Annotation of parser3/src/main/pa_http.C, revision 1.49
1.1 paf 1: /** @file
2: Parser: http support functions.
3:
1.25 misha 4: Copyright(c) 2001-2009 ArtLebedev Group (http://www.artlebedev.com)
1.1 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
6: */
7:
1.49 ! moko 8: static const char * const IDENT_HTTP_C="$Date: 2010-10-16 22:24:20 $";
1.1 paf 9:
10: #include "pa_http.h"
11: #include "pa_common.h"
12: #include "pa_charsets.h"
13: #include "pa_request_charsets.h"
1.22 misha 14: #include "pa_request.h"
15: #include "pa_vfile.h"
16: #include "pa_random.h"
1.1 paf 17:
18: // defines
19:
1.19 misha 20: #define HTTP_METHOD_NAME "method"
21: #define HTTP_FORM_NAME "form"
22: #define HTTP_BODY_NAME "body"
23: #define HTTP_TIMEOUT_NAME "timeout"
24: #define HTTP_HEADERS_NAME "headers"
25: #define HTTP_COOKIES_NAME "cookies"
1.22 misha 26: #define HTTP_FORM_ENCTYPE_NAME "enctype"
1.19 misha 27: #define HTTP_ANY_STATUS_NAME "any-status"
1.20 misha 28: #define HTTP_OMIT_POST_CHARSET_NAME "omit-post-charset" // ^file::load[...;http://...;$.form[...]$.method[post]]
1.12 misha 29: // by default add charset to content-type
30:
1.1 paf 31: #define HTTP_TABLES_NAME "tables"
1.12 misha 32:
1.1 paf 33: #define HTTP_USER "user"
34: #define HTTP_PASSWORD "password"
35:
36: #define DEFAULT_USER_AGENT "parser3"
37:
38: # ifndef INADDR_NONE
39: # define INADDR_NONE ((ulong) -1)
40: # endif
41:
42: #undef CRLF
43: #define CRLF "\r\n"
44:
45: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
1.22 misha 46: memset(addr, 0, sizeof(*addr));
47: addr->sin_family=AF_INET;
48: addr->sin_port=htons(port);
49: if(host) {
1.1 paf 50: ulong packed_ip=inet_addr(host);
51: if(packed_ip!=INADDR_NONE)
52: memcpy(&addr->sin_addr, &packed_ip, sizeof(packed_ip));
53: else {
54: struct hostent *hostIP=gethostbyname(host);
55: if(hostIP)
56: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
57: else
58: return false;
59: }
1.22 misha 60: } else
1.1 paf 61: addr->sin_addr.s_addr=INADDR_ANY;
1.22 misha 62: return true;
1.1 paf 63: }
64:
65: size_t guess_content_length(char* buf) {
66: char* ptr;
67: if((ptr=strstr(buf, "Content-Length:"))) // Apache
68: goto found;
1.37 misha 69: if((ptr=strstr(buf, "content-length:"))) // Parser 3 before 3.4.0
1.1 paf 70: goto found;
71: if((ptr=strstr(buf, "Content-length:"))) // maybe 1
72: goto found;
73: if((ptr=strstr(buf, "CONTENT-LENGTH:"))) // maybe 2
74: goto found;
75: return 0;
76: found:
77: char *error_pos;
1.37 misha 78: size_t result=(size_t)strtol(ptr+15/*strlen("Content-Length:")*/, &error_pos, 0);
1.1 paf 79:
80: const size_t reasonable_initial_max=0x400*0x400*10 /*10M*/;
81: if(result>reasonable_initial_max) // sanity check
82: return reasonable_initial_max;
83: return 0;//result;
84: }
85:
86: static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200) {
87: int result=0;
1.37 misha 88: // fetching some to local buffer, guessing on possible Content-Length
89: response_size=0x400*20; // initial size if Content-Length could not be determined
1.1 paf 90: const size_t preview_size=0x400*20;
91: char preview_buf[preview_size+1/*terminator*/]; // 20K buffer to preview headers
92: ssize_t received_size=recv(sock, preview_buf, preview_size, 0);
93: if(received_size==0)
94: goto done;
95: if(received_size<0) {
96: if(int no=pa_socks_errno())
97: throw Exception("http.timeout",
98: 0,
99: "error receiving response header: %s (%d)", pa_socks_strerr(no), no);
100: goto done;
101: }
1.2 paf 102: // terminator [helps futher string searches]
103: preview_buf[received_size]=0;
104: // checking status
105: if(char* EOLat=strstr(preview_buf, "\n")) {
106: const String status_line(pa_strdup(preview_buf, EOLat-preview_buf));
107: ArrayString astatus;
108: size_t pos_after=0;
109: status_line.split(astatus, pos_after, " ");
110: const String& status_code=*astatus.get(astatus.count()>1?1:0);
111: result=status_code.as_int();
112:
113: if(fail_on_status_ne_200 && result!=200)
114: throw Exception("http.status",
115: &status_code,
116: "invalid HTTP response status");
117: }
1.1 paf 118: // detecting response_size
119: {
120: if(size_t content_length=guess_content_length(preview_buf))
121: response_size=preview_size+content_length; // a little more than needed, will adjust response_size by actual received size later
122: }
123:
124: // [gcc is happier this way, see goto above]
125: {
126: // allocating initial buf
127: response=(char*)pa_malloc_atomic(response_size+1/*terminator*/); // just setting memory block type
128: char* ptr=response;
129: size_t todo_size=response_size;
130: // coping part of already received body
131: memcpy(ptr, preview_buf, received_size);
132: ptr+=received_size;
133: todo_size-=received_size;
134:
135: // we use terminator byte for two purposes here:
136: // 1. we return there zero always, not knowing: maybe they would want to create String form $file.body?
137: // invariant: all Strings should have zero-terminated buffers
1.37 misha 138: // 2. we use that out-of-size byte to detect if our Content-Length guess was wrong
1.1 paf 139: // when recv gets more than we expected
1.37 misha 140: // a) we know that the Content-Length guess was wrong
1.1 paf 141: // b) we have space to put the first byte of extra data
142: // c) we use less code to detect normal situation: on last while-cycle recv expected to just return 0
143: while(true) {
144: received_size=recv(sock, ptr, todo_size+1/*there is always a place for terminator*/, 0);
145: if(received_size==0) {
146: response_size-=todo_size; // in case we received less than expected, cut down the reported size
147: break;
148: }
149: if(received_size<0) {
150: if(int no=pa_socks_errno())
151: throw Exception("http.timeout",
152: 0,
153: "error receiving response body: %s (%d)", pa_socks_strerr(no), no);
154: break;
155: }
156: // they've touched the terminator?
157: if((size_t)received_size>todo_size)
158: {
159: // that means that our guessed response_size was not big enough
160: const size_t grow_chunk_size=0x400*0x400; // 1M
161: response_size+=grow_chunk_size;
162: size_t ptr_offset=ptr-response;
163: response=(char*)pa_realloc(response, response_size+1/*terminator*/);
164: ptr=response+ptr_offset;
165: todo_size+=grow_chunk_size;
166: }
167: // can't do this before realloc: we need <todo_size check
168: ptr+=received_size;
169: todo_size-=received_size;
170: }
171: }
172: done:
173: if(result)
174: {
175: response[response_size]=0;
176: return result;
177: }
178: else
179: throw Exception("http.response",
180: 0,
181: "bad response from host - no status found (size=%u)", response_size);
182: }
183:
184: /* ********************** request *************************** */
185:
186: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
187: # define PA_USE_ALARM
188: #endif
189:
190: #ifdef PA_USE_ALARM
191: static sigjmp_buf timeout_env;
192: static void timeout_handler(int /*sig*/){
1.22 misha 193: siglongjmp(timeout_env, 1);
1.1 paf 194: }
195: #endif
196:
197: static int http_request(char*& response, size_t& response_size,
198: const char* host, short port,
1.22 misha 199: const char* request, size_t request_size,
1.1 paf 200: int timeout_secs,
201: bool fail_on_status_ne_200) {
202: if(!host)
203: throw Exception("http.host",
204: 0,
205: "zero hostname"); //never
206:
207: volatile // to prevent makeing it register variable, because it will be clobbered by longjmp [thanks gcc warning]
208: int sock=-1;
209: #ifdef PA_USE_ALARM
210: signal(SIGALRM, timeout_handler);
211: #endif
212: #ifdef PA_USE_ALARM
213: if(sigsetjmp(timeout_env, 1)) {
214: // stupid gcc [2.95.4] generated bad code
215: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
216: // rewritten simplier [athough duplicating closesocket code]
217: if(sock>=0)
218: closesocket(sock);
219: throw Exception("http.timeout",
220: 0,
221: "timeout occured while retrieving document");
222: return 0; // never
223: } else {
224: alarm(timeout_secs);
225: #endif
226: try {
227: int result;
228: struct sockaddr_in dest;
229:
230: if(!set_addr(&dest, host, port))
231: throw Exception("http.host",
232: 0,
233: "can not resolve hostname \"%s\"", host);
234:
235: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) {
236: int no=pa_socks_errno();
237: throw Exception("http.connect",
238: 0,
239: "can not make socket: %s (%d)", pa_socks_strerr(no), no);
240: }
241:
242: // To enable SO_DONTLINGER (that is, disable SO_LINGER)
243: // l_onoff should be set to zero and setsockopt should be called
244: linger dont_linger={0,0};
245: setsockopt(sock, SOL_SOCKET, SO_LINGER, (const char *)&dont_linger, sizeof(dont_linger));
246:
247: #ifdef WIN32
248: // SO_*TIMEO can be defined in .h but not implemlemented in protocol,
249: // failing subsequently with Option not supported by protocol (99) message
250: // could not suppress that, so leaving this only for win32
251: int timeout_ms=timeout_secs*1000;
252: setsockopt(sock, SOL_SOCKET, SO_SNDTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms));
253: setsockopt(sock, SOL_SOCKET, SO_RCVTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms));
254: #endif
255:
256: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) {
257: int no=pa_socks_errno();
258: throw Exception("http.connect",
259: 0,
260: "can not connect to host \"%s\": %s (%d)", host, pa_socks_strerr(no), no);
261: }
1.22 misha 262:
1.1 paf 263: if(send(sock, request, request_size, 0)!=(ssize_t)request_size) {
264: int no=pa_socks_errno();
265: throw Exception("http.timeout",
266: 0,
267: "error sending request: %s (%d)", pa_socks_strerr(no), no);
268: }
269:
270: result=http_read_response(response, response_size, sock, fail_on_status_ne_200);
271: closesocket(sock);
272: #ifdef PA_USE_ALARM
273: alarm(0);
274: #endif
275: return result;
276: } catch(...) {
277: #ifdef PA_USE_ALARM
278: alarm(0);
279: #endif
280: if(sock>=0)
281: closesocket(sock);
282: rethrow;
283: }
284: #ifdef PA_USE_ALARM
285: }
286: #endif
287: }
288:
289: #ifndef DOXYGEN
290: struct Http_pass_header_info {
291: Request_charsets* charsets;
292: String* request;
1.35 misha 293: bool* user_agent_specified;
294: bool* content_type_specified;
295: bool* content_type_url_encoded;
1.1 paf 296: };
297: #endif
1.35 misha 298: static void http_pass_header(HashStringValue::key_type aname,
299: HashStringValue::value_type avalue,
1.22 misha 300: Http_pass_header_info *info) {
1.9 misha 301:
1.41 misha 302: const char* name_cstr=aname.cstr();
303:
1.38 misha 304: if(strcasecmp(name_cstr, HTTP_CONTENT_LENGTH)==0)
305: return;
306:
1.43 misha 307: String name=String(capitalize(name_cstr), String::L_URI);
1.42 misha 308: String value=attributed_meaning_to_string(*avalue, String::L_URI, true);
1.9 misha 309:
1.35 misha 310: *info->request << name << ": " << value << CRLF;
1.1 paf 311:
1.38 misha 312: if(strcasecmp(name_cstr, HTTP_USER_AGENT)==0)
1.35 misha 313: *info->user_agent_specified=true;
1.38 misha 314: if(strcasecmp(name_cstr, HTTP_CONTENT_TYPE)==0){
1.35 misha 315: *info->content_type_specified=true;
316: *info->content_type_url_encoded=StrStartFromNC(value.cstr(), HTTP_CONTENT_TYPE_FORM_URLENCODED);
317: }
1.1 paf 318: }
319:
1.10 misha 320: static void http_pass_cookie(HashStringValue::key_type name,
1.20 misha 321: HashStringValue::value_type value,
322: Http_pass_header_info *info) {
1.10 misha 323:
1.17 misha 324: *info->request << String(name, String::L_HTTP_COOKIE) << "="
1.31 misha 325: << attributed_meaning_to_string(*value, String::L_HTTP_COOKIE, true)
1.10 misha 326: << "; ";
327:
328: }
1.1 paf 329:
330: static const String* basic_authorization_field(const char* user, const char* pass) {
331: if(!user&& !pass)
332: return 0;
333:
334: String combined;
335: if(user)
336: combined<<user;
337: combined<<":";
338: if(pass)
339: combined<<pass;
340:
1.20 misha 341: String* result=new String("Basic ");
342: *result<<pa_base64_encode(combined.cstr(), combined.length());
1.1 paf 343: return result;
344: }
345:
346: static void form_string_value2string(
1.20 misha 347: HashStringValue::key_type key,
348: const String& value,
349: String& result)
1.1 paf 350: {
1.30 misha 351: result << String(key, String::L_URI) << "=" << String(value, String::L_URI) << "&";
1.1 paf 352: }
1.20 misha 353:
1.1 paf 354: #ifndef DOXYGEN
355: struct Form_table_value2string_info {
356: HashStringValue::key_type key;
357: String& result;
358:
359: Form_table_value2string_info(HashStringValue::key_type akey, String& aresult):
360: key(akey), result(aresult) {}
361: };
362: #endif
363: static void form_table_value2string(Table::element_type row, Form_table_value2string_info* info) {
364: form_string_value2string(info->key, *row->get(0), info->result);
365: }
366: static void form_value2string(
1.20 misha 367: HashStringValue::key_type key,
368: HashStringValue::value_type value,
369: String* result)
1.1 paf 370: {
371: if(const String* svalue=value->get_string())
372: form_string_value2string(key, *svalue, *result);
373: else if(Table* tvalue=value->get_table()) {
374: Form_table_value2string_info info(key, *result);
375: tvalue->for_each(form_table_value2string, &info);
376: } else
1.18 misha 377: throw Exception(PARSER_RUNTIME,
1.1 paf 378: new String(key, String::L_TAINTED),
1.22 misha 379: "is %s, "HTTP_FORM_NAME" option value can be string or table only (file is allowed for $."HTTP_METHOD_NAME"[POST] + $."HTTP_FORM_ENCTYPE_NAME"["HTTP_CONTENT_TYPE_MULTIPART_FORMDATA"])", value->type());
1.1 paf 380: }
1.20 misha 381:
1.5 misha 382: const char* pa_form2string(HashStringValue& form, Request_charsets& charsets) {
1.1 paf 383: String string;
1.3 paf 384: form.for_each<String*>(form_value2string, &string);
1.44 misha 385: return string.untaint_and_transcode_cstr(String::L_URI, &charsets);
1.1 paf 386: }
1.22 misha 387:
388: struct FormPart {
389: Request* r;
390: const char* boundary;
1.48 moko 391: String* string;
1.22 misha 392: Form_table_value2string_info* info;
1.48 moko 393:
394: struct BinaryBlock{
395: const char* ptr;
396: size_t length;
397:
398: BinaryBlock(String* astring, Request* r): ptr(astring->untaint_and_transcode_cstr(String::L_AS_IS, &r->charsets)), length(strlen(ptr)){}
399: BinaryBlock(const char* aptr, size_t alength): ptr(aptr), length(alength){}
400: };
401:
402: Array<BinaryBlock> blocks;
403:
404: FormPart(Request* ar, const char* aboundary): r(ar), boundary(aboundary), string(new String()){}
405:
406: const char *post(size_t &length){
407: if(blocks.count()){
408: blocks+=BinaryBlock(string, r);
409:
410: length=0;
411: for(size_t i=0; i<blocks.count(); i++)
412: length+=blocks[i].length;
413:
414: char *result=(char *)pa_malloc_atomic(length);
415: char *ptr=result;
416:
417: for(size_t i=0; i<blocks.count(); i++){
418: memcpy(ptr, blocks[i].ptr, blocks[i].length);
419: ptr+=blocks[i].length;
420: }
421:
422: return result;
423: } else {
424: BinaryBlock result(string, r);
425: length=result.length;
426: return result.ptr;
427: }
428: }
1.22 misha 429: };
430:
1.28 misha 431: static void form_part_boundary_header(FormPart& part, String::Body name, const char* file_name=0){
1.48 moko 432: *part.string << "--" << part.boundary
1.41 misha 433: << CRLF CONTENT_DISPOSITION_CAPITALIZED ": form-data; name=\""
1.48 moko 434: << name
1.28 misha 435: << "\"";
1.22 misha 436: if(file_name){
437: if(strcmp(file_name, NONAME_DAT)!=0)
1.48 moko 438: *part.string << "; filename=\"" << file_name << "\"";
439: *part.string << CRLF HTTP_CONTENT_TYPE_CAPITALIZED ": " << part.r->mime_type_of(file_name);
1.22 misha 440: }
1.48 moko 441: *part.string << CRLF CRLF;
1.22 misha 442: }
443:
444: static void form_string_value2part(
1.28 misha 445: HashStringValue::key_type key,
446: const String& value,
447: FormPart& part)
1.22 misha 448: {
1.28 misha 449: form_part_boundary_header(part, key);
1.48 moko 450: *part.string << value << CRLF;
1.22 misha 451: }
452:
453: static void form_file_value2part(
1.28 misha 454: HashStringValue::key_type key,
455: VFile& vfile,
456: FormPart& part)
1.22 misha 457: {
1.28 misha 458: form_part_boundary_header(part, key, vfile.fields().get(name_name)->as_string().cstr());
1.48 moko 459: part.blocks+=FormPart::BinaryBlock(part.string, part.r);
460: part.blocks+=FormPart::BinaryBlock(vfile.value_ptr(), vfile.value_size());
461: part.string=new String();
462: *part.string << CRLF;
1.22 misha 463: }
464:
465: static void form_table_value2part(Table::element_type row, FormPart* part) {
466: form_string_value2part(part->info->key, *row->get(0), *part);
467: }
468:
469: static void form_value2part(
1.28 misha 470: HashStringValue::key_type key,
471: HashStringValue::value_type value,
472: FormPart& part)
1.22 misha 473: {
474: if(const String* svalue=value->get_string())
475: form_string_value2part(key, *svalue, part);
476: else if(Table* tvalue=value->get_table()) {
1.48 moko 477: Form_table_value2string_info info(key, *part.string);
1.22 misha 478: part.info = &info;
479: tvalue->for_each(form_table_value2part, &part);
1.33 misha 480: } else if(VFile* vfile=static_cast<VFile *>(value->as("file"))){
1.22 misha 481: form_file_value2part(key, *vfile, part);
482: } else
483: throw Exception(PARSER_RUNTIME,
484: new String(key, String::L_TAINTED),
485: "is %s, "HTTP_FORM_NAME" option value can be string, table or file only", value->type());
486: }
487:
488: const char* pa_form2string_multipart(HashStringValue& form, Request& r, const char* boundary, size_t& post_size){
1.48 moko 489: FormPart formpart(&r, boundary);
1.22 misha 490: form.for_each<FormPart&>(form_value2part, formpart);
1.48 moko 491: *formpart.string << "--" << boundary << "--";
492: // @todo: return binary blocks here to save memory in pa_internal_file_read_http
493: return formpart.post(post_size);
1.22 misha 494: }
495:
1.1 paf 496: static void find_headers_end(char* p,
497: char*& headers_end_at,
498: char*& raw_body)
499: {
500: raw_body=p;
501: // \n\n
502: // \r\n\r\n
503: while((p=strchr(p, '\n'))) {
504: headers_end_at=++p; // \n>.<
505: if(*p=='\r') // \r\n>\r?<\n
506: p++;
507: if(*p=='\n') { // \r\n\r>\n?<
508: raw_body=p+1;
509: return;
510: }
511: }
512: headers_end_at=0;
513: }
514:
515: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.22 misha 516: File_read_http_result pa_internal_file_read_http(Request& r,
517: const String& file_spec,
1.20 misha 518: bool as_text,
1.15 misha 519: HashStringValue *options,
520: bool transcode_text_result) {
1.1 paf 521: File_read_http_result result;
1.20 misha 522: char host[MAX_STRING];
1.1 paf 523: const char* uri;
1.49 ! moko 524: short port=80;
1.10 misha 525: const char* method="GET";
1.21 misha 526: bool method_is_get=true;
1.1 paf 527: HashStringValue* form=0;
528: int timeout_secs=2;
529: bool fail_on_status_ne_200=true;
1.12 misha 530: bool omit_post_charset=false;
1.1 paf 531: Value* vheaders=0;
1.10 misha 532: Value* vcookies=0;
1.11 misha 533: Value* vbody=0;
1.1 paf 534: Charset *asked_remote_charset=0;
535: const char* user_cstr=0;
536: const char* password_cstr=0;
1.22 misha 537: const char* encode=0;
538: bool multipart=false;
1.1 paf 539:
540: if(options) {
541: int valid_options=pa_get_valid_file_options_count(*options);
542:
543: if(Value* vmethod=options->get(HTTP_METHOD_NAME)) {
544: valid_options++;
1.21 misha 545: method=vmethod->as_string().change_case(r.charsets.source(), String::CC_UPPER).cstr();
546: method_is_get=strcmp(method, "GET")==0;
1.1 paf 547: }
1.22 misha 548: if(Value* vencode=options->get(HTTP_FORM_ENCTYPE_NAME)) {
549: valid_options++;
550: encode=vencode->as_string().cstr();
551: }
1.1 paf 552: if(Value* vform=options->get(HTTP_FORM_NAME)) {
553: valid_options++;
554: form=vform->get_hash();
555: }
1.11 misha 556: if(vbody=options->get(HTTP_BODY_NAME)) {
1.1 paf 557: valid_options++;
558: }
559: if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) {
560: valid_options++;
561: timeout_secs=vtimeout->as_int();
562: }
1.11 misha 563: if(vheaders=options->get(HTTP_HEADERS_NAME)) {
1.1 paf 564: valid_options++;
565: }
1.11 misha 566: if(vcookies=options->get(HTTP_COOKIES_NAME)) {
1.10 misha 567: valid_options++;
568: }
1.1 paf 569: if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) {
570: valid_options++;
571: fail_on_status_ne_200=!vany_status->as_bool();
1.12 misha 572: }
1.20 misha 573: if(Value* vomit_post_charset=options->get(HTTP_OMIT_POST_CHARSET_NAME)){
1.12 misha 574: valid_options++;
575: omit_post_charset=vomit_post_charset->as_bool();
576: }
1.6 misha 577: if(Value* vcharset_name=options->get(PA_CHARSET_NAME)) {
1.23 misha 578: asked_remote_charset=&charsets.get(vcharset_name->as_string().
1.22 misha 579: change_case(r.charsets.source(), String::CC_UPPER));
1.1 paf 580: }
581: if(Value* vuser=options->get(HTTP_USER)) {
582: valid_options++;
583: user_cstr=vuser->as_string().cstr();
584: }
585: if(Value* vpassword=options->get(HTTP_PASSWORD)) {
586: valid_options++;
587: password_cstr=vpassword->as_string().cstr();
588: }
589:
590: if(valid_options!=options->count())
1.46 misha 591: throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION);
1.1 paf 592: }
593: if(!asked_remote_charset) // defaulting to $request:charset
1.22 misha 594: asked_remote_charset=&(r.charsets).source();
595:
596: if(encode){
597: if(method_is_get)
598: throw Exception(PARSER_RUNTIME,
599: 0,
600: "you can not use $."HTTP_FORM_ENCTYPE_NAME" option with method GET");
601:
602: multipart=strcasecmp(encode, HTTP_CONTENT_TYPE_MULTIPART_FORMDATA)==0;
603:
604: if(!multipart && strcasecmp(encode, HTTP_CONTENT_TYPE_FORM_URLENCODED)!=0)
605: throw Exception(PARSER_RUNTIME,
606: 0,
607: "$."HTTP_FORM_ENCTYPE_NAME" option value can be "HTTP_CONTENT_TYPE_FORM_URLENCODED" or "HTTP_CONTENT_TYPE_MULTIPART_FORMDATA" only");
608: }
1.1 paf 609:
1.11 misha 610: if(vbody){
611: if(method_is_get)
612: throw Exception(PARSER_RUNTIME,
613: 0,
614: "you can not use $."HTTP_BODY_NAME" option with method GET");
615:
616: if(form)
617: throw Exception(PARSER_RUNTIME,
618: 0,
619: "you can not use options $."HTTP_BODY_NAME" and $."HTTP_FORM_NAME" together");
620: }
1.1 paf 621:
622: //preparing request
1.29 misha 623: String& connect_string=*new String(file_spec);
1.1 paf 624:
1.48 moko 625: const char* request;
626: size_t request_size;
1.1 paf 627: {
628: // influence URLencoding of tainted pieces to String::L_URI lang
1.22 misha 629: Temp_client_charset temp(r.charsets, *asked_remote_charset);
1.1 paf 630:
1.44 misha 631: const char* connect_string_cstr=connect_string.untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.1 paf 632:
633: const char* current=connect_string_cstr;
634: if(strncmp(current, "http://", 7)!=0)
1.18 misha 635: throw Exception(PARSER_RUNTIME,
1.1 paf 636: &connect_string,
637: "does not start with http://"); //never
638: current+=7;
639:
640: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
1.34 misha 641: char* host_uri=lsplit(host, '/');
642: uri=host_uri?current+(host_uri-1-host):"/";
643: char* port_cstr=lsplit(host, ':');
1.49 ! moko 644:
! 645: if (port_cstr){
! 646: char* error_pos=0;
! 647: port=(short)strtol(port_cstr, &error_pos, 10);
! 648: if(port==0 || *error_pos)
! 649: throw Exception(PARSER_RUNTIME, &connect_string, "invalid port number '%s'", port_cstr);
! 650: }
1.1 paf 651:
1.11 misha 652: // making request head
1.1 paf 653: String head;
1.11 misha 654: head << method << " " << uri;
1.28 misha 655: if(method_is_get && form)
656: head << (strchr(uri, '?')!=0?"&":"?") << pa_form2string(*form, r.charsets);
1.11 misha 657:
1.49 ! moko 658: head <<" HTTP/1.0" CRLF "Host: "<< host;
! 659: if (port != 80)
! 660: head << ":" << port_cstr;
! 661: head << CRLF;
1.11 misha 662:
1.28 misha 663: char* boundary=0;
1.22 misha 664:
665: if(multipart){
666: uuid uuid=get_uuid();
667: const int boundary_bufsize=10+32+1/*for zero-teminator*/+1/*for faulty snprintfs*/;
668: boundary=new(PointerFreeGC) char[boundary_bufsize];
669: snprintf(boundary, boundary_bufsize,
670: "----------%08X%04X%04X%02X%02X%02X%02X%02X%02X%02X%02X",
671: uuid.time_low, uuid.time_mid, uuid.time_hi_and_version,
672: uuid.clock_seq >> 8, uuid.clock_seq & 0xFF,
673: uuid.node[0], uuid.node[1], uuid.node[2],
674: uuid.node[3], uuid.node[4], uuid.node[5]);
675: }
676:
1.35 misha 677: String user_headers;
678: bool user_agent_specified=false;
679: bool content_type_specified=false;
680: bool content_type_url_encoded=false;
681: if(vheaders && !vheaders->is_string()) { // allow empty
682: if(HashStringValue *headers=vheaders->get_hash()) {
683: Http_pass_header_info info={
684: &(r.charsets),
685: &user_headers,
686: &user_agent_specified,
687: &content_type_specified,
688: &content_type_url_encoded};
689: headers->for_each<Http_pass_header_info*>(http_pass_header, &info);
690: } else
691: throw Exception(PARSER_RUNTIME,
692: 0,
693: "headers param must be hash");
694: };
695:
1.48 moko 696: const char* request_body=0;
1.22 misha 697: size_t post_size=0;
698: if(form && !method_is_get) {
1.38 misha 699: head << "Content-Type: " << (multipart ? HTTP_CONTENT_TYPE_MULTIPART_FORMDATA : HTTP_CONTENT_TYPE_FORM_URLENCODED);
1.28 misha 700:
701: if(!omit_post_charset)
702: head << "; charset=" << asked_remote_charset->NAME_CSTR();
703:
1.22 misha 704: if(multipart) {
1.28 misha 705: head << "; boundary=" << boundary;
1.48 moko 706: request_body=pa_form2string_multipart(*form, r/*charsets & mime_type needed*/, boundary, post_size/*correct post_size returned here*/);
1.22 misha 707: } else {
1.48 moko 708: request_body=pa_form2string(*form, r.charsets);
709: post_size=strlen(request_body);
1.22 misha 710: }
1.28 misha 711: head << CRLF;
1.35 misha 712: } else if(vbody) {
1.38 misha 713: // $.body was specified
1.35 misha 714: if(content_type_url_encoded){
1.36 misha 715: // transcode + url-encode
1.48 moko 716: request_body=vbody->as_string().untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.35 misha 717: } else {
1.36 misha 718: // content-type != application/x-www-form-urlencoded -> transcode only, don't url-encode!
1.48 moko 719: request_body=Charset::transcode(
1.35 misha 720: String::C(vbody->as_string().cstr(), vbody->as_string().length()),
721: r.charsets.source(),
722: *asked_remote_charset
723: );
724: }
1.48 moko 725: post_size=strlen(request_body);
1.1 paf 726: }
727:
728: // http://www.ietf.org/rfc/rfc2617.txt
729: if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr))
1.38 misha 730: head << "Authorization: " << *authorization_field_value << CRLF;
1.1 paf 731:
1.35 misha 732: head << user_headers;
733:
1.1 paf 734: if(!user_agent_specified) // defaulting
1.38 misha 735: head << "User-Agent: " DEFAULT_USER_AGENT CRLF;
1.1 paf 736:
1.12 misha 737: if(form && !method_is_get && content_type_specified) // POST + form + content-type was specified
738: throw Exception(PARSER_RUNTIME,
1.35 misha 739: 0,
1.12 misha 740: "$.content-type can't be specified with method POST");
741:
1.11 misha 742: if(vcookies && !vcookies->is_string()){ // allow empty
1.10 misha 743: if(HashStringValue* cookies=vcookies->get_hash()) {
1.37 misha 744: head << "Cookie: ";
1.35 misha 745: Http_pass_header_info info={&(r.charsets), &head, 0, 0, 0};
1.10 misha 746: cookies->for_each<Http_pass_header_info*>(http_pass_cookie, &info);
747: head << CRLF;
748: } else
749: throw Exception(PARSER_RUNTIME,
1.35 misha 750: 0,
1.44 misha 751: "cookies param must be hash");
1.10 misha 752: }
753:
1.48 moko 754: if(request_body)
1.38 misha 755: head << "Content-Length: " << format(post_size, "%u") << CRLF;
1.48 moko 756:
757: head << CRLF;
758:
759: const char *request_head=head.untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.1 paf 760:
1.48 moko 761: if(request_body){
762: size_t head_size = strlen(request_head);
763: request_size=post_size + head_size;
764: char *ptr=(char *)pa_malloc_atomic(request_size);
765: memcpy(ptr, request_head, head_size);
766: memcpy(ptr+head_size, request_body, post_size);
767: request=ptr;
768: } else {
769: request_size=strlen(request_head);
770: request=request_head;
771: }
1.1 paf 772: }
773:
774: char* response;
775: size_t response_size;
1.22 misha 776:
1.28 misha 777: // sending request
1.1 paf 778: int status_code=http_request(response, response_size,
1.48 moko 779: host, port, request, request_size,
1.1 paf 780: timeout_secs, fail_on_status_ne_200);
781:
1.28 misha 782: // processing results
1.1 paf 783: char* raw_body; size_t raw_body_size;
784: char* headers_end_at;
785: find_headers_end(response,
786: headers_end_at,
787: raw_body);
788: raw_body_size=response_size-(raw_body-response);
789:
790: result.headers=new HashStringValue;
791: VHash* vtables=new VHash;
792: result.headers->put(HTTP_TABLES_NAME, vtables);
793: Charset* real_remote_charset=0; // undetected, yet
794:
795: if(headers_end_at) {
796: *headers_end_at=0;
1.25 misha 797: const String header_block(String::C(response, headers_end_at-response), String::L_TAINTED);
1.1 paf 798:
799: ArrayString aheaders;
800: HashStringValue& tables=vtables->hash();
801:
802: size_t pos_after=0;
803: header_block.split(aheaders, pos_after, "\n");
804:
1.28 misha 805: // processing headers
1.1 paf 806: size_t aheaders_count=aheaders.count();
807: for(size_t i=1; i<aheaders_count; i++) {
808: const String& line=*aheaders.get(i);
809: size_t pos=line.pos(':');
810: if(pos==STRING_NOT_FOUND || pos<1)
811: throw Exception("http.response",
812: &connect_string,
813: "bad response from host - bad header \"%s\"", line.cstr());
1.22 misha 814: const String::Body HEADER_NAME=line.mid(0, pos).change_case(r.charsets.source(), String::CC_UPPER);
1.14 misha 815: const String& HEADER_VALUE=line.mid(pos+1, line.length()).trim(String::TRIM_BOTH, " \t\r");
1.20 misha 816: if(as_text && HEADER_NAME==HTTP_CONTENT_TYPE_UPPER)
1.32 misha 817: real_remote_charset=detect_charset(HEADER_VALUE.cstr());
1.1 paf 818:
819: // tables
820: {
821: Value *valready=(Value *)tables.get(HEADER_NAME);
822: bool existed=valready!=0;
823: Table *table;
824: if(existed) {
825: // second+ appearence
826: table=valready->get_table();
827: } else {
828: // first appearence
1.14 misha 829: Table::columns_type columns=new ArrayString(1);
1.1 paf 830: *columns+=new String("value");
831: table=new Table(columns);
832: }
833: // this string becomes next row
834: ArrayString& row=*new ArrayString(1);
1.14 misha 835: row+=&HEADER_VALUE;
1.1 paf 836: *table+=&row;
837: // not existed before? add it
838: if(!existed)
839: tables.put(HEADER_NAME, new VTable(table));
840: }
841:
1.14 misha 842: result.headers->put(HEADER_NAME, new VString(HEADER_VALUE));
1.1 paf 843: }
844: }
845:
1.16 misha 846: if(as_text && raw_body_size>=3 && strncmp(raw_body, "\xEF\xBB\xBF", 3)==0){
1.20 misha 847: // skip UTF-8 signature (BOM code)
1.16 misha 848: raw_body+=3;
849: raw_body_size-=3;
1.47 misha 850: if(!real_remote_charset)
851: real_remote_charset=&UTF8_charset;
1.16 misha 852: }
853:
1.1 paf 854: // output response
855: String::C real_body=String::C(raw_body, raw_body_size);
1.16 misha 856:
857: 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 858: // defaulting to used-asked charset [it's never empty!]
859: if(!real_remote_charset)
860: real_remote_charset=asked_remote_charset;
1.16 misha 861:
1.22 misha 862: real_body=Charset::transcode(real_body, *real_remote_charset, r.charsets.source());
1.16 misha 863:
1.1 paf 864: }
865:
866: result.str=const_cast<char *>(real_body.str); // hacking a little
867: result.length=real_body.length;
1.16 misha 868:
1.22 misha 869: if(as_text && result.length)
870: fix_line_breaks(result.str, result.length);
871:
1.1 paf 872: result.headers->put(file_status_name, new VInt(status_code));
1.16 misha 873:
1.1 paf 874: return result;
875: }
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