Annotation of parser3/src/main/pa_common.C, revision 1.157
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:
1.157 ! paf 8: static const char* IDENT_COMMON_C="$Date: 2003/09/25 09:15:03 $";
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"
1.155 paf 22: #include "pa_vhash.h"
23: #include "pa_vtable.h"
1.154 paf 24:
25: #ifdef CYGWIN
26: #define _GNU_H_WINDOWS32_SOCKETS
27: // for PASCAL
28: #include <windows.h>
29: // SOCKET
30: typedef u_int SOCKET;
31: int PASCAL closesocket(SOCKET);
32: #else
33: # if defined(WIN32)
34: # include <windows.h>
35: # else
36: # define closesocket close
37: # endif
38: #endif
1.1 paf 39:
1.126 paf 40: #else
1.154 paf 41:
42: # if defined(WIN32)
43: # include <windows.h>
44: # endif
45:
1.98 paf 46: #endif
47:
1.93 paf 48: // some maybe-undefined constants
49:
1.82 paf 50: #ifndef _O_TEXT
51: # define _O_TEXT 0
52: #endif
53: #ifndef _O_BINARY
54: # define _O_BINARY 0
1.47 paf 55: #endif
1.80 paf 56:
1.138 paf 57: #ifdef HAVE_FTRUNCATE
58: # define PA_O_TRUNC 0
59: #else
60: # ifdef _O_TRUNC
61: # define PA_O_TRUNC _O_TRUNC
62: # else
63: # error you must have either ftruncate function or _O_TRUNC bit declared
64: # endif
65: #endif
66:
1.93 paf 67: // locking constants
1.154 paf 68: //#define PA_DEBUG_NO_LOCKING
69:
70: #ifdef PA_DEBUG_NO_LOCKING
1.93 paf 71:
1.99 paf 72: #ifdef HAVE_FLOCK
73:
74: static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); }
75: static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); }
76: static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); }
77: static int unlock(int fd) { return flock(fd, LOCK_UN); }
78:
1.98 paf 79: #else
1.99 paf 80: #ifdef HAVE__LOCKING
1.98 paf 81:
1.126 paf 82: #define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1)
1.99 paf 83: static int lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); }
84: static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); }
85: static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); }
86: static int unlock(int fd) { FLOCK(_LK_UNLCK); }
1.93 paf 87:
1.99 paf 88: #else
89: #ifdef HAVE_FCNTL
1.93 paf 90:
1.126 paf 91: #define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls)
1.99 paf 92: static int lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); }
93: static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); }
94: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); }
95: static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); }
1.93 paf 96:
97: #else
98: #ifdef HAVE_LOCKF
1.99 paf 99:
1.126 paf 100: #define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1)
1.99 paf 101: static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking
102: static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); }
103: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); }
104: static int unlock(int fd) { FLOCK(F_TLOCK); }
105:
1.93 paf 106: #else
1.99 paf 107:
108: #error unable to find file locking func
109:
110: #endif
1.93 paf 111: #endif
112: #endif
113: #endif
114:
1.154 paf 115: #else
116: static int lock_shared_blocking(int fd) { return 0; }
117: static int lock_exclusive_blocking(int fd) { return 0; }
118: static int lock_exclusive_nonblocking(int fd) { return 0; }
119: static int unlock(int fd) { return 0; }
120:
121: #endif
122:
123: // defines for globals
124:
125: #define FILE_STATUS_NAME "status"
126:
127: // globals
128:
129: const String file_status_name(FILE_STATUS_NAME);
130:
131: // defines for statics
132:
133: #define HTTP_METHOD_NAME "method"
134: #define HTTP_TIMEOUT_NAME "timeout"
135: #define HTTP_HEADERS_NAME "headers"
136: #define HTTP_ANY_STATUS_NAME "any-status"
137: #define HTTP_CHARSET_NAME "charset"
1.155 paf 138: #define HTTP_TABLES_NAME "tables"
1.154 paf 139:
140: // statics
141:
142: static const String http_method_name(HTTP_METHOD_NAME);
143: static const String http_timeout_name(HTTP_TIMEOUT_NAME);
144: static const String http_headers_name(HTTP_HEADERS_NAME);
145: static const String http_any_status_name(HTTP_ANY_STATUS_NAME);
146: static const String http_charset_name(HTTP_CHARSET_NAME);
1.155 paf 147: static const String http_tables_name(HTTP_TABLES_NAME);
1.154 paf 148:
149: // defines
150:
1.127 paf 151: #define DEFAULT_USER_AGENT "parser3"
152:
1.154 paf 153: // functions
1.127 paf 154:
1.154 paf 155: void fix_line_breaks(char *str, size_t& length) {
1.87 paf 156: //_asm int 3;
1.154 paf 157: const char* const eob=str+length;
158: char* dest=str;
1.72 parser 159: // fix DOS: \r\n -> \n
160: // fix Macintosh: \r -> \n
1.154 paf 161: char* bol=str;
1.137 paf 162: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 163: size_t len=eol-bol;
164: if(dest!=bol)
1.126 paf 165: memcpy(dest, bol, len);
1.72 parser 166: dest+=len;
1.126 paf 167: *dest++='\n';
1.72 parser 168:
1.126 paf 169: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 170: bol=eol+2;
1.154 paf 171: length--;
1.126 paf 172: } else // \r, not \n = Macintosh
1.72 parser 173: bol=eol+1;
174: }
1.154 paf 175: // last piece without \r
1.72 parser 176: if(dest!=bol)
1.126 paf 177: memcpy(dest, bol, eob-bol);
1.154 paf 178: str[length]=0; // terminating
1.72 parser 179: }
1.18 paf 180:
1.154 paf 181: char* file_read_text(Request_charsets& charsets,
182: const String& file_spec,
183: bool fail_on_read_problem,
184: HashStringValue* params/*, HashStringValue* * out_fields*/) {
185: File_read_result file=
186: file_read(charsets, file_spec, true, params, fail_on_read_problem);
187: return file.success?file.str:0;
1.126 paf 188: }
189:
1.154 paf 190: #ifdef PA_HTTP
1.126 paf 191: //http request stuff
1.154 paf 192:
193: #undef CRLF
194: #define CRLF "\r\n"
1.126 paf 195:
196: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
197: memset(addr, 0, sizeof(*addr));
198: addr->sin_family=AF_INET;
199: addr->sin_port=htons(port);
200: if(host) {
201: if(struct hostent *hostIP=gethostbyname(host))
202: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
203: else
204: return false;
205: } else
206: addr->sin_addr.s_addr=INADDR_ANY;
207: return true;
208: }
209:
1.142 paf 210: static int http_read_response(String& response, int sock, bool fail_on_status_ne_200){
1.154 paf 211: int result=0;
212: size_t EOLat=0;
1.126 paf 213: while(true) {
1.154 paf 214: char *buf=new(PointerFreeGC) char[MAX_STRING];
1.126 paf 215: ssize_t size=recv(sock, buf, MAX_STRING, 0);
216: if(size<=0)
217: break;
1.154 paf 218: response.append_strdup(buf, size, String::L_TAINTED);
219: if(!result && (EOLat=response.pos(CRLF, 2))!=STRING_NOT_FOUND) { // checking status in first response
1.142 paf 220: const String& status_line=response.mid(0, (size_t)EOLat);
1.154 paf 221: ArrayString astatus;
222: size_t pos_after=0;
223: status_line.split(astatus, pos_after, " ");
224: const String& status_code=*astatus.get(1);
225: result=status_code.as_int();
1.142 paf 226:
1.154 paf 227: if(fail_on_status_ne_200 && result!=200)
1.142 paf 228: throw Exception("http.status",
1.154 paf 229: &status_code,
1.142 paf 230: "invalid HTTP response status");
231: }
232: }
1.154 paf 233: if(result)
234: return result;
1.142 paf 235: else
236: throw Exception("http.response",
237: 0,
1.154 paf 238: "bad response from host - no status found (size=%lu)", response.length());
1.126 paf 239: }
240:
241: /* ********************** request *************************** */
242:
243: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
1.145 paf 244: # define PA_USE_ALARM
1.126 paf 245: #endif
246:
1.145 paf 247: #ifdef PA_USE_ALARM
1.126 paf 248: static sigjmp_buf timeout_env;
249: static void timeout_handler(int sig){
250: siglongjmp(timeout_env, 1);
251: }
252: #endif
253:
1.142 paf 254: static int http_request(String& response,
1.152 paf 255: const char* host, int port,
256: const char* request,
257: int timeout,
258: bool fail_on_status_ne_200) {
1.126 paf 259: if(!host)
260: throw Exception("http.host",
1.154 paf 261: 0,
1.126 paf 262: "zero hostname"); //never
263:
1.145 paf 264: #ifdef PA_USE_ALARM
1.146 paf 265: signal(SIGALRM, timeout_handler);
1.126 paf 266: #endif
267: int sock=-1;
1.145 paf 268: #ifdef PA_USE_ALARM
269: if(sigsetjmp(timeout_env, 1)) {
270: // stupid gcc [2.95.4] generated bad code
271: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
272: // rewritten simplier [though duplicating closesocket code]
273: if(sock>=0)
274: closesocket(sock);
275: throw Exception("http.timeout",
276: origin_string,
277: "timeout occured while retrieving document");
1.146 paf 278: return 0; // never
1.145 paf 279: } else {
280: alarm(timeout);
281: #endif
1.146 paf 282: try {
283: int result;
1.126 paf 284: struct sockaddr_in dest;
1.154 paf 285:
286: if(!set_addr(&dest, host, port))
1.126 paf 287: throw Exception("http.host",
1.154 paf 288: 0,
1.127 paf 289: "can not resolve hostname \"%s\"", host);
1.126 paf 290:
291: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
292: throw Exception("http.connect",
1.154 paf 293: 0,
1.127 paf 294: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 295: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
296: throw Exception("http.connect",
1.154 paf 297: 0,
1.127 paf 298: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 299: size_t request_size=strlen(request);
300: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
301: throw Exception("http.connect",
1.154 paf 302: 0,
1.127 paf 303: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 304:
1.142 paf 305: result=http_read_response(response, sock, fail_on_status_ne_200);
306: closesocket(sock);
1.145 paf 307: #ifdef PA_USE_ALARM
1.142 paf 308: alarm(0);
1.126 paf 309: #endif
1.147 paf 310: return result;
1.146 paf 311: } catch(...) {
1.145 paf 312: #ifdef PA_USE_ALARM
1.146 paf 313: alarm(0);
1.126 paf 314: #endif
1.146 paf 315: if(sock>=0)
316: closesocket(sock);
1.154 paf 317: rethrow;
1.146 paf 318: }
1.148 paf 319: #ifdef PA_USE_ALARM
1.126 paf 320: }
1.148 paf 321: #endif
1.126 paf 322: }
323:
1.127 paf 324: #ifndef DOXYGEN
325: struct Http_pass_header_info {
1.154 paf 326: Request_charsets* charsets;
1.127 paf 327: String* request;
328: bool user_agent_specified;
329: };
330: #endif
1.154 paf 331: static void http_pass_header(HashStringValue::key_type key,
332: HashStringValue::value_type value,
333: Http_pass_header_info *info) {
334: *info->request <<key<<": "
335: << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
336: << CRLF;
1.135 paf 337:
1.154 paf 338: if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT")
339: info->user_agent_specified=true;
1.126 paf 340: }
1.154 paf 341:
342:
343: static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
1.156 paf 344: const String::Body CONTENT_TYPE_VALUE=
1.154 paf 345: content_type_value.change_case(source_charset, String::CC_UPPER);
346: // content-type: xxx/xxx; source_charset=WE-NEED-THIS
347: // content-type: xxx/xxx; source_charset="WE-NEED-THIS"
348: // content-type: xxx/xxx; source_charset="WE-NEED-THIS";
349: size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET=");
350: if(before_charseteq_pos!=STRING_NOT_FOUND) {
351: size_t charset_begin=before_charseteq_pos+8/*CHARSET="*/;
352: size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
353: bool quoted=open_quote_pos==charset_begin;
354: if(quoted)
355: charset_begin++; // skip opening '"'
356: size_t charset_end=CONTENT_TYPE_VALUE.length();
357: if(quoted) {
358: size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
359: if(close_quote_pos!=STRING_NOT_FOUND)
360: charset_end=close_quote_pos;
361: } else {
362: size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin);
363: if(delim_pos!=STRING_NOT_FOUND)
364: charset_end=delim_pos;
365: }
1.156 paf 366: const String::Body CHARSET_NAME_BODY=
1.154 paf 367: CONTENT_TYPE_VALUE.mid(charset_begin, charset_end);
368:
369: return &charsets.get(CHARSET_NAME_BODY);
370: }
371:
372: return 0;
373: }
374:
375:
376: #ifndef DOXYGEN
377: struct File_read_http_result {
378: char *str; size_t length;
379: HashStringValue* headers;
380: };
381: #endif
1.155 paf 382: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.154 paf 383: static File_read_http_result file_read_http(Request_charsets& charsets,
384: const String& file_spec,
385: HashStringValue *options=0) {
386: File_read_http_result result;
1.126 paf 387: char host[MAX_STRING];
1.129 paf 388: const char* uri;
1.126 paf 389: int port;
1.154 paf 390: const char* method=0;
1.126 paf 391: int timeout=2;
1.142 paf 392: bool fail_on_status_ne_200=true;
1.154 paf 393: Value* vheaders=0;
394: Charset *asked_remote_charset=0;
1.126 paf 395:
1.127 paf 396: if(options) {
397: int valid_options=0;
1.154 paf 398: if(Value* vmethod=options->get(http_method_name)) {
1.127 paf 399: valid_options++;
1.154 paf 400: method=vmethod->as_string().cstr();
1.127 paf 401: }
1.154 paf 402: if(Value* vtimeout=options->get(http_timeout_name)) {
1.127 paf 403: valid_options++;
404: timeout=vtimeout->as_int();
405: }
1.154 paf 406: if(vheaders=options->get(http_headers_name)) {
1.127 paf 407: valid_options++;
408: }
1.154 paf 409: if(Value* vany_status=options->get(http_any_status_name)) {
1.142 paf 410: valid_options++;
411: fail_on_status_ne_200=!vany_status->as_bool();
412: }
1.154 paf 413: if(Value* vcharset_name=options->get(http_charset_name)) {
414: valid_options++;
415: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
416: change_case(charsets.source(), String::CC_UPPER));
417: }
1.142 paf 418:
1.154 paf 419: if(valid_options!=options->count())
1.127 paf 420: throw Exception("parser.runtime",
421: 0,
422: "invalid option passed");
1.154 paf 423: }
424: if(!asked_remote_charset) // defaulting to $request:charset
425: asked_remote_charset=&charsets.source();
426:
427: //preparing request
428: String& connect_string=*new String;
429: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
430: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
431:
432: const char* request_cstr;
433: {
434: // influence URLencoding of tainted pieces to String::L_URI lang
435: Temp_client_charset temp(charsets, *asked_remote_charset);
436:
437: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
1.126 paf 438:
1.154 paf 439: const char* current=connect_string_cstr;
440: if(strncmp(current, "http://", 7)!=0)
441: throw Exception(0,
442: &connect_string,
443: "does not start with http://"); //never
444: current+=7;
445:
446: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
447: char* host_uri=lsplit(host, '/');
448: uri=host_uri?current+(host_uri-1-host):"/";
449: char* port_cstr=lsplit(host, ':');
450: char* error_pos=0;
451: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
452:
453:
454: //making request
455: String request;
456: if(method)
457: request<<method;
458: else
459: request<<"GET";
460: request<< " "<< uri <<" HTTP/1.0" CRLF
461: "host: "<< host << CRLF;
462: bool user_agent_specified=false;
463: if(vheaders && !vheaders->is_string()) { // allow empty
464: if(HashStringValue *headers=vheaders->get_hash()) {
465: Http_pass_header_info info={&charsets, &request};
466: headers->for_each(http_pass_header, &info);
467: user_agent_specified=info.user_agent_specified;
468: } else
469: throw Exception("parser.runtime",
470: &connect_string,
471: "headers param must be hash");
472: };
473: if(!user_agent_specified) // defaulting
474: request << "user-agent: " DEFAULT_USER_AGENT CRLF;
475: request << CRLF;
476:
477: request_cstr=request.cstr(String::L_UNSPECIFIED);
478: }
479: // recode those pieces which are not in String::L_URI lang
480: // [those violating HTTP standard, but widly used]
481: request_cstr=Charset::transcode(
482: String::C(request_cstr, strlen(request_cstr)),
483: charsets.source(),
484: *asked_remote_charset);
1.126 paf 485:
486: //sending request
1.154 paf 487: String response;
1.142 paf 488: int status_code=http_request(response,
1.154 paf 489: host, port, request_cstr,
1.142 paf 490: timeout, fail_on_status_ne_200);
1.126 paf 491:
492: //processing results
1.154 paf 493: size_t pos=response.pos(CRLF CRLF, 4);
494: if(pos==STRING_NOT_FOUND || pos<1) {
1.126 paf 495: throw Exception("http.response",
1.133 paf 496: &connect_string,
1.126 paf 497: "bad response from host - no headers found");
498: }
1.154 paf 499: const String& header_block=response.mid(0, pos);
500: const String& raw_body=response.mid(pos+4, response.length());
1.126 paf 501:
1.154 paf 502: ArrayString aheaders;
503: result.headers=new HashStringValue;
1.155 paf 504: VHash* vtables=new VHash;
505: result.headers->put(http_tables_name, vtables);
506: HashStringValue& tables=vtables->hash();
507:
1.154 paf 508: size_t pos_after=0;
509: header_block.split(aheaders, pos_after, CRLF);
1.126 paf 510:
511: //processing headers
1.154 paf 512: size_t aheaders_count=aheaders.count();
513: Charset* real_remote_charset=0; // undetected, yet
514: for(size_t i=1; i<aheaders_count; i++) {
515: const String& line=*aheaders.get(i);
516: pos=line.pos(": ", 2);
517: if(pos==STRING_NOT_FOUND || pos<1)
1.126 paf 518: throw Exception("http.response",
1.154 paf 519: &connect_string,
520: "bad response from host - bad header \"%s\"", line.cstr());
1.156 paf 521: const String::Body HEADER_NAME=
1.154 paf 522: line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
523: const String& header_value=line.mid(pos+2, line.length());
524: if(HEADER_NAME=="CONTENT-TYPE")
525: real_remote_charset=detect_charset(charsets.source(), header_value);
1.155 paf 526:
527: // tables
528: {
529: Value *valready=(Value *)tables.get(HEADER_NAME);
530: bool existed=valready!=0;
531: Table *table;
532: if(existed) {
533: // second+ appearence
534: table=valready->get_table();
535: } else {
536: // first appearence
537: Table::columns_type columns =new ArrayString(1);
538: *columns+=new String("value");
539: table=new Table(columns);
540: }
541: // this string becomes next row
542: ArrayString& row=*new ArrayString(1);
543: row+=&header_value;
544: *table+=&row;
545: // not existed before? add it
546: if(!existed)
547: tables.put(HEADER_NAME, new VTable(table));
548: }
549:
1.154 paf 550: result.headers->put(HEADER_NAME, new VString(header_value));
1.126 paf 551: }
1.154 paf 552: // defaulting to used-asked charset [it's never empty!]
553: if(!real_remote_charset)
554: real_remote_charset=asked_remote_charset;
1.126 paf 555:
556: // output response
1.154 paf 557: String::C real_body=Charset::transcode(
558: String::C(raw_body.cstrm()/*must be modifiable*/, raw_body.length()),
559: *real_remote_charset,
560: charsets.source());
561:
562: result.str=const_cast<char *>(real_body.str); // hacking a little
563: result.length=real_body.length;
564: result.headers->put(file_status_name, new VInt(status_code));
565: return result;
1.34 paf 566: }
1.123 paf 567:
1.154 paf 568: #endif
569:
1.123 paf 570: #ifndef DOXYGEN
571: struct File_read_action_info {
1.154 paf 572: char **data; size_t *data_size;
1.126 paf 573: };
1.123 paf 574: #endif
1.154 paf 575: static void file_read_action(
576: struct stat& finfo,
577: int f,
578: const String& file_spec, const char* fname, bool as_text,
579: void *context) {
1.126 paf 580: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 581: if(size_t to_read_size=(size_t)finfo.st_size) {
1.154 paf 582: *info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 583: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 584:
585: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 586: throw Exception(0,
1.123 paf 587: &file_spec,
588: "read failed: actually read %lu bytes count not in [0..%lu] valid range",
1.126 paf 589: *info.data_size, to_read_size);
1.123 paf 590: } else { // empty file
591: if(as_text) {
1.154 paf 592: *info.data=new(PointerFreeGC) char[1];
1.123 paf 593: *(char*)(*info.data)=0;
594: } else
595: *info.data=0;
596: *info.data_size=0;
597: return;
598: }
1.126 paf 599: }
1.154 paf 600: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
601: bool as_text, HashStringValue *params,
1.126 paf 602: bool fail_on_read_problem) {
1.154 paf 603: File_read_result result={false};
604: #ifdef PA_HTTP
605: if(file_spec.starts_with("http://")) {
1.126 paf 606: // fail on read problem
1.154 paf 607: File_read_http_result http=file_read_http(charsets, file_spec, params);
608: result.success=true;
609: result.str=http.str;
610: result.length=http.length;
611: result.headers=http.headers;
1.126 paf 612: } else {
1.154 paf 613: #endif
614: File_read_action_info info={&result.str, &result.length};
615: result.success=file_read_action_under_lock(file_spec,
1.126 paf 616: "read", file_read_action, &info,
617: as_text, fail_on_read_problem);
1.154 paf 618: #ifdef PA_HTTP
1.126 paf 619: }
1.154 paf 620: #endif
1.123 paf 621:
1.154 paf 622: if(result.success && as_text) {
1.131 paf 623: // UTF-8 signature: EF BB BF
1.154 paf 624: if(result.length>=3) {
625: char *in=(char *)result.str;
626: if(strncmp(in, "\xEB\xBB\xBF", 3)==0) {
627: result.str=in+3; result.length-=3;// skip prefix
1.131 paf 628: }
629: }
630:
1.154 paf 631: fix_line_breaks((char *)(result.str), result.length);
1.123 paf 632: // note: after fixing
1.154 paf 633: ((char*&)(result.str))[result.length]=0;
1.123 paf 634: }
1.126 paf 635:
636: return result;
1.123 paf 637: }
638:
1.154 paf 639: #ifdef PA_SAFE_MODE
640: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
641: if(finfo.st_uid/*foreign?*/!=geteuid()
642: && finfo.st_gid/*foreign?*/!=getegid())
643: throw Exception("parser.runtime",
644: &file_spec,
645: "parser is in safe mode: "
646: "reading files of foreign group and user disabled "
647: "[recompile parser with --disable-safe-mode configure option], "
648: "actual filename '%s', "
649: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
650: fname,
651: finfo.st_uid, geteuid(),
652: finfo.st_gid, getegid());
653: }
654: #endif
1.149 paf 655:
1.154 paf 656: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 657: const char* action_name, File_read_action action, void *context,
658: bool as_text,
1.123 paf 659: bool fail_on_read_problem) {
1.154 paf 660: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 661: int f;
662:
663: // first open, next stat:
1.45 paf 664: // directory update of NTFS hard links performed on open.
1.33 paf 665: // ex:
666: // a.html:^test[] and b.html hardlink to a.html
667: // user inserts ! before ^test in a.html
1.126 paf 668: // directory entry of b.html in NTFS not updated at once,
1.35 paf 669: // they delay update till open, so we would receive "!^test[" string
670: // if would do stat, next open.
1.123 paf 671: // later: it seems, even this does not help sometimes
1.98 paf 672: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 673: try {
674: if(lock_shared_blocking(f)!=0)
1.126 paf 675: throw Exception("file.lock",
1.123 paf 676: &file_spec,
677: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 678: strerror(errno), errno, fname);
1.123 paf 679:
1.124 paf 680: struct stat finfo;
681: if(stat(fname, &finfo)!=0)
682: throw Exception("file.missing", // hardly possible: we just opened it OK
683: &file_spec,
684: "stat failed: %s (%d), actual filename '%s'",
1.154 paf 685: strerror(errno), errno, fname);
1.124 paf 686:
1.140 paf 687: #ifdef PA_SAFE_MODE
1.149 paf 688: check_safe_mode(finfo, file_spec, fname);
1.105 paf 689: #endif
1.32 paf 690:
1.154 paf 691: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 692: } catch(...) {
1.126 paf 693: unlock(f);close(f);
1.123 paf 694: if(fail_on_read_problem)
1.154 paf 695: rethrow;
1.123 paf 696: return false;
697: }
1.87 paf 698:
1.126 paf 699: unlock(f);close(f);
1.72 parser 700: return true;
1.118 paf 701: } else {
702: if(fail_on_read_problem)
1.126 paf 703: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 704: &file_spec,
1.123 paf 705: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 706: action_name, strerror(errno), errno, fname);
1.118 paf 707: return false;
708: }
1.8 paf 709: }
710:
1.63 parser 711: static void create_dir_for_file(const String& file_spec) {
712: size_t pos_after=1;
1.154 paf 713: size_t pos_before;
714: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
715: mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 716: pos_after=pos_before+1;
717: }
718: }
719:
1.98 paf 720: bool file_write_action_under_lock(
1.28 paf 721: const String& file_spec,
1.126 paf 722: const char* action_name, File_write_action action, void *context,
723: bool as_text,
724: bool do_append,
725: bool do_block,
1.110 paf 726: bool fail_on_lock_problem) {
1.154 paf 727: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 728: int f;
1.80 paf 729: if(access(fname, W_OK)!=0) // no
1.126 paf 730: create_dir_for_file(file_spec);
1.50 paf 731:
1.80 paf 732: if((f=open(fname,
733: O_CREAT|O_RDWR
734: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 735: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.99 paf 736: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.126 paf 737: Exception e("file.lock",
1.110 paf 738: &file_spec,
739: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 740: strerror(errno), errno, fname);
1.126 paf 741: close(f);
1.110 paf 742: if(fail_on_lock_problem)
743: throw e;
1.98 paf 744: return false;
745: }
1.96 paf 746:
1.157 ! paf 747: /// try {
1.126 paf 748: action(f, context);
1.157 ! paf 749: /*} catch(...) {
1.138 paf 750: #ifdef HAVE_FTRUNCATE
1.104 paf 751: if(!do_append)
1.125 paf 752: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 753: #endif
1.126 paf 754: unlock(f);close(f);
1.154 paf 755: rethrow;
1.157 ! paf 756: }*/
1.80 paf 757:
1.138 paf 758: #ifdef HAVE_FTRUNCATE
1.104 paf 759: if(!do_append)
1.125 paf 760: 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 761: #endif
1.126 paf 762: unlock(f);close(f);
1.98 paf 763: return true;
1.80 paf 764: } else
1.126 paf 765: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 766: &file_spec,
1.96 paf 767: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 768: action_name, strerror(errno), errno, fname);
1.96 paf 769: // here should be nothing, see rethrow above
770: }
771:
772: #ifndef DOXYGEN
773: struct File_write_action_info {
1.154 paf 774: const char* str; size_t length;
1.126 paf 775: };
1.96 paf 776: #endif
777: static void file_write_action(int f, void *context) {
1.126 paf 778: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 779: if(info.length) {
780: int written=write(f, info.str, info.length);
1.116 paf 781: if(written<0)
1.126 paf 782: throw Exception(0,
783: 0,
784: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 785: }
1.96 paf 786: }
787: void file_write(
788: const String& file_spec,
1.154 paf 789: const char* data, size_t size,
1.126 paf 790: bool as_text,
1.96 paf 791: bool do_append) {
1.126 paf 792: File_write_action_info info={data, size};
1.98 paf 793: file_write_action_under_lock(
1.154 paf 794: file_spec,
795: "write", file_write_action, &info,
796: as_text,
797: do_append);
1.30 paf 798: }
799:
1.63 parser 800: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 801: static void rmdir(const String& file_spec, size_t pos_after) {
1.154 paf 802: size_t pos_before;
803: if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
1.126 paf 804: rmdir(file_spec, pos_before+1);
1.50 paf 805:
1.154 paf 806: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 807: }
1.95 paf 808: bool file_delete(const String& file_spec, bool fail_on_read_problem) {
1.154 paf 809: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 810: if(unlink(fname)!=0)
1.93 paf 811: if(fail_on_read_problem)
1.126 paf 812: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 813: &file_spec,
814: "unlink failed: %s (%d), actual filename '%s'",
1.154 paf 815: strerror(errno), errno, fname);
1.93 paf 816: else
817: return false;
1.50 paf 818:
1.126 paf 819: rmdir(file_spec, 1);
1.93 paf 820: return true;
1.60 parser 821: }
1.95 paf 822: void file_move(const String& old_spec, const String& new_spec) {
1.154 paf 823: const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
824: const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 825:
1.126 paf 826: create_dir_for_file(new_spec);
1.63 parser 827:
1.60 parser 828: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 829: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 830: &old_spec,
831: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 paf 832: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 833:
1.126 paf 834: rmdir(old_spec, 1);
1.31 paf 835: }
836:
1.51 paf 837:
1.126 paf 838: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 839: struct stat lfinfo;
840: bool result=stat(fname, &lfinfo)==0;
841: if(afinfo)
842: *afinfo=lfinfo;
843: return result;
1.119 paf 844: }
845:
846: bool entry_exists(const String& file_spec) {
1.154 paf 847: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 848: return entry_exists(fname, 0);
1.118 paf 849: }
850:
1.51 paf 851: static bool entry_readable(const String& file_spec, bool need_dir) {
1.154 paf 852: char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 853: if(need_dir) {
1.126 paf 854: size_t size=strlen(fname);
1.120 paf 855: while(size) {
1.126 paf 856: char c=fname[size-1];
1.120 paf 857: if(c=='/' || c=='\\')
858: fname[--size]=0;
859: else
860: break;
861: }
862: }
1.51 paf 863: struct stat finfo;
1.118 paf 864: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 865: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 866: return is_dir==need_dir;
867: }
868: return false;
869: }
1.31 paf 870: bool file_readable(const String& file_spec) {
1.126 paf 871: return entry_readable(file_spec, false);
1.51 paf 872: }
873: bool dir_readable(const String& file_spec) {
1.126 paf 874: return entry_readable(file_spec, true);
1.65 parser 875: }
1.154 paf 876: const String* file_readable(const String& path, const String& name) {
877: String& result=*new String(path);
878: result << "/";
879: result << name;
880: return file_readable(result)?&result:0;
1.43 paf 881: }
882: bool file_executable(const String& file_spec) {
1.154 paf 883: return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 884: }
885:
1.64 parser 886: bool file_stat(const String& file_spec,
1.58 parser 887: size_t& rsize,
1.126 paf 888: time_t& ratime,
889: time_t& rmtime,
890: time_t& rctime,
1.64 parser 891: bool fail_on_read_problem) {
1.154 paf 892: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
893: struct stat finfo;
1.44 paf 894: if(stat(fname, &finfo)!=0)
1.64 parser 895: if(fail_on_read_problem)
1.126 paf 896: throw Exception("file.missing",
1.67 parser 897: &file_spec,
898: "getting file size failed: %s (%d), real filename '%s'",
1.154 paf 899: strerror(errno), errno, fname);
1.64 parser 900: else
901: return false;
1.58 parser 902: rsize=finfo.st_size;
903: ratime=finfo.st_atime;
904: rmtime=finfo.st_mtime;
905: rctime=finfo.st_ctime;
1.64 parser 906: return true;
1.18 paf 907: }
908:
1.126 paf 909: char* getrow(char* *row_ref, char delim) {
910: char* result=*row_ref;
1.8 paf 911: if(result) {
1.126 paf 912: *row_ref=strchr(result, delim);
1.8 paf 913: if(*row_ref)
914: *((*row_ref)++)=0;
915: else if(!*result)
916: return 0;
917: }
918: return result;
919: }
920:
1.126 paf 921: char* lsplit(char* string, char delim) {
1.23 paf 922: if(string) {
1.126 paf 923: char* v=strchr(string, delim);
1.8 paf 924: if(v) {
925: *v=0;
926: return v+1;
927: }
928: }
929: return 0;
930: }
931:
1.126 paf 932: char* lsplit(char* *string_ref, char delim) {
933: char* result=*string_ref;
934: char* next=lsplit(*string_ref, delim);
1.8 paf 935: *string_ref=next;
936: return result;
1.9 paf 937: }
938:
1.126 paf 939: char* rsplit(char* string, char delim) {
1.18 paf 940: if(string) {
1.126 paf 941: char* v=strrchr(string, delim);
1.18 paf 942: if(v) {
1.9 paf 943: *v=0;
944: return v+1;
945: }
946: }
947: return NULL;
1.10 paf 948: }
949:
1.37 paf 950: /// @todo less stupid type detection
1.154 paf 951: const char* format(double value, char* fmt) {
1.126 paf 952: char local_buf[MAX_NUMBER];
1.108 paf 953: size_t size;
954:
1.10 paf 955: if(fmt)
956: if(strpbrk(fmt, "diouxX"))
957: if(strpbrk(fmt, "ouxX"))
1.126 paf 958: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 959: else
1.126 paf 960: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 961: else
1.126 paf 962: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 963: else
1.126 paf 964: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 965:
1.154 paf 966: return pa_strdup(local_buf, size);
1.12 paf 967: }
968:
1.36 paf 969: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 970: #ifdef WIN32
971: do{
1.154 paf 972: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 973: if(chunk_written<=0)
974: break;
975: size-=chunk_written;
1.36 paf 976: buf=((const char*)buf)+chunk_written;
1.126 paf 977: } while(size>0);
1.12 paf 978:
979: return size;
980: #else
1.126 paf 981: return fwrite(buf, 1, size, stdout);
1.12 paf 982: #endif
1.2 paf 983: }
1.14 paf 984:
1.154 paf 985: char* unescape_chars(const char* cp, int len) {
986: char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 987: enum EscapeState {
1.33 paf 988: EscapeRest,
989: EscapeFirst,
1.14 paf 990: EscapeSecond
991: } escapeState=EscapeRest;
992: int escapedValue=0;
993: int srcPos=0;
994: int dstPos=0;
995: while(srcPos < len) {
1.126 paf 996: int ch=cp[srcPos];
1.14 paf 997: switch(escapeState) {
998: case EscapeRest:
999: if(ch=='%') {
1000: escapeState=EscapeFirst;
1001: } else if(ch=='+') {
1.126 paf 1002: s[dstPos++]=' ';
1.14 paf 1003: } else {
1004: s[dstPos++]=ch;
1005: }
1006: break;
1007: case EscapeFirst:
1008: escapedValue=hex_value[ch] << 4;
1009: escapeState=EscapeSecond;
1010: break;
1011: case EscapeSecond:
1.126 paf 1012: escapedValue +=hex_value[ch];
1.14 paf 1013: s[dstPos++]=escapedValue;
1014: escapeState=EscapeRest;
1015: break;
1016: }
1.126 paf 1017: srcPos++;
1.14 paf 1018: }
1019: s[dstPos]=0;
1020: return s;
1.24 paf 1021: }
1022:
1023: #ifdef WIN32
1.126 paf 1024: void back_slashes_to_slashes(char* s) {
1.24 paf 1025: if(s)
1026: for(; *s; s++)
1027: if(*s=='\\')
1.126 paf 1028: *s='/';
1.24 paf 1029: }
1.42 paf 1030: /*
1.126 paf 1031: void slashes_to_back_slashes(char* s) {
1.42 paf 1032: if(s)
1033: for(; *s; s++)
1034: if(*s=='/')
1.126 paf 1035: *s='\\';
1.42 paf 1036: }
1037: */
1.24 paf 1038: #endif
1.41 paf 1039:
1.126 paf 1040: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 1041: while(true) {
1042: if(!(*s1)) {
1043: if(!(*s2))
1044: return true;
1045: else
1046: return !strict;
1047: } else if(!(*s2))
1048: return !strict;
1049: if(isalpha(*s1)) {
1050: if(tolower(*s1) !=tolower(*s2))
1051: return false;
1052: } else if((*s1) !=(*s2))
1053: return false;
1.126 paf 1054: s1++;
1055: s2++;
1.41 paf 1056: }
1.57 parser 1057: }
1058:
1.84 paf 1059: static bool isLeap(int year) {
1.57 parser 1060: return !(
1061: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 1062: );
1.57 parser 1063: }
1064:
1065: int getMonthDays(int year, int month) {
1066: int monthDays[]={
1.126 paf 1067: 31,
1068: isLeap(year) ? 29 : 28,
1069: 31,
1070: 30,
1071: 31,
1072: 30,
1073: 31,
1074: 31,
1075: 30,
1076: 31,
1077: 30,
1.57 parser 1078: 31
1.126 paf 1079: };
1080: return monthDays[month];
1.41 paf 1081: }
1.69 parser 1082:
1.126 paf 1083: void remove_crlf(char* start, char* end) {
1084: for(char* p=start; p<end; p++)
1.69 parser 1085: switch(*p) {
1.126 paf 1086: case '\n': *p='|'; break;
1087: case '\r': *p=' '; break;
1.69 parser 1088: }
1.91 paf 1089: }
1090:
1091:
1092: /// must be last in this file
1093: #undef vsnprintf
1.126 paf 1094: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 1095: if(!s)
1096: return 0;
1097:
1098: int r;
1099: // note: on win32& maybe somewhere else
1100: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1101: --s;
1102: #if _MSC_VER
1103: /*
1104: win32:
1105: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
1106:
1.154 paf 1107: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 1108: */
1.126 paf 1109: r=_vsnprintf(b, s, f, l);
1.91 paf 1110: if(r<0)
1111: r=s;
1112: #else
1.126 paf 1113: r=vsnprintf(b, s, f, l);
1.91 paf 1114: /*
1115: solaris:
1116: man vsnprintf
1117:
1118: The snprintf() function returns the number of characters
1119: formatted, that is, the number of characters that would have
1120: been written to the buffer if it were large enough. If the
1121: value of n is 0 on a call to snprintf(), an unspecified
1122: value less than 1 is returned.
1123: */
1124:
1125: if(r<0)
1126: r=0;
1127: else if(r>s)
1128: r=s;
1129: #endif
1130: b[r]=0;
1131: return r;
1132: }
1133:
1.126 paf 1134: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1135: va_list l;
1.126 paf 1136: va_start(l, f);
1137: int r=__vsnprintf(b, s, f, l);
1138: va_end(l);
1.91 paf 1139: return r;
1.98 paf 1140: }
1141:
1142: int pa_sleep(unsigned long secs, unsigned long usecs) {
1.126 paf 1143: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
1.98 paf 1144:
1145: #ifdef WIN32
1.126 paf 1146: Sleep(secs * 1000 + usecs / 1000);
1.98 paf 1147: return 0;
1148: #else
1149: struct timeval t;
1150: t.tv_sec = secs;
1151: t.tv_usec = usecs;
1.126 paf 1152: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
1.98 paf 1153: #endif
1.135 paf 1154: }
1155:
1156:
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