Charset& charset,
1.143.2.1 paf 118: ConstStringPtr file_spec,
1.126 paf 119: bool fail_on_read_problem,
1.143.2.3! paf 120: HashStringValue *params, HashStringValue** out_fields) {
1.72 parser 121: void *result; size_t size;
1.143.2.3! paf 122: return file_read(pool, charset, file_spec, result, size, true, params, out_fields, fail_on_read_problem)?
1.143.2.1 paf 123: (char *)result:0;
1.126 paf 124: }
125:
126: //http request stuff
127: /* ************************ http stuff *********************** */
128:
129: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
130: memset(addr, 0, sizeof(*addr));
131: addr->sin_family=AF_INET;
132: addr->sin_port=htons(port);
133: if(host) {
134: if(struct hostent *hostIP=gethostbyname(host))
135: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
136: else
137: return false;
138: } else
139: addr->sin_addr.s_addr=INADDR_ANY;
140: return true;
141: }
142:
1.143.2.1 paf 143: static int http_read_response(Pool& pool, String& response, int sock, bool fail_on_status_ne_200){
1.143.2.2 paf 144: int result=0;
1.130 paf 145: ssize_t EOLat=0;
1.126 paf 146: while(true) {
1.143.2.1 paf 147: char *buf=(char *)pool.malloc(MAX_STRING);
1.126 paf 148: ssize_t size=recv(sock, buf, MAX_STRING, 0);
149: if(size<=0)
150: break;
1.130 paf 151: response.APPEND_TAINTED(buf, size, "remote HTTP server response", 0);
1.143.2.2 paf 152: if(!result && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response
1.143.2.3! paf 153: ConstStringPtr status_line=response.mid(0, (size_t)EOLat);
! 154: Array<ConstStringPtr> astatus;
! 155: status_line->split(astatus, 0, " ", 1);
! 156: ConstStringPtr status_code=astatus.get(1);
! 157: result=status_code->as_int();
1.142 paf 158:
1.143.2.3! paf 159: if(fail_on_status_ne_200 && result!=200)
1.142 paf 160: throw Exception("http.status",
161: status_code,
162: "invalid HTTP response status");
163: }
164: }
1.143.2.2 paf 165: if(result)
166: return result;
1.142 paf 167: else
168: throw Exception("http.response",
1.143.2.3! paf 169: Exception::undefined_source,
1.142 paf 170: "bad response from host - no status found (size=%lu)", response.size());
1.126 paf 171: }
172:
173: /* ********************** request *************************** */
174:
175: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
176: # define WE_CAN_USE_ALARM
177: #endif
178:
179: #ifdef WE_CAN_USE_ALARM
180: static sigjmp_buf timeout_env;
181: static void timeout_handler(int sig){
182: siglongjmp(timeout_env, 1);
183: }
184: #endif
185:
1.143.2.3! paf 186: static int http_request(Pool& pool,
! 187: String& response,
! 188: ConstStringPtr origin_string,
! 189: const char* host, int port,
! 190: const char* request,
! 191: int timeout,
! 192: bool fail_on_status_ne_200) {
1.126 paf 193: if(!host)
194: throw Exception("http.host",
195: origin_string,
196: "zero hostname"); //never
197:
198: #ifdef WE_CAN_USE_ALARM
199: signal(SIGALRM, timeout_handler);
200: #endif
201: int sock=-1;
202: try {
1.142 paf 203: int result;
1.126 paf 204: #ifdef WE_CAN_USE_ALARM
205: if(sigsetjmp(timeout_env, 1))
206: throw Exception("http.timeout",
207: origin_string,
208: "timeout occured while retrieving document");
209: else {
210: alarm(timeout);
211: #endif
212: struct sockaddr_in dest;
213:
214: if(!set_addr(&dest, host, port))
215: throw Exception("http.host",
216: origin_string,
1.127 paf 217: "can not resolve hostname \"%s\"", host);
1.126 paf 218:
219: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
220: throw Exception("http.connect",
221: origin_string,
1.127 paf 222: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 223: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
224: throw Exception("http.connect",
225: origin_string,
1.127 paf 226: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 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,
1.127 paf 231: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 232:
1.143.2.3! paf 233: result=http_read_response(pool, response, sock, fail_on_status_ne_200);
1.142 paf 234: closesocket(sock);
1.126 paf 235: #ifdef WE_CAN_USE_ALARM
1.142 paf 236: alarm(0);
1.126 paf 237: }
238: #endif
1.142 paf 239: return result;
1.126 paf 240: } catch(...) {
241: if(sock>=0)
242: closesocket(sock);
243: #ifdef WE_CAN_USE_ALARM
244: alarm(0);
245: #endif
246: /*re*/throw;
247: }
248: }
249:
1.127 paf 250: #ifndef DOXYGEN
251: struct Http_pass_header_info {
1.143.2.3! paf 252: Pool* pool;
! 253: Charset* charset;
1.127 paf 254: String* request;
255: bool user_agent_specified;
256: };
257: #endif
1.143.2.3! paf 258: static void http_pass_header(HashStringValue::key_type key, HashStringValue::value_type value,
! 259: Http_pass_header_info *info) {
! 260: *(info->request)<<*key<<": "
! 261: << *attributed_meaning_to_string(value, String::UL_HTTP_HEADER, false)
1.135 paf 262: <<"\n";
263:
1.143.2.3! paf 264: if(*key->change_case(*info->pool, *info->charset, String::CC_UPPER)=="USER-AGENT")
! 265: info->user_agent_specified=true;
1.126 paf 266: }
1.143.2.3! paf 267: static HashStringValuePtr file_read_http(Pool& pool, Charset& charset, const String& file_spec,
! 268: char*& data, size_t& data_size,
! 269: HashStringValue *options=0, bool return_outfields=false) {
! 270: HashStringValuePtr result(0);
1.126 paf 271: char host[MAX_STRING];
1.129 paf 272: const char* uri;
1.126 paf 273: int port;
1.143.2.3! paf 274: ConstCharPtr method(0);
1.126 paf 275: int timeout=2;
1.142 paf 276: bool fail_on_status_ne_200=true;
1.143.2.3! paf 277: ValuePtr vheaders(0);
1.126 paf 278:
1.143.2.3! paf 279: StringPtr connect_string(new String());
1.133 paf 280: // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one
1.143.2.3! paf 281: connect_string->append(file_spec, String::UL_URI); // tainted pieces -> URI pieces
1.133 paf 282:
1.143.2.3! paf 283: CharPtr connect_string_cstr=connect_string->cstr(String::UL_UNSPECIFIED);
! 284: const char *current=connect_string_cstr.get();
! 285: if(strncmp(current, "http://", 7)!=0)
1.126 paf 286: throw Exception(0,
1.143.2.3! paf 287: connect_string,
1.126 paf 288: "does not start with http://"); //never
1.143.2.3! paf 289: current+=7;
1.126 paf 290:
1.143.2.3! paf 291: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
1.126 paf 292: char* host_uri=lsplit(host, '/');
1.143.2.3! paf 293: uri=host_uri?current+(host_uri-1-host):"/";
1.126 paf 294: char* port_cstr=lsplit(host, ':');
295: char* error_pos=0;
296: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
297:
1.127 paf 298: if(options) {
299: int valid_options=0;
1.143.2.3! paf 300: if(ValuePtr vmethod=options->get(http_method_name)) {
1.127 paf 301: valid_options++;
1.143.2.3! paf 302: method=vmethod->as_string(pool)->cstr();
1.127 paf 303: }
1.143.2.3! paf 304: if(ValuePtr vtimeout=options->get(http_timeout_name)) {
1.127 paf 305: valid_options++;
306: timeout=vtimeout->as_int();
307: }
1.143.2.3! paf 308: if(vheaders=options->get(http_headers_name)) {
1.127 paf 309: valid_options++;
310: }
1.143.2.3! paf 311: if(ValuePtr vany_status=options->get(http_any_status_name)) {
1.142 paf 312: valid_options++;
313: fail_on_status_ne_200=!vany_status->as_bool();
314: }
315:
1.143.2.3! paf 316: if(valid_options!=options->count())
1.127 paf 317: throw Exception("parser.runtime",
1.143.2.3! paf 318: Exception::undefined_source,
1.127 paf 319: "invalid option passed");
1.133 paf 320: }
1.126 paf 321:
322: //making request
1.143.2.3! paf 323: String request;
! 324: if(method)
! 325: request<<method.get();
! 326: else
! 327: request<<"GET";
! 328: request<< " "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";
1.127 paf 329: bool user_agent_specified=false;
330: if(vheaders && !vheaders->is_string()) { // allow empty
1.143.2.3! paf 331: if(HashStringValuePtr headers=vheaders->get_hash(connect_string)) {
! 332: Http_pass_header_info info={&pool, &charset, &request};
1.127 paf 333: headers->for_each(http_pass_header, &info);
334: user_agent_specified=info.user_agent_specified;
335: } else
336: throw Exception("parser.runtime",
1.143.2.3! paf 337: connect_string,
1.127 paf 338: "headers param must be hash");
339: };
340: if(!user_agent_specified) // defaulting
341: request << "user-agent: " DEFAULT_USER_AGENT "\n";
1.126 paf 342: request<<"\n";
343:
344: //sending request
1.143.2.3! paf 345: String response;
! 346: ConstCharPtr request_cstr=request.cstr(String::UL_UNSPECIFIED);
! 347: int status_code=http_request(pool, response,
! 348: connect_string, host, port, request_cstr.get(),
1.142 paf 349: timeout, fail_on_status_ne_200);
1.126 paf 350:
351: //processing results
352: int pos=response.pos("\r\n\r\n", 4);
353: if(pos<1){
354: throw Exception("http.response",
1.143.2.3! paf 355: connect_string,
1.126 paf 356: "bad response from host - no headers found");
357: }
1.143.2.3! paf 358: StringPtr header_block=response.mid(0, pos);
! 359: StringPtr body=response.mid(pos+4, response.size());
1.126 paf 360:
1.143.2.3! paf 361: Array<ConstStringPtr> aheaders;
! 362: HashStringValuePtr headers(new HashStringValue);
! 363: header_block->split(aheaders, 0, "\r\n", 2);
1.126 paf 364:
365: //processing headers
1.143.2.3! paf 366: for(int i=1; i<aheaders.count(); i++) {
! 367: ConstStringPtr line=aheaders.get(i);
! 368: pos=line->pos(": ", 2);
! 369: if(pos<1)
1.126 paf 370: throw Exception("http.response",
1.143.2.3! paf 371: connect_string,
! 372: "bad response from host - bad header \"%s\"", line->cstr());
! 373:
! 374: headers->put(
! 375: line->mid(0, pos)->change_case(pool, charset, String::CC_UPPER),
! 376: ValuePtr(new VString(line->mid(pos+2, line->size()))));
1.126 paf 377: }
378:
379: // output response
1.143.2.3! paf 380: data=new(pool) char[data_size=body->size()];
! 381: body->store_to(data, String::UL_AS_IS);
! 382: if(return_outfields) {
! 383: headers.put(file_status_name, VIntPtr<new VInt(pool, status_code)>);
1.126 paf 384: *out_fields=&headers;
1.141 paf 385: }
1.143.2.3! paf 386:
! 387: return result;
1.34 paf 388: }
1.123 paf 389:
390: #ifndef DOXYGEN
391: struct File_read_action_info {
392: void **data; size_t *data_size;
1.126 paf 393: };
1.123 paf 394: #endif
1.143.2.1 paf 395: static void file_read_action(
1.126 paf 396: struct stat& finfo,
1.123 paf 397: int f,
1.126 paf 398: const String& file_spec, const char* fname, bool as_text,
1.123 paf 399: void *context) {
1.126 paf 400: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 401: if(size_t to_read_size=(size_t)finfo.st_size) {
1.126 paf 402: *info.data=pool.malloc(to_read_size+(as_text?1:0), 3);
403: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 404:
405: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 406: throw Exception(0,
1.123 paf 407: &file_spec,
408: "read failed: actually read %lu bytes count not in [0..%lu] valid range",
1.126 paf 409: *info.data_size, to_read_size);
1.123 paf 410: } else { // empty file
411: if(as_text) {
1.126 paf 412: *info.data=pool.malloc(1);
1.123 paf 413: *(char*)(*info.data)=0;
414: } else
415: *info.data=0;
416: *info.data_size=0;
417: return;
418: }
1.126 paf 419: }
1.143.2.1 paf 420: bool file_read(ConstStringPtr file_spec,
1.126 paf 421: void*& data, size_t& data_size,
1.143.2.3! paf 422: bool as_text, HashStringValue *params, HashStringValue** out_fields,
1.126 paf 423: bool fail_on_read_problem) {
424: bool result;
425: if(file_spec.starts_with("http://", 7)) {
426: // fail on read problem
427: file_read_http(pool, file_spec, data, data_size, params, out_fields);
428: result=true;
429: } else {
430: File_read_action_info info={&data, &data_size};
431: result=file_read_action_under_lock(pool, file_spec,
432: "read", file_read_action, &info,
433: as_text, fail_on_read_problem);
434: }
1.123 paf 435:
1.126 paf 436: if(result && as_text) {
1.131 paf 437: // UTF-8 signature: EF BB BF
438: if(data_size>=3) {
439: char *in=(char *)data;
440: if((in[0] == '\xEF') && (in[1] == '\xBB') &&
441: (in[2] == '\xBF')) {
442: data=in+3; data_size-=3;// skip prefix
443: }
444: }
445:
1.126 paf 446: fix_line_breaks((char *)(data), data_size);
1.123 paf 447: // note: after fixing
1.126 paf 448: ((char*&)(data))[data_size]=0;
1.123 paf 449: }
1.126 paf 450:
451: return result;
1.123 paf 452: }
453:
1.143.2.1 paf 454: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 455: const char* action_name, File_read_action action, void *context,
456: bool as_text,
1.123 paf 457: bool fail_on_read_problem) {
1.126 paf 458: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.33 paf 459: int f;
460:
461: // first open, next stat:
1.45 paf 462: // directory update of NTFS hard links performed on open.
1.33 paf 463: // ex:
464: // a.html:^test[] and b.html hardlink to a.html
465: // user inserts ! before ^test in a.html
1.126 paf 466: // directory entry of b.html in NTFS not updated at once,
1.35 paf 467: // they delay update till open, so we would receive "!^test[" string
468: // if would do stat, next open.
1.123 paf 469: // later: it seems, even this does not help sometimes
1.98 paf 470: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 471: try {
472: if(lock_shared_blocking(f)!=0)
1.126 paf 473: throw Exception("file.lock",
1.123 paf 474: &file_spec,
475: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 476: strerror(errno), errno, fname);
1.123 paf 477:
1.124 paf 478: struct stat finfo;
479: if(stat(fname, &finfo)!=0)
480: throw Exception("file.missing", // hardly possible: we just opened it OK
481: &file_spec,
482: "stat failed: %s (%d), actual filename '%s'",
1.126 paf 483: strerror(errno), errno, fname);
1.124 paf 484:
1.140 paf 485: #ifdef PA_SAFE_MODE
486: if(finfo.st_uid/*foreign?*/!=geteuid()
487: && finfo.st_gid/*foreign?*/!=getegid())
1.126 paf 488: throw Exception("parser.runtime",
489: &file_spec,
1.140 paf 490: "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 491: fname);
1.105 paf 492: #endif
1.32 paf 493:
1.126 paf 494: action(pool, finfo, f, file_spec, fname, as_text, context);
1.123 paf 495: } catch(...) {
1.126 paf 496: unlock(f);close(f);
1.123 paf 497: if(fail_on_read_problem)
498: /*re*/throw;
499: return false;
500: }
1.87 paf 501:
1.126 paf 502: unlock(f);close(f);
1.72 parser 503: return true;
1.118 paf 504: } else {
505: if(fail_on_read_problem)
1.126 paf 506: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 507: &file_spec,
1.123 paf 508: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 509: action_name, strerror(errno), errno, fname);
1.118 paf 510: return false;
511: }
1.8 paf 512: }
513:
1.63 parser 514: static void create_dir_for_file(const String& file_spec) {
515: size_t pos_after=1;
516: int pos_before;
517: while((pos_before=file_spec.pos("/", 1, pos_after))>=0) {
1.126 paf 518: mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775);
1.63 parser 519: pos_after=pos_before+1;
520: }
521: }
522:
1.98 paf 523: bool file_write_action_under_lock(
1.28 paf 524: const String& file_spec,
1.126 paf 525: const char* action_name, File_write_action action, void *context,
526: bool as_text,
527: bool do_append,
528: bool do_block,
1.110 paf 529: bool fail_on_lock_problem) {
1.126 paf 530: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.28 paf 531: int f;
1.80 paf 532: if(access(fname, W_OK)!=0) // no
1.126 paf 533: create_dir_for_file(file_spec);
1.50 paf 534:
1.80 paf 535: if((f=open(fname,
536: O_CREAT|O_RDWR
537: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 538: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.99 paf 539: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.126 paf 540: Exception e("file.lock",
1.110 paf 541: &file_spec,
542: "shared lock failed: %s (%d), actual filename '%s'",
1.126 paf 543: strerror(errno), errno, fname);
544: close(f);
1.110 paf 545: if(fail_on_lock_problem)
546: throw e;
1.98 paf 547: return false;
548: }
1.96 paf 549:
550: try {
1.126 paf 551: action(f, context);
1.96 paf 552: } catch(...) {
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)); // one can not use O_TRUNC, read lower
1.138 paf 556: #endif
1.126 paf 557: unlock(f);close(f);
1.96 paf 558: /*re*/throw;
559: }
1.80 paf 560:
1.138 paf 561: #ifdef HAVE_FTRUNCATE
1.104 paf 562: if(!do_append)
1.125 paf 563: 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 564: #endif
1.126 paf 565: unlock(f);close(f);
1.98 paf 566: return true;
1.80 paf 567: } else
1.126 paf 568: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 569: &file_spec,
1.96 paf 570: "%s failed: %s (%d), actual filename '%s'",
1.126 paf 571: action_name, strerror(errno), errno, fname);
1.96 paf 572: // here should be nothing, see rethrow above
573: }
574:
575: #ifndef DOXYGEN
576: struct File_write_action_info {
577: const void *data; size_t size;
1.126 paf 578: };
1.96 paf 579: #endif
580: static void file_write_action(int f, void *context) {
1.126 paf 581: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.113 paf 582: if(info.size) {
1.126 paf 583: int written=write(f, info.data, info.size);
1.116 paf 584: if(written<0)
1.126 paf 585: throw Exception(0,
586: 0,
587: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 588: }
1.96 paf 589: }
590: void file_write(
591: const String& file_spec,
592: const void *data, size_t size,
1.126 paf 593: bool as_text,
1.96 paf 594: bool do_append) {
1.126 paf 595: File_write_action_info info={data, size};
1.98 paf 596: file_write_action_under_lock(
1.96 paf 597: file_spec,
1.126 paf 598: "write", file_write_action, &info,
599: as_text,
600: do_append);
1.30 paf 601: }
602:
1.63 parser 603: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 604: static void rmdir(const String& file_spec, size_t pos_after) {
605: int pos_before;
606: if((pos_before=file_spec.pos("/", 1, pos_after))>=0)
1.126 paf 607: rmdir(file_spec, pos_before+1);
1.50 paf 608:
1.126 paf 609: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_FILE_SPEC));
1.50 paf 610: }
1.95 paf 611: bool file_delete(const String& file_spec, bool fail_on_read_problem) {
1.126 paf 612: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.54 parser 613: if(unlink(fname)!=0)
1.93 paf 614: if(fail_on_read_problem)
1.126 paf 615: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 616: &file_spec,
617: "unlink failed: %s (%d), actual filename '%s'",
1.126 paf 618: strerror(errno), errno, fname);
1.93 paf 619: else
620: return false;
1.50 paf 621:
1.126 paf 622: rmdir(file_spec, 1);
1.93 paf 623: return true;
1.60 parser 624: }
1.95 paf 625: void file_move(const String& old_spec, const String& new_spec) {
1.126 paf 626: const char* old_spec_cstr=old_spec.cstr(String::UL_FILE_SPEC);
627: const char* new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC);
1.63 parser 628:
1.126 paf 629: create_dir_for_file(new_spec);
1.63 parser 630:
1.60 parser 631: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 632: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 633: &old_spec,
634: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.126 paf 635: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 636:
1.126 paf 637: rmdir(old_spec, 1);
1.31 paf 638: }
639:
1.51 paf 640:
1.126 paf 641: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 642: struct stat lfinfo;
643: bool result=stat(fname, &lfinfo)==0;
644: if(afinfo)
645: *afinfo=lfinfo;
646: return result;
1.119 paf 647: }
648:
649: bool entry_exists(const String& file_spec) {
1.126 paf 650: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
651: return entry_exists(fname, 0);
1.118 paf 652: }
653:
1.51 paf 654: static bool entry_readable(const String& file_spec, bool need_dir) {
1.126 paf 655: char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.120 paf 656: if(need_dir) {
1.126 paf 657: size_t size=strlen(fname);
1.120 paf 658: while(size) {
1.126 paf 659: char c=fname[size-1];
1.120 paf 660: if(c=='/' || c=='\\')
661: fname[--size]=0;
662: else
663: break;
664: }
665: }
1.51 paf 666: struct stat finfo;
1.118 paf 667: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 668: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 669: return is_dir==need_dir;
670: }
671: return false;
672: }
1.31 paf 673: bool file_readable(const String& file_spec) {
1.126 paf 674: return entry_readable(file_spec, false);
1.51 paf 675: }
676: bool dir_readable(const String& file_spec) {
1.126 paf 677: return entry_readable(file_spec, true);
1.65 parser 678: }
679: String *file_readable(const String& path, const String& name) {
1.126 paf 680: String *result=new(path.pool()) String(path);
681: *result << "/";
1.65 parser 682: *result << name;
683: return file_readable(*result)?result:0;
1.43 paf 684: }
685: bool file_executable(const String& file_spec) {
1.64 parser 686: return access(file_spec.cstr(String::UL_FILE_SPEC), X_OK)==0;
1.44 paf 687: }
688:
1.64 parser 689: bool file_stat(const String& file_spec,
1.58 parser 690: size_t& rsize,
1.126 paf 691: time_t& ratime,
692: time_t& rmtime,
693: time_t& rctime,
1.64 parser 694: bool fail_on_read_problem) {
1.126 paf 695: Pool& pool=file_spec.pool();
696: const char* fname=file_spec.cstr(String::UL_FILE_SPEC);
1.44 paf 697: struct stat finfo;
698: if(stat(fname, &finfo)!=0)
1.64 parser 699: if(fail_on_read_problem)
1.126 paf 700: throw Exception("file.missing",
1.67 parser 701: &file_spec,
702: "getting file size failed: %s (%d), real filename '%s'",
1.126 paf 703: strerror(errno), errno, fname);
1.64 parser 704: else
705: return false;
1.58 parser 706: rsize=finfo.st_size;
707: ratime=finfo.st_atime;
708: rmtime=finfo.st_mtime;
709: rctime=finfo.st_ctime;
1.64 parser 710: return true;
1.18 paf 711: }
712:
1.126 paf 713: char* getrow(char* *row_ref, char delim) {
714: char* result=*row_ref;
1.8 paf 715: if(result) {
1.126 paf 716: *row_ref=strchr(result, delim);
1.8 paf 717: if(*row_ref)
718: *((*row_ref)++)=0;
719: else if(!*result)
720: return 0;
721: }
722: return result;
723: }
724:
1.126 paf 725: char* lsplit(char* string, char delim) {
1.23 paf 726: if(string) {
1.126 paf 727: char* v=strchr(string, delim);
1.8 paf 728: if(v) {
729: *v=0;
730: return v+1;
731: }
732: }
733: return 0;
734: }
735:
1.126 paf 736: char* lsplit(char* *string_ref, char delim) {
737: char* result=*string_ref;
738: char* next=lsplit(*string_ref, delim);
1.8 paf 739: *string_ref=next;
740: return result;
1.9 paf 741: }
742:
1.126 paf 743: char* rsplit(char* string, char delim) {
1.18 paf 744: if(string) {
1.126 paf 745: char* v=strrchr(string, delim);
1.18 paf 746: if(v) {
1.9 paf 747: *v=0;
748: return v+1;
749: }
750: }
751: return NULL;
1.10 paf 752: }
753:
1.37 paf 754: /// @todo less stupid type detection
1.143.2.1 paf 755: char* format(double value, char* fmt) {
1.126 paf 756: char local_buf[MAX_NUMBER];
1.108 paf 757: size_t size;
758:
1.10 paf 759: if(fmt)
760: if(strpbrk(fmt, "diouxX"))
761: if(strpbrk(fmt, "ouxX"))
1.126 paf 762: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 763: else
1.126 paf 764: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 765: else
1.126 paf 766: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 767: else
1.126 paf 768: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 769:
1.126 paf 770: char* pool_buf=(char *)pool.malloc(size+1, 4);
771: memcpy(pool_buf, local_buf, size+1);
1.108 paf 772: return pool_buf;
1.12 paf 773: }
774:
1.36 paf 775: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 776: #ifdef WIN32
777: do{
1.126 paf 778: int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout);
1.12 paf 779: if(chunk_written<=0)
780: break;
781: size-=chunk_written;
1.36 paf 782: buf=((const char*)buf)+chunk_written;
1.126 paf 783: } while(size>0);
1.12 paf 784:
785: return size;
786: #else
1.126 paf 787: return fwrite(buf, 1, size, stdout);
1.12 paf 788: #endif
1.2 paf 789: }
1.14 paf 790:
1.143.2.1 paf 791: char* unescape_chars(const char* cp, int len) {
1.126 paf 792: char* s=(char *)pool.malloc(len + 1, 5);
1.14 paf 793: enum EscapeState {
1.33 paf 794: EscapeRest,
795: EscapeFirst,
1.14 paf 796: EscapeSecond
797: } escapeState=EscapeRest;
798: int escapedValue=0;
799: int srcPos=0;
800: int dstPos=0;
801: while(srcPos < len) {
1.126 paf 802: int ch=cp[srcPos];
1.14 paf 803: switch(escapeState) {
804: case EscapeRest:
805: if(ch=='%') {
806: escapeState=EscapeFirst;
807: } else if(ch=='+') {
1.126 paf 808: s[dstPos++]=' ';
1.14 paf 809: } else {
810: s[dstPos++]=ch;
811: }
812: break;
813: case EscapeFirst:
814: escapedValue=hex_value[ch] << 4;
815: escapeState=EscapeSecond;
816: break;
817: case EscapeSecond:
1.126 paf 818: escapedValue +=hex_value[ch];
1.14 paf 819: s[dstPos++]=escapedValue;
820: escapeState=EscapeRest;
821: break;
822: }
1.126 paf 823: srcPos++;
1.14 paf 824: }
825: s[dstPos]=0;
826: return s;
1.24 paf 827: }
828:
829: #ifdef WIN32
1.126 paf 830: void back_slashes_to_slashes(char* s) {
1.24 paf 831: if(s)
832: for(; *s; s++)
833: if(*s=='\\')
1.126 paf 834: *s='/';
1.24 paf 835: }
1.42 paf 836: /*
1.126 paf 837: void slashes_to_back_slashes(char* s) {
1.42 paf 838: if(s)
839: for(; *s; s++)
840: if(*s=='/')
1.126 paf 841: *s='\\';
1.42 paf 842: }
843: */
1.24 paf 844: #endif
1.41 paf 845:
1.126 paf 846: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 847: while(true) {
848: if(!(*s1)) {
849: if(!(*s2))
850: return true;
851: else
852: return !strict;
853: } else if(!(*s2))
854: return !strict;
855: if(isalpha(*s1)) {
856: if(tolower(*s1) !=tolower(*s2))
857: return false;
858: } else if((*s1) !=(*s2))
859: return false;
1.126 paf 860: s1++;
861: s2++;
1.41 paf 862: }
1.57 parser 863: }
864:
1.84 paf 865: static bool isLeap(int year) {
1.57 parser 866: return !(
867: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 868: );
1.57 parser 869: }
870:
871: int getMonthDays(int year, int month) {
872: int monthDays[]={
1.126 paf 873: 31,
874: isLeap(year) ? 29 : 28,
875: 31,
876: 30,
877: 31,
878: 30,
879: 31,
880: 31,
881: 30,
882: 31,
883: 30,
1.57 parser 884: 31
1.126 paf 885: };
886: return monthDays[month];
1.41 paf 887: }
1.69 parser 888:
1.126 paf 889: void remove_crlf(char* start, char* end) {
890: for(char* p=start; p<end; p++)
1.69 parser 891: switch(*p) {
1.126 paf 892: case '\n': *p='|'; break;
893: case '\r': *p=' '; break;
1.69 parser 894: }
1.91 paf 895: }
896:
897:
898: /// must be last in this file
899: #undef vsnprintf
1.126 paf 900: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 901: if(!s)
902: return 0;
903:
904: int r;
905: // note: on win32& maybe somewhere else
906: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
907: --s;
908: #if _MSC_VER
909: /*
910: win32:
911: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
912:
913: if the number of bytes to write exceeds buffer, then count bytes are written and –1 is returned
914: */
1.126 paf 915: r=_vsnprintf(b, s, f, l);
1.91 paf 916: if(r<0)
917: r=s;
918: #else
1.126 paf 919: r=vsnprintf(b, s, f, l);
1.91 paf 920: /*
921: solaris:
922: man vsnprintf
923:
924: The snprintf() function returns the number of characters
925: formatted, that is, the number of characters that would have
926: been written to the buffer if it were large enough. If the
927: value of n is 0 on a call to snprintf(), an unspecified
928: value less than 1 is returned.
929: */
930:
931: if(r<0)
932: r=0;
933: else if(r>s)
934: r=s;
935: #endif
936: b[r]=0;
937: return r;
938: }
939:
1.126 paf 940: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 941: va_list l;
1.126 paf 942: va_start(l, f);
943: int r=__vsnprintf(b, s, f, l);
944: va_end(l);
1.91 paf 945: return r;
1.98 paf 946: }
947:
948: int pa_sleep(unsigned long secs, unsigned long usecs) {
1.126 paf 949: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
1.98 paf 950:
951: #ifdef WIN32
1.126 paf 952: Sleep(secs * 1000 + usecs / 1000);
1.98 paf 953: return 0;
954: #else
955: struct timeval t;
956: t.tv_sec = secs;
957: t.tv_usec = usecs;
1.126 paf 958: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
1.98 paf 959: #endif
1.135 paf 960: }
961:
962:
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