Annotation of parser3/src/main/pa_http.C, revision 1.21
1.1 paf 1: /** @file
2: Parser: http support functions.
3:
4: Copyright(c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com)
5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
6: */
7:
1.21 ! misha 8: static const char * const IDENT_HTTP_C="$Date: 2009-01-12 07:47:26 $";
1.1 paf 9:
10: #include "pa_http.h"
11: #include "pa_common.h"
12: #include "pa_charsets.h"
13: #include "pa_request_charsets.h"
14:
15: // defines
16:
1.19 misha 17: #define HTTP_METHOD_NAME "method"
18: #define HTTP_FORM_NAME "form"
19: #define HTTP_BODY_NAME "body"
20: #define HTTP_TIMEOUT_NAME "timeout"
21: #define HTTP_HEADERS_NAME "headers"
22: #define HTTP_COOKIES_NAME "cookies"
23:
24: #define HTTP_ANY_STATUS_NAME "any-status"
1.20 misha 25: #define HTTP_OMIT_POST_CHARSET_NAME "omit-post-charset" // ^file::load[...;http://...;$.form[...]$.method[post]]
1.12 misha 26: // by default add charset to content-type
27:
1.1 paf 28: #define HTTP_TABLES_NAME "tables"
1.12 misha 29:
1.1 paf 30: #define HTTP_USER "user"
31: #define HTTP_PASSWORD "password"
32:
33: #define DEFAULT_USER_AGENT "parser3"
34:
35: # ifndef INADDR_NONE
36: # define INADDR_NONE ((ulong) -1)
37: # endif
38:
39: #undef CRLF
40: #define CRLF "\r\n"
41:
42: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
43: memset(addr, 0, sizeof(*addr));
44: addr->sin_family=AF_INET;
45: addr->sin_port=htons(port);
46: if(host) {
47: ulong packed_ip=inet_addr(host);
48: if(packed_ip!=INADDR_NONE)
49: memcpy(&addr->sin_addr, &packed_ip, sizeof(packed_ip));
50: else {
51: struct hostent *hostIP=gethostbyname(host);
52: if(hostIP)
53: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
54: else
55: return false;
56: }
57: } else
58: addr->sin_addr.s_addr=INADDR_ANY;
59: return true;
60: }
61:
62: size_t guess_content_length(char* buf) {
63: char* ptr;
64: if((ptr=strstr(buf, "Content-Length:"))) // Apache
65: goto found;
66: if((ptr=strstr(buf, "content-length:"))) // Parser 3
67: goto found;
68: if((ptr=strstr(buf, "Content-length:"))) // maybe 1
69: goto found;
70: if((ptr=strstr(buf, "CONTENT-LENGTH:"))) // maybe 2
71: goto found;
72: return 0;
73: found:
74: char *error_pos;
75: size_t result=(size_t)strtol(ptr+15/*strlen("CONTENT-LENGTH:")*/, &error_pos, 0);
76:
77: const size_t reasonable_initial_max=0x400*0x400*10 /*10M*/;
78: if(result>reasonable_initial_max) // sanity check
79: return reasonable_initial_max;
80: return 0;//result;
81: }
82:
83: static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200) {
84: int result=0;
85: // fetching some to local buffer, guessing on possible content-length
86: response_size=0x400*20; // initial size if content-length could not be determined
87: const size_t preview_size=0x400*20;
88: char preview_buf[preview_size+1/*terminator*/]; // 20K buffer to preview headers
89: ssize_t received_size=recv(sock, preview_buf, preview_size, 0);
90: if(received_size==0)
91: goto done;
92: if(received_size<0) {
93: if(int no=pa_socks_errno())
94: throw Exception("http.timeout",
95: 0,
96: "error receiving response header: %s (%d)", pa_socks_strerr(no), no);
97: goto done;
98: }
1.2 paf 99: // terminator [helps futher string searches]
100: preview_buf[received_size]=0;
101: // checking status
102: if(char* EOLat=strstr(preview_buf, "\n")) {
103: const String status_line(pa_strdup(preview_buf, EOLat-preview_buf));
104: ArrayString astatus;
105: size_t pos_after=0;
106: status_line.split(astatus, pos_after, " ");
107: const String& status_code=*astatus.get(astatus.count()>1?1:0);
108: result=status_code.as_int();
109:
110: if(fail_on_status_ne_200 && result!=200)
111: throw Exception("http.status",
112: &status_code,
113: "invalid HTTP response status");
114: }
1.1 paf 115: // detecting response_size
116: {
117: if(size_t content_length=guess_content_length(preview_buf))
118: response_size=preview_size+content_length; // a little more than needed, will adjust response_size by actual received size later
119: }
120:
121: // [gcc is happier this way, see goto above]
122: {
123: // allocating initial buf
124: response=(char*)pa_malloc_atomic(response_size+1/*terminator*/); // just setting memory block type
125: char* ptr=response;
126: size_t todo_size=response_size;
127: // coping part of already received body
128: memcpy(ptr, preview_buf, received_size);
129: ptr+=received_size;
130: todo_size-=received_size;
131:
132: // we use terminator byte for two purposes here:
133: // 1. we return there zero always, not knowing: maybe they would want to create String form $file.body?
134: // invariant: all Strings should have zero-terminated buffers
135: // 2. we use that out-of-size byte to detect if our content-length guess was wrong
136: // when recv gets more than we expected
137: // a) we know that the content-length guess was wrong
138: // b) we have space to put the first byte of extra data
139: // c) we use less code to detect normal situation: on last while-cycle recv expected to just return 0
140: while(true) {
141: received_size=recv(sock, ptr, todo_size+1/*there is always a place for terminator*/, 0);
142: if(received_size==0) {
143: response_size-=todo_size; // in case we received less than expected, cut down the reported size
144: break;
145: }
146: if(received_size<0) {
147: if(int no=pa_socks_errno())
148: throw Exception("http.timeout",
149: 0,
150: "error receiving response body: %s (%d)", pa_socks_strerr(no), no);
151: break;
152: }
153: // they've touched the terminator?
154: if((size_t)received_size>todo_size)
155: {
156: // that means that our guessed response_size was not big enough
157: const size_t grow_chunk_size=0x400*0x400; // 1M
158: response_size+=grow_chunk_size;
159: size_t ptr_offset=ptr-response;
160: response=(char*)pa_realloc(response, response_size+1/*terminator*/);
161: ptr=response+ptr_offset;
162: todo_size+=grow_chunk_size;
163: }
164: // can't do this before realloc: we need <todo_size check
165: ptr+=received_size;
166: todo_size-=received_size;
167: }
168: }
169: done:
170: if(result)
171: {
172: response[response_size]=0;
173: return result;
174: }
175: else
176: throw Exception("http.response",
177: 0,
178: "bad response from host - no status found (size=%u)", response_size);
179: }
180:
181: /* ********************** request *************************** */
182:
183: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
184: # define PA_USE_ALARM
185: #endif
186:
187: #ifdef PA_USE_ALARM
188: static sigjmp_buf timeout_env;
189: static void timeout_handler(int /*sig*/){
190: siglongjmp(timeout_env, 1);
191: }
192: #endif
193:
194: static int http_request(char*& response, size_t& response_size,
195: const char* host, short port,
196: const char* request,
197: int timeout_secs,
198: bool fail_on_status_ne_200) {
199: if(!host)
200: throw Exception("http.host",
201: 0,
202: "zero hostname"); //never
203:
204: volatile // to prevent makeing it register variable, because it will be clobbered by longjmp [thanks gcc warning]
205: int sock=-1;
206: #ifdef PA_USE_ALARM
207: signal(SIGALRM, timeout_handler);
208: #endif
209: #ifdef PA_USE_ALARM
210: if(sigsetjmp(timeout_env, 1)) {
211: // stupid gcc [2.95.4] generated bad code
212: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
213: // rewritten simplier [athough duplicating closesocket code]
214: if(sock>=0)
215: closesocket(sock);
216: throw Exception("http.timeout",
217: 0,
218: "timeout occured while retrieving document");
219: return 0; // never
220: } else {
221: alarm(timeout_secs);
222: #endif
223: try {
224: int result;
225: struct sockaddr_in dest;
226:
227: if(!set_addr(&dest, host, port))
228: throw Exception("http.host",
229: 0,
230: "can not resolve hostname \"%s\"", host);
231:
232: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) {
233: int no=pa_socks_errno();
234: throw Exception("http.connect",
235: 0,
236: "can not make socket: %s (%d)", pa_socks_strerr(no), no);
237: }
238:
239: // To enable SO_DONTLINGER (that is, disable SO_LINGER)
240: // l_onoff should be set to zero and setsockopt should be called
241: linger dont_linger={0,0};
242: setsockopt(sock, SOL_SOCKET, SO_LINGER, (const char *)&dont_linger, sizeof(dont_linger));
243:
244: #ifdef WIN32
245: // SO_*TIMEO can be defined in .h but not implemlemented in protocol,
246: // failing subsequently with Option not supported by protocol (99) message
247: // could not suppress that, so leaving this only for win32
248: int timeout_ms=timeout_secs*1000;
249: setsockopt(sock, SOL_SOCKET, SO_SNDTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms));
250: setsockopt(sock, SOL_SOCKET, SO_RCVTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms));
251: #endif
252:
253: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) {
254: int no=pa_socks_errno();
255: throw Exception("http.connect",
256: 0,
257: "can not connect to host \"%s\": %s (%d)", host, pa_socks_strerr(no), no);
258: }
259: size_t request_size=strlen(request);
260: if(send(sock, request, request_size, 0)!=(ssize_t)request_size) {
261: int no=pa_socks_errno();
262: throw Exception("http.timeout",
263: 0,
264: "error sending request: %s (%d)", pa_socks_strerr(no), no);
265: }
266:
267: result=http_read_response(response, response_size, sock, fail_on_status_ne_200);
268: closesocket(sock);
269: #ifdef PA_USE_ALARM
270: alarm(0);
271: #endif
272: return result;
273: } catch(...) {
274: #ifdef PA_USE_ALARM
275: alarm(0);
276: #endif
277: if(sock>=0)
278: closesocket(sock);
279: rethrow;
280: }
281: #ifdef PA_USE_ALARM
282: }
283: #endif
284: }
285:
286: #ifndef DOXYGEN
287: struct Http_pass_header_info {
288: Request_charsets* charsets;
289: String* request;
290: bool user_agent_specified;
1.12 misha 291: bool content_type_specified;
1.1 paf 292: };
293: #endif
1.9 misha 294: static void http_pass_header(HashStringValue::key_type name,
1.1 paf 295: HashStringValue::value_type value,
296: Http_pass_header_info *info) {
1.9 misha 297:
1.10 misha 298: String aname=String(name, String::L_URI);
1.9 misha 299:
1.21 ! misha 300: *info->request << aname << ": "
1.10 misha 301: << attributed_meaning_to_string(*value, String::L_URI, false)
1.9 misha 302: << CRLF;
1.1 paf 303:
1.12 misha 304: const String::Body name_upper=aname.change_case(info->charsets->source(), String::CC_UPPER);
1.20 misha 305: if(name_upper==HTTP_USER_AGENT_UPPER)
1.9 misha 306: info->user_agent_specified=true;
1.20 misha 307: if(name_upper==HTTP_CONTENT_TYPE_UPPER)
1.12 misha 308: info->content_type_specified=true;
1.1 paf 309: }
310:
1.10 misha 311: static void http_pass_cookie(HashStringValue::key_type name,
1.20 misha 312: HashStringValue::value_type value,
313: Http_pass_header_info *info) {
1.10 misha 314:
1.17 misha 315: *info->request << String(name, String::L_HTTP_COOKIE) << "="
316: << attributed_meaning_to_string(*value, String::L_HTTP_COOKIE, false)
1.10 misha 317: << "; ";
318:
319: }
1.1 paf 320:
321: static const String* basic_authorization_field(const char* user, const char* pass) {
322: if(!user&& !pass)
323: return 0;
324:
325: String combined;
326: if(user)
327: combined<<user;
328: combined<<":";
329: if(pass)
330: combined<<pass;
331:
1.20 misha 332: String* result=new String("Basic ");
333: *result<<pa_base64_encode(combined.cstr(), combined.length());
1.1 paf 334: return result;
335: }
336:
337: static void form_string_value2string(
1.20 misha 338: HashStringValue::key_type key,
339: const String& value,
340: String& result)
1.1 paf 341: {
342: result << String(key, String::L_URI) << "=";
343: result.append(value, String::L_URI, true);
1.20 misha 344: result << "&";
1.1 paf 345: }
1.20 misha 346:
1.1 paf 347: #ifndef DOXYGEN
348: struct Form_table_value2string_info {
349: HashStringValue::key_type key;
350: String& result;
351:
352: Form_table_value2string_info(HashStringValue::key_type akey, String& aresult):
353: key(akey), result(aresult) {}
354: };
355: #endif
356: static void form_table_value2string(Table::element_type row, Form_table_value2string_info* info) {
357: form_string_value2string(info->key, *row->get(0), info->result);
358: }
359: static void form_value2string(
1.20 misha 360: HashStringValue::key_type key,
361: HashStringValue::value_type value,
362: String* result)
1.1 paf 363: {
364: if(const String* svalue=value->get_string())
365: form_string_value2string(key, *svalue, *result);
366: else if(Table* tvalue=value->get_table()) {
367: Form_table_value2string_info info(key, *result);
368: tvalue->for_each(form_table_value2string, &info);
369: } else
1.18 misha 370: throw Exception(PARSER_RUNTIME,
1.1 paf 371: new String(key, String::L_TAINTED),
372: "is %s, "HTTP_FORM_NAME" option value must either string or table", value->type());
373: }
1.20 misha 374:
1.5 misha 375: const char* pa_form2string(HashStringValue& form, Request_charsets& charsets) {
1.1 paf 376: String string;
1.3 paf 377: form.for_each<String*>(form_value2string, &string);
1.5 misha 378: return string.cstr(String::L_UNSPECIFIED, 0, &charsets);
1.1 paf 379: }
380: static void find_headers_end(char* p,
381: char*& headers_end_at,
382: char*& raw_body)
383: {
384: raw_body=p;
385: // \n\n
386: // \r\n\r\n
387: while((p=strchr(p, '\n'))) {
388: headers_end_at=++p; // \n>.<
389: if(*p=='\r') // \r\n>\r?<\n
390: p++;
391: if(*p=='\n') { // \r\n\r>\n?<
392: raw_body=p+1;
393: return;
394: }
395: }
396: headers_end_at=0;
397: }
398:
399: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
400: File_read_http_result pa_internal_file_read_http(Request_charsets& charsets,
1.20 misha 401: const String& file_spec,
402: bool as_text,
1.15 misha 403: HashStringValue *options,
404: bool transcode_text_result) {
1.1 paf 405: File_read_http_result result;
1.20 misha 406: char host[MAX_STRING];
1.1 paf 407: const char* uri;
408: short port;
1.10 misha 409: const char* method="GET";
1.21 ! misha 410: bool method_is_get=true;
1.1 paf 411: HashStringValue* form=0;
412: const char* body_cstr=0;
413: int timeout_secs=2;
414: bool fail_on_status_ne_200=true;
1.12 misha 415: bool omit_post_charset=false;
1.1 paf 416: Value* vheaders=0;
1.10 misha 417: Value* vcookies=0;
1.11 misha 418: Value* vbody=0;
1.1 paf 419: Charset *asked_remote_charset=0;
420: const char* user_cstr=0;
421: const char* password_cstr=0;
422:
423: if(options) {
424: int valid_options=pa_get_valid_file_options_count(*options);
425:
426: if(Value* vmethod=options->get(HTTP_METHOD_NAME)) {
427: valid_options++;
1.21 ! misha 428: method=vmethod->as_string().change_case(r.charsets.source(), String::CC_UPPER).cstr();
! 429: method_is_get=strcmp(method, "GET")==0;
1.1 paf 430: }
431: if(Value* vform=options->get(HTTP_FORM_NAME)) {
432: valid_options++;
433: form=vform->get_hash();
434: }
1.11 misha 435: if(vbody=options->get(HTTP_BODY_NAME)) {
1.1 paf 436: valid_options++;
437: }
438: if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) {
439: valid_options++;
440: timeout_secs=vtimeout->as_int();
441: }
1.11 misha 442: if(vheaders=options->get(HTTP_HEADERS_NAME)) {
1.1 paf 443: valid_options++;
444: }
1.11 misha 445: if(vcookies=options->get(HTTP_COOKIES_NAME)) {
1.10 misha 446: valid_options++;
447: }
1.1 paf 448: if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) {
449: valid_options++;
450: fail_on_status_ne_200=!vany_status->as_bool();
1.12 misha 451: }
1.20 misha 452: if(Value* vomit_post_charset=options->get(HTTP_OMIT_POST_CHARSET_NAME)){
1.12 misha 453: valid_options++;
454: omit_post_charset=vomit_post_charset->as_bool();
455: }
1.6 misha 456: if(Value* vcharset_name=options->get(PA_CHARSET_NAME)) {
1.1 paf 457: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
458: change_case(charsets.source(), String::CC_UPPER));
459: }
460: if(Value* vuser=options->get(HTTP_USER)) {
461: valid_options++;
462: user_cstr=vuser->as_string().cstr();
463: }
464: if(Value* vpassword=options->get(HTTP_PASSWORD)) {
465: valid_options++;
466: password_cstr=vpassword->as_string().cstr();
467: }
468:
469: if(valid_options!=options->count())
1.7 misha 470: throw Exception(PARSER_RUNTIME,
1.1 paf 471: 0,
472: "invalid option passed");
473: }
474: if(!asked_remote_charset) // defaulting to $request:charset
475: asked_remote_charset=&charsets.source();
476:
1.11 misha 477: if(vbody){
478: if(method_is_get)
479: throw Exception(PARSER_RUNTIME,
480: 0,
481: "you can not use $."HTTP_BODY_NAME" option with method GET");
482:
483: if(form)
484: throw Exception(PARSER_RUNTIME,
485: 0,
486: "you can not use options $."HTTP_BODY_NAME" and $."HTTP_FORM_NAME" together");
487: }
1.1 paf 488:
489: //preparing request
490: String& connect_string=*new String;
491: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
492: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
493:
494: String request_head_and_body;
495: {
496: // influence URLencoding of tainted pieces to String::L_URI lang
497: Temp_client_charset temp(charsets, *asked_remote_charset);
498:
1.20 misha 499: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED, 0, &charsets);
1.1 paf 500:
501: const char* current=connect_string_cstr;
502: if(strncmp(current, "http://", 7)!=0)
1.18 misha 503: throw Exception(PARSER_RUNTIME,
1.1 paf 504: &connect_string,
505: "does not start with http://"); //never
506: current+=7;
507:
508: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
509: char* host_uri=lsplit(host, '/');
510: uri=host_uri?current+(host_uri-1-host):"/";
511: char* port_cstr=lsplit(host, ':');
512: char* error_pos=0;
513: port=port_cstr?(short)strtol(port_cstr, &error_pos, 0):80;
514:
515: bool uri_has_query_string=strchr(uri, '?')!=0;
516:
1.11 misha 517: // making request head
1.1 paf 518: String head;
1.11 misha 519: head << method << " " << uri;
520: if(form && method_is_get)
521: head << (uri_has_query_string?"&":"?") << pa_form2string(*form, charsets);
522:
523: head <<" HTTP/1.0" CRLF "host: "<< host << CRLF;
524:
1.12 misha 525: if(form && !method_is_get) { // POST
526: head << "content-type: " << HTTP_CONTENT_TYPE_FORM_URLENCODED;
527: if(!omit_post_charset)
528: head << "; charset=" << asked_remote_charset->NAME_CSTR() << ";";
529: head << CRLF;
1.11 misha 530: body_cstr=pa_form2string(*form, charsets);
531: } else if (vbody) {
532: body_cstr=vbody->as_string().cstr(String::L_UNSPECIFIED, 0, &charsets);
533: // needed for transcoded $.body[] first of all
534: body_cstr=Charset::transcode(
535: String::C(body_cstr, strlen(body_cstr)),
536: charsets.source(),
537: *asked_remote_charset
538: );
1.1 paf 539: }
540:
541: // http://www.ietf.org/rfc/rfc2617.txt
542: if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr))
543: head<<"authorization: "<<*authorization_field_value<<CRLF;
544:
545: bool user_agent_specified=false;
1.12 misha 546: bool content_type_specified=false;
1.1 paf 547: if(vheaders && !vheaders->is_string()) { // allow empty
548: if(HashStringValue *headers=vheaders->get_hash()) {
549: Http_pass_header_info info={&charsets, &head, false};
1.3 paf 550: headers->for_each<Http_pass_header_info*>(http_pass_header, &info);
1.1 paf 551: user_agent_specified=info.user_agent_specified;
1.12 misha 552: content_type_specified=info.content_type_specified;
1.1 paf 553: } else
1.7 misha 554: throw Exception(PARSER_RUNTIME,
1.1 paf 555: &connect_string,
556: "headers param must be hash");
557: };
558: if(!user_agent_specified) // defaulting
1.20 misha 559: head << HTTP_USER_AGENT ": " DEFAULT_USER_AGENT CRLF;
1.1 paf 560:
1.12 misha 561: if(form && !method_is_get && content_type_specified) // POST + form + content-type was specified
562: throw Exception(PARSER_RUNTIME,
563: &connect_string,
564: "$.content-type can't be specified with method POST");
565:
1.11 misha 566: if(vcookies && !vcookies->is_string()){ // allow empty
1.10 misha 567: if(HashStringValue* cookies=vcookies->get_hash()) {
568: head << "cookie: ";
569: Http_pass_header_info info={&charsets, &head, false};
570: cookies->for_each<Http_pass_header_info*>(http_pass_cookie, &info);
571: head << CRLF;
572: } else
573: throw Exception(PARSER_RUNTIME,
574: &connect_string,
575: "cookies param must be hash");
576: }
577:
1.1 paf 578: if(body_cstr) {
579: head << "content-length: " << format(strlen(body_cstr), "%u") << CRLF;
580: }
581:
1.6 misha 582: const char* head_cstr=head.cstr(String::L_UNSPECIFIED, 0, &charsets);
1.1 paf 583:
584: // head + end of header
585: request_head_and_body << head_cstr << CRLF;
1.8 misha 586:
1.1 paf 587: // body
588: if(body_cstr)
589: request_head_and_body << body_cstr;
590: }
591:
592: //sending request
593: char* response;
594: size_t response_size;
595: int status_code=http_request(response, response_size,
596: host, port, request_head_and_body.cstr(),
597: timeout_secs, fail_on_status_ne_200);
598:
599: //processing results
600: char* raw_body; size_t raw_body_size;
601: char* headers_end_at;
602: find_headers_end(response,
603: headers_end_at,
604: raw_body);
605: raw_body_size=response_size-(raw_body-response);
606:
607: result.headers=new HashStringValue;
608: VHash* vtables=new VHash;
609: result.headers->put(HTTP_TABLES_NAME, vtables);
610: Charset* real_remote_charset=0; // undetected, yet
611:
612: if(headers_end_at) {
613: *headers_end_at=0;
614: const String header_block(String::C(response, headers_end_at-response), true);
615:
616: ArrayString aheaders;
617: HashStringValue& tables=vtables->hash();
618:
619: size_t pos_after=0;
620: header_block.split(aheaders, pos_after, "\n");
621:
622: //processing headers
623: size_t aheaders_count=aheaders.count();
624: for(size_t i=1; i<aheaders_count; i++) {
625: const String& line=*aheaders.get(i);
626: size_t pos=line.pos(':');
627: if(pos==STRING_NOT_FOUND || pos<1)
628: throw Exception("http.response",
629: &connect_string,
630: "bad response from host - bad header \"%s\"", line.cstr());
1.14 misha 631: const String::Body HEADER_NAME=line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
632: const String& HEADER_VALUE=line.mid(pos+1, line.length()).trim(String::TRIM_BOTH, " \t\r");
1.20 misha 633: if(as_text && HEADER_NAME==HTTP_CONTENT_TYPE_UPPER)
1.14 misha 634: real_remote_charset=detect_charset(charsets.source(), HEADER_VALUE);
1.1 paf 635:
636: // tables
637: {
638: Value *valready=(Value *)tables.get(HEADER_NAME);
639: bool existed=valready!=0;
640: Table *table;
641: if(existed) {
642: // second+ appearence
643: table=valready->get_table();
644: } else {
645: // first appearence
1.14 misha 646: Table::columns_type columns=new ArrayString(1);
1.1 paf 647: *columns+=new String("value");
648: table=new Table(columns);
649: }
650: // this string becomes next row
651: ArrayString& row=*new ArrayString(1);
1.14 misha 652: row+=&HEADER_VALUE;
1.1 paf 653: *table+=&row;
654: // not existed before? add it
655: if(!existed)
656: tables.put(HEADER_NAME, new VTable(table));
657: }
658:
1.14 misha 659: result.headers->put(HEADER_NAME, new VString(HEADER_VALUE));
1.1 paf 660: }
661: }
662:
1.16 misha 663: if(as_text && raw_body_size>=3 && strncmp(raw_body, "\xEF\xBB\xBF", 3)==0){
1.20 misha 664: // skip UTF-8 signature (BOM code)
1.16 misha 665: raw_body+=3;
666: raw_body_size-=3;
667: }
668:
1.1 paf 669: // output response
670: String::C real_body=String::C(raw_body, raw_body_size);
1.16 misha 671:
672: if(as_text && transcode_text_result && raw_body_size) { // raw_body_size must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below
1.1 paf 673: // defaulting to used-asked charset [it's never empty!]
674: if(!real_remote_charset)
675: real_remote_charset=asked_remote_charset;
1.16 misha 676:
1.1 paf 677: real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source());
1.16 misha 678:
1.1 paf 679: }
680:
681: result.str=const_cast<char *>(real_body.str); // hacking a little
682: result.length=real_body.length;
1.16 misha 683:
1.1 paf 684: result.headers->put(file_status_name, new VInt(status_code));
1.16 misha 685:
1.1 paf 686: return result;
687: }
E-mail:
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to pa_http.C CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [parser3project] / parser3 / src / main