Annotation of parser3/src/main/pa_common.C, revision 1.143.2.12
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:
1.143.2.12! paf 8: static const char* IDENT_COMMON_C="$Date: 2003/01/31 14:03:54 $";
1.1 paf 9:
1.143.2.12! paf 10: //#include "pa_value_includes.h"
1.1 paf 11: #include "pa_common.h"
1.4 paf 12: #include "pa_exception.h"
1.143.2.8 paf 13: #include "pa_value_includes.h"
1.126 paf 14: #include "pa_hash.h"
1.143.2.3 paf 15: #include "pa_vstring.h"
1.143.2.4 paf 16: #include "pa_vint.h"
1.1 paf 17:
1.143.2.12! paf 18: #ifdef CYGWIN
! 19: #define _GNU_H_WINDOWS32_SOCKETS
! 20: // for PASCAL
! 21: #include <windows.h>
! 22: // SOCKET
! 23: typedef u_int SOCKET;
! 24: int PASCAL closesocket(SOCKET);
1.126 paf 25: #else
1.143.2.12! paf 26: # if defined(WIN32)
! 27: # include <windows.h>
! 28: # else
! 29: # define closesocket close
! 30: # endif
1.98 paf 31: #endif
32:
1.93 paf 33: // some maybe-undefined constants
34:
1.82 paf 35: #ifndef _O_TEXT
36: # define _O_TEXT 0
37: #endif
38: #ifndef _O_BINARY
39: # define _O_BINARY 0
1.47 paf 40: #endif
1.80 paf 41:
1.138 paf 42: #ifdef HAVE_FTRUNCATE
43: # define PA_O_TRUNC 0
44: #else
45: # ifdef _O_TRUNC
46: # define PA_O_TRUNC _O_TRUNC
47: # else
48: # error you must have either ftruncate function or _O_TRUNC bit declared
49: # endif
50: #endif
51:
1.93 paf 52: // locking constants
53:
1.99 paf 54: #ifdef HAVE_FLOCK
55:
56: static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); }
57: static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); }
58: static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); }
59: static int unlock(int fd) { return flock(fd, LOCK_UN); }
60:
1.98 paf 61: #else
1.99 paf 62: #ifdef HAVE__LOCKING
1.98 paf 63:
1.126 paf 64: #define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1)
1.99 paf 65: static int lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); }
66: static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); }
67: static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); }
68: static int unlock(int fd) { FLOCK(_LK_UNLCK); }
1.93 paf 69:
1.99 paf 70: #else
71: #ifdef HAVE_FCNTL
1.93 paf 72:
1.126 paf 73: #define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls)
1.99 paf 74: static int lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); }
75: static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); }
76: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); }
77: static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); }
1.93 paf 78:
79: #else
80: #ifdef HAVE_LOCKF
1.99 paf 81:
1.126 paf 82: #define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1)
1.99 paf 83: static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking
84: static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); }
85: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); }
86: static int unlock(int fd) { FLOCK(F_TLOCK); }
87:
1.93 paf 88: #else
1.99 paf 89:
90: #error unable to find file locking func
91:
92: #endif
1.93 paf 93: #endif
94: #endif
95: #endif
96:
1.127 paf 97: #define DEFAULT_USER_AGENT "parser3"
98:
99:
1.126 paf 100: void fix_line_breaks(char* buf, size_t& size) {
1.139 paf 101: if(size==0)
102: return;
103:
1.87 paf 104: //_asm int 3;
1.126 paf 105: const char* const eob=buf+size;
106: char* dest=buf;
1.72 parser 107: // fix DOS: \r\n -> \n
108: // fix Macintosh: \r -> \n
1.126 paf 109: char* bol=buf;
1.137 paf 110: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 111: size_t len=eol-bol;
112: if(dest!=bol)
1.126 paf 113: memcpy(dest, bol, len);
1.72 parser 114: dest+=len;
1.126 paf 115: *dest++='\n';
1.72 parser 116:
1.126 paf 117: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 118: bol=eol+2;
1.126 paf 119: size--;
120: } else // \r, not \n = Macintosh
1.72 parser 121: bol=eol+1;
122: }
123: // last piece without \r, including terminating 0
124: if(dest!=bol)
1.126 paf 125: memcpy(dest, bol, eob-bol);
1.72 parser 126: }
1.18 paf 127:
1.143.2.3 paf 128: char* file_read_text(Pool& pool, Charset& charset,
1.143.2.9 paf 129: StringPtr file_spec,
1.126 paf 130: bool fail_on_read_problem,
1.143.2.3 paf 131: HashStringValue *params, HashStringValue** out_fields) {
1.143.2.4 paf 132: char *result; size_t size;
133: File_read_result file_read_result=file_read(pool, charset, file_spec, result, size, true, params, out_fields, fail_on_read_problem);
134: return file_read_result.success?result:0;
1.126 paf 135: }
136:
137: //http request stuff
138: /* ************************ http stuff *********************** */
139:
140: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
141: memset(addr, 0, sizeof(*addr));
142: addr->sin_family=AF_INET;
143: addr->sin_port=htons(port);
144: if(host) {
145: if(struct hostent *hostIP=gethostbyname(host))
146: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
147: else
148: return false;
149: } else
150: addr->sin_addr.s_addr=INADDR_ANY;
151: return true;
152: }
153:
1.143.2.1 paf 154: static int http_read_response(Pool& pool, String& response, int sock, bool fail_on_status_ne_200){
1.143.2.2 paf 155: int result=0;
1.130 paf 156: ssize_t EOLat=0;
1.126 paf 157: while(true) {
1.143.2.1 paf 158: char *buf=(char *)pool.malloc(MAX_STRING);
1.126 paf 159: ssize_t size=recv(sock, buf, MAX_STRING, 0);
160: if(size<=0)
161: break;
1.130 paf 162: response.APPEND_TAINTED(buf, size, "remote HTTP server response", 0);
1.143.2.2 paf 163: if(!result && (EOLat=response.pos("\r\n", 2))>=0) { // checking status in first response
1.143.2.9 paf 164: StringPtr status_line=response.mid(0, (size_t)EOLat);
165: Array<StringPtr> astatus;
1.143.2.3 paf 166: status_line->split(astatus, 0, " ", 1);
1.143.2.9 paf 167: StringPtr status_code=astatus.get(1);
1.143.2.3 paf 168: result=status_code->as_int();
1.142 paf 169:
1.143.2.3 paf 170: if(fail_on_status_ne_200 && result!=200)
1.142 paf 171: throw Exception("http.status",
172: status_code,
173: "invalid HTTP response status");
174: }
175: }
1.143.2.2 paf 176: if(result)
177: return result;
1.142 paf 178: else
179: throw Exception("http.response",
1.143.2.3 paf 180: Exception::undefined_source,
1.142 paf 181: "bad response from host - no status found (size=%lu)", response.size());
1.126 paf 182: }
183:
184: /* ********************** request *************************** */
185:
186: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
187: # define WE_CAN_USE_ALARM
188: #endif
189:
190: #ifdef WE_CAN_USE_ALARM
191: static sigjmp_buf timeout_env;
192: static void timeout_handler(int sig){
193: siglongjmp(timeout_env, 1);
194: }
195: #endif
196:
1.143.2.3 paf 197: static int http_request(Pool& pool,
198: String& response,
1.143.2.9 paf 199: StringPtr origin_string,
1.143.2.3 paf 200: const char* host, int port,
201: const char* request,
202: int timeout,
203: bool fail_on_status_ne_200) {
1.126 paf 204: if(!host)
205: throw Exception("http.host",
206: origin_string,
207: "zero hostname"); //never
208:
209: #ifdef WE_CAN_USE_ALARM
210: signal(SIGALRM, timeout_handler);
211: #endif
212: int sock=-1;
213: try {
1.142 paf 214: int result;
1.126 paf 215: #ifdef WE_CAN_USE_ALARM
216: if(sigsetjmp(timeout_env, 1))
217: throw Exception("http.timeout",
218: origin_string,
219: "timeout occured while retrieving document");
220: else {
221: alarm(timeout);
222: #endif
223: struct sockaddr_in dest;
224:
225: if(!set_addr(&dest, host, port))
226: throw Exception("http.host",
227: origin_string,
1.127 paf 228: "can not resolve hostname \"%s\"", host);
1.126 paf 229:
230: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
231: throw Exception("http.connect",
232: origin_string,
1.127 paf 233: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 234: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
235: throw Exception("http.connect",
236: origin_string,
1.127 paf 237: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 238: size_t request_size=strlen(request);
239: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
240: throw Exception("http.connect",
241: origin_string,
1.127 paf 242: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 243:
1.143.2.3 paf 244: result=http_read_response(pool, response, sock, fail_on_status_ne_200);
1.142 paf 245: closesocket(sock);
1.126 paf 246: #ifdef WE_CAN_USE_ALARM
1.142 paf 247: alarm(0);
1.126 paf 248: }
249: #endif
1.142 paf 250: return result;
1.126 paf 251: } catch(...) {
252: if(sock>=0)
253: closesocket(sock);
254: #ifdef WE_CAN_USE_ALARM
255: alarm(0);
256: #endif
1.143.2.11 paf 257: rethrow;
1.126 paf 258: }
259: }
260:
1.127 paf 261: #ifndef DOXYGEN
262: struct Http_pass_header_info {
1.143.2.3 paf 263: Pool* pool;
264: Charset* charset;
1.127 paf 265: String* request;
266: bool user_agent_specified;
267: };
268: #endif
1.143.2.3 paf 269: static void http_pass_header(HashStringValue::key_type key, HashStringValue::value_type value,
270: Http_pass_header_info *info) {
271: *(info->request)<<*key<<": "
1.143.2.9 paf 272: << *attributed_meaning_to_string(*info->pool, value, String::UL_HTTP_HEADER, false)
1.135 paf 273: <<"\n";
274:
1.143.2.3 paf 275: if(*key->change_case(*info->pool, *info->charset, String::CC_UPPER)=="USER-AGENT")
276: info->user_agent_specified=true;
1.126 paf 277: }
1.143.2.9 paf 278: static HashStringValuePtr file_read_http(Pool& pool, Charset& charset, StringPtr file_spec,
1.143.2.3 paf 279: char*& data, size_t& data_size,
280: HashStringValue *options=0, bool return_outfields=false) {
281: HashStringValuePtr result(0);
1.126 paf 282: char host[MAX_STRING];
1.129 paf 283: const char* uri;
1.126 paf 284: int port;
1.143.2.5 paf 285: CharPtr method(0);
1.126 paf 286: int timeout=2;
1.142 paf 287: bool fail_on_status_ne_200=true;
1.143.2.3 paf 288: ValuePtr vheaders(0);
1.126 paf 289:
1.143.2.3 paf 290: StringPtr connect_string(new String());
1.133 paf 291: // not in ^sql{... UL_SQL ...} spirit, but closer to ^file::load one
1.143.2.4 paf 292: connect_string->append(*file_spec, String::UL_URI); // tainted pieces -> URI pieces
1.133 paf 293:
1.143.2.3 paf 294: CharPtr connect_string_cstr=connect_string->cstr(String::UL_UNSPECIFIED);
1.143.2.10 paf 295: const char* current=connect_string_cstr;
1.143.2.3 paf 296: if(strncmp(current, "http://", 7)!=0)
1.126 paf 297: throw Exception(0,
1.143.2.3 paf 298: connect_string,
1.126 paf 299: "does not start with http://"); //never
1.143.2.3 paf 300: current+=7;
1.126 paf 301:
1.143.2.3 paf 302: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
1.126 paf 303: char* host_uri=lsplit(host, '/');
1.143.2.3 paf 304: uri=host_uri?current+(host_uri-1-host):"/";
1.126 paf 305: char* port_cstr=lsplit(host, ':');
306: char* error_pos=0;
307: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
308:
1.127 paf 309: if(options) {
310: int valid_options=0;
1.143.2.3 paf 311: if(ValuePtr vmethod=options->get(http_method_name)) {
1.127 paf 312: valid_options++;
1.143.2.8 paf 313: method=vmethod->as_string(&pool)->cstr();
1.127 paf 314: }
1.143.2.3 paf 315: if(ValuePtr vtimeout=options->get(http_timeout_name)) {
1.127 paf 316: valid_options++;
317: timeout=vtimeout->as_int();
318: }
1.143.2.3 paf 319: if(vheaders=options->get(http_headers_name)) {
1.127 paf 320: valid_options++;
321: }
1.143.2.3 paf 322: if(ValuePtr vany_status=options->get(http_any_status_name)) {
1.142 paf 323: valid_options++;
324: fail_on_status_ne_200=!vany_status->as_bool();
325: }
326:
1.143.2.3 paf 327: if(valid_options!=options->count())
1.127 paf 328: throw Exception("parser.runtime",
1.143.2.3 paf 329: Exception::undefined_source,
1.127 paf 330: "invalid option passed");
1.133 paf 331: }
1.126 paf 332:
333: //making request
1.143.2.3 paf 334: String request;
335: if(method)
1.143.2.4 paf 336: request<<method;
1.143.2.3 paf 337: else
338: request<<"GET";
339: request<< " "<< uri <<" HTTP/1.0\nHost: "<< host<<"\n";
1.127 paf 340: bool user_agent_specified=false;
341: if(vheaders && !vheaders->is_string()) { // allow empty
1.143.2.8 paf 342: if(HashStringValue *headers=vheaders->get_hash(connect_string)) {
1.143.2.3 paf 343: Http_pass_header_info info={&pool, &charset, &request};
1.127 paf 344: headers->for_each(http_pass_header, &info);
345: user_agent_specified=info.user_agent_specified;
346: } else
347: throw Exception("parser.runtime",
1.143.2.3 paf 348: connect_string,
1.127 paf 349: "headers param must be hash");
350: };
351: if(!user_agent_specified) // defaulting
352: request << "user-agent: " DEFAULT_USER_AGENT "\n";
1.126 paf 353: request<<"\n";
354:
355: //sending request
1.143.2.3 paf 356: String response;
1.143.2.5 paf 357: CharPtr request_cstr=request.cstr(String::UL_UNSPECIFIED);
1.143.2.3 paf 358: int status_code=http_request(pool, response,
1.143.2.4 paf 359: connect_string, host, port, request_cstr,
1.142 paf 360: timeout, fail_on_status_ne_200);
1.126 paf 361:
362: //processing results
363: int pos=response.pos("\r\n\r\n", 4);
364: if(pos<1){
365: throw Exception("http.response",
1.143.2.3 paf 366: connect_string,
1.126 paf 367: "bad response from host - no headers found");
368: }
1.143.2.3 paf 369: StringPtr header_block=response.mid(0, pos);
370: StringPtr body=response.mid(pos+4, response.size());
1.126 paf 371:
1.143.2.9 paf 372: Array<StringPtr> aheaders;
1.143.2.3 paf 373: HashStringValuePtr headers(new HashStringValue);
374: header_block->split(aheaders, 0, "\r\n", 2);
1.126 paf 375:
376: //processing headers
1.143.2.3 paf 377: for(int i=1; i<aheaders.count(); i++) {
1.143.2.9 paf 378: StringPtr line=aheaders.get(i);
1.143.2.3 paf 379: pos=line->pos(": ", 2);
380: if(pos<1)
1.126 paf 381: throw Exception("http.response",
1.143.2.3 paf 382: connect_string,
1.143.2.12! paf 383: "bad response from host - bad header \"%s\"", line->cstr().get());
1.143.2.3 paf 384:
385: headers->put(
386: line->mid(0, pos)->change_case(pool, charset, String::CC_UPPER),
387: ValuePtr(new VString(line->mid(pos+2, line->size()))));
1.126 paf 388: }
389:
390: // output response
1.143.2.3 paf 391: data=new(pool) char[data_size=body->size()];
392: body->store_to(data, String::UL_AS_IS);
393: if(return_outfields) {
1.143.2.4 paf 394: headers->put(file_status_name, ValuePtr(new VInt(status_code)));
395: return headers;
396: } else
397: return HashStringValuePtr(0);
1.34 paf 398: }
1.123 paf 399:
400: #ifndef DOXYGEN
401: struct File_read_action_info {
1.143.2.4 paf 402: char **data; size_t *data_size;
1.126 paf 403: };
1.123 paf 404: #endif
1.143.2.4 paf 405: static void file_read_action(Pool& pool,
1.126 paf 406: struct stat& finfo,
1.123 paf 407: int f,
1.143.2.9 paf 408: StringPtr file_spec, const char* fname, bool as_text,
1.123 paf 409: void *context) {
1.126 paf 410: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 411: if(size_t to_read_size=(size_t)finfo.st_size) {
1.143.2.4 paf 412: *info.data=pool.malloc(to_read_size+(as_text?1:0));
1.126 paf 413: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 414:
415: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 416: throw Exception(0,
1.143.2.4 paf 417: file_spec,
1.123 paf 418: "read failed: actually read %lu bytes count not in [0..%lu] valid range",
1.126 paf 419: *info.data_size, to_read_size);
1.123 paf 420: } else { // empty file
421: if(as_text) {
1.126 paf 422: *info.data=pool.malloc(1);
1.123 paf 423: *(char*)(*info.data)=0;
424: } else
425: *info.data=0;
426: *info.data_size=0;
427: return;
428: }
1.126 paf 429: }
1.143.2.9 paf 430: File_read_result file_read(Pool& pool, Charset& charset, StringPtr file_spec,
1.143.2.4 paf 431: char*& data, size_t& data_size,
1.143.2.3 paf 432: bool as_text, HashStringValue *params, HashStringValue** out_fields,
1.126 paf 433: bool fail_on_read_problem) {
1.143.2.4 paf 434: File_read_result result;
435: if(file_spec->starts_with("http://", 7)) {
1.126 paf 436: // fail on read problem
1.143.2.4 paf 437: result.fields=file_read_http(pool, charset, file_spec, data, data_size, params);
438: result.success=true;
1.126 paf 439: } else {
440: File_read_action_info info={&data, &data_size};
1.143.2.4 paf 441: result.success=file_read_action_under_lock(pool, file_spec,
1.126 paf 442: "read", file_read_action, &info,
443: as_text, fail_on_read_problem);
444: }
1.123 paf 445:
1.143.2.4 paf 446: if(result.success && as_text) {
1.131 paf 447: // UTF-8 signature: EF BB BF
448: if(data_size>=3) {
449: char *in=(char *)data;
450: if((in[0] == '\xEF') && (in[1] == '\xBB') &&
451: (in[2] == '\xBF')) {
452: data=in+3; data_size-=3;// skip prefix
453: }
454: }
455:
1.126 paf 456: fix_line_breaks((char *)(data), data_size);
1.123 paf 457: // note: after fixing
1.126 paf 458: ((char*&)(data))[data_size]=0;
1.123 paf 459: }
1.126 paf 460:
461: return result;
1.123 paf 462: }
463:
1.143.2.9 paf 464: bool file_read_action_under_lock(Pool& pool, StringPtr file_spec,
1.126 paf 465: const char* action_name, File_read_action action, void *context,
466: bool as_text,
1.123 paf 467: bool fail_on_read_problem) {
1.143.2.5 paf 468: CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC);
1.33 paf 469: int f;
470:
471: // first open, next stat:
1.45 paf 472: // directory update of NTFS hard links performed on open.
1.33 paf 473: // ex:
474: // a.html:^test[] and b.html hardlink to a.html
475: // user inserts ! before ^test in a.html
1.126 paf 476: // directory entry of b.html in NTFS not updated at once,
1.35 paf 477: // they delay update till open, so we would receive "!^test[" string
478: // if would do stat, next open.
1.123 paf 479: // later: it seems, even this does not help sometimes
1.98 paf 480: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 481: try {
482: if(lock_shared_blocking(f)!=0)
1.126 paf 483: throw Exception("file.lock",
1.143.2.4 paf 484: file_spec,
1.123 paf 485: "shared lock failed: %s (%d), actual filename '%s'",
1.143.2.12! paf 486: strerror(errno), errno, fname.get());
1.123 paf 487:
1.124 paf 488: struct stat finfo;
489: if(stat(fname, &finfo)!=0)
490: throw Exception("file.missing", // hardly possible: we just opened it OK
1.143.2.4 paf 491: file_spec,
1.124 paf 492: "stat failed: %s (%d), actual filename '%s'",
1.143.2.12! paf 493: strerror(errno), errno, fname.get());
1.124 paf 494:
1.140 paf 495: #ifdef PA_SAFE_MODE
496: if(finfo.st_uid/*foreign?*/!=geteuid()
497: && finfo.st_gid/*foreign?*/!=getegid())
1.126 paf 498: throw Exception("parser.runtime",
1.143.2.4 paf 499: file_spec,
1.140 paf 500: "parser is in safe mode: reading files of foreign group and user disabled [recompile parser with --disable-safe-mode configure option], actual filename '%s'",
1.143.2.12! paf 501: fname.get());
1.105 paf 502: #endif
1.32 paf 503:
1.126 paf 504: action(pool, finfo, f, file_spec, fname, as_text, context);
1.123 paf 505: } catch(...) {
1.126 paf 506: unlock(f);close(f);
1.123 paf 507: if(fail_on_read_problem)
1.143.2.11 paf 508: rethrow;
1.123 paf 509: return false;
510: }
1.87 paf 511:
1.126 paf 512: unlock(f);close(f);
1.72 parser 513: return true;
1.118 paf 514: } else {
515: if(fail_on_read_problem)
1.126 paf 516: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.143.2.4 paf 517: file_spec,
1.123 paf 518: "%s failed: %s (%d), actual filename '%s'",
1.143.2.12! paf 519: action_name, strerror(errno), errno, fname.get());
1.118 paf 520: return false;
521: }
1.8 paf 522: }
523:
1.143.2.9 paf 524: static void create_dir_for_file(StringPtr file_spec) {
1.63 parser 525: size_t pos_after=1;
526: int pos_before;
1.143.2.4 paf 527: while((pos_before=file_spec->pos("/", 1, pos_after))>=0) {
528: mkdir(file_spec->mid(0, pos_before)->cstr(String::UL_FILE_SPEC), 0775);
1.63 parser 529: pos_after=pos_before+1;
530: }
531: }
532:
1.98 paf 533: bool file_write_action_under_lock(
1.143.2.9 paf 534: StringPtr file_spec,
1.126 paf 535: const char* action_name, File_write_action action, void *context,
536: bool as_text,
537: bool do_append,
538: bool do_block,
1.110 paf 539: bool fail_on_lock_problem) {
1.143.2.5 paf 540: CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC);
1.28 paf 541: int f;
1.80 paf 542: if(access(fname, W_OK)!=0) // no
1.126 paf 543: create_dir_for_file(file_spec);
1.50 paf 544:
1.80 paf 545: if((f=open(fname,
546: O_CREAT|O_RDWR
547: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 548: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.99 paf 549: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.126 paf 550: Exception e("file.lock",
1.143.2.4 paf 551: file_spec,
1.110 paf 552: "shared lock failed: %s (%d), actual filename '%s'",
1.143.2.12! paf 553: strerror(errno), errno, fname.get());
1.126 paf 554: close(f);
1.110 paf 555: if(fail_on_lock_problem)
556: throw e;
1.98 paf 557: return false;
558: }
1.96 paf 559:
560: try {
1.126 paf 561: action(f, context);
1.96 paf 562: } catch(...) {
1.138 paf 563: #ifdef HAVE_FTRUNCATE
1.104 paf 564: if(!do_append)
1.125 paf 565: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 566: #endif
1.126 paf 567: unlock(f);close(f);
1.143.2.11 paf 568: rethrow;
1.96 paf 569: }
1.80 paf 570:
1.138 paf 571: #ifdef HAVE_FTRUNCATE
1.104 paf 572: if(!do_append)
1.125 paf 573: 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 574: #endif
1.126 paf 575: unlock(f);close(f);
1.98 paf 576: return true;
1.80 paf 577: } else
1.126 paf 578: throw Exception(errno==EACCES?"file.access":0,
1.143.2.4 paf 579: file_spec,
1.96 paf 580: "%s failed: %s (%d), actual filename '%s'",
1.143.2.12! paf 581: action_name, strerror(errno), errno, fname.get());
1.96 paf 582: // here should be nothing, see rethrow above
583: }
584:
585: #ifndef DOXYGEN
586: struct File_write_action_info {
587: const void *data; size_t size;
1.126 paf 588: };
1.96 paf 589: #endif
590: static void file_write_action(int f, void *context) {
1.126 paf 591: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.113 paf 592: if(info.size) {
1.126 paf 593: int written=write(f, info.data, info.size);
1.116 paf 594: if(written<0)
1.126 paf 595: throw Exception(0,
1.143.2.4 paf 596: Exception::undefined_source,
1.126 paf 597: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 598: }
1.96 paf 599: }
600: void file_write(
1.143.2.9 paf 601: StringPtr file_spec,
1.96 paf 602: const void *data, size_t size,
1.126 paf 603: bool as_text,
1.96 paf 604: bool do_append) {
1.126 paf 605: File_write_action_info info={data, size};
1.98 paf 606: file_write_action_under_lock(
1.143.2.4 paf 607: file_spec,
608: "write", file_write_action, &info,
609: as_text,
610: do_append);
1.30 paf 611: }
612:
1.63 parser 613: // throws nothing! [this is required in file_move & file_delete]
1.143.2.9 paf 614: static void rmdir(StringPtr file_spec, size_t pos_after) {
1.50 paf 615: int pos_before;
1.143.2.4 paf 616: if((pos_before=file_spec->pos("/", 1, pos_after))>=0)
1.126 paf 617: rmdir(file_spec, pos_before+1);
1.50 paf 618:
1.143.2.4 paf 619: rmdir(file_spec->mid(0, pos_after-1/* / */)->cstr(String::UL_FILE_SPEC));
1.50 paf 620: }
1.143.2.9 paf 621: bool file_delete(StringPtr file_spec, bool fail_on_read_problem) {
1.143.2.5 paf 622: CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC);
1.54 parser 623: if(unlink(fname)!=0)
1.93 paf 624: if(fail_on_read_problem)
1.126 paf 625: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.143.2.4 paf 626: file_spec,
1.93 paf 627: "unlink failed: %s (%d), actual filename '%s'",
1.143.2.12! paf 628: strerror(errno), errno, fname.get());
1.93 paf 629: else
630: return false;
1.50 paf 631:
1.126 paf 632: rmdir(file_spec, 1);
1.93 paf 633: return true;
1.60 parser 634: }
1.143.2.9 paf 635: void file_move(StringPtr old_spec, StringPtr new_spec) {
1.143.2.5 paf 636: CharPtr old_spec_cstr=old_spec->cstr(String::UL_FILE_SPEC);
637: CharPtr new_spec_cstr=new_spec->cstr(String::UL_FILE_SPEC);
1.63 parser 638:
1.126 paf 639: create_dir_for_file(new_spec);
1.63 parser 640:
1.60 parser 641: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 642: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.143.2.4 paf 643: old_spec,
1.60 parser 644: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.143.2.12! paf 645: strerror(errno), errno, old_spec_cstr.get(), new_spec_cstr.get());
1.63 parser 646:
1.126 paf 647: rmdir(old_spec, 1);
1.31 paf 648: }
649:
1.51 paf 650:
1.126 paf 651: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 652: struct stat lfinfo;
653: bool result=stat(fname, &lfinfo)==0;
654: if(afinfo)
655: *afinfo=lfinfo;
656: return result;
1.119 paf 657: }
658:
1.143.2.9 paf 659: bool entry_exists(StringPtr file_spec) {
1.143.2.5 paf 660: CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC);
1.126 paf 661: return entry_exists(fname, 0);
1.118 paf 662: }
663:
1.143.2.9 paf 664: static bool entry_readable(StringPtr file_spec, bool need_dir) {
1.143.2.4 paf 665: CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC);
1.120 paf 666: if(need_dir) {
1.126 paf 667: size_t size=strlen(fname);
1.120 paf 668: while(size) {
1.126 paf 669: char c=fname[size-1];
1.120 paf 670: if(c=='/' || c=='\\')
671: fname[--size]=0;
672: else
673: break;
674: }
675: }
1.51 paf 676: struct stat finfo;
1.118 paf 677: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 678: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 679: return is_dir==need_dir;
680: }
681: return false;
682: }
1.143.2.9 paf 683: bool file_readable(StringPtr file_spec) {
1.126 paf 684: return entry_readable(file_spec, false);
1.51 paf 685: }
1.143.2.9 paf 686: bool dir_readable(StringPtr file_spec) {
1.126 paf 687: return entry_readable(file_spec, true);
1.65 parser 688: }
1.143.2.9 paf 689: StringPtr file_readable(StringPtr path, const String& name) {
1.143.2.4 paf 690: StringPtr result(new String(*path));
1.126 paf 691: *result << "/";
1.65 parser 692: *result << name;
1.143.2.9 paf 693: return file_readable(result)?result:StringPtr(0);
1.43 paf 694: }
1.143.2.9 paf 695: bool file_executable(StringPtr file_spec) {
1.143.2.4 paf 696: return access(file_spec->cstr(String::UL_FILE_SPEC), X_OK)==0;
1.44 paf 697: }
698:
1.143.2.9 paf 699: bool file_stat(StringPtr file_spec,
1.58 parser 700: size_t& rsize,
1.126 paf 701: time_t& ratime,
702: time_t& rmtime,
703: time_t& rctime,
1.64 parser 704: bool fail_on_read_problem) {
1.143.2.5 paf 705: CharPtr fname=file_spec->cstr(String::UL_FILE_SPEC);
1.44 paf 706: struct stat finfo;
707: if(stat(fname, &finfo)!=0)
1.64 parser 708: if(fail_on_read_problem)
1.126 paf 709: throw Exception("file.missing",
1.143.2.4 paf 710: file_spec,
1.67 parser 711: "getting file size failed: %s (%d), real filename '%s'",
1.143.2.12! paf 712: strerror(errno), errno, fname.get());
1.64 parser 713: else
714: return false;
1.58 parser 715: rsize=finfo.st_size;
716: ratime=finfo.st_atime;
717: rmtime=finfo.st_mtime;
718: rctime=finfo.st_ctime;
1.64 parser 719: return true;
1.18 paf 720: }
721:
1.126 paf 722: char* getrow(char* *row_ref, char delim) {
723: char* result=*row_ref;
1.8 paf 724: if(result) {
1.126 paf 725: *row_ref=strchr(result, delim);
1.8 paf 726: if(*row_ref)
727: *((*row_ref)++)=0;
728: else if(!*result)
729: return 0;
730: }
731: return result;
732: }
733:
1.126 paf 734: char* lsplit(char* string, char delim) {
1.23 paf 735: if(string) {
1.126 paf 736: char* v=strchr(string, delim);
1.8 paf 737: if(v) {
738: *v=0;
739: return v+1;
740: }
741: }
742: return 0;
743: }
744:
1.126 paf 745: char* lsplit(char* *string_ref, char delim) {
746: char* result=*string_ref;
747: char* next=lsplit(*string_ref, delim);
1.8 paf 748: *string_ref=next;
749: return result;
1.9 paf 750: }
751:
1.126 paf 752: char* rsplit(char* string, char delim) {
1.18 paf 753: if(string) {
1.126 paf 754: char* v=strrchr(string, delim);
1.18 paf 755: if(v) {
1.9 paf 756: *v=0;
757: return v+1;
758: }
759: }
760: return NULL;
1.10 paf 761: }
762:
1.37 paf 763: /// @todo less stupid type detection
1.143.2.10 paf 764: const char* format(Pool& pool, double value, char* fmt) {
1.126 paf 765: char local_buf[MAX_NUMBER];
1.108 paf 766: size_t size;
767:
1.10 paf 768: if(fmt)
769: if(strpbrk(fmt, "diouxX"))
770: if(strpbrk(fmt, "ouxX"))
1.126 paf 771: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 772: else
1.126 paf 773: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 774: else
1.126 paf 775: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 776: else
1.126 paf 777: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 778:
1.143.2.7 paf 779: return pool.copy(local_buf, size+1);
1.12 paf 780: }
781:
1.36 paf 782: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 783: #ifdef WIN32
784: do{
1.126 paf 785: int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout);
1.12 paf 786: if(chunk_written<=0)
787: break;
788: size-=chunk_written;
1.36 paf 789: buf=((const char*)buf)+chunk_written;
1.126 paf 790: } while(size>0);
1.12 paf 791:
792: return size;
793: #else
1.126 paf 794: return fwrite(buf, 1, size, stdout);
1.12 paf 795: #endif
1.2 paf 796: }
1.14 paf 797:
1.143.2.4 paf 798: char* unescape_chars(Pool& pool, const char* cp, int len) {
799: char* s=pool.malloc(len + 1);
1.14 paf 800: enum EscapeState {
1.33 paf 801: EscapeRest,
802: EscapeFirst,
1.14 paf 803: EscapeSecond
804: } escapeState=EscapeRest;
805: int escapedValue=0;
806: int srcPos=0;
807: int dstPos=0;
808: while(srcPos < len) {
1.126 paf 809: int ch=cp[srcPos];
1.14 paf 810: switch(escapeState) {
811: case EscapeRest:
812: if(ch=='%') {
813: escapeState=EscapeFirst;
814: } else if(ch=='+') {
1.126 paf 815: s[dstPos++]=' ';
1.14 paf 816: } else {
817: s[dstPos++]=ch;
818: }
819: break;
820: case EscapeFirst:
821: escapedValue=hex_value[ch] << 4;
822: escapeState=EscapeSecond;
823: break;
824: case EscapeSecond:
1.126 paf 825: escapedValue +=hex_value[ch];
1.14 paf 826: s[dstPos++]=escapedValue;
827: escapeState=EscapeRest;
828: break;
829: }
1.126 paf 830: srcPos++;
1.14 paf 831: }
832: s[dstPos]=0;
833: return s;
1.24 paf 834: }
835:
836: #ifdef WIN32
1.126 paf 837: void back_slashes_to_slashes(char* s) {
1.24 paf 838: if(s)
839: for(; *s; s++)
840: if(*s=='\\')
1.126 paf 841: *s='/';
1.24 paf 842: }
1.42 paf 843: /*
1.126 paf 844: void slashes_to_back_slashes(char* s) {
1.42 paf 845: if(s)
846: for(; *s; s++)
847: if(*s=='/')
1.126 paf 848: *s='\\';
1.42 paf 849: }
850: */
1.24 paf 851: #endif
1.41 paf 852:
1.126 paf 853: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 854: while(true) {
855: if(!(*s1)) {
856: if(!(*s2))
857: return true;
858: else
859: return !strict;
860: } else if(!(*s2))
861: return !strict;
862: if(isalpha(*s1)) {
863: if(tolower(*s1) !=tolower(*s2))
864: return false;
865: } else if((*s1) !=(*s2))
866: return false;
1.126 paf 867: s1++;
868: s2++;
1.41 paf 869: }
1.57 parser 870: }
871:
1.84 paf 872: static bool isLeap(int year) {
1.57 parser 873: return !(
874: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 875: );
1.57 parser 876: }
877:
878: int getMonthDays(int year, int month) {
879: int monthDays[]={
1.126 paf 880: 31,
881: isLeap(year) ? 29 : 28,
882: 31,
883: 30,
884: 31,
885: 30,
886: 31,
887: 31,
888: 30,
889: 31,
890: 30,
1.57 parser 891: 31
1.126 paf 892: };
893: return monthDays[month];
1.41 paf 894: }
1.69 parser 895:
1.126 paf 896: void remove_crlf(char* start, char* end) {
897: for(char* p=start; p<end; p++)
1.69 parser 898: switch(*p) {
1.126 paf 899: case '\n': *p='|'; break;
900: case '\r': *p=' '; break;
1.69 parser 901: }
1.91 paf 902: }
903:
904:
905: /// must be last in this file
906: #undef vsnprintf
1.126 paf 907: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 908: if(!s)
909: return 0;
910:
911: int r;
912: // note: on win32& maybe somewhere else
913: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
914: --s;
915: #if _MSC_VER
916: /*
917: win32:
918: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
919:
1.143.2.12! paf 920: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 921: */
1.126 paf 922: r=_vsnprintf(b, s, f, l);
1.91 paf 923: if(r<0)
924: r=s;
925: #else
1.126 paf 926: r=vsnprintf(b, s, f, l);
1.91 paf 927: /*
928: solaris:
929: man vsnprintf
930:
931: The snprintf() function returns the number of characters
932: formatted, that is, the number of characters that would have
933: been written to the buffer if it were large enough. If the
934: value of n is 0 on a call to snprintf(), an unspecified
935: value less than 1 is returned.
936: */
937:
938: if(r<0)
939: r=0;
940: else if(r>s)
941: r=s;
942: #endif
943: b[r]=0;
944: return r;
945: }
946:
1.126 paf 947: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 948: va_list l;
1.126 paf 949: va_start(l, f);
950: int r=__vsnprintf(b, s, f, l);
951: va_end(l);
1.91 paf 952: return r;
1.98 paf 953: }
954:
955: int pa_sleep(unsigned long secs, unsigned long usecs) {
1.126 paf 956: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
1.98 paf 957:
958: #ifdef WIN32
1.126 paf 959: Sleep(secs * 1000 + usecs / 1000);
1.98 paf 960: return 0;
961: #else
962: struct timeval t;
963: t.tv_sec = secs;
964: t.tv_usec = usecs;
1.126 paf 965: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
1.98 paf 966: #endif
1.135 paf 967: }
968:
969:
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