Annotation of parser3/src/main/pa_http.C, revision 1.76
1.1 paf 1: /** @file
2: Parser: http support functions.
3:
1.69 moko 4: Copyright (c) 2001-2015 Art. Lebedev Studio (http://www.artlebedev.com)
1.1 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
6: */
7:
8: #include "pa_http.h"
9: #include "pa_common.h"
10: #include "pa_charsets.h"
11: #include "pa_request_charsets.h"
1.22 misha 12: #include "pa_request.h"
13: #include "pa_vfile.h"
14: #include "pa_random.h"
1.1 paf 15:
1.76 ! moko 16: volatile const char * IDENT_PA_HTTP_C="$Id: pa_http.C,v 1.75 2016/09/01 13:57:05 moko Exp $" IDENT_PA_HTTP_H;
1.59 moko 17:
18: #ifdef _MSC_VER
19: #include <windows.h>
20: #else
21: #define closesocket close
22: #endif
1.53 moko 23:
1.1 paf 24: // defines
25:
1.19 misha 26: #define HTTP_METHOD_NAME "method"
27: #define HTTP_FORM_NAME "form"
28: #define HTTP_BODY_NAME "body"
29: #define HTTP_TIMEOUT_NAME "timeout"
30: #define HTTP_HEADERS_NAME "headers"
1.22 misha 31: #define HTTP_FORM_ENCTYPE_NAME "enctype"
1.19 misha 32: #define HTTP_ANY_STATUS_NAME "any-status"
1.59 moko 33: #define HTTP_OMIT_POST_CHARSET_NAME "omit-post-charset" // ^file::load[...;http://...;$.method[post]] by default adds charset to content-type
1.12 misha 34:
1.1 paf 35: #define HTTP_USER "user"
36: #define HTTP_PASSWORD "password"
37:
1.70 moko 38: #define HTTP_USER_AGENT "user-agent"
1.1 paf 39: #define DEFAULT_USER_AGENT "parser3"
40:
1.59 moko 41: #ifndef INADDR_NONE
42: #define INADDR_NONE ((ulong) -1)
43: #endif
1.1 paf 44:
45: #undef CRLF
46: #define CRLF "\r\n"
47:
1.54 misha 48: // helpers
1.56 misha 49:
1.54 misha 50: class Cookies_table_template_columns: public ArrayString {
51: public:
52: Cookies_table_template_columns() {
53: *this+=new String("name");
54: *this+=new String("value");
55: *this+=new String("expires");
56: *this+=new String("max-age");
57: *this+=new String("domain");
58: *this+=new String("path");
59: *this+=new String("httponly");
60: *this+=new String("secure");
61: }
62: };
63:
64:
1.1 paf 65: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
1.22 misha 66: memset(addr, 0, sizeof(*addr));
67: addr->sin_family=AF_INET;
68: addr->sin_port=htons(port);
69: if(host) {
1.65 moko 70: struct hostent *hostIP=gethostbyname(host);
71: if(hostIP && hostIP->h_addrtype == AF_INET){
72: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
73: return true;
74: }
75: }
76: return false;
1.1 paf 77: }
78:
79: size_t guess_content_length(char* buf) {
80: char* ptr;
81: if((ptr=strstr(buf, "Content-Length:"))) // Apache
82: goto found;
1.37 misha 83: if((ptr=strstr(buf, "content-length:"))) // Parser 3 before 3.4.0
1.1 paf 84: goto found;
85: if((ptr=strstr(buf, "Content-length:"))) // maybe 1
86: goto found;
87: if((ptr=strstr(buf, "CONTENT-LENGTH:"))) // maybe 2
88: goto found;
89: return 0;
90: found:
91: char *error_pos;
1.37 misha 92: size_t result=(size_t)strtol(ptr+15/*strlen("Content-Length:")*/, &error_pos, 0);
1.1 paf 93:
94: const size_t reasonable_initial_max=0x400*0x400*10 /*10M*/;
95: if(result>reasonable_initial_max) // sanity check
96: return reasonable_initial_max;
97: return 0;//result;
98: }
99:
100: static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200) {
101: int result=0;
1.37 misha 102: // fetching some to local buffer, guessing on possible Content-Length
103: response_size=0x400*20; // initial size if Content-Length could not be determined
1.1 paf 104: const size_t preview_size=0x400*20;
105: char preview_buf[preview_size+1/*terminator*/]; // 20K buffer to preview headers
106: ssize_t received_size=recv(sock, preview_buf, preview_size, 0);
107: if(received_size==0)
108: goto done;
109: if(received_size<0) {
110: if(int no=pa_socks_errno())
1.73 moko 111: throw Exception("http.timeout", 0, "error receiving response header: %s (%d)", pa_socks_strerr(no), no);
1.1 paf 112: goto done;
113: }
1.2 paf 114: // terminator [helps futher string searches]
115: preview_buf[received_size]=0;
116: // checking status
117: if(char* EOLat=strstr(preview_buf, "\n")) {
118: const String status_line(pa_strdup(preview_buf, EOLat-preview_buf));
119: ArrayString astatus;
1.76 ! moko 120: status_line.split(astatus, 0, " ");
1.2 paf 121: const String& status_code=*astatus.get(astatus.count()>1?1:0);
122: result=status_code.as_int();
123:
124: if(fail_on_status_ne_200 && result!=200)
1.73 moko 125: throw Exception("http.status", &status_code, "invalid HTTP response status");
1.2 paf 126: }
1.1 paf 127: // detecting response_size
128: {
129: if(size_t content_length=guess_content_length(preview_buf))
130: response_size=preview_size+content_length; // a little more than needed, will adjust response_size by actual received size later
131: }
132:
133: // [gcc is happier this way, see goto above]
134: {
135: // allocating initial buf
136: response=(char*)pa_malloc_atomic(response_size+1/*terminator*/); // just setting memory block type
137: char* ptr=response;
138: size_t todo_size=response_size;
139: // coping part of already received body
140: memcpy(ptr, preview_buf, received_size);
141: ptr+=received_size;
142: todo_size-=received_size;
143:
144: // we use terminator byte for two purposes here:
145: // 1. we return there zero always, not knowing: maybe they would want to create String form $file.body?
146: // invariant: all Strings should have zero-terminated buffers
1.37 misha 147: // 2. we use that out-of-size byte to detect if our Content-Length guess was wrong
1.1 paf 148: // when recv gets more than we expected
1.37 misha 149: // a) we know that the Content-Length guess was wrong
1.1 paf 150: // b) we have space to put the first byte of extra data
151: // c) we use less code to detect normal situation: on last while-cycle recv expected to just return 0
152: while(true) {
153: received_size=recv(sock, ptr, todo_size+1/*there is always a place for terminator*/, 0);
154: if(received_size==0) {
155: response_size-=todo_size; // in case we received less than expected, cut down the reported size
156: break;
157: }
158: if(received_size<0) {
159: if(int no=pa_socks_errno())
1.73 moko 160: throw Exception("http.timeout", 0, "error receiving response body: %s (%d)", pa_socks_strerr(no), no);
1.1 paf 161: break;
162: }
163: // they've touched the terminator?
164: if((size_t)received_size>todo_size)
165: {
166: // that means that our guessed response_size was not big enough
167: const size_t grow_chunk_size=0x400*0x400; // 1M
168: response_size+=grow_chunk_size;
169: size_t ptr_offset=ptr-response;
170: response=(char*)pa_realloc(response, response_size+1/*terminator*/);
171: ptr=response+ptr_offset;
172: todo_size+=grow_chunk_size;
173: }
174: // can't do this before realloc: we need <todo_size check
175: ptr+=received_size;
176: todo_size-=received_size;
177: }
178: }
179: done:
180: if(result)
181: {
182: response[response_size]=0;
183: return result;
184: }
185: else
1.73 moko 186: throw Exception("http.response", 0, "bad response from host - no status found (size=%u)", response_size);
1.1 paf 187: }
188:
189: /* ********************** request *************************** */
190:
191: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
192: # define PA_USE_ALARM
193: #endif
194:
195: #ifdef PA_USE_ALARM
196: static sigjmp_buf timeout_env;
197: static void timeout_handler(int /*sig*/){
1.22 misha 198: siglongjmp(timeout_env, 1);
1.1 paf 199: }
200: #endif
201:
202: static int http_request(char*& response, size_t& response_size,
203: const char* host, short port,
1.22 misha 204: const char* request, size_t request_size,
1.1 paf 205: int timeout_secs,
206: bool fail_on_status_ne_200) {
207: if(!host)
1.73 moko 208: throw Exception("http.host", 0, "zero hostname"); //never
1.1 paf 209:
210: volatile // to prevent makeing it register variable, because it will be clobbered by longjmp [thanks gcc warning]
211: int sock=-1;
212: #ifdef PA_USE_ALARM
213: signal(SIGALRM, timeout_handler);
214: #endif
215: #ifdef PA_USE_ALARM
216: if(sigsetjmp(timeout_env, 1)) {
217: // stupid gcc [2.95.4] generated bad code
218: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
219: // rewritten simplier [athough duplicating closesocket code]
220: if(sock>=0)
221: closesocket(sock);
1.73 moko 222: throw Exception("http.timeout", 0, "timeout occured while retrieving document");
1.1 paf 223: return 0; // never
224: } else {
225: alarm(timeout_secs);
226: #endif
227: try {
228: int result;
229: struct sockaddr_in dest;
230:
231: if(!set_addr(&dest, host, port))
1.73 moko 232: throw Exception("http.host", 0, "can not resolve hostname \"%s\"", host);
1.1 paf 233:
234: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) {
235: int no=pa_socks_errno();
1.73 moko 236: throw Exception("http.connect", 0, "can not make socket: %s (%d)", pa_socks_strerr(no), no);
1.1 paf 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();
1.73 moko 255: throw Exception("http.connect", 0, "can not connect to host \"%s\": %s (%d)", host, pa_socks_strerr(no), no);
1.1 paf 256: }
1.22 misha 257:
1.1 paf 258: if(send(sock, request, request_size, 0)!=(ssize_t)request_size) {
259: int no=pa_socks_errno();
1.73 moko 260: throw Exception("http.timeout", 0, "error sending request: %s (%d)", pa_socks_strerr(no), no);
1.1 paf 261: }
262:
263: result=http_read_response(response, response_size, sock, fail_on_status_ne_200);
264: closesocket(sock);
265: #ifdef PA_USE_ALARM
266: alarm(0);
267: #endif
268: return result;
269: } catch(...) {
270: #ifdef PA_USE_ALARM
271: alarm(0);
272: #endif
273: if(sock>=0)
274: closesocket(sock);
275: rethrow;
276: }
277: #ifdef PA_USE_ALARM
278: }
279: #endif
280: }
281:
282: #ifndef DOXYGEN
283: struct Http_pass_header_info {
284: Request_charsets* charsets;
285: String* request;
1.35 misha 286: bool* user_agent_specified;
287: bool* content_type_specified;
288: bool* content_type_url_encoded;
1.1 paf 289: };
290: #endif
1.50 moko 291:
292: char *pa_http_safe_header_name(const char *name) {
293: char *result=pa_strdup(name);
294: char *n=result;
1.52 misha 295: if(!pa_isalpha((unsigned char)*n))
1.50 moko 296: *n++ = '_';
297: for(; *n; ++n) {
1.52 misha 298: if (!pa_isalnum((unsigned char)*n) && *n != '-' && *n != '_')
1.50 moko 299: *n = '_';
300: }
301: return result;
302: }
303:
1.35 misha 304: static void http_pass_header(HashStringValue::key_type aname,
305: HashStringValue::value_type avalue,
1.22 misha 306: Http_pass_header_info *info) {
1.9 misha 307:
1.41 misha 308: const char* name_cstr=aname.cstr();
309:
1.38 misha 310: if(strcasecmp(name_cstr, HTTP_CONTENT_LENGTH)==0)
311: return;
312:
1.50 moko 313: String name=String(pa_http_safe_header_name(capitalize(name_cstr)), String::L_AS_IS);
314: String value=attributed_meaning_to_string(*avalue, String::L_HTTP_HEADER, true);
1.9 misha 315:
1.35 misha 316: *info->request << name << ": " << value << CRLF;
1.1 paf 317:
1.38 misha 318: if(strcasecmp(name_cstr, HTTP_USER_AGENT)==0)
1.35 misha 319: *info->user_agent_specified=true;
1.38 misha 320: if(strcasecmp(name_cstr, HTTP_CONTENT_TYPE)==0){
1.35 misha 321: *info->content_type_specified=true;
1.62 moko 322: *info->content_type_url_encoded=pa_strncasecmp(value.cstr(), HTTP_CONTENT_TYPE_FORM_URLENCODED)==0;
1.35 misha 323: }
1.1 paf 324: }
325:
1.10 misha 326: static void http_pass_cookie(HashStringValue::key_type name,
1.20 misha 327: HashStringValue::value_type value,
328: Http_pass_header_info *info) {
1.10 misha 329:
1.17 misha 330: *info->request << String(name, String::L_HTTP_COOKIE) << "="
1.31 misha 331: << attributed_meaning_to_string(*value, String::L_HTTP_COOKIE, true)
1.10 misha 332: << "; ";
333:
334: }
1.1 paf 335:
336: static const String* basic_authorization_field(const char* user, const char* pass) {
337: if(!user&& !pass)
338: return 0;
339:
340: String combined;
341: if(user)
342: combined<<user;
343: combined<<":";
344: if(pass)
345: combined<<pass;
346:
1.20 misha 347: String* result=new String("Basic ");
348: *result<<pa_base64_encode(combined.cstr(), combined.length());
1.1 paf 349: return result;
350: }
351:
1.73 moko 352: static void form_string_value2string(HashStringValue::key_type key, const String& value, String& result) {
1.30 misha 353: result << String(key, String::L_URI) << "=" << String(value, String::L_URI) << "&";
1.1 paf 354: }
1.20 misha 355:
1.1 paf 356: #ifndef DOXYGEN
357: struct Form_table_value2string_info {
358: HashStringValue::key_type key;
359: String& result;
360:
361: Form_table_value2string_info(HashStringValue::key_type akey, String& aresult):
362: key(akey), result(aresult) {}
363: };
364: #endif
365: static void form_table_value2string(Table::element_type row, Form_table_value2string_info* info) {
366: form_string_value2string(info->key, *row->get(0), info->result);
367: }
1.73 moko 368:
369: static void form_value2string(HashStringValue::key_type key, HashStringValue::value_type value, String* result) {
1.1 paf 370: if(const String* svalue=value->get_string())
371: form_string_value2string(key, *svalue, *result);
372: else if(Table* tvalue=value->get_table()) {
373: Form_table_value2string_info info(key, *result);
374: tvalue->for_each(form_table_value2string, &info);
375: } else
1.73 moko 376: throw Exception(PARSER_RUNTIME, new String(key, String::L_TAINTED),
1.63 moko 377: "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 378: }
1.20 misha 379:
1.5 misha 380: const char* pa_form2string(HashStringValue& form, Request_charsets& charsets) {
1.1 paf 381: String string;
1.3 paf 382: form.for_each<String*>(form_value2string, &string);
1.44 misha 383: return string.untaint_and_transcode_cstr(String::L_URI, &charsets);
1.1 paf 384: }
1.22 misha 385:
386: struct FormPart {
387: Request* r;
388: const char* boundary;
1.48 moko 389: String* string;
1.22 misha 390: Form_table_value2string_info* info;
1.48 moko 391:
392: struct BinaryBlock{
393: const char* ptr;
394: size_t length;
395:
396: BinaryBlock(String* astring, Request* r): ptr(astring->untaint_and_transcode_cstr(String::L_AS_IS, &r->charsets)), length(strlen(ptr)){}
397: BinaryBlock(const char* aptr, size_t alength): ptr(aptr), length(alength){}
398: };
399:
400: Array<BinaryBlock> blocks;
401:
402: FormPart(Request* ar, const char* aboundary): r(ar), boundary(aboundary), string(new String()){}
403:
404: const char *post(size_t &length){
405: if(blocks.count()){
406: blocks+=BinaryBlock(string, r);
407:
408: length=0;
409: for(size_t i=0; i<blocks.count(); i++)
410: length+=blocks[i].length;
411:
412: char *result=(char *)pa_malloc_atomic(length);
413: char *ptr=result;
414:
415: for(size_t i=0; i<blocks.count(); i++){
416: memcpy(ptr, blocks[i].ptr, blocks[i].length);
417: ptr+=blocks[i].length;
418: }
419:
420: return result;
421: } else {
422: BinaryBlock result(string, r);
423: length=result.length;
424: return result.ptr;
425: }
426: }
1.22 misha 427: };
428:
1.73 moko 429: static void form_part_boundary_header(FormPart& part, String::Body name, const char* file_name=0) {
430: *part.string << "--" << part.boundary << CRLF CONTENT_DISPOSITION_CAPITALIZED ": form-data; name=\"" << name << "\"";
1.22 misha 431: if(file_name){
432: if(strcmp(file_name, NONAME_DAT)!=0)
1.48 moko 433: *part.string << "; filename=\"" << file_name << "\"";
434: *part.string << CRLF HTTP_CONTENT_TYPE_CAPITALIZED ": " << part.r->mime_type_of(file_name);
1.22 misha 435: }
1.48 moko 436: *part.string << CRLF CRLF;
1.22 misha 437: }
438:
1.73 moko 439: static void form_string_value2part(HashStringValue::key_type key, const String& value, FormPart& part) {
1.28 misha 440: form_part_boundary_header(part, key);
1.48 moko 441: *part.string << value << CRLF;
1.22 misha 442: }
443:
1.73 moko 444: static void form_file_value2part(HashStringValue::key_type key, VFile& vfile, FormPart& part) {
1.28 misha 445: form_part_boundary_header(part, key, vfile.fields().get(name_name)->as_string().cstr());
1.48 moko 446: part.blocks+=FormPart::BinaryBlock(part.string, part.r);
447: part.blocks+=FormPart::BinaryBlock(vfile.value_ptr(), vfile.value_size());
448: part.string=new String();
449: *part.string << CRLF;
1.22 misha 450: }
451:
452: static void form_table_value2part(Table::element_type row, FormPart* part) {
453: form_string_value2part(part->info->key, *row->get(0), *part);
454: }
455:
1.73 moko 456: static void form_value2part(HashStringValue::key_type key, HashStringValue::value_type value, FormPart& part) {
1.22 misha 457: if(const String* svalue=value->get_string())
458: form_string_value2part(key, *svalue, part);
459: else if(Table* tvalue=value->get_table()) {
1.48 moko 460: Form_table_value2string_info info(key, *part.string);
1.22 misha 461: part.info = &info;
462: tvalue->for_each(form_table_value2part, &part);
1.33 misha 463: } else if(VFile* vfile=static_cast<VFile *>(value->as("file"))){
1.22 misha 464: form_file_value2part(key, *vfile, part);
465: } else
1.73 moko 466: throw Exception(PARSER_RUNTIME, new String(key, String::L_TAINTED), "is %s, " HTTP_FORM_NAME " option value can be string, table or file only", value->type());
1.22 misha 467: }
468:
469: const char* pa_form2string_multipart(HashStringValue& form, Request& r, const char* boundary, size_t& post_size){
1.48 moko 470: FormPart formpart(&r, boundary);
1.22 misha 471: form.for_each<FormPart&>(form_value2part, formpart);
1.48 moko 472: *formpart.string << "--" << boundary << "--";
473: // @todo: return binary blocks here to save memory in pa_internal_file_read_http
474: return formpart.post(post_size);
1.22 misha 475: }
476:
1.73 moko 477: static void find_headers_end(char* p, char*& headers_end_at, char*& raw_body) {
1.1 paf 478: raw_body=p;
479: // \n\n
480: // \r\n\r\n
481: while((p=strchr(p, '\n'))) {
482: headers_end_at=++p; // \n>.<
483: if(*p=='\r') // \r\n>\r?<\n
484: p++;
485: if(*p=='\n') { // \r\n\r>\n?<
486: raw_body=p+1;
487: return;
488: }
489: }
490: headers_end_at=0;
491: }
492:
1.54 misha 493: // Set-Cookie: name=value; Domain=docs.foo.com; Path=/accounts; Expires=Wed, 13-Jan-2021 22:23:01 GMT; Secure; HttpOnly
494: static ArrayString* parse_cookie(Request& r, const String& cookie) {
1.64 moko 495: char *current=pa_strdup(cookie.cstr());
1.54 misha 496:
497: const String* name=0;
1.55 moko 498: const String* value=&String::Empty;
499: const String* expires=&String::Empty;
500: const String* max_age=&String::Empty;
501: const String* path=&String::Empty;
502: const String* domain=&String::Empty;
503: const String* httponly=&String::Empty;
504: const String* secure=&String::Empty;
1.54 misha 505:
506: bool first_pair=true;
507:
508: do {
509: if(char *meaning=search_stop(current, ';'))
510: if(char *attribute=search_stop(meaning, '=')) {
511: const String* sname=new String(unescape_chars(attribute, strlen(attribute), &r.charsets.source(), true/*don't convert '"' to space*/), String::L_TAINTED);
512: const String* smeaning=0;
513: if(meaning)
514: smeaning=new String(unescape_chars(meaning, strlen(meaning), &r.charsets.source(), true/*don't convert '"' to space*/), String::L_TAINTED);
515:
516: if(first_pair) {
517: // name + value
518: name=sname;
519: value=smeaning;
520: first_pair=false;
521: } else {
522: const String& slower=sname->change_case(r.charsets.source(), String::CC_LOWER);
523:
524: if(slower == "expires")
525: expires=smeaning;
526: else if(slower == "max-age")
527: max_age=smeaning;
528: else if(slower == "domain")
529: domain=smeaning;
530: else if(slower == "path")
531: path=smeaning;
532: else if(slower == "httponly")
533: httponly=new String("1", String::L_CLEAN);
534: else if(slower == "secure")
535: secure=new String("1", String::L_CLEAN);
536: else {
537: // todo@ ?
538: }
539: }
540: }
541: } while(current);
542:
543: if(!name)
544: return 0;
545:
546: ArrayString* result=new ArrayString(8);
547: *result+=name;
548: *result+=value;
549: *result+=expires;
550: *result+=max_age;
551: *result+=domain;
552: *result+=path;
553: *result+=httponly;
554: *result+=secure;
555:
556: return result;
557: }
558:
1.56 misha 559: Table* parse_cookies(Request& r, Table *cookies){
560: Table& result=*new Table(new Cookies_table_template_columns);
561:
562: for(Array_iterator<Table::element_type> i(*cookies); i.has_next(); )
563: if(ArrayString* row=parse_cookie(r, *i.next()->get(0)))
564: result+=row;
565:
566: return &result;
567: }
568:
1.75 moko 569: void tables_update(HashStringValue& tables, const String::Body name, const String& value){
1.72 moko 570: Table *table;
571: if(Value *valready=tables.get(name)) {
572: // second+ appearence
573: table=valready->get_table();
574: } else {
575: // first appearence
576: Table::columns_type columns=new ArrayString(1);
577: *columns+=new String("value");
578: table=new Table(columns);
579: tables.put(name, new VTable(table));
580: }
581: // this string becomes next row
582: ArrayString& row=*new ArrayString(1);
583: row+=&value;
584: *table+=&row;
585: }
586:
1.1 paf 587: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.72 moko 588: File_read_http_result pa_internal_file_read_http(Request& r, const String& file_spec, bool as_text, HashStringValue *options, bool transcode_text_result) {
1.1 paf 589: File_read_http_result result;
1.20 misha 590: char host[MAX_STRING];
1.66 moko 591: const char *idna_host;
1.1 paf 592: const char* uri;
1.49 moko 593: short port=80;
1.10 misha 594: const char* method="GET";
1.21 misha 595: bool method_is_get=true;
1.1 paf 596: HashStringValue* form=0;
597: int timeout_secs=2;
598: bool fail_on_status_ne_200=true;
1.12 misha 599: bool omit_post_charset=false;
1.1 paf 600: Value* vheaders=0;
1.10 misha 601: Value* vcookies=0;
1.11 misha 602: Value* vbody=0;
1.72 moko 603: Charset* asked_remote_charset=0;
1.58 moko 604: Charset* real_remote_charset=0;
1.1 paf 605: const char* user_cstr=0;
606: const char* password_cstr=0;
1.22 misha 607: const char* encode=0;
608: bool multipart=false;
1.1 paf 609:
610: if(options) {
611: int valid_options=pa_get_valid_file_options_count(*options);
612:
613: if(Value* vmethod=options->get(HTTP_METHOD_NAME)) {
614: valid_options++;
1.21 misha 615: method=vmethod->as_string().change_case(r.charsets.source(), String::CC_UPPER).cstr();
616: method_is_get=strcmp(method, "GET")==0;
1.1 paf 617: }
1.22 misha 618: if(Value* vencode=options->get(HTTP_FORM_ENCTYPE_NAME)) {
619: valid_options++;
620: encode=vencode->as_string().cstr();
621: }
1.1 paf 622: if(Value* vform=options->get(HTTP_FORM_NAME)) {
623: valid_options++;
624: form=vform->get_hash();
625: }
1.11 misha 626: if(vbody=options->get(HTTP_BODY_NAME)) {
1.1 paf 627: valid_options++;
628: }
629: if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) {
630: valid_options++;
631: timeout_secs=vtimeout->as_int();
632: }
1.11 misha 633: if(vheaders=options->get(HTTP_HEADERS_NAME)) {
1.1 paf 634: valid_options++;
635: }
1.11 misha 636: if(vcookies=options->get(HTTP_COOKIES_NAME)) {
1.10 misha 637: valid_options++;
638: }
1.1 paf 639: if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) {
640: valid_options++;
641: fail_on_status_ne_200=!vany_status->as_bool();
1.12 misha 642: }
1.20 misha 643: if(Value* vomit_post_charset=options->get(HTTP_OMIT_POST_CHARSET_NAME)){
1.12 misha 644: valid_options++;
645: omit_post_charset=vomit_post_charset->as_bool();
646: }
1.6 misha 647: if(Value* vcharset_name=options->get(PA_CHARSET_NAME)) {
1.74 moko 648: asked_remote_charset=&charsets.get(vcharset_name->as_string());
1.58 moko 649: }
650: if(Value* vresponse_charset_name=options->get(PA_RESPONSE_CHARSET_NAME)) {
1.61 moko 651: valid_options++;
1.74 moko 652: real_remote_charset=&charsets.get(vresponse_charset_name->as_string());
1.1 paf 653: }
654: if(Value* vuser=options->get(HTTP_USER)) {
655: valid_options++;
656: user_cstr=vuser->as_string().cstr();
657: }
658: if(Value* vpassword=options->get(HTTP_PASSWORD)) {
659: valid_options++;
660: password_cstr=vpassword->as_string().cstr();
661: }
662:
663: if(valid_options!=options->count())
1.46 misha 664: throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION);
1.1 paf 665: }
666: if(!asked_remote_charset) // defaulting to $request:charset
1.22 misha 667: asked_remote_charset=&(r.charsets).source();
668:
669: if(encode){
670: if(method_is_get)
1.72 moko 671: throw Exception(PARSER_RUNTIME, 0, "you can not use $." HTTP_FORM_ENCTYPE_NAME " option with method GET");
1.22 misha 672:
673: multipart=strcasecmp(encode, HTTP_CONTENT_TYPE_MULTIPART_FORMDATA)==0;
674:
675: if(!multipart && strcasecmp(encode, HTTP_CONTENT_TYPE_FORM_URLENCODED)!=0)
1.72 moko 676: throw Exception(PARSER_RUNTIME, 0, "$." HTTP_FORM_ENCTYPE_NAME " option value can be " HTTP_CONTENT_TYPE_FORM_URLENCODED " or " HTTP_CONTENT_TYPE_MULTIPART_FORMDATA " only");
1.22 misha 677: }
1.1 paf 678:
1.11 misha 679: if(vbody){
680: if(method_is_get)
1.72 moko 681: throw Exception(PARSER_RUNTIME, 0, "you can not use $." HTTP_BODY_NAME " option with method GET");
1.11 misha 682:
683: if(form)
1.72 moko 684: throw Exception(PARSER_RUNTIME, 0, "you can not use options $." HTTP_BODY_NAME " and $." HTTP_FORM_NAME " together");
1.11 misha 685: }
1.1 paf 686:
687: //preparing request
1.29 misha 688: String& connect_string=*new String(file_spec);
1.1 paf 689:
1.48 moko 690: const char* request;
691: size_t request_size;
1.1 paf 692: {
693: // influence URLencoding of tainted pieces to String::L_URI lang
1.22 misha 694: Temp_client_charset temp(r.charsets, *asked_remote_charset);
1.1 paf 695:
1.44 misha 696: const char* connect_string_cstr=connect_string.untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.1 paf 697:
698: const char* current=connect_string_cstr;
699: if(strncmp(current, "http://", 7)!=0)
1.72 moko 700: throw Exception(PARSER_RUNTIME, &connect_string, "does not start with http://"); //never
1.1 paf 701: current+=7;
702:
703: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
1.34 misha 704: char* host_uri=lsplit(host, '/');
705: uri=host_uri?current+(host_uri-1-host):"/";
706: char* port_cstr=lsplit(host, ':');
1.49 moko 707:
708: if (port_cstr){
709: char* error_pos=0;
710: port=(short)strtol(port_cstr, &error_pos, 10);
711: if(port==0 || *error_pos)
712: throw Exception(PARSER_RUNTIME, &connect_string, "invalid port number '%s'", port_cstr);
713: }
1.1 paf 714:
1.66 moko 715: idna_host=pa_idna_encode(host, r.charsets.source());
716:
1.11 misha 717: // making request head
1.1 paf 718: String head;
1.11 misha 719: head << method << " " << uri;
1.28 misha 720: if(method_is_get && form)
721: head << (strchr(uri, '?')!=0?"&":"?") << pa_form2string(*form, r.charsets);
1.11 misha 722:
1.66 moko 723: head <<" HTTP/1.0" CRLF "Host: "<< idna_host;
1.49 moko 724: if (port != 80)
725: head << ":" << port_cstr;
726: head << CRLF;
1.11 misha 727:
1.71 moko 728: char* boundary= multipart ? get_uuid_boundary() : 0;
1.22 misha 729:
1.35 misha 730: String user_headers;
731: bool user_agent_specified=false;
732: bool content_type_specified=false;
733: bool content_type_url_encoded=false;
734: if(vheaders && !vheaders->is_string()) { // allow empty
735: if(HashStringValue *headers=vheaders->get_hash()) {
736: Http_pass_header_info info={
737: &(r.charsets),
738: &user_headers,
739: &user_agent_specified,
740: &content_type_specified,
741: &content_type_url_encoded};
742: headers->for_each<Http_pass_header_info*>(http_pass_header, &info);
743: } else
1.72 moko 744: throw Exception(PARSER_RUNTIME, 0, "headers param must be hash");
1.35 misha 745: };
746:
1.48 moko 747: const char* request_body=0;
1.22 misha 748: size_t post_size=0;
749: if(form && !method_is_get) {
1.38 misha 750: head << "Content-Type: " << (multipart ? HTTP_CONTENT_TYPE_MULTIPART_FORMDATA : HTTP_CONTENT_TYPE_FORM_URLENCODED);
1.28 misha 751:
752: if(!omit_post_charset)
753: head << "; charset=" << asked_remote_charset->NAME_CSTR();
754:
1.22 misha 755: if(multipart) {
1.28 misha 756: head << "; boundary=" << boundary;
1.48 moko 757: request_body=pa_form2string_multipart(*form, r/*charsets & mime_type needed*/, boundary, post_size/*correct post_size returned here*/);
1.22 misha 758: } else {
1.48 moko 759: request_body=pa_form2string(*form, r.charsets);
760: post_size=strlen(request_body);
1.22 misha 761: }
1.28 misha 762: head << CRLF;
1.35 misha 763: } else if(vbody) {
1.38 misha 764: // $.body was specified
1.35 misha 765: if(content_type_url_encoded){
1.36 misha 766: // transcode + url-encode
1.48 moko 767: request_body=vbody->as_string().untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.35 misha 768: } else {
1.36 misha 769: // content-type != application/x-www-form-urlencoded -> transcode only, don't url-encode!
1.72 moko 770: const String &sbody=vbody->as_string();
771: request_body=Charset::transcode(String::C(sbody.cstr(), sbody.length()), r.charsets.source(), *asked_remote_charset).str;
1.35 misha 772: }
1.48 moko 773: post_size=strlen(request_body);
1.1 paf 774: }
775:
776: // http://www.ietf.org/rfc/rfc2617.txt
777: if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr))
1.38 misha 778: head << "Authorization: " << *authorization_field_value << CRLF;
1.1 paf 779:
1.35 misha 780: head << user_headers;
781:
1.1 paf 782: if(!user_agent_specified) // defaulting
1.38 misha 783: head << "User-Agent: " DEFAULT_USER_AGENT CRLF;
1.1 paf 784:
1.12 misha 785: if(form && !method_is_get && content_type_specified) // POST + form + content-type was specified
1.72 moko 786: throw Exception(PARSER_RUNTIME, 0, "$.content-type can't be specified with method POST");
1.12 misha 787:
1.11 misha 788: if(vcookies && !vcookies->is_string()){ // allow empty
1.10 misha 789: if(HashStringValue* cookies=vcookies->get_hash()) {
1.37 misha 790: head << "Cookie: ";
1.35 misha 791: Http_pass_header_info info={&(r.charsets), &head, 0, 0, 0};
1.10 misha 792: cookies->for_each<Http_pass_header_info*>(http_pass_cookie, &info);
793: head << CRLF;
794: } else
1.72 moko 795: throw Exception(PARSER_RUNTIME, 0, "cookies param must be hash");
1.10 misha 796: }
797:
1.48 moko 798: if(request_body)
1.38 misha 799: head << "Content-Length: " << format(post_size, "%u") << CRLF;
1.48 moko 800:
801: head << CRLF;
802:
803: const char *request_head=head.untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.1 paf 804:
1.48 moko 805: if(request_body){
806: size_t head_size = strlen(request_head);
807: request_size=post_size + head_size;
808: char *ptr=(char *)pa_malloc_atomic(request_size);
809: memcpy(ptr, request_head, head_size);
810: memcpy(ptr+head_size, request_body, post_size);
811: request=ptr;
812: } else {
813: request_size=strlen(request_head);
814: request=request_head;
815: }
1.1 paf 816: }
817:
1.72 moko 818: char* response_str;
1.1 paf 819: size_t response_size;
1.22 misha 820:
1.28 misha 821: // sending request
1.72 moko 822: int status_code=http_request(response_str, response_size, idna_host, port, request, request_size, timeout_secs, fail_on_status_ne_200);
1.1 paf 823:
1.72 moko 824: // processing results
1.1 paf 825: char* raw_body; size_t raw_body_size;
826: char* headers_end_at;
1.72 moko 827: find_headers_end(response_str, headers_end_at, raw_body);
828: raw_body_size=response_size-(raw_body-response_str);
1.1 paf 829:
830: result.headers=new HashStringValue;
831: VHash* vtables=new VHash;
1.72 moko 832: result.headers->put("tables", vtables);
833:
834: ResponseHeaders response;
1.1 paf 835:
836: if(headers_end_at) {
837: *headers_end_at=0;
1.72 moko 838: const String header_block(String::C(response_str, headers_end_at-response_str), String::L_TAINTED);
1.1 paf 839:
840: ArrayString aheaders;
1.76 ! moko 841: header_block.split(aheaders, 0, "\n");
1.1 paf 842:
1.72 moko 843: Array_iterator<const String*> i(aheaders);
844: i.next(); // skipping status
845: for(;i.has_next();){
846: const char *line=i.next()->cstr();
847: if(!response.add_header(line))
848: throw Exception("http.response", &connect_string, "bad response from host - bad header \"%s\"", line);
1.1 paf 849: }
1.72 moko 850: }
1.54 misha 851:
1.72 moko 852: if (!real_remote_charset && !response.content_type.is_empty())
853: real_remote_charset= detect_charset(response.content_type.cstr());
1.1 paf 854:
1.72 moko 855: if(as_text)
1.68 moko 856: real_remote_charset=charsets.checkBOM(raw_body, raw_body_size, real_remote_charset);
1.72 moko 857:
858: if (!real_remote_charset)
859: real_remote_charset=asked_remote_charset; // never null
860:
861: for(Array_iterator<ResponseHeaders::Header> i(response.headers); i.has_next(); ){
862: ResponseHeaders::Header header=i.next();
863:
864: header.transcode(*real_remote_charset, r.charsets.source());
865:
866: String &header_value=*new String(header.value, String::L_TAINTED);
867:
868: tables_update(vtables->hash(), header.name, header_value);
869: result.headers->put(header.name, new VString(header_value));
1.16 misha 870: }
871:
1.72 moko 872: // filling $.cookies
873: if(Value *vcookies=vtables->hash().get("SET-COOKIE"))
874: result.headers->put(HTTP_COOKIES_NAME, new VTable(parse_cookies(r, vcookies->get_table())));
875:
1.1 paf 876: // output response
877: String::C real_body=String::C(raw_body, raw_body_size);
1.16 misha 878:
879: 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
1.22 misha 880: real_body=Charset::transcode(real_body, *real_remote_charset, r.charsets.source());
1.1 paf 881: }
882:
883: result.str=const_cast<char *>(real_body.str); // hacking a little
884: result.length=real_body.length;
1.16 misha 885:
1.22 misha 886: if(as_text && result.length)
887: fix_line_breaks(result.str, result.length);
888:
1.1 paf 889: result.headers->put(file_status_name, new VInt(status_code));
1.16 misha 890:
1.1 paf 891: return result;
892: }
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