Annotation of parser3/src/main/pa_common.C, revision 1.154
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.154 ! 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:
1.154 ! paf 8: static const char* IDENT_COMMON_C="$Date: 2003/07/21 07:09:02 $";
1.1 paf 9:
10: #include "pa_common.h"
1.4 paf 11: #include "pa_exception.h"
1.154 ! paf 12: #include "pa_hash.h"
1.14 paf 13: #include "pa_globals.h"
1.154 ! paf 14: #include "pa_request_charsets.h"
! 15: #include "pa_charsets.h"
! 16:
! 17: #define PA_HTTP
! 18:
! 19: #ifdef PA_HTTP
1.126 paf 20: #include "pa_vstring.h"
1.154 ! paf 21: #include "pa_vint.h"
! 22:
! 23: #ifdef CYGWIN
! 24: #define _GNU_H_WINDOWS32_SOCKETS
! 25: // for PASCAL
! 26: #include <windows.h>
! 27: // SOCKET
! 28: typedef u_int SOCKET;
! 29: int PASCAL closesocket(SOCKET);
! 30: #else
! 31: # if defined(WIN32)
! 32: # include <windows.h>
! 33: # else
! 34: # define closesocket close
! 35: # endif
! 36: #endif
1.1 paf 37:
1.126 paf 38: #else
1.154 ! paf 39:
! 40: # if defined(WIN32)
! 41: # include <windows.h>
! 42: # endif
! 43:
1.98 paf 44: #endif
45:
1.93 paf 46: // some maybe-undefined constants
47:
1.82 paf 48: #ifndef _O_TEXT
49: # define _O_TEXT 0
50: #endif
51: #ifndef _O_BINARY
52: # define _O_BINARY 0
1.47 paf 53: #endif
1.80 paf 54:
1.138 paf 55: #ifdef HAVE_FTRUNCATE
56: # define PA_O_TRUNC 0
57: #else
58: # ifdef _O_TRUNC
59: # define PA_O_TRUNC _O_TRUNC
60: # else
61: # error you must have either ftruncate function or _O_TRUNC bit declared
62: # endif
63: #endif
64:
1.93 paf 65: // locking constants
1.154 ! paf 66: //#define PA_DEBUG_NO_LOCKING
! 67:
! 68: #ifdef PA_DEBUG_NO_LOCKING
1.93 paf 69:
1.99 paf 70: #ifdef HAVE_FLOCK
71:
72: static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); }
73: static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); }
74: static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); }
75: static int unlock(int fd) { return flock(fd, LOCK_UN); }
76:
1.98 paf 77: #else
1.99 paf 78: #ifdef HAVE__LOCKING
1.98 paf 79:
1.126 paf 80: #define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1)
1.99 paf 81: static int lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); }
82: static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); }
83: static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); }
84: static int unlock(int fd) { FLOCK(_LK_UNLCK); }
1.93 paf 85:
1.99 paf 86: #else
87: #ifdef HAVE_FCNTL
1.93 paf 88:
1.126 paf 89: #define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls)
1.99 paf 90: static int lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); }
91: static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); }
92: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); }
93: static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); }
1.93 paf 94:
95: #else
96: #ifdef HAVE_LOCKF
1.99 paf 97:
1.126 paf 98: #define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1)
1.99 paf 99: static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking
100: static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); }
101: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); }
102: static int unlock(int fd) { FLOCK(F_TLOCK); }
103:
1.93 paf 104: #else
1.99 paf 105:
106: #error unable to find file locking func
107:
108: #endif
1.93 paf 109: #endif
110: #endif
111: #endif
112:
1.154 ! paf 113: #else
! 114: static int lock_shared_blocking(int fd) { return 0; }
! 115: static int lock_exclusive_blocking(int fd) { return 0; }
! 116: static int lock_exclusive_nonblocking(int fd) { return 0; }
! 117: static int unlock(int fd) { return 0; }
! 118:
! 119: #endif
! 120:
! 121: // defines for globals
! 122:
! 123: #define FILE_STATUS_NAME "status"
! 124:
! 125: // globals
! 126:
! 127: const String file_status_name(FILE_STATUS_NAME);
! 128:
! 129: // defines for statics
! 130:
! 131: #define HTTP_METHOD_NAME "method"
! 132: #define HTTP_TIMEOUT_NAME "timeout"
! 133: #define HTTP_HEADERS_NAME "headers"
! 134: #define HTTP_ANY_STATUS_NAME "any-status"
! 135: #define HTTP_CHARSET_NAME "charset"
! 136:
! 137: // statics
! 138:
! 139: static const String http_method_name(HTTP_METHOD_NAME);
! 140: static const String http_timeout_name(HTTP_TIMEOUT_NAME);
! 141: static const String http_headers_name(HTTP_HEADERS_NAME);
! 142: static const String http_any_status_name(HTTP_ANY_STATUS_NAME);
! 143: static const String http_charset_name(HTTP_CHARSET_NAME);
! 144:
! 145: // defines
! 146:
1.127 paf 147: #define DEFAULT_USER_AGENT "parser3"
148:
1.154 ! paf 149: // functions
1.127 paf 150:
1.154 ! paf 151: void fix_line_breaks(char *str, size_t& length) {
1.87 paf 152: //_asm int 3;
1.154 ! paf 153: const char* const eob=str+length;
! 154: char* dest=str;
1.72 parser 155: // fix DOS: \r\n -> \n
156: // fix Macintosh: \r -> \n
1.154 ! paf 157: char* bol=str;
1.137 paf 158: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 159: size_t len=eol-bol;
160: if(dest!=bol)
1.126 paf 161: memcpy(dest, bol, len);
1.72 parser 162: dest+=len;
1.126 paf 163: *dest++='\n';
1.72 parser 164:
1.126 paf 165: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 166: bol=eol+2;
1.154 ! paf 167: length--;
1.126 paf 168: } else // \r, not \n = Macintosh
1.72 parser 169: bol=eol+1;
170: }
1.154 ! paf 171: // last piece without \r
1.72 parser 172: if(dest!=bol)
1.126 paf 173: memcpy(dest, bol, eob-bol);
1.154 ! paf 174: str[length]=0; // terminating
1.72 parser 175: }
1.18 paf 176:
1.154 ! paf 177: char* file_read_text(Request_charsets& charsets,
! 178: const String& file_spec,
! 179: bool fail_on_read_problem,
! 180: HashStringValue* params/*, HashStringValue* * out_fields*/) {
! 181: File_read_result file=
! 182: file_read(charsets, file_spec, true, params, fail_on_read_problem);
! 183: return file.success?file.str:0;
1.126 paf 184: }
185:
1.154 ! paf 186: #ifdef PA_HTTP
1.126 paf 187: //http request stuff
1.154 ! paf 188:
! 189: #undef CRLF
! 190: #define CRLF "\r\n"
1.126 paf 191:
192: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
193: memset(addr, 0, sizeof(*addr));
194: addr->sin_family=AF_INET;
195: addr->sin_port=htons(port);
196: if(host) {
197: if(struct hostent *hostIP=gethostbyname(host))
198: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
199: else
200: return false;
201: } else
202: addr->sin_addr.s_addr=INADDR_ANY;
203: return true;
204: }
205:
1.142 paf 206: static int http_read_response(String& response, int sock, bool fail_on_status_ne_200){
1.154 ! paf 207: int result=0;
! 208: size_t EOLat=0;
1.126 paf 209: while(true) {
1.154 ! paf 210: char *buf=new(PointerFreeGC) char[MAX_STRING];
1.126 paf 211: ssize_t size=recv(sock, buf, MAX_STRING, 0);
212: if(size<=0)
213: break;
1.154 ! paf 214: response.append_strdup(buf, size, String::L_TAINTED);
! 215: if(!result && (EOLat=response.pos(CRLF, 2))!=STRING_NOT_FOUND) { // checking status in first response
1.142 paf 216: const String& status_line=response.mid(0, (size_t)EOLat);
1.154 ! paf 217: ArrayString astatus;
! 218: size_t pos_after=0;
! 219: status_line.split(astatus, pos_after, " ");
! 220: const String& status_code=*astatus.get(1);
! 221: result=status_code.as_int();
1.142 paf 222:
1.154 ! paf 223: if(fail_on_status_ne_200 && result!=200)
1.142 paf 224: throw Exception("http.status",
1.154 ! paf 225: &status_code,
1.142 paf 226: "invalid HTTP response status");
227: }
228: }
1.154 ! paf 229: if(result)
! 230: return result;
1.142 paf 231: else
232: throw Exception("http.response",
233: 0,
1.154 ! paf 234: "bad response from host - no status found (size=%lu)", response.length());
1.126 paf 235: }
236:
237: /* ********************** request *************************** */
238:
239: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
1.145 paf 240: # define PA_USE_ALARM
1.126 paf 241: #endif
242:
1.145 paf 243: #ifdef PA_USE_ALARM
1.126 paf 244: static sigjmp_buf timeout_env;
245: static void timeout_handler(int sig){
246: siglongjmp(timeout_env, 1);
247: }
248: #endif
249:
1.142 paf 250: static int http_request(String& response,
1.152 paf 251: const char* host, int port,
252: const char* request,
253: int timeout,
254: bool fail_on_status_ne_200) {
1.126 paf 255: if(!host)
256: throw Exception("http.host",
1.154 ! paf 257: 0,
1.126 paf 258: "zero hostname"); //never
259:
1.145 paf 260: #ifdef PA_USE_ALARM
1.146 paf 261: signal(SIGALRM, timeout_handler);
1.126 paf 262: #endif
263: int sock=-1;
1.145 paf 264: #ifdef PA_USE_ALARM
265: if(sigsetjmp(timeout_env, 1)) {
266: // stupid gcc [2.95.4] generated bad code
267: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
268: // rewritten simplier [though duplicating closesocket code]
269: if(sock>=0)
270: closesocket(sock);
271: throw Exception("http.timeout",
272: origin_string,
273: "timeout occured while retrieving document");
1.146 paf 274: return 0; // never
1.145 paf 275: } else {
276: alarm(timeout);
277: #endif
1.146 paf 278: try {
279: int result;
1.126 paf 280: struct sockaddr_in dest;
1.154 ! paf 281:
! 282: if(!set_addr(&dest, host, port))
1.126 paf 283: throw Exception("http.host",
1.154 ! paf 284: 0,
1.127 paf 285: "can not resolve hostname \"%s\"", host);
1.126 paf 286:
287: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
288: throw Exception("http.connect",
1.154 ! paf 289: 0,
1.127 paf 290: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 291: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
292: throw Exception("http.connect",
1.154 ! paf 293: 0,
1.127 paf 294: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 295: size_t request_size=strlen(request);
296: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
297: throw Exception("http.connect",
1.154 ! paf 298: 0,
1.127 paf 299: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 300:
1.142 paf 301: result=http_read_response(response, sock, fail_on_status_ne_200);
302: closesocket(sock);
1.145 paf 303: #ifdef PA_USE_ALARM
1.142 paf 304: alarm(0);
1.126 paf 305: #endif
1.147 paf 306: return result;
1.146 paf 307: } catch(...) {
1.145 paf 308: #ifdef PA_USE_ALARM
1.146 paf 309: alarm(0);
1.126 paf 310: #endif
1.146 paf 311: if(sock>=0)
312: closesocket(sock);
1.154 ! paf 313: rethrow;
1.146 paf 314: }
1.148 paf 315: #ifdef PA_USE_ALARM
1.126 paf 316: }
1.148 paf 317: #endif
1.126 paf 318: }
319:
1.127 paf 320: #ifndef DOXYGEN
321: struct Http_pass_header_info {
1.154 ! paf 322: Request_charsets* charsets;
1.127 paf 323: String* request;
324: bool user_agent_specified;
325: };
326: #endif
1.154 ! paf 327: static void http_pass_header(HashStringValue::key_type key,
! 328: HashStringValue::value_type value,
! 329: Http_pass_header_info *info) {
! 330: *info->request <<key<<": "
! 331: << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
! 332: << CRLF;
1.135 paf 333:
1.154 ! paf 334: if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT")
! 335: info->user_agent_specified=true;
1.126 paf 336: }
1.154 ! paf 337:
! 338:
! 339: static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
! 340: const StringBody CONTENT_TYPE_VALUE=
! 341: content_type_value.change_case(source_charset, String::CC_UPPER);
! 342: // content-type: xxx/xxx; source_charset=WE-NEED-THIS
! 343: // content-type: xxx/xxx; source_charset="WE-NEED-THIS"
! 344: // content-type: xxx/xxx; source_charset="WE-NEED-THIS";
! 345: size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET=");
! 346: if(before_charseteq_pos!=STRING_NOT_FOUND) {
! 347: size_t charset_begin=before_charseteq_pos+8/*CHARSET="*/;
! 348: size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
! 349: bool quoted=open_quote_pos==charset_begin;
! 350: if(quoted)
! 351: charset_begin++; // skip opening '"'
! 352: size_t charset_end=CONTENT_TYPE_VALUE.length();
! 353: if(quoted) {
! 354: size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
! 355: if(close_quote_pos!=STRING_NOT_FOUND)
! 356: charset_end=close_quote_pos;
! 357: } else {
! 358: size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin);
! 359: if(delim_pos!=STRING_NOT_FOUND)
! 360: charset_end=delim_pos;
! 361: }
! 362: const StringBody CHARSET_NAME_BODY=
! 363: CONTENT_TYPE_VALUE.mid(charset_begin, charset_end);
! 364:
! 365: return &charsets.get(CHARSET_NAME_BODY);
! 366: }
! 367:
! 368: return 0;
! 369: }
! 370:
! 371:
! 372: #ifndef DOXYGEN
! 373: struct File_read_http_result {
! 374: char *str; size_t length;
! 375: HashStringValue* headers;
! 376: };
! 377: #endif
! 378: static File_read_http_result file_read_http(Request_charsets& charsets,
! 379: const String& file_spec,
! 380: HashStringValue *options=0) {
! 381: File_read_http_result result;
1.126 paf 382: char host[MAX_STRING];
1.129 paf 383: const char* uri;
1.126 paf 384: int port;
1.154 ! paf 385: const char* method=0;
1.126 paf 386: int timeout=2;
1.142 paf 387: bool fail_on_status_ne_200=true;
1.154 ! paf 388: Value* vheaders=0;
! 389: Charset *asked_remote_charset=0;
1.126 paf 390:
1.127 paf 391: if(options) {
392: int valid_options=0;
1.154 ! paf 393: if(Value* vmethod=options->get(http_method_name)) {
1.127 paf 394: valid_options++;
1.154 ! paf 395: method=vmethod->as_string().cstr();
1.127 paf 396: }
1.154 ! paf 397: if(Value* vtimeout=options->get(http_timeout_name)) {
1.127 paf 398: valid_options++;
399: timeout=vtimeout->as_int();
400: }
1.154 ! paf 401: if(vheaders=options->get(http_headers_name)) {
1.127 paf 402: valid_options++;
403: }
1.154 ! paf 404: if(Value* vany_status=options->get(http_any_status_name)) {
1.142 paf 405: valid_options++;
406: fail_on_status_ne_200=!vany_status->as_bool();
407: }
1.154 ! paf 408: if(Value* vcharset_name=options->get(http_charset_name)) {
! 409: valid_options++;
! 410: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
! 411: change_case(charsets.source(), String::CC_UPPER));
! 412: }
1.142 paf 413:
1.154 ! paf 414: if(valid_options!=options->count())
1.127 paf 415: throw Exception("parser.runtime",
416: 0,
417: "invalid option passed");
1.154 ! paf 418: }
! 419: if(!asked_remote_charset) // defaulting to $request:charset
! 420: asked_remote_charset=&charsets.source();
! 421:
! 422: //preparing request
! 423: String& connect_string=*new String;
! 424: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
! 425: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
! 426:
! 427: const char* request_cstr;
! 428: {
! 429: // influence URLencoding of tainted pieces to String::L_URI lang
! 430: Temp_client_charset temp(charsets, *asked_remote_charset);
! 431:
! 432: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
1.126 paf 433:
1.154 ! paf 434: const char* current=connect_string_cstr;
! 435: if(strncmp(current, "http://", 7)!=0)
! 436: throw Exception(0,
! 437: &connect_string,
! 438: "does not start with http://"); //never
! 439: current+=7;
! 440:
! 441: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
! 442: char* host_uri=lsplit(host, '/');
! 443: uri=host_uri?current+(host_uri-1-host):"/";
! 444: char* port_cstr=lsplit(host, ':');
! 445: char* error_pos=0;
! 446: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
! 447:
! 448:
! 449: //making request
! 450: String request;
! 451: if(method)
! 452: request<<method;
! 453: else
! 454: request<<"GET";
! 455: request<< " "<< uri <<" HTTP/1.0" CRLF
! 456: "host: "<< host << CRLF;
! 457: bool user_agent_specified=false;
! 458: if(vheaders && !vheaders->is_string()) { // allow empty
! 459: if(HashStringValue *headers=vheaders->get_hash()) {
! 460: Http_pass_header_info info={&charsets, &request};
! 461: headers->for_each(http_pass_header, &info);
! 462: user_agent_specified=info.user_agent_specified;
! 463: } else
! 464: throw Exception("parser.runtime",
! 465: &connect_string,
! 466: "headers param must be hash");
! 467: };
! 468: if(!user_agent_specified) // defaulting
! 469: request << "user-agent: " DEFAULT_USER_AGENT CRLF;
! 470: request << CRLF;
! 471:
! 472: request_cstr=request.cstr(String::L_UNSPECIFIED);
! 473: }
! 474: // recode those pieces which are not in String::L_URI lang
! 475: // [those violating HTTP standard, but widly used]
! 476: request_cstr=Charset::transcode(
! 477: String::C(request_cstr, strlen(request_cstr)),
! 478: charsets.source(),
! 479: *asked_remote_charset);
1.126 paf 480:
481: //sending request
1.154 ! paf 482: String response;
1.142 paf 483: int status_code=http_request(response,
1.154 ! paf 484: host, port, request_cstr,
1.142 paf 485: timeout, fail_on_status_ne_200);
1.126 paf 486:
487: //processing results
1.154 ! paf 488: size_t pos=response.pos(CRLF CRLF, 4);
! 489: if(pos==STRING_NOT_FOUND || pos<1) {
1.126 paf 490: throw Exception("http.response",
1.133 paf 491: &connect_string,
1.126 paf 492: "bad response from host - no headers found");
493: }
1.154 ! paf 494: const String& header_block=response.mid(0, pos);
! 495: const String& raw_body=response.mid(pos+4, response.length());
1.126 paf 496:
1.154 ! paf 497: ArrayString aheaders;
! 498: result.headers=new HashStringValue;
! 499: size_t pos_after=0;
! 500: header_block.split(aheaders, pos_after, CRLF);
1.126 paf 501:
502: //processing headers
1.154 ! paf 503: size_t aheaders_count=aheaders.count();
! 504: Charset* real_remote_charset=0; // undetected, yet
! 505: for(size_t i=1; i<aheaders_count; i++) {
! 506: const String& line=*aheaders.get(i);
! 507: pos=line.pos(": ", 2);
! 508: if(pos==STRING_NOT_FOUND || pos<1)
1.126 paf 509: throw Exception("http.response",
1.154 ! paf 510: &connect_string,
! 511: "bad response from host - bad header \"%s\"", line.cstr());
! 512: const StringBody HEADER_NAME=
! 513: line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
! 514: const String& header_value=line.mid(pos+2, line.length());
! 515: if(HEADER_NAME=="CONTENT-TYPE")
! 516: real_remote_charset=detect_charset(charsets.source(), header_value);
! 517: result.headers->put(HEADER_NAME, new VString(header_value));
1.126 paf 518: }
1.154 ! paf 519: // defaulting to used-asked charset [it's never empty!]
! 520: if(!real_remote_charset)
! 521: real_remote_charset=asked_remote_charset;
1.126 paf 522:
523: // output response
1.154 ! paf 524: String::C real_body=Charset::transcode(
! 525: String::C(raw_body.cstrm()/*must be modifiable*/, raw_body.length()),
! 526: *real_remote_charset,
! 527: charsets.source());
! 528:
! 529: result.str=const_cast<char *>(real_body.str); // hacking a little
! 530: result.length=real_body.length;
! 531: result.headers->put(file_status_name, new VInt(status_code));
! 532: return result;
1.34 paf 533: }
1.123 paf 534:
1.154 ! paf 535: #endif
! 536:
1.123 paf 537: #ifndef DOXYGEN
538: struct File_read_action_info {
1.154 ! paf 539: char **data; size_t *data_size;
1.126 paf 540: };
1.123 paf 541: #endif
1.154 ! paf 542: static void file_read_action(
! 543: struct stat& finfo,
! 544: int f,
! 545: const String& file_spec, const char* fname, bool as_text,
! 546: void *context) {
1.126 paf 547: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 548: if(size_t to_read_size=(size_t)finfo.st_size) {
1.154 ! paf 549: *info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 550: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 551:
552: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 553: throw Exception(0,
1.123 paf 554: &file_spec,
555: "read failed: actually read %lu bytes count not in [0..%lu] valid range",
1.126 paf 556: *info.data_size, to_read_size);
1.123 paf 557: } else { // empty file
558: if(as_text) {
1.154 ! paf 559: *info.data=new(PointerFreeGC) char[1];
1.123 paf 560: *(char*)(*info.data)=0;
561: } else
562: *info.data=0;
563: *info.data_size=0;
564: return;
565: }
1.126 paf 566: }
1.154 ! paf 567: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
! 568: bool as_text, HashStringValue *params,
1.126 paf 569: bool fail_on_read_problem) {
1.154 ! paf 570: File_read_result result={false};
! 571: #ifdef PA_HTTP
! 572: if(file_spec.starts_with("http://")) {
1.126 paf 573: // fail on read problem
1.154 ! paf 574: File_read_http_result http=file_read_http(charsets, file_spec, params);
! 575: result.success=true;
! 576: result.str=http.str;
! 577: result.length=http.length;
! 578: result.headers=http.headers;
1.126 paf 579: } else {
1.154 ! paf 580: #endif
! 581: File_read_action_info info={&result.str, &result.length};
! 582: result.success=file_read_action_under_lock(file_spec,
1.126 paf 583: "read", file_read_action, &info,
584: as_text, fail_on_read_problem);
1.154 ! paf 585: #ifdef PA_HTTP
1.126 paf 586: }
1.154 ! paf 587: #endif
1.123 paf 588:
1.154 ! paf 589: if(result.success && as_text) {
1.131 paf 590: // UTF-8 signature: EF BB BF
1.154 ! paf 591: if(result.length>=3) {
! 592: char *in=(char *)result.str;
! 593: if(strncmp(in, "\xEB\xBB\xBF", 3)==0) {
! 594: result.str=in+3; result.length-=3;// skip prefix
1.131 paf 595: }
596: }
597:
1.154 ! paf 598: fix_line_breaks((char *)(result.str), result.length);
1.123 paf 599: // note: after fixing
1.154 ! paf 600: ((char*&)(result.str))[result.length]=0;
1.123 paf 601: }
1.126 paf 602:
603: return result;
1.123 paf 604: }
605:
1.154 ! paf 606: #ifdef PA_SAFE_MODE
! 607: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
! 608: if(finfo.st_uid/*foreign?*/!=geteuid()
! 609: && finfo.st_gid/*foreign?*/!=getegid())
! 610: throw Exception("parser.runtime",
! 611: &file_spec,
! 612: "parser is in safe mode: "
! 613: "reading files of foreign group and user disabled "
! 614: "[recompile parser with --disable-safe-mode configure option], "
! 615: "actual filename '%s', "
! 616: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
! 617: fname,
! 618: finfo.st_uid, geteuid(),
! 619: finfo.st_gid, getegid());
! 620: }
! 621: #endif
1.149 paf 622:
1.154 ! paf 623: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 624: const char* action_name, File_read_action action, void *context,
625: bool as_text,
1.123 paf 626: bool fail_on_read_problem) {
1.154 ! paf 627: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 628: int f;
629:
630: // first open, next stat:
1.45 paf 631: // directory update of NTFS hard links performed on open.
1.33 paf 632: // ex:
633: // a.html:^test[] and b.html hardlink to a.html
634: // user inserts ! before ^test in a.html
1.126 paf 635: // directory entry of b.html in NTFS not updated at once,
1.35 paf 636: // they delay update till open, so we would receive "!^test[" string
637: // if would do stat, next open.
1.123 paf 638: // later: it seems, even this does not help sometimes
1.98 paf 639: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 640: try {
641: if(lock_shared_blocking(f)!=0)
1.126 paf 642: throw Exception("file.lock",
1.123 paf 643: &file_spec,
644: "shared lock failed: %s (%d), actual filename '%s'",
1.154 ! paf 645: strerror(errno), errno, fname);
1.123 paf 646:
1.124 paf 647: struct stat finfo;
648: if(stat(fname, &finfo)!=0)
649: throw Exception("file.missing", // hardly possible: we just opened it OK
650: &file_spec,
651: "stat failed: %s (%d), actual filename '%s'",
1.154 ! paf 652: strerror(errno), errno, fname);
1.124 paf 653:
1.140 paf 654: #ifdef PA_SAFE_MODE
1.149 paf 655: check_safe_mode(finfo, file_spec, fname);
1.105 paf 656: #endif
1.32 paf 657:
1.154 ! paf 658: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 659: } catch(...) {
1.126 paf 660: unlock(f);close(f);
1.123 paf 661: if(fail_on_read_problem)
1.154 ! paf 662: rethrow;
1.123 paf 663: return false;
664: }
1.87 paf 665:
1.126 paf 666: unlock(f);close(f);
1.72 parser 667: return true;
1.118 paf 668: } else {
669: if(fail_on_read_problem)
1.126 paf 670: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 671: &file_spec,
1.123 paf 672: "%s failed: %s (%d), actual filename '%s'",
1.154 ! paf 673: action_name, strerror(errno), errno, fname);
1.118 paf 674: return false;
675: }
1.8 paf 676: }
677:
1.63 parser 678: static void create_dir_for_file(const String& file_spec) {
679: size_t pos_after=1;
1.154 ! paf 680: size_t pos_before;
! 681: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
! 682: mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 683: pos_after=pos_before+1;
684: }
685: }
686:
1.98 paf 687: bool file_write_action_under_lock(
1.28 paf 688: const String& file_spec,
1.126 paf 689: const char* action_name, File_write_action action, void *context,
690: bool as_text,
691: bool do_append,
692: bool do_block,
1.110 paf 693: bool fail_on_lock_problem) {
1.154 ! paf 694: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 695: int f;
1.80 paf 696: if(access(fname, W_OK)!=0) // no
1.126 paf 697: create_dir_for_file(file_spec);
1.50 paf 698:
1.80 paf 699: if((f=open(fname,
700: O_CREAT|O_RDWR
701: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 702: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.99 paf 703: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.126 paf 704: Exception e("file.lock",
1.110 paf 705: &file_spec,
706: "shared lock failed: %s (%d), actual filename '%s'",
1.154 ! paf 707: strerror(errno), errno, fname);
1.126 paf 708: close(f);
1.110 paf 709: if(fail_on_lock_problem)
710: throw e;
1.98 paf 711: return false;
712: }
1.96 paf 713:
714: try {
1.126 paf 715: action(f, context);
1.96 paf 716: } catch(...) {
1.138 paf 717: #ifdef HAVE_FTRUNCATE
1.104 paf 718: if(!do_append)
1.125 paf 719: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 720: #endif
1.126 paf 721: unlock(f);close(f);
1.154 ! paf 722: rethrow;
1.96 paf 723: }
1.80 paf 724:
1.138 paf 725: #ifdef HAVE_FTRUNCATE
1.104 paf 726: if(!do_append)
1.125 paf 727: 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 728: #endif
1.126 paf 729: unlock(f);close(f);
1.98 paf 730: return true;
1.80 paf 731: } else
1.126 paf 732: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 733: &file_spec,
1.96 paf 734: "%s failed: %s (%d), actual filename '%s'",
1.154 ! paf 735: action_name, strerror(errno), errno, fname);
1.96 paf 736: // here should be nothing, see rethrow above
737: }
738:
739: #ifndef DOXYGEN
740: struct File_write_action_info {
1.154 ! paf 741: const char* str; size_t length;
1.126 paf 742: };
1.96 paf 743: #endif
744: static void file_write_action(int f, void *context) {
1.126 paf 745: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 ! paf 746: if(info.length) {
! 747: int written=write(f, info.str, info.length);
1.116 paf 748: if(written<0)
1.126 paf 749: throw Exception(0,
750: 0,
751: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 752: }
1.96 paf 753: }
754: void file_write(
755: const String& file_spec,
1.154 ! paf 756: const char* data, size_t size,
1.126 paf 757: bool as_text,
1.96 paf 758: bool do_append) {
1.126 paf 759: File_write_action_info info={data, size};
1.98 paf 760: file_write_action_under_lock(
1.154 ! paf 761: file_spec,
! 762: "write", file_write_action, &info,
! 763: as_text,
! 764: do_append);
1.30 paf 765: }
766:
1.63 parser 767: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 768: static void rmdir(const String& file_spec, size_t pos_after) {
1.154 ! paf 769: size_t pos_before;
! 770: if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
1.126 paf 771: rmdir(file_spec, pos_before+1);
1.50 paf 772:
1.154 ! paf 773: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 774: }
1.95 paf 775: bool file_delete(const String& file_spec, bool fail_on_read_problem) {
1.154 ! paf 776: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 777: if(unlink(fname)!=0)
1.93 paf 778: if(fail_on_read_problem)
1.126 paf 779: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 780: &file_spec,
781: "unlink failed: %s (%d), actual filename '%s'",
1.154 ! paf 782: strerror(errno), errno, fname);
1.93 paf 783: else
784: return false;
1.50 paf 785:
1.126 paf 786: rmdir(file_spec, 1);
1.93 paf 787: return true;
1.60 parser 788: }
1.95 paf 789: void file_move(const String& old_spec, const String& new_spec) {
1.154 ! paf 790: const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
! 791: const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 792:
1.126 paf 793: create_dir_for_file(new_spec);
1.63 parser 794:
1.60 parser 795: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 796: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 797: &old_spec,
798: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 ! paf 799: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 800:
1.126 paf 801: rmdir(old_spec, 1);
1.31 paf 802: }
803:
1.51 paf 804:
1.126 paf 805: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 806: struct stat lfinfo;
807: bool result=stat(fname, &lfinfo)==0;
808: if(afinfo)
809: *afinfo=lfinfo;
810: return result;
1.119 paf 811: }
812:
813: bool entry_exists(const String& file_spec) {
1.154 ! paf 814: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 815: return entry_exists(fname, 0);
1.118 paf 816: }
817:
1.51 paf 818: static bool entry_readable(const String& file_spec, bool need_dir) {
1.154 ! paf 819: char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 820: if(need_dir) {
1.126 paf 821: size_t size=strlen(fname);
1.120 paf 822: while(size) {
1.126 paf 823: char c=fname[size-1];
1.120 paf 824: if(c=='/' || c=='\\')
825: fname[--size]=0;
826: else
827: break;
828: }
829: }
1.51 paf 830: struct stat finfo;
1.118 paf 831: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 832: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 833: return is_dir==need_dir;
834: }
835: return false;
836: }
1.31 paf 837: bool file_readable(const String& file_spec) {
1.126 paf 838: return entry_readable(file_spec, false);
1.51 paf 839: }
840: bool dir_readable(const String& file_spec) {
1.126 paf 841: return entry_readable(file_spec, true);
1.65 parser 842: }
1.154 ! paf 843: const String* file_readable(const String& path, const String& name) {
! 844: String& result=*new String(path);
! 845: result << "/";
! 846: result << name;
! 847: return file_readable(result)?&result:0;
1.43 paf 848: }
849: bool file_executable(const String& file_spec) {
1.154 ! paf 850: return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 851: }
852:
1.64 parser 853: bool file_stat(const String& file_spec,
1.58 parser 854: size_t& rsize,
1.126 paf 855: time_t& ratime,
856: time_t& rmtime,
857: time_t& rctime,
1.64 parser 858: bool fail_on_read_problem) {
1.154 ! paf 859: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
! 860: struct stat finfo;
1.44 paf 861: if(stat(fname, &finfo)!=0)
1.64 parser 862: if(fail_on_read_problem)
1.126 paf 863: throw Exception("file.missing",
1.67 parser 864: &file_spec,
865: "getting file size failed: %s (%d), real filename '%s'",
1.154 ! paf 866: strerror(errno), errno, fname);
1.64 parser 867: else
868: return false;
1.58 parser 869: rsize=finfo.st_size;
870: ratime=finfo.st_atime;
871: rmtime=finfo.st_mtime;
872: rctime=finfo.st_ctime;
1.64 parser 873: return true;
1.18 paf 874: }
875:
1.126 paf 876: char* getrow(char* *row_ref, char delim) {
877: char* result=*row_ref;
1.8 paf 878: if(result) {
1.126 paf 879: *row_ref=strchr(result, delim);
1.8 paf 880: if(*row_ref)
881: *((*row_ref)++)=0;
882: else if(!*result)
883: return 0;
884: }
885: return result;
886: }
887:
1.126 paf 888: char* lsplit(char* string, char delim) {
1.23 paf 889: if(string) {
1.126 paf 890: char* v=strchr(string, delim);
1.8 paf 891: if(v) {
892: *v=0;
893: return v+1;
894: }
895: }
896: return 0;
897: }
898:
1.126 paf 899: char* lsplit(char* *string_ref, char delim) {
900: char* result=*string_ref;
901: char* next=lsplit(*string_ref, delim);
1.8 paf 902: *string_ref=next;
903: return result;
1.9 paf 904: }
905:
1.126 paf 906: char* rsplit(char* string, char delim) {
1.18 paf 907: if(string) {
1.126 paf 908: char* v=strrchr(string, delim);
1.18 paf 909: if(v) {
1.9 paf 910: *v=0;
911: return v+1;
912: }
913: }
914: return NULL;
1.10 paf 915: }
916:
1.37 paf 917: /// @todo less stupid type detection
1.154 ! paf 918: const char* format(double value, char* fmt) {
1.126 paf 919: char local_buf[MAX_NUMBER];
1.108 paf 920: size_t size;
921:
1.10 paf 922: if(fmt)
923: if(strpbrk(fmt, "diouxX"))
924: if(strpbrk(fmt, "ouxX"))
1.126 paf 925: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 926: else
1.126 paf 927: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 928: else
1.126 paf 929: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 930: else
1.126 paf 931: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 932:
1.154 ! paf 933: return pa_strdup(local_buf, size);
1.12 paf 934: }
935:
1.36 paf 936: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 937: #ifdef WIN32
938: do{
1.154 ! paf 939: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 940: if(chunk_written<=0)
941: break;
942: size-=chunk_written;
1.36 paf 943: buf=((const char*)buf)+chunk_written;
1.126 paf 944: } while(size>0);
1.12 paf 945:
946: return size;
947: #else
1.126 paf 948: return fwrite(buf, 1, size, stdout);
1.12 paf 949: #endif
1.2 paf 950: }
1.14 paf 951:
1.154 ! paf 952: char* unescape_chars(const char* cp, int len) {
! 953: char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 954: enum EscapeState {
1.33 paf 955: EscapeRest,
956: EscapeFirst,
1.14 paf 957: EscapeSecond
958: } escapeState=EscapeRest;
959: int escapedValue=0;
960: int srcPos=0;
961: int dstPos=0;
962: while(srcPos < len) {
1.126 paf 963: int ch=cp[srcPos];
1.14 paf 964: switch(escapeState) {
965: case EscapeRest:
966: if(ch=='%') {
967: escapeState=EscapeFirst;
968: } else if(ch=='+') {
1.126 paf 969: s[dstPos++]=' ';
1.14 paf 970: } else {
971: s[dstPos++]=ch;
972: }
973: break;
974: case EscapeFirst:
975: escapedValue=hex_value[ch] << 4;
976: escapeState=EscapeSecond;
977: break;
978: case EscapeSecond:
1.126 paf 979: escapedValue +=hex_value[ch];
1.14 paf 980: s[dstPos++]=escapedValue;
981: escapeState=EscapeRest;
982: break;
983: }
1.126 paf 984: srcPos++;
1.14 paf 985: }
986: s[dstPos]=0;
987: return s;
1.24 paf 988: }
989:
990: #ifdef WIN32
1.126 paf 991: void back_slashes_to_slashes(char* s) {
1.24 paf 992: if(s)
993: for(; *s; s++)
994: if(*s=='\\')
1.126 paf 995: *s='/';
1.24 paf 996: }
1.42 paf 997: /*
1.126 paf 998: void slashes_to_back_slashes(char* s) {
1.42 paf 999: if(s)
1000: for(; *s; s++)
1001: if(*s=='/')
1.126 paf 1002: *s='\\';
1.42 paf 1003: }
1004: */
1.24 paf 1005: #endif
1.41 paf 1006:
1.126 paf 1007: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 1008: while(true) {
1009: if(!(*s1)) {
1010: if(!(*s2))
1011: return true;
1012: else
1013: return !strict;
1014: } else if(!(*s2))
1015: return !strict;
1016: if(isalpha(*s1)) {
1017: if(tolower(*s1) !=tolower(*s2))
1018: return false;
1019: } else if((*s1) !=(*s2))
1020: return false;
1.126 paf 1021: s1++;
1022: s2++;
1.41 paf 1023: }
1.57 parser 1024: }
1025:
1.84 paf 1026: static bool isLeap(int year) {
1.57 parser 1027: return !(
1028: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 1029: );
1.57 parser 1030: }
1031:
1032: int getMonthDays(int year, int month) {
1033: int monthDays[]={
1.126 paf 1034: 31,
1035: isLeap(year) ? 29 : 28,
1036: 31,
1037: 30,
1038: 31,
1039: 30,
1040: 31,
1041: 31,
1042: 30,
1043: 31,
1044: 30,
1.57 parser 1045: 31
1.126 paf 1046: };
1047: return monthDays[month];
1.41 paf 1048: }
1.69 parser 1049:
1.126 paf 1050: void remove_crlf(char* start, char* end) {
1051: for(char* p=start; p<end; p++)
1.69 parser 1052: switch(*p) {
1.126 paf 1053: case '\n': *p='|'; break;
1054: case '\r': *p=' '; break;
1.69 parser 1055: }
1.91 paf 1056: }
1057:
1058:
1059: /// must be last in this file
1060: #undef vsnprintf
1.126 paf 1061: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 1062: if(!s)
1063: return 0;
1064:
1065: int r;
1066: // note: on win32& maybe somewhere else
1067: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1068: --s;
1069: #if _MSC_VER
1070: /*
1071: win32:
1072: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
1073:
1.154 ! paf 1074: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 1075: */
1.126 paf 1076: r=_vsnprintf(b, s, f, l);
1.91 paf 1077: if(r<0)
1078: r=s;
1079: #else
1.126 paf 1080: r=vsnprintf(b, s, f, l);
1.91 paf 1081: /*
1082: solaris:
1083: man vsnprintf
1084:
1085: The snprintf() function returns the number of characters
1086: formatted, that is, the number of characters that would have
1087: been written to the buffer if it were large enough. If the
1088: value of n is 0 on a call to snprintf(), an unspecified
1089: value less than 1 is returned.
1090: */
1091:
1092: if(r<0)
1093: r=0;
1094: else if(r>s)
1095: r=s;
1096: #endif
1097: b[r]=0;
1098: return r;
1099: }
1100:
1.126 paf 1101: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1102: va_list l;
1.126 paf 1103: va_start(l, f);
1104: int r=__vsnprintf(b, s, f, l);
1105: va_end(l);
1.91 paf 1106: return r;
1.98 paf 1107: }
1108:
1109: int pa_sleep(unsigned long secs, unsigned long usecs) {
1.126 paf 1110: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
1.98 paf 1111:
1112: #ifdef WIN32
1.126 paf 1113: Sleep(secs * 1000 + usecs / 1000);
1.98 paf 1114: return 0;
1115: #else
1116: struct timeval t;
1117: t.tv_sec = secs;
1118: t.tv_usec = usecs;
1.126 paf 1119: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
1.98 paf 1120: #endif
1.135 paf 1121: }
1122:
1123:
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