Annotation of parser3/src/main/pa_common.C, revision 1.143.2.21.2.19
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.143.2.10 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:
1.143.2.21.2. 9(paf 8:3): static const char* IDENT_COMMON_C="$Date: 2003/04/01 12:05:46 $";
1.1 paf 9:
10: #include "pa_common.h"
1.4 paf 11: #include "pa_exception.h"
1.126 paf 12: #include "pa_hash.h"
1.143.2.21.2. (paf 13:): #include "pa_globals.h"
14:):
4(paf 15:3): #define PA_HTTP
(paf 16:):
17:): #ifdef PA_HTTP
1.143.2.3 paf 18: #include "pa_vstring.h"
1.143.2.4 paf 19: #include "pa_vint.h"
1.1 paf 20:
1.143.2.12 paf 21: #ifdef CYGWIN
22: #define _GNU_H_WINDOWS32_SOCKETS
23: // for PASCAL
24: #include <windows.h>
25: // SOCKET
26: typedef u_int SOCKET;
27: int PASCAL closesocket(SOCKET);
1.126 paf 28: #else
1.143.2.12 paf 29: # if defined(WIN32)
30: # include <windows.h>
31: # else
32: # define closesocket close
33: # endif
1.98 paf 34: #endif
35:
1.143.2.21.2. (paf 36:): #else
37:):
38:): # if defined(WIN32)
39:): # include <windows.h>
40:): # endif
41:):
42:): #endif
43:):
1.93 paf 44: // some maybe-undefined constants
45:
1.82 paf 46: #ifndef _O_TEXT
47: # define _O_TEXT 0
48: #endif
49: #ifndef _O_BINARY
50: # define _O_BINARY 0
1.47 paf 51: #endif
1.80 paf 52:
1.138 paf 53: #ifdef HAVE_FTRUNCATE
54: # define PA_O_TRUNC 0
55: #else
56: # ifdef _O_TRUNC
57: # define PA_O_TRUNC _O_TRUNC
58: # else
59: # error you must have either ftruncate function or _O_TRUNC bit declared
60: # endif
61: #endif
62:
1.93 paf 63: // locking constants
1.143.2.21.2. 9(paf 64:3): //#define PA_DEBUG_NO_LOCKING
65:3):
66:3): #ifdef PA_DEBUG_NO_LOCKING
67:3):
1.99 paf 68: #ifdef HAVE_FLOCK
69:
70: static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); }
71: static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); }
72: static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); }
73: static int unlock(int fd) { return flock(fd, LOCK_UN); }
74:
1.98 paf 75: #else
1.99 paf 76: #ifdef HAVE__LOCKING
1.98 paf 77:
1.126 paf 78: #define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1)
1.99 paf 79: static int lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); }
80: static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); }
81: static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); }
82: static int unlock(int fd) { FLOCK(_LK_UNLCK); }
1.93 paf 83:
1.99 paf 84: #else
85: #ifdef HAVE_FCNTL
1.93 paf 86:
1.126 paf 87: #define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls)
1.99 paf 88: static int lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); }
89: static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); }
90: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); }
91: static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); }
1.93 paf 92:
93: #else
94: #ifdef HAVE_LOCKF
1.99 paf 95:
1.126 paf 96: #define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1)
1.99 paf 97: static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking
98: static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); }
99: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); }
100: static int unlock(int fd) { FLOCK(F_TLOCK); }
101:
1.93 paf 102: #else
1.99 paf 103:
104: #error unable to find file locking func
105:
106: #endif
1.93 paf 107: #endif
108: #endif
109: #endif
110:
1.143.2.21.2. 6(paf 111:3): #else
112:3): static int lock_shared_blocking(int fd) { return 0; }
113:3): static int lock_exclusive_blocking(int fd) { return 0; }
114:3): static int lock_exclusive_nonblocking(int fd) { return 0; }
115:3): static int unlock(int fd) { return 0; }
116:3):
117:3): #endif
118:3):
1.143.2.16 paf 119: // defines for globals
120:
121: #define FILE_STATUS_NAME "status"
122:
123: // globals
124:
1.143.2.21.2. (paf 125:): const String file_status_name(FILE_STATUS_NAME);
1.143.2.16 paf 126:
1.143.2.14 paf 127: // defines for statics
128:
129: #define HTTP_METHOD_NAME "method"
130: #define HTTP_TIMEOUT_NAME "timeout"
131: #define HTTP_HEADERS_NAME "headers"
132: #define HTTP_ANY_STATUS_NAME "any-status"
133:
134: // statics
135:
1.143.2.21.2. (paf 136:): static const String http_method_name(HTTP_METHOD_NAME);
137:): static const String http_timeout_name(HTTP_TIMEOUT_NAME);
138:): static const String http_headers_name(HTTP_HEADERS_NAME);
139:): static const String http_any_status_name(HTTP_ANY_STATUS_NAME);
1.143.2.14 paf 140:
141: // defines
142:
1.127 paf 143: #define DEFAULT_USER_AGENT "parser3"
144:
1.143.2.14 paf 145: // functions
1.127 paf 146:
1.126 paf 147: void fix_line_breaks(char* buf, size_t& size) {
1.139 paf 148: if(size==0)
149: return;
150:
1.87 paf 151: //_asm int 3;
1.126 paf 152: const char* const eob=buf+size;
153: char* dest=buf;
1.72 parser 154: // fix DOS: \r\n -> \n
155: // fix Macintosh: \r -> \n
1.126 paf 156: char* bol=buf;
1.137 paf 157: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 158: size_t len=eol-bol;
159: if(dest!=bol)
1.126 paf 160: memcpy(dest, bol, len);
1.72 parser 161: dest+=len;
1.126 paf 162: *dest++='\n';
1.72 parser 163:
1.126 paf 164: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 165: bol=eol+2;
1.126 paf 166: size--;
167: } else // \r, not \n = Macintosh
1.72 parser 168: bol=eol+1;
169: }
170: // last piece without \r, including terminating 0
171: if(dest!=bol)
1.126 paf 172: memcpy(dest, bol, eob-bol);
1.72 parser 173: }
1.18 paf 174:
1.143.2.21.2. (paf 175:): char* file_read_text(Charset& charset,
176:): const String& file_spec,
1.126 paf 177: bool fail_on_read_problem,
1.143.2.21.2. (paf 178:): HashStringValue* params/*, HashStringValue* * out_fields*/) {
1.143.2.18 paf 179: File_read_result file=
1.143.2.21.2. (paf 180:): file_read(charset, file_spec, true, params, fail_on_read_problem);
181:): return file.success?file.str:0;
1.126 paf 182: }
183:
1.143.2.21.2. (paf 184:): #ifdef PA_HTTP
1.126 paf 185: //http request stuff
186:
187: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
188: memset(addr, 0, sizeof(*addr));
189: addr->sin_family=AF_INET;
190: addr->sin_port=htons(port);
191: if(host) {
192: if(struct hostent *hostIP=gethostbyname(host))
193: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
194: else
195: return false;
196: } else
197: addr->sin_addr.s_addr=INADDR_ANY;
198: return true;
199: }
200:
1.143.2.21.2. 4(paf 201:3): static int http_read_response(String& response, int sock, bool fail_on_status_ne_200){
1.143.2.2 paf 202: int result=0;
1.143.2.21.2. 2(paf 203:3): size_t EOLat=0;
1.126 paf 204: while(true) {
1.143.2.21.2. (paf 205:): char *buf=new(PointerFreeGC) char[MAX_STRING];
1.126 paf 206: ssize_t size=recv(sock, buf, MAX_STRING, 0);
207: if(size<=0)
208: break;
1.143.2.21.2. 7(paf 209:3): response.append_strdup(buf, size, String::L_TAINTED);
2(paf 210:3): if(!result && (EOLat=response.pos("\r\n", 2))!=STRING_NOT_FOUND) { // checking status in first response
(paf 211:): const String& status_line=response.mid(0, (size_t)EOLat);
1.143.2.15 paf 212: ArrayString astatus;
1.143.2.21.2. 4(paf 213:3): size_t pos_after=0;
214:3): status_line.split(astatus, pos_after, " ");
215:3): const String& status_code=*astatus.get(1);
216:3): result=status_code.as_int();
1.142 paf 217:
1.143.2.3 paf 218: if(fail_on_status_ne_200 && result!=200)
1.142 paf 219: throw Exception("http.status",
1.143.2.21.2. 4(paf 220:3): &status_code,
1.142 paf 221: "invalid HTTP response status");
222: }
223: }
1.143.2.2 paf 224: if(result)
225: return result;
1.142 paf 226: else
227: throw Exception("http.response",
1.143.2.21.2. (paf 228:): 0,
229:): "bad response from host - no status found (size=%lu)", response.length());
1.126 paf 230: }
231:
232: /* ********************** request *************************** */
233:
234: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
235: # define WE_CAN_USE_ALARM
236: #endif
237:
238: #ifdef WE_CAN_USE_ALARM
239: static sigjmp_buf timeout_env;
240: static void timeout_handler(int sig){
241: siglongjmp(timeout_env, 1);
242: }
243: #endif
244:
1.143.2.21.2. (paf 245:): static int http_request(
4(paf 246:3): String& response,
247:3): const char* host, int port,
248:3): const char* request,
249:3): int timeout,
250:3): bool fail_on_status_ne_200) {
1.126 paf 251: if(!host)
252: throw Exception("http.host",
1.143.2.21.2. 4(paf 253:3): 0,
1.126 paf 254: "zero hostname"); //never
255:
256: #ifdef WE_CAN_USE_ALARM
257: signal(SIGALRM, timeout_handler);
258: #endif
259: int sock=-1;
260: try {
1.142 paf 261: int result;
1.126 paf 262: #ifdef WE_CAN_USE_ALARM
263: if(sigsetjmp(timeout_env, 1))
264: throw Exception("http.timeout",
1.143.2.21.2. 4(paf 265:3): 0,
1.126 paf 266: "timeout occured while retrieving document");
267: else {
268: alarm(timeout);
269: #endif
270: struct sockaddr_in dest;
271:
272: if(!set_addr(&dest, host, port))
273: throw Exception("http.host",
1.143.2.21.2. 4(paf 274:3): 0,
1.127 paf 275: "can not resolve hostname \"%s\"", host);
1.126 paf 276:
277: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
278: throw Exception("http.connect",
1.143.2.21.2. 4(paf 279:3): 0,
1.127 paf 280: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 281: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
282: throw Exception("http.connect",
1.143.2.21.2. 4(paf 283:3): 0,
1.127 paf 284: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 285: size_t request_size=strlen(request);
286: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
287: throw Exception("http.connect",
1.143.2.21.2. 4(paf 288:3): 0,
1.127 paf 289: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 290:
1.143.2.21.2. (paf 291:): result=http_read_response(response, sock, fail_on_status_ne_200);
1.142 paf 292: closesocket(sock);
1.126 paf 293: #ifdef WE_CAN_USE_ALARM
1.142 paf 294: alarm(0);
1.126 paf 295: }
296: #endif
1.142 paf 297: return result;
1.126 paf 298: } catch(...) {
299: if(sock>=0)
300: closesocket(sock);
301: #ifdef WE_CAN_USE_ALARM
302: alarm(0);
303: #endif
1.143.2.11 paf 304: rethrow;
1.126 paf 305: }
306: }
307:
1.127 paf 308: #ifndef DOXYGEN
309: struct Http_pass_header_info {
1.143.2.3 paf 310: Charset* charset;
1.127 paf 311: String* request;
312: bool user_agent_specified;
313: };
314: #endif
1.143.2.3 paf 315: static void http_pass_header(HashStringValue::key_type key, HashStringValue::value_type value,
1.143.2.21.2. 4(paf 316:3): Http_pass_header_info *info) {
317:3): *info->request <<key<<": "
318:3): << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
319:3): << "\n";
1.135 paf 320:
1.143.2.21.2. 4(paf 321:3): if(String(key, String::L_TAINTED).change_case(*info->charset, String::CC_UPPER)=="USER-AGENT")
1.143.2.3 paf 322: info->user_agent_specified=true;
1.126 paf 323: }
1.143.2.16 paf 324: #ifndef DOXYGEN
1.143.2.21 paf 325: struct File_read_http_result {
1.143.2.21.2. 4(paf 326:3): char *str; size_t length;
(paf 327:): HashStringValue* headers;
1.143.2.16 paf 328: };
329: #endif
1.143.2.21.2. 4(paf 330:3): static File_read_http_result file_read_http(Charset& charset, const String& file_spec,
331:3): HashStringValue *options=0) {
1.143.2.21 paf 332: File_read_http_result result;
1.126 paf 333: char host[MAX_STRING];
1.129 paf 334: const char* uri;
1.126 paf 335: int port;
1.143.2.21.2. 4(paf 336:3): const char* method=0;
1.126 paf 337: int timeout=2;
1.142 paf 338: bool fail_on_status_ne_200=true;
1.143.2.21.2. 4(paf 339:3): Value* vheaders=0;
1.126 paf 340:
1.143.2.21.2. 4(paf 341:3): String& connect_string=*new String;
(paf 342:): // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
4(paf 343:3): connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
1.133 paf 344:
1.143.2.21.2. 4(paf 345:3): const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
1.143.2.10 paf 346: const char* current=connect_string_cstr;
1.143.2.3 paf 347: if(strncmp(current, "http://", 7)!=0)
1.126 paf 348: throw Exception(0,
1.143.2.21.2. 4(paf 349:3): &connect_string,
1.126 paf 350: "does not start with http://"); //never
1.143.2.3 paf 351: current+=7;
1.126 paf 352:
1.143.2.3 paf 353: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
1.126 paf 354: char* host_uri=lsplit(host, '/');
1.143.2.3 paf 355: uri=host_uri?current+(host_uri-1-host):"/";
1.126 paf 356: char* port_cstr=lsplit(host, ':');
357: char* error_pos=0;
358: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
359:
1.127 paf 360: if(options) {
361: int valid_options=0;
1.143.2.21.2. (paf 362:): if(Value* vmethod=options->get(http_method_name)) {
1.127 paf 363: valid_options++;
1.143.2.21.2. (paf 364:): method=vmethod->as_string().cstr();
1.127 paf 365: }
1.143.2.21.2. (paf 366:): if(Value* vtimeout=options->get(http_timeout_name)) {
1.127 paf 367: valid_options++;
368: timeout=vtimeout->as_int();
369: }
1.143.2.3 paf 370: if(vheaders=options->get(http_headers_name)) {
1.127 paf 371: valid_options++;
372: }
1.143.2.21.2. (paf 373:): if(Value* vany_status=options->get(http_any_status_name)) {
1.142 paf 374: valid_options++;
375: fail_on_status_ne_200=!vany_status->as_bool();
376: }
377:
1.143.2.3 paf 378: if(valid_options!=options->count())
1.127 paf 379: throw Exception("parser.runtime",
1.143.2.21.2. (paf 380:): 0,
1.127 paf 381: "invalid option passed");
1.133 paf 382: }
1.126 paf 383:
384: //making request
1.143.2.3 paf 385: String request;
386: if(method)
1.143.2.4 paf 387: request<<method;
1.143.2.3 paf 388: else
389: request<<"GET";
390: request<< " "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";
1.127 paf 391: bool user_agent_specified=false;
392: if(vheaders && !vheaders->is_string()) { // allow empty
1.143.2.21.2. 4(paf 393:3): if(HashStringValue *headers=vheaders->get_hash()) {
394:3): Http_pass_header_info info={&charset, &request};
1.127 paf 395: headers->for_each(http_pass_header, &info);
396: user_agent_specified=info.user_agent_specified;
397: } else
398: throw Exception("parser.runtime",
1.143.2.21.2. 4(paf 399:3): &connect_string,
1.127 paf 400: "headers param must be hash");
401: };
402: if(!user_agent_specified) // defaulting
403: request << "user-agent: " DEFAULT_USER_AGENT "\n";
1.126 paf 404: request<<"\n";
405:
406: //sending request
1.143.2.3 paf 407: String response;
1.143.2.21.2. (paf 408:): const char* request_cstr=request.cstr(String::L_UNSPECIFIED);
409:): int status_code=http_request(response,
4(paf 410:3): host, port, request_cstr,
1.142 paf 411: timeout, fail_on_status_ne_200);
1.126 paf 412:
413: //processing results
1.143.2.21.2. 2(paf 414:3): size_t pos=response.pos("\r\n\r\n", 4);
415:3): if(pos==STRING_NOT_FOUND || pos<1){
1.126 paf 416: throw Exception("http.response",
1.143.2.21.2. 4(paf 417:3): &connect_string,
1.126 paf 418: "bad response from host - no headers found");
419: }
1.143.2.21.2. (paf 420:): const String& header_block=response.mid(0, pos);
421:): const String& body=response.mid(pos+4, response.length());
1.126 paf 422:
1.143.2.15 paf 423: ArrayString aheaders;
1.143.2.21.2. 4(paf 424:3): result.headers=new HashStringValue;
425:3): size_t pos_after=0;
426:3): header_block.split(aheaders, pos_after, "\r\n");
1.126 paf 427:
428: //processing headers
1.143.2.21.2. 6(paf 429:3): size_t aheaders_count=aheaders.count();
430:3): for(size_t i=1; i<aheaders_count; i++) {
4(paf 431:3): const String& line=*aheaders.get(i);
432:3): pos=line.pos(": ", 2);
2(paf 433:3): if(pos==STRING_NOT_FOUND || pos<1)
1.126 paf 434: throw Exception("http.response",
1.143.2.21.2. 4(paf 435:3): &connect_string,
436:3): "bad response from host - bad header \"%s\"", line.cstr());
1.143.2.3 paf 437:
1.143.2.16 paf 438: result.headers->put(
1.143.2.21.2. 4(paf 439:3): line.mid(0, pos).change_case(charset, String::CC_UPPER),
440:3): new VString(line.mid(pos+2, line.length())));
1.126 paf 441: }
442:
443: // output response
1.143.2.21.2. 4(paf 444:3): result.str=body.cstrm();
445:3): result.length=body.length();
446:3): result.headers->put(file_status_name, new VInt(status_code));
1.143.2.16 paf 447: return result;
1.34 paf 448: }
1.123 paf 449:
1.143.2.21.2. (paf 450:): #endif
451:):
1.123 paf 452: #ifndef DOXYGEN
453: struct File_read_action_info {
1.143.2.4 paf 454: char **data; size_t *data_size;
1.126 paf 455: };
1.123 paf 456: #endif
1.143.2.21.2. (paf 457:): static void file_read_action(
1.126 paf 458: struct stat& finfo,
1.123 paf 459: int f,
1.143.2.21.2. (paf 460:): const String& file_spec, const char* fname, bool as_text,
1.123 paf 461: void *context) {
1.126 paf 462: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 463: if(size_t to_read_size=(size_t)finfo.st_size) {
1.143.2.21.2. (paf 464:): *info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 465: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 466:
467: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 468: throw Exception(0,
1.143.2.21.2. (paf 469:): &file_spec,
1.123 paf 470: "read failed: actually read %lu bytes count not in [0..%lu] valid range",
1.126 paf 471: *info.data_size, to_read_size);
1.123 paf 472: } else { // empty file
473: if(as_text) {
1.143.2.21.2. (paf 474:): *info.data=new(PointerFreeGC) char[1];
1.123 paf 475: *(char*)(*info.data)=0;
476: } else
477: *info.data=0;
478: *info.data_size=0;
479: return;
480: }
1.126 paf 481: }
1.143.2.21.2. (paf 482:): File_read_result file_read(Charset& charset, const String& file_spec,
1.143.2.16 paf 483: bool as_text, HashStringValue *params,
1.126 paf 484: bool fail_on_read_problem) {
1.143.2.21.2. 1(paf 485:3): File_read_result result={false};
(paf 486:): #ifdef PA_HTTP
4(paf 487:3): if(file_spec.starts_with("http://")) {
1.126 paf 488: // fail on read problem
1.143.2.21.2. 8(paf 489:3): File_read_http_result http;//=file_read_http(charset, file_spec, params);
1.143.2.4 paf 490: result.success=true;
1.143.2.21.2. 8(paf 491:3): result.str="HELLO";//http.str;
492:3): result.length=5;//http.length;
493:3): result.headers=new HashStringValue;//http.headers;
1.126 paf 494: } else {
1.143.2.21.2. (paf 495:): #endif
496:): File_read_action_info info={&result.str, &result.length};
497:): result.success=file_read_action_under_lock(file_spec,
1.126 paf 498: "read", file_read_action, &info,
499: as_text, fail_on_read_problem);
1.143.2.21.2. (paf 500:): #ifdef PA_HTTP
1.126 paf 501: }
1.143.2.21.2. (paf 502:): #endif
1.123 paf 503:
1.143.2.4 paf 504: if(result.success && as_text) {
1.131 paf 505: // UTF-8 signature: EF BB BF
1.143.2.21.2. (paf 506:): if(result.length>=3) {
507:): char *in=(char *)result.str;
1.131 paf 508: if((in[0] == '\xEF') && (in[1] == '\xBB') &&
509: (in[2] == '\xBF')) {
1.143.2.21.2. (paf 510:): result.str=in+3; result.length-=3;// skip prefix
1.131 paf 511: }
512: }
513:
1.143.2.21.2. (paf 514:): fix_line_breaks((char *)(result.str), result.length);
1.123 paf 515: // note: after fixing
1.143.2.21.2. (paf 516:): ((char*&)(result.str))[result.length]=0;
1.123 paf 517: }
1.126 paf 518:
519: return result;
1.123 paf 520: }
521:
1.143.2.21.2. (paf 522:): bool file_read_action_under_lock(const String& file_spec,
1.126 paf 523: const char* action_name, File_read_action action, void *context,
524: bool as_text,
1.123 paf 525: bool fail_on_read_problem) {
1.143.2.21.2. (paf 526:): const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 527: int f;
528:
529: // first open, next stat:
1.45 paf 530: // directory update of NTFS hard links performed on open.
1.33 paf 531: // ex:
532: // a.html:^test[] and b.html hardlink to a.html
533: // user inserts ! before ^test in a.html
1.126 paf 534: // directory entry of b.html in NTFS not updated at once,
1.35 paf 535: // they delay update till open, so we would receive "!^test[" string
536: // if would do stat, next open.
1.123 paf 537: // later: it seems, even this does not help sometimes
1.98 paf 538: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 539: try {
540: if(lock_shared_blocking(f)!=0)
1.126 paf 541: throw Exception("file.lock",
1.143.2.21.2. (paf 542:): &file_spec,
1.123 paf 543: "shared lock failed: %s (%d), actual filename '%s'",
1.143.2.21.2. (paf 544:): strerror(errno), errno, fname);
1.123 paf 545:
1.124 paf 546: struct stat finfo;
547: if(stat(fname, &finfo)!=0)
548: throw Exception("file.missing", // hardly possible: we just opened it OK
1.143.2.21.2. (paf 549:): &file_spec,
1.124 paf 550: "stat failed: %s (%d), actual filename '%s'",
1.143.2.21.2. (paf 551:): strerror(errno), errno, fname);
1.124 paf 552:
1.140 paf 553: #ifdef PA_SAFE_MODE
554: if(finfo.st_uid/*foreign?*/!=geteuid()
555: && finfo.st_gid/*foreign?*/!=getegid())
1.126 paf 556: throw Exception("parser.runtime",
1.143.2.21.2. (paf 557:): &file_spec,
1.140 paf 558: "parser is in safe mode: reading files of foreign group and user disabled [recompile parser with --disable-safe-mode configure option], actual filename '%s'",
1.143.2.21.2. (paf 559:): fname);
1.105 paf 560: #endif
1.32 paf 561:
1.143.2.21.2. (paf 562:): action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 563: } catch(...) {
1.126 paf 564: unlock(f);close(f);
1.123 paf 565: if(fail_on_read_problem)
1.143.2.11 paf 566: rethrow;
1.123 paf 567: return false;
568: }
1.87 paf 569:
1.126 paf 570: unlock(f);close(f);
1.72 parser 571: return true;
1.118 paf 572: } else {
573: if(fail_on_read_problem)
1.126 paf 574: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.143.2.21.2. (paf 575:): &file_spec,
1.123 paf 576: "%s failed: %s (%d), actual filename '%s'",
1.143.2.21.2. (paf 577:): action_name, strerror(errno), errno, fname);
1.118 paf 578: return false;
579: }
1.8 paf 580: }
581:
1.143.2.21.2. (paf 582:): static void create_dir_for_file(const String& file_spec) {
1.63 parser 583: size_t pos_after=1;
1.143.2.21.2. 2(paf 584:3): size_t pos_before;
3(paf 585:3): while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
(paf 586:): mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 587: pos_after=pos_before+1;
588: }
589: }
590:
1.98 paf 591: bool file_write_action_under_lock(
1.143.2.21.2. (paf 592:): const String& file_spec,
1.126 paf 593: const char* action_name, File_write_action action, void *context,
594: bool as_text,
595: bool do_append,
596: bool do_block,
1.110 paf 597: bool fail_on_lock_problem) {
1.143.2.21.2. (paf 598:): const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 599: int f;
1.80 paf 600: if(access(fname, W_OK)!=0) // no
1.126 paf 601: create_dir_for_file(file_spec);
1.50 paf 602:
1.80 paf 603: if((f=open(fname,
604: O_CREAT|O_RDWR
605: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 606: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.99 paf 607: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.126 paf 608: Exception e("file.lock",
1.143.2.21.2. (paf 609:): &file_spec,
1.110 paf 610: "shared lock failed: %s (%d), actual filename '%s'",
1.143.2.21.2. (paf 611:): strerror(errno), errno, fname);
1.126 paf 612: close(f);
1.110 paf 613: if(fail_on_lock_problem)
614: throw e;
1.98 paf 615: return false;
616: }
1.96 paf 617:
618: try {
1.126 paf 619: action(f, context);
1.96 paf 620: } catch(...) {
1.138 paf 621: #ifdef HAVE_FTRUNCATE
1.104 paf 622: if(!do_append)
1.125 paf 623: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 624: #endif
1.126 paf 625: unlock(f);close(f);
1.143.2.11 paf 626: rethrow;
1.96 paf 627: }
1.80 paf 628:
1.138 paf 629: #ifdef HAVE_FTRUNCATE
1.104 paf 630: if(!do_append)
1.125 paf 631: 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 632: #endif
1.126 paf 633: unlock(f);close(f);
1.98 paf 634: return true;
1.80 paf 635: } else
1.126 paf 636: throw Exception(errno==EACCES?"file.access":0,
1.143.2.21.2. (paf 637:): &file_spec,
1.96 paf 638: "%s failed: %s (%d), actual filename '%s'",
1.143.2.21.2. (paf 639:): action_name, strerror(errno), errno, fname);
1.96 paf 640: // here should be nothing, see rethrow above
641: }
642:
643: #ifndef DOXYGEN
644: struct File_write_action_info {
1.143.2.21.2. (paf 645:): const char* str; size_t length;
1.126 paf 646: };
1.96 paf 647: #endif
648: static void file_write_action(int f, void *context) {
1.126 paf 649: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.143.2.21.2. (paf 650:): if(info.length) {
651:): int written=write(f, info.str, info.length);
1.116 paf 652: if(written<0)
1.126 paf 653: throw Exception(0,
1.143.2.21.2. (paf 654:): 0,
1.126 paf 655: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 656: }
1.96 paf 657: }
658: void file_write(
1.143.2.21.2. (paf 659:): const String& file_spec,
660:): const char* data, size_t size,
1.126 paf 661: bool as_text,
1.96 paf 662: bool do_append) {
1.126 paf 663: File_write_action_info info={data, size};
1.98 paf 664: file_write_action_under_lock(
1.143.2.4 paf 665: file_spec,
666: "write", file_write_action, &info,
667: as_text,
668: do_append);
1.30 paf 669: }
670:
1.63 parser 671: // throws nothing! [this is required in file_move & file_delete]
1.143.2.21.2. (paf 672:): static void rmdir(const String& file_spec, size_t pos_after) {
2(paf 673:3): size_t pos_before;
3(paf 674:3): if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
1.126 paf 675: rmdir(file_spec, pos_before+1);
1.50 paf 676:
1.143.2.21.2. (paf 677:): rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 678: }
1.143.2.21.2. (paf 679:): bool file_delete(const String& file_spec, bool fail_on_read_problem) {
680:): const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 681: if(unlink(fname)!=0)
1.93 paf 682: if(fail_on_read_problem)
1.126 paf 683: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.143.2.21.2. (paf 684:): &file_spec,
1.93 paf 685: "unlink failed: %s (%d), actual filename '%s'",
1.143.2.21.2. (paf 686:): strerror(errno), errno, fname);
1.93 paf 687: else
688: return false;
1.50 paf 689:
1.126 paf 690: rmdir(file_spec, 1);
1.93 paf 691: return true;
1.60 parser 692: }
1.143.2.21.2. (paf 693:): void file_move(const String& old_spec, const String& new_spec) {
694:): const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
695:): const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 696:
1.126 paf 697: create_dir_for_file(new_spec);
1.63 parser 698:
1.60 parser 699: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 700: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.143.2.21.2. (paf 701:): &old_spec,
1.60 parser 702: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.143.2.21.2. (paf 703:): strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 704:
1.126 paf 705: rmdir(old_spec, 1);
1.31 paf 706: }
707:
1.51 paf 708:
1.126 paf 709: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 710: struct stat lfinfo;
711: bool result=stat(fname, &lfinfo)==0;
712: if(afinfo)
713: *afinfo=lfinfo;
714: return result;
1.119 paf 715: }
716:
1.143.2.21.2. (paf 717:): bool entry_exists(const String& file_spec) {
718:): const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 719: return entry_exists(fname, 0);
1.118 paf 720: }
721:
1.143.2.21.2. (paf 722:): static bool entry_readable(const String& file_spec, bool need_dir) {
723:): char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 724: if(need_dir) {
1.126 paf 725: size_t size=strlen(fname);
1.120 paf 726: while(size) {
1.126 paf 727: char c=fname[size-1];
1.120 paf 728: if(c=='/' || c=='\\')
729: fname[--size]=0;
730: else
731: break;
732: }
733: }
1.51 paf 734: struct stat finfo;
1.118 paf 735: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 736: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 737: return is_dir==need_dir;
738: }
739: return false;
740: }
1.143.2.21.2. (paf 741:): bool file_readable(const String& file_spec) {
1.126 paf 742: return entry_readable(file_spec, false);
1.51 paf 743: }
1.143.2.21.2. (paf 744:): bool dir_readable(const String& file_spec) {
1.126 paf 745: return entry_readable(file_spec, true);
1.65 parser 746: }
1.143.2.21.2. 0(paf 747:3): const String* file_readable(const String& path, const String& name) {
(paf 748:): String& result=*new String(path);
749:): result << "/";
750:): result << name;
751:): return file_readable(result)?&result:0;
1.43 paf 752: }
1.143.2.21.2. (paf 753:): bool file_executable(const String& file_spec) {
754:): return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 755: }
756:
1.143.2.21.2. (paf 757:): bool file_stat(const String& file_spec,
1.58 parser 758: size_t& rsize,
1.126 paf 759: time_t& ratime,
760: time_t& rmtime,
761: time_t& rctime,
1.64 parser 762: bool fail_on_read_problem) {
1.143.2.21.2. (paf 763:): const char* fname=file_spec.cstr(String::L_FILE_SPEC);
5(paf 764:3): struct stat finfo;
1.44 paf 765: if(stat(fname, &finfo)!=0)
1.64 parser 766: if(fail_on_read_problem)
1.126 paf 767: throw Exception("file.missing",
1.143.2.21.2. (paf 768:): &file_spec,
1.67 parser 769: "getting file size failed: %s (%d), real filename '%s'",
1.143.2.21.2. (paf 770:): strerror(errno), errno, fname);
1.64 parser 771: else
772: return false;
1.58 parser 773: rsize=finfo.st_size;
774: ratime=finfo.st_atime;
775: rmtime=finfo.st_mtime;
776: rctime=finfo.st_ctime;
1.64 parser 777: return true;
1.18 paf 778: }
779:
1.126 paf 780: char* getrow(char* *row_ref, char delim) {
781: char* result=*row_ref;
1.8 paf 782: if(result) {
1.126 paf 783: *row_ref=strchr(result, delim);
1.8 paf 784: if(*row_ref)
785: *((*row_ref)++)=0;
786: else if(!*result)
787: return 0;
788: }
789: return result;
790: }
791:
1.126 paf 792: char* lsplit(char* string, char delim) {
1.23 paf 793: if(string) {
1.126 paf 794: char* v=strchr(string, delim);
1.8 paf 795: if(v) {
796: *v=0;
797: return v+1;
798: }
799: }
800: return 0;
801: }
802:
1.126 paf 803: char* lsplit(char* *string_ref, char delim) {
804: char* result=*string_ref;
805: char* next=lsplit(*string_ref, delim);
1.8 paf 806: *string_ref=next;
807: return result;
1.9 paf 808: }
809:
1.126 paf 810: char* rsplit(char* string, char delim) {
1.18 paf 811: if(string) {
1.126 paf 812: char* v=strrchr(string, delim);
1.18 paf 813: if(v) {
1.9 paf 814: *v=0;
815: return v+1;
816: }
817: }
818: return NULL;
1.10 paf 819: }
820:
1.37 paf 821: /// @todo less stupid type detection
1.143.2.21.2. (paf 822:): const char* format(double value, char* fmt) {
1.126 paf 823: char local_buf[MAX_NUMBER];
1.108 paf 824: size_t size;
825:
1.10 paf 826: if(fmt)
827: if(strpbrk(fmt, "diouxX"))
828: if(strpbrk(fmt, "ouxX"))
1.126 paf 829: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 830: else
1.126 paf 831: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 832: else
1.126 paf 833: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 834: else
1.126 paf 835: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 836:
1.143.2.21.2. (paf 837:): return pa_strdup(local_buf, size);
1.12 paf 838: }
839:
1.36 paf 840: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 841: #ifdef WIN32
842: do{
1.143.2.21.2. (paf 843:): int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 844: if(chunk_written<=0)
845: break;
846: size-=chunk_written;
1.36 paf 847: buf=((const char*)buf)+chunk_written;
1.126 paf 848: } while(size>0);
1.12 paf 849:
850: return size;
851: #else
1.126 paf 852: return fwrite(buf, 1, size, stdout);
1.12 paf 853: #endif
1.2 paf 854: }
1.14 paf 855:
1.143.2.21.2. (paf 856:): char* unescape_chars(const char* cp, int len) {
857:): char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 858: enum EscapeState {
1.33 paf 859: EscapeRest,
860: EscapeFirst,
1.14 paf 861: EscapeSecond
862: } escapeState=EscapeRest;
863: int escapedValue=0;
864: int srcPos=0;
865: int dstPos=0;
866: while(srcPos < len) {
1.126 paf 867: int ch=cp[srcPos];
1.14 paf 868: switch(escapeState) {
869: case EscapeRest:
870: if(ch=='%') {
871: escapeState=EscapeFirst;
872: } else if(ch=='+') {
1.126 paf 873: s[dstPos++]=' ';
1.14 paf 874: } else {
875: s[dstPos++]=ch;
876: }
877: break;
878: case EscapeFirst:
879: escapedValue=hex_value[ch] << 4;
880: escapeState=EscapeSecond;
881: break;
882: case EscapeSecond:
1.126 paf 883: escapedValue +=hex_value[ch];
1.14 paf 884: s[dstPos++]=escapedValue;
885: escapeState=EscapeRest;
886: break;
887: }
1.126 paf 888: srcPos++;
1.14 paf 889: }
890: s[dstPos]=0;
891: return s;
1.24 paf 892: }
893:
894: #ifdef WIN32
1.126 paf 895: void back_slashes_to_slashes(char* s) {
1.24 paf 896: if(s)
897: for(; *s; s++)
898: if(*s=='\\')
1.126 paf 899: *s='/';
1.24 paf 900: }
1.42 paf 901: /*
1.126 paf 902: void slashes_to_back_slashes(char* s) {
1.42 paf 903: if(s)
904: for(; *s; s++)
905: if(*s=='/')
1.126 paf 906: *s='\\';
1.42 paf 907: }
908: */
1.24 paf 909: #endif
1.41 paf 910:
1.126 paf 911: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 912: while(true) {
913: if(!(*s1)) {
914: if(!(*s2))
915: return true;
916: else
917: return !strict;
918: } else if(!(*s2))
919: return !strict;
920: if(isalpha(*s1)) {
921: if(tolower(*s1) !=tolower(*s2))
922: return false;
923: } else if((*s1) !=(*s2))
924: return false;
1.126 paf 925: s1++;
926: s2++;
1.41 paf 927: }
1.57 parser 928: }
929:
1.84 paf 930: static bool isLeap(int year) {
1.57 parser 931: return !(
932: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 933: );
1.57 parser 934: }
935:
936: int getMonthDays(int year, int month) {
937: int monthDays[]={
1.126 paf 938: 31,
939: isLeap(year) ? 29 : 28,
940: 31,
941: 30,
942: 31,
943: 30,
944: 31,
945: 31,
946: 30,
947: 31,
948: 30,
1.57 parser 949: 31
1.126 paf 950: };
951: return monthDays[month];
1.41 paf 952: }
1.69 parser 953:
1.126 paf 954: void remove_crlf(char* start, char* end) {
955: for(char* p=start; p<end; p++)
1.69 parser 956: switch(*p) {
1.126 paf 957: case '\n': *p='|'; break;
958: case '\r': *p=' '; break;
1.69 parser 959: }
1.91 paf 960: }
961:
962:
963: /// must be last in this file
964: #undef vsnprintf
1.126 paf 965: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 966: if(!s)
967: return 0;
968:
969: int r;
970: // note: on win32& maybe somewhere else
971: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
972: --s;
973: #if _MSC_VER
974: /*
975: win32:
976: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
977:
1.143.2.12 paf 978: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 979: */
1.126 paf 980: r=_vsnprintf(b, s, f, l);
1.91 paf 981: if(r<0)
982: r=s;
983: #else
1.126 paf 984: r=vsnprintf(b, s, f, l);
1.91 paf 985: /*
986: solaris:
987: man vsnprintf
988:
989: The snprintf() function returns the number of characters
990: formatted, that is, the number of characters that would have
991: been written to the buffer if it were large enough. If the
992: value of n is 0 on a call to snprintf(), an unspecified
993: value less than 1 is returned.
994: */
995:
996: if(r<0)
997: r=0;
998: else if(r>s)
999: r=s;
1000: #endif
1001: b[r]=0;
1002: return r;
1003: }
1004:
1.126 paf 1005: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1006: va_list l;
1.126 paf 1007: va_start(l, f);
1008: int r=__vsnprintf(b, s, f, l);
1009: va_end(l);
1.91 paf 1010: return r;
1.98 paf 1011: }
1012:
1013: int pa_sleep(unsigned long secs, unsigned long usecs) {
1.126 paf 1014: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
1.98 paf 1015:
1016: #ifdef WIN32
1.126 paf 1017: Sleep(secs * 1000 + usecs / 1000);
1.98 paf 1018: return 0;
1019: #else
1020: struct timeval t;
1021: t.tv_sec = secs;
1022: t.tv_usec = usecs;
1.126 paf 1023: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
1.98 paf 1024: #endif
1.135 paf 1025: }
1026:
1027:
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