Annotation of parser3/src/main/pa_http.C, revision 1.12
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:
1.12 ! misha 8: static const char * const IDENT_HTTP_C="$Date: 2008-02-22 17:29:38 $";
1.1 paf 9:
10: #include "pa_http.h"
11: #include "pa_common.h"
12: #include "pa_charsets.h"
13: #include "pa_request_charsets.h"
14:
15: // defines
16:
17: #define HTTP_METHOD_NAME "method"
18: #define HTTP_FORM_NAME "form"
19: #define HTTP_BODY_NAME "body"
20: #define HTTP_TIMEOUT_NAME "timeout"
21: #define HTTP_HEADERS_NAME "headers"
1.10 misha 22: #define HTTP_COOKIES_NAME "cookies"
1.1 paf 23: #define HTTP_ANY_STATUS_NAME "any-status"
1.12 ! misha 24: #define HTTP_OMIT_POST_CHARSET "omit-post-charset" // ^file::load[...;http://...;$.form[...]$.method[post]]
! 25: // by default add charset to content-type
! 26:
1.1 paf 27: #define HTTP_TABLES_NAME "tables"
1.12 ! misha 28:
1.1 paf 29: #define HTTP_USER "user"
30: #define HTTP_PASSWORD "password"
31:
32: #define DEFAULT_USER_AGENT "parser3"
33:
34: # ifndef INADDR_NONE
35: # define INADDR_NONE ((ulong) -1)
36: # endif
37:
38: #undef CRLF
39: #define CRLF "\r\n"
40:
41: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
42: memset(addr, 0, sizeof(*addr));
43: addr->sin_family=AF_INET;
44: addr->sin_port=htons(port);
45: if(host) {
46: ulong packed_ip=inet_addr(host);
47: if(packed_ip!=INADDR_NONE)
48: memcpy(&addr->sin_addr, &packed_ip, sizeof(packed_ip));
49: else {
50: struct hostent *hostIP=gethostbyname(host);
51: if(hostIP)
52: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
53: else
54: return false;
55: }
56: } else
57: addr->sin_addr.s_addr=INADDR_ANY;
58: return true;
59: }
60:
61: size_t guess_content_length(char* buf) {
62: char* ptr;
63: if((ptr=strstr(buf, "Content-Length:"))) // Apache
64: goto found;
65: if((ptr=strstr(buf, "content-length:"))) // Parser 3
66: goto found;
67: if((ptr=strstr(buf, "Content-length:"))) // maybe 1
68: goto found;
69: if((ptr=strstr(buf, "CONTENT-LENGTH:"))) // maybe 2
70: goto found;
71: return 0;
72: found:
73: char *error_pos;
74: size_t result=(size_t)strtol(ptr+15/*strlen("CONTENT-LENGTH:")*/, &error_pos, 0);
75:
76: const size_t reasonable_initial_max=0x400*0x400*10 /*10M*/;
77: if(result>reasonable_initial_max) // sanity check
78: return reasonable_initial_max;
79: return 0;//result;
80: }
81:
82: static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200) {
83: int result=0;
84: // fetching some to local buffer, guessing on possible content-length
85: response_size=0x400*20; // initial size if content-length could not be determined
86: const size_t preview_size=0x400*20;
87: char preview_buf[preview_size+1/*terminator*/]; // 20K buffer to preview headers
88: ssize_t received_size=recv(sock, preview_buf, preview_size, 0);
89: if(received_size==0)
90: goto done;
91: if(received_size<0) {
92: if(int no=pa_socks_errno())
93: throw Exception("http.timeout",
94: 0,
95: "error receiving response header: %s (%d)", pa_socks_strerr(no), no);
96: goto done;
97: }
1.2 paf 98: // terminator [helps futher string searches]
99: preview_buf[received_size]=0;
100: // checking status
101: if(char* EOLat=strstr(preview_buf, "\n")) {
102: const String status_line(pa_strdup(preview_buf, EOLat-preview_buf));
103: ArrayString astatus;
104: size_t pos_after=0;
105: status_line.split(astatus, pos_after, " ");
106: const String& status_code=*astatus.get(astatus.count()>1?1:0);
107: result=status_code.as_int();
108:
109: if(fail_on_status_ne_200 && result!=200)
110: throw Exception("http.status",
111: &status_code,
112: "invalid HTTP response status");
113: }
1.1 paf 114: // detecting response_size
115: {
116: if(size_t content_length=guess_content_length(preview_buf))
117: response_size=preview_size+content_length; // a little more than needed, will adjust response_size by actual received size later
118: }
119:
120: // [gcc is happier this way, see goto above]
121: {
122: // allocating initial buf
123: response=(char*)pa_malloc_atomic(response_size+1/*terminator*/); // just setting memory block type
124: char* ptr=response;
125: size_t todo_size=response_size;
126: // coping part of already received body
127: memcpy(ptr, preview_buf, received_size);
128: ptr+=received_size;
129: todo_size-=received_size;
130:
131: // we use terminator byte for two purposes here:
132: // 1. we return there zero always, not knowing: maybe they would want to create String form $file.body?
133: // invariant: all Strings should have zero-terminated buffers
134: // 2. we use that out-of-size byte to detect if our content-length guess was wrong
135: // when recv gets more than we expected
136: // a) we know that the content-length guess was wrong
137: // b) we have space to put the first byte of extra data
138: // c) we use less code to detect normal situation: on last while-cycle recv expected to just return 0
139: while(true) {
140: received_size=recv(sock, ptr, todo_size+1/*there is always a place for terminator*/, 0);
141: if(received_size==0) {
142: response_size-=todo_size; // in case we received less than expected, cut down the reported size
143: break;
144: }
145: if(received_size<0) {
146: if(int no=pa_socks_errno())
147: throw Exception("http.timeout",
148: 0,
149: "error receiving response body: %s (%d)", pa_socks_strerr(no), no);
150: break;
151: }
152: // they've touched the terminator?
153: if((size_t)received_size>todo_size)
154: {
155: // that means that our guessed response_size was not big enough
156: const size_t grow_chunk_size=0x400*0x400; // 1M
157: response_size+=grow_chunk_size;
158: size_t ptr_offset=ptr-response;
159: response=(char*)pa_realloc(response, response_size+1/*terminator*/);
160: ptr=response+ptr_offset;
161: todo_size+=grow_chunk_size;
162: }
163: // can't do this before realloc: we need <todo_size check
164: ptr+=received_size;
165: todo_size-=received_size;
166: }
167: }
168: done:
169: if(result)
170: {
171: response[response_size]=0;
172: return result;
173: }
174: else
175: throw Exception("http.response",
176: 0,
177: "bad response from host - no status found (size=%u)", response_size);
178: }
179:
180: /* ********************** request *************************** */
181:
182: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
183: # define PA_USE_ALARM
184: #endif
185:
186: #ifdef PA_USE_ALARM
187: static sigjmp_buf timeout_env;
188: static void timeout_handler(int /*sig*/){
189: siglongjmp(timeout_env, 1);
190: }
191: #endif
192:
193: static int http_request(char*& response, size_t& response_size,
194: const char* host, short port,
195: const char* request,
196: int timeout_secs,
197: bool fail_on_status_ne_200) {
198: if(!host)
199: throw Exception("http.host",
200: 0,
201: "zero hostname"); //never
202:
203: volatile // to prevent makeing it register variable, because it will be clobbered by longjmp [thanks gcc warning]
204: int sock=-1;
205: #ifdef PA_USE_ALARM
206: signal(SIGALRM, timeout_handler);
207: #endif
208: #ifdef PA_USE_ALARM
209: if(sigsetjmp(timeout_env, 1)) {
210: // stupid gcc [2.95.4] generated bad code
211: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
212: // rewritten simplier [athough duplicating closesocket code]
213: if(sock>=0)
214: closesocket(sock);
215: throw Exception("http.timeout",
216: 0,
217: "timeout occured while retrieving document");
218: return 0; // never
219: } else {
220: alarm(timeout_secs);
221: #endif
222: try {
223: int result;
224: struct sockaddr_in dest;
225:
226: if(!set_addr(&dest, host, port))
227: throw Exception("http.host",
228: 0,
229: "can not resolve hostname \"%s\"", host);
230:
231: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) {
232: int no=pa_socks_errno();
233: throw Exception("http.connect",
234: 0,
235: "can not make socket: %s (%d)", pa_socks_strerr(no), no);
236: }
237:
238: // To enable SO_DONTLINGER (that is, disable SO_LINGER)
239: // l_onoff should be set to zero and setsockopt should be called
240: linger dont_linger={0,0};
241: setsockopt(sock, SOL_SOCKET, SO_LINGER, (const char *)&dont_linger, sizeof(dont_linger));
242:
243: #ifdef WIN32
244: // SO_*TIMEO can be defined in .h but not implemlemented in protocol,
245: // failing subsequently with Option not supported by protocol (99) message
246: // could not suppress that, so leaving this only for win32
247: int timeout_ms=timeout_secs*1000;
248: setsockopt(sock, SOL_SOCKET, SO_SNDTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms));
249: setsockopt(sock, SOL_SOCKET, SO_RCVTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms));
250: #endif
251:
252: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) {
253: int no=pa_socks_errno();
254: throw Exception("http.connect",
255: 0,
256: "can not connect to host \"%s\": %s (%d)", host, pa_socks_strerr(no), no);
257: }
258: size_t request_size=strlen(request);
259: if(send(sock, request, request_size, 0)!=(ssize_t)request_size) {
260: int no=pa_socks_errno();
261: throw Exception("http.timeout",
262: 0,
263: "error sending request: %s (%d)", pa_socks_strerr(no), no);
264: }
265:
266: result=http_read_response(response, response_size, sock, fail_on_status_ne_200);
267: closesocket(sock);
268: #ifdef PA_USE_ALARM
269: alarm(0);
270: #endif
271: return result;
272: } catch(...) {
273: #ifdef PA_USE_ALARM
274: alarm(0);
275: #endif
276: if(sock>=0)
277: closesocket(sock);
278: rethrow;
279: }
280: #ifdef PA_USE_ALARM
281: }
282: #endif
283: }
284:
285: #ifndef DOXYGEN
286: struct Http_pass_header_info {
287: Request_charsets* charsets;
288: String* request;
289: bool user_agent_specified;
1.12 ! misha 290: bool content_type_specified;
1.1 paf 291: };
292: #endif
1.9 misha 293: static void http_pass_header(HashStringValue::key_type name,
1.1 paf 294: HashStringValue::value_type value,
295: Http_pass_header_info *info) {
1.9 misha 296:
1.10 misha 297: String aname=String(name, String::L_URI);
1.9 misha 298:
299: *info->request <<aname<<": "
1.10 misha 300: << attributed_meaning_to_string(*value, String::L_URI, false)
1.9 misha 301: << CRLF;
1.1 paf 302:
1.12 ! misha 303: const String::Body name_upper=aname.change_case(info->charsets->source(), String::CC_UPPER);
! 304: if(name_upper==HTTP_USER_AGENT)
1.9 misha 305: info->user_agent_specified=true;
1.12 ! misha 306: if(name_upper==HTTP_CONTENT_TYPE)
! 307: info->content_type_specified=true;
1.1 paf 308: }
309:
1.10 misha 310: static void http_pass_cookie(HashStringValue::key_type name,
311: HashStringValue::value_type value,
312: Http_pass_header_info *info) {
313:
314: *info->request << String(name, String::L_HTTP_HEADER) << "="
315: << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
316: << "; ";
317:
318: }
1.1 paf 319:
320: static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
321: const String::Body CONTENT_TYPE_VALUE=
322: content_type_value.change_case(source_charset, String::CC_UPPER);
323: // content-type: xxx/xxx; source_charset=WE-NEED-THIS
324: // content-type: xxx/xxx; source_charset="WE-NEED-THIS"
325: // content-type: xxx/xxx; source_charset="WE-NEED-THIS";
326: size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET=");
327: if(before_charseteq_pos!=STRING_NOT_FOUND) {
1.5 misha 328: size_t charset_begin=before_charseteq_pos+8/*CHARSET=*/;
1.1 paf 329: size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
330: bool quoted=open_quote_pos==charset_begin;
331: if(quoted)
332: charset_begin++; // skip opening '"'
333: size_t charset_end=CONTENT_TYPE_VALUE.length();
334: if(quoted) {
335: size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
336: if(close_quote_pos!=STRING_NOT_FOUND)
337: charset_end=close_quote_pos;
338: } else {
339: size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin);
340: if(delim_pos!=STRING_NOT_FOUND)
341: charset_end=delim_pos;
342: }
343: const String::Body CHARSET_NAME_BODY=
1.4 misha 344: CONTENT_TYPE_VALUE.mid(charset_begin, charset_end - charset_begin);
1.1 paf 345:
346: return &charsets.get(CHARSET_NAME_BODY);
347: }
348:
349: return 0;
350: }
351:
352: static const String* basic_authorization_field(const char* user, const char* pass) {
353: if(!user&& !pass)
354: return 0;
355:
356: String combined;
357: if(user)
358: combined<<user;
359: combined<<":";
360: if(pass)
361: combined<<pass;
362:
363: String* result=new String("Basic "); *result<<pa_base64_encode(combined.cstr(), combined.length());
364: return result;
365: }
366:
367: static void form_string_value2string(
368: HashStringValue::key_type key,
369: const String& value,
370: String& result)
371: {
372: result << String(key, String::L_URI) << "=";
373: result.append(value, String::L_URI, true);
374: result<< "&";
375: }
376: #ifndef DOXYGEN
377: struct Form_table_value2string_info {
378: HashStringValue::key_type key;
379: String& result;
380:
381: Form_table_value2string_info(HashStringValue::key_type akey, String& aresult):
382: key(akey), result(aresult) {}
383: };
384: #endif
385: static void form_table_value2string(Table::element_type row, Form_table_value2string_info* info) {
386: form_string_value2string(info->key, *row->get(0), info->result);
387: }
388: static void form_value2string(
389: HashStringValue::key_type key,
390: HashStringValue::value_type value,
391: String* result)
392: {
393: if(const String* svalue=value->get_string())
394: form_string_value2string(key, *svalue, *result);
395: else if(Table* tvalue=value->get_table()) {
396: Form_table_value2string_info info(key, *result);
397: tvalue->for_each(form_table_value2string, &info);
398: } else
399: throw Exception(0,
400: new String(key, String::L_TAINTED),
401: "is %s, "HTTP_FORM_NAME" option value must either string or table", value->type());
402: }
1.5 misha 403: const char* pa_form2string(HashStringValue& form, Request_charsets& charsets) {
1.1 paf 404: String string;
1.3 paf 405: form.for_each<String*>(form_value2string, &string);
1.5 misha 406: return string.cstr(String::L_UNSPECIFIED, 0, &charsets);
1.1 paf 407: }
408: static void find_headers_end(char* p,
409: char*& headers_end_at,
410: char*& raw_body)
411: {
412: raw_body=p;
413: // \n\n
414: // \r\n\r\n
415: while((p=strchr(p, '\n'))) {
416: headers_end_at=++p; // \n>.<
417: if(*p=='\r') // \r\n>\r?<\n
418: p++;
419: if(*p=='\n') { // \r\n\r>\n?<
420: raw_body=p+1;
421: return;
422: }
423: }
424: headers_end_at=0;
425: }
426:
427: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
428: File_read_http_result pa_internal_file_read_http(Request_charsets& charsets,
429: const String& file_spec,
430: bool as_text,
431: HashStringValue *options) {
432: File_read_http_result result;
433: char host[MAX_STRING];
434: const char* uri;
435: short port;
1.10 misha 436: const char* method="GET";
1.1 paf 437: HashStringValue* form=0;
438: const char* body_cstr=0;
439: int timeout_secs=2;
440: bool fail_on_status_ne_200=true;
1.12 ! misha 441: bool omit_post_charset=false;
1.1 paf 442: Value* vheaders=0;
1.10 misha 443: Value* vcookies=0;
1.11 misha 444: Value* vbody=0;
1.1 paf 445: Charset *asked_remote_charset=0;
446: const char* user_cstr=0;
447: const char* password_cstr=0;
448:
449: if(options) {
450: int valid_options=pa_get_valid_file_options_count(*options);
451:
452: if(Value* vmethod=options->get(HTTP_METHOD_NAME)) {
453: valid_options++;
454: method=vmethod->as_string().cstr();
455: }
456: if(Value* vform=options->get(HTTP_FORM_NAME)) {
457: valid_options++;
458: form=vform->get_hash();
459: }
1.11 misha 460: if(vbody=options->get(HTTP_BODY_NAME)) {
1.1 paf 461: valid_options++;
462: }
463: if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) {
464: valid_options++;
465: timeout_secs=vtimeout->as_int();
466: }
1.11 misha 467: if(vheaders=options->get(HTTP_HEADERS_NAME)) {
1.1 paf 468: valid_options++;
469: }
1.11 misha 470: if(vcookies=options->get(HTTP_COOKIES_NAME)) {
1.10 misha 471: valid_options++;
472: }
1.1 paf 473: if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) {
474: valid_options++;
475: fail_on_status_ne_200=!vany_status->as_bool();
1.12 ! misha 476: }
! 477: if(Value* vomit_post_charset=options->get(HTTP_OMIT_POST_CHARSET)){
! 478: valid_options++;
! 479: omit_post_charset=vomit_post_charset->as_bool();
! 480: }
1.6 misha 481: if(Value* vcharset_name=options->get(PA_CHARSET_NAME)) {
1.1 paf 482: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
483: change_case(charsets.source(), String::CC_UPPER));
484: }
485: if(Value* vuser=options->get(HTTP_USER)) {
486: valid_options++;
487: user_cstr=vuser->as_string().cstr();
488: }
489: if(Value* vpassword=options->get(HTTP_PASSWORD)) {
490: valid_options++;
491: password_cstr=vpassword->as_string().cstr();
492: }
493:
494: if(valid_options!=options->count())
1.7 misha 495: throw Exception(PARSER_RUNTIME,
1.1 paf 496: 0,
497: "invalid option passed");
498: }
499: if(!asked_remote_charset) // defaulting to $request:charset
500: asked_remote_charset=&charsets.source();
501:
1.10 misha 502: bool method_is_get=strcmp(method, "GET")==0;
1.11 misha 503: if(vbody){
504: if(method_is_get)
505: throw Exception(PARSER_RUNTIME,
506: 0,
507: "you can not use $."HTTP_BODY_NAME" option with method GET");
508:
509: if(form)
510: throw Exception(PARSER_RUNTIME,
511: 0,
512: "you can not use options $."HTTP_BODY_NAME" and $."HTTP_FORM_NAME" together");
513: }
1.1 paf 514:
515: //preparing request
516: String& connect_string=*new String;
517: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
518: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
519:
520: String request_head_and_body;
521: {
522: // influence URLencoding of tainted pieces to String::L_URI lang
523: Temp_client_charset temp(charsets, *asked_remote_charset);
524:
1.5 misha 525: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED, 0, &charsets);
1.1 paf 526:
527: const char* current=connect_string_cstr;
528: if(strncmp(current, "http://", 7)!=0)
529: throw Exception(0,
530: &connect_string,
531: "does not start with http://"); //never
532: current+=7;
533:
534: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
535: char* host_uri=lsplit(host, '/');
536: uri=host_uri?current+(host_uri-1-host):"/";
537: char* port_cstr=lsplit(host, ':');
538: char* error_pos=0;
539: port=port_cstr?(short)strtol(port_cstr, &error_pos, 0):80;
540:
541: bool uri_has_query_string=strchr(uri, '?')!=0;
542:
1.11 misha 543: // making request head
1.1 paf 544: String head;
1.11 misha 545: head << method << " " << uri;
546: if(form && method_is_get)
547: head << (uri_has_query_string?"&":"?") << pa_form2string(*form, charsets);
548:
549: head <<" HTTP/1.0" CRLF "host: "<< host << CRLF;
550:
1.12 ! misha 551: if(form && !method_is_get) { // POST
! 552: head << "content-type: " << HTTP_CONTENT_TYPE_FORM_URLENCODED;
! 553: if(!omit_post_charset)
! 554: head << "; charset=" << asked_remote_charset->NAME_CSTR() << ";";
! 555: head << CRLF;
1.11 misha 556: body_cstr=pa_form2string(*form, charsets);
557: } else if (vbody) {
558: body_cstr=vbody->as_string().cstr(String::L_UNSPECIFIED, 0, &charsets);
559: // needed for transcoded $.body[] first of all
560: body_cstr=Charset::transcode(
561: String::C(body_cstr, strlen(body_cstr)),
562: charsets.source(),
563: *asked_remote_charset
564: );
1.1 paf 565: }
566:
567: // http://www.ietf.org/rfc/rfc2617.txt
568: if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr))
569: head<<"authorization: "<<*authorization_field_value<<CRLF;
570:
571: bool user_agent_specified=false;
1.12 ! misha 572: bool content_type_specified=false;
1.1 paf 573: if(vheaders && !vheaders->is_string()) { // allow empty
574: if(HashStringValue *headers=vheaders->get_hash()) {
575: Http_pass_header_info info={&charsets, &head, false};
1.3 paf 576: headers->for_each<Http_pass_header_info*>(http_pass_header, &info);
1.1 paf 577: user_agent_specified=info.user_agent_specified;
1.12 ! misha 578: content_type_specified=info.content_type_specified;
1.1 paf 579: } else
1.7 misha 580: throw Exception(PARSER_RUNTIME,
1.1 paf 581: &connect_string,
582: "headers param must be hash");
583: };
584: if(!user_agent_specified) // defaulting
585: head << "user-agent: " DEFAULT_USER_AGENT CRLF;
586:
1.12 ! misha 587: if(form && !method_is_get && content_type_specified) // POST + form + content-type was specified
! 588: throw Exception(PARSER_RUNTIME,
! 589: &connect_string,
! 590: "$.content-type can't be specified with method POST");
! 591:
1.11 misha 592: if(vcookies && !vcookies->is_string()){ // allow empty
1.10 misha 593: if(HashStringValue* cookies=vcookies->get_hash()) {
594: head << "cookie: ";
595: Http_pass_header_info info={&charsets, &head, false};
596: cookies->for_each<Http_pass_header_info*>(http_pass_cookie, &info);
597: head << CRLF;
598: } else
599: throw Exception(PARSER_RUNTIME,
600: &connect_string,
601: "cookies param must be hash");
602: }
603:
1.1 paf 604: if(body_cstr) {
605: head << "content-length: " << format(strlen(body_cstr), "%u") << CRLF;
606: }
607:
1.6 misha 608: const char* head_cstr=head.cstr(String::L_UNSPECIFIED, 0, &charsets);
1.1 paf 609:
610: // head + end of header
611: request_head_and_body << head_cstr << CRLF;
1.8 misha 612:
1.1 paf 613: // body
614: if(body_cstr)
615: request_head_and_body << body_cstr;
616: }
617:
618: //sending request
619: char* response;
620: size_t response_size;
621: int status_code=http_request(response, response_size,
622: host, port, request_head_and_body.cstr(),
623: timeout_secs, fail_on_status_ne_200);
624:
625: //processing results
626: char* raw_body; size_t raw_body_size;
627: char* headers_end_at;
628: find_headers_end(response,
629: headers_end_at,
630: raw_body);
631: raw_body_size=response_size-(raw_body-response);
632:
633: result.headers=new HashStringValue;
634: VHash* vtables=new VHash;
635: result.headers->put(HTTP_TABLES_NAME, vtables);
636: Charset* real_remote_charset=0; // undetected, yet
637:
638: if(headers_end_at) {
639: *headers_end_at=0;
640: const String header_block(String::C(response, headers_end_at-response), true);
641:
642: ArrayString aheaders;
643: HashStringValue& tables=vtables->hash();
644:
645: size_t pos_after=0;
646: header_block.split(aheaders, pos_after, "\n");
647:
648: //processing headers
649: size_t aheaders_count=aheaders.count();
650: for(size_t i=1; i<aheaders_count; i++) {
651: const String& line=*aheaders.get(i);
652: size_t pos=line.pos(':');
653: if(pos==STRING_NOT_FOUND || pos<1)
654: throw Exception("http.response",
655: &connect_string,
656: "bad response from host - bad header \"%s\"", line.cstr());
657: const String::Body HEADER_NAME=
658: line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
659: const String& header_value=line.mid(pos+1, line.length()).trim(String::TRIM_BOTH, " \t\r");
1.12 ! misha 660: if(as_text && HEADER_NAME==HTTP_CONTENT_TYPE)
1.1 paf 661: real_remote_charset=detect_charset(charsets.source(), header_value);
662:
663: // tables
664: {
665: Value *valready=(Value *)tables.get(HEADER_NAME);
666: bool existed=valready!=0;
667: Table *table;
668: if(existed) {
669: // second+ appearence
670: table=valready->get_table();
671: } else {
672: // first appearence
673: Table::columns_type columns =new ArrayString(1);
674: *columns+=new String("value");
675: table=new Table(columns);
676: }
677: // this string becomes next row
678: ArrayString& row=*new ArrayString(1);
679: row+=&header_value;
680: *table+=&row;
681: // not existed before? add it
682: if(!existed)
683: tables.put(HEADER_NAME, new VTable(table));
684: }
685:
686: result.headers->put(HEADER_NAME, new VString(header_value));
687: }
688: }
689:
690: // output response
691: String::C real_body=String::C(raw_body, raw_body_size);
692: if(as_text && raw_body_size) { // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below
693: // defaulting to used-asked charset [it's never empty!]
694: if(!real_remote_charset)
695: real_remote_charset=asked_remote_charset;
696: real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source());
697: }
698:
699: result.str=const_cast<char *>(real_body.str); // hacking a little
700: result.length=real_body.length;
701: result.headers->put(file_status_name, new VInt(status_code));
702: return result;
703: }
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