Annotation of parser3/src/main/pa_http.C, revision 1.90
1.1 paf 1: /** @file
2: Parser: http support functions.
3:
1.79 moko 4: Copyright (c) 2001-2017 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"
1.81 moko 10: #include "pa_base64.h"
1.1 paf 11: #include "pa_charsets.h"
12: #include "pa_request_charsets.h"
1.22 misha 13: #include "pa_request.h"
14: #include "pa_vfile.h"
15: #include "pa_random.h"
1.1 paf 16:
1.90 ! moko 17: volatile const char * IDENT_PA_HTTP_C="$Id: pa_http.C,v 1.89 2020/10/12 21:15:07 moko Exp $" IDENT_PA_HTTP_H;
1.59 moko 18:
19: #ifdef _MSC_VER
20: #include <windows.h>
1.89 moko 21: #define socklen_t int
1.59 moko 22: #else
23: #define closesocket close
24: #endif
1.53 moko 25:
1.1 paf 26: // defines
27:
1.19 misha 28: #define HTTP_METHOD_NAME "method"
29: #define HTTP_FORM_NAME "form"
30: #define HTTP_BODY_NAME "body"
31: #define HTTP_TIMEOUT_NAME "timeout"
32: #define HTTP_HEADERS_NAME "headers"
1.22 misha 33: #define HTTP_FORM_ENCTYPE_NAME "enctype"
1.19 misha 34: #define HTTP_ANY_STATUS_NAME "any-status"
1.59 moko 35: #define HTTP_OMIT_POST_CHARSET_NAME "omit-post-charset" // ^file::load[...;http://...;$.method[post]] by default adds charset to content-type
1.12 misha 36:
1.1 paf 37: #define HTTP_USER "user"
38: #define HTTP_PASSWORD "password"
39:
1.70 moko 40: #define HTTP_USER_AGENT "user-agent"
1.1 paf 41: #define DEFAULT_USER_AGENT "parser3"
42:
1.59 moko 43: #ifndef INADDR_NONE
44: #define INADDR_NONE ((ulong) -1)
45: #endif
1.1 paf 46:
47: #undef CRLF
48: #define CRLF "\r\n"
49:
1.54 misha 50: // helpers
1.56 misha 51:
1.85 moko 52: bool HTTP_Headers::add_header(const char *line){
1.78 moko 53: const char *value=strchr(line, ':');
54:
55: if(value && value != line){ // we need only headers, not the response code
56: Header header(str_upper(line, value-line), String::Body(value+1).trim(String::TRIM_BOTH, " \t\n\r"));
57:
58: if(header.name == String::Body(HTTP_CONTENT_TYPE_UPPER) && content_type.is_empty())
59: content_type=header.value;
60:
61: if(header.name == String::Body("CONTENT-LENGTH") && content_length==0)
62: content_length=pa_atoul(header.value.cstr(), 10);
63:
64: headers+=header;
65:
66: return true;
67: }
68: return false;
69: }
70:
1.54 misha 71: class Cookies_table_template_columns: public ArrayString {
72: public:
73: Cookies_table_template_columns() {
74: *this+=new String("name");
75: *this+=new String("value");
76: *this+=new String("expires");
77: *this+=new String("max-age");
78: *this+=new String("domain");
79: *this+=new String("path");
80: *this+=new String("httponly");
81: *this+=new String("secure");
82: }
83: };
84:
85:
1.1 paf 86: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
1.22 misha 87: memset(addr, 0, sizeof(*addr));
88: addr->sin_family=AF_INET;
89: addr->sin_port=htons(port);
90: if(host) {
1.65 moko 91: struct hostent *hostIP=gethostbyname(host);
92: if(hostIP && hostIP->h_addrtype == AF_INET){
93: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
94: return true;
95: }
96: }
97: return false;
1.1 paf 98: }
99:
1.84 moko 100: class HTTP_response : public PA_Allocated {
1.78 moko 101: public:
102: char *buf;
103: size_t length;
104: size_t buf_size;
105: size_t body_offset;
106:
1.85 moko 107: HTTP_Headers headers;
1.84 moko 108: const String &url;
1.78 moko 109:
1.84 moko 110: HTTP_response(const String& aurl) : buf(NULL), length(0), buf_size(0), body_offset(0), url(aurl){}
1.78 moko 111:
112: void resize(size_t size){
113: buf_size=size;
114: buf=(char *)pa_realloc(buf, size + 1);
115: }
116:
117: bool read(int sock, size_t size){
118: if(length+size>buf_size)
119: resize(buf_size*2 + size);
120: ssize_t received_size=recv(sock, buf + length, size, 0);
121: if(received_size==0)
122: return false;
123: if(received_size<0) {
124: if(int no=pa_socks_errno())
125: throw Exception("http.timeout", &url, "error receiving response body: %s (%d)", pa_socks_strerr(no), no);
126: return false;
127: }
128: length+=received_size;
129: buf[length]='\0';
130: return true;
131: }
132:
1.83 moko 133: size_t first_line(){
1.89 moko 134: char *header=strchr(buf, '\n');
135: if(!header)
1.78 moko 136: return false;
137:
1.89 moko 138: return header-buf;
1.78 moko 139: }
140:
141: const char *status_code(char *status_line, int &result){
142: char* status_start = strchr(status_line, ' ');
143:
144: if(!(status_start++))
145: return status_line;
146:
147: char* status_end=strchr(status_start, ' ');
148:
149: if(!status_end)
150: return status_line;
151:
152: if(status_end==status_start)
153: return status_line;
1.1 paf 154:
1.78 moko 155: const char *result_str=pa_strdup(status_start, status_end-status_start);
1.84 moko 156: result=pa_atoui(result_str, 10);
1.78 moko 157: return result_str;
158: }
1.2 paf 159:
1.78 moko 160: bool body_start(){
161: char *p=buf;
162: while((p=strchr(p, '\n'))) {
163: if(p[1]=='\r' && p[2]=='\n'){ // \r\n\r\n
164: *p='\0';
165: body_offset=p-buf+3;
166: return true;
167: }
168: if(p[1]=='\n') { // \n\n
169: *p='\0';
170: body_offset=p-buf+2;
171: return true;
172: }
173: p++;
174: }
175: return false;
1.2 paf 176: }
1.78 moko 177:
178: void parse_headers(){
179: const String header_block(buf, String::L_TAINTED);
180:
181: ArrayString aheaders;
182: header_block.split(aheaders, 0, "\n");
183:
184: Array_iterator<const String*> i(aheaders);
185: i.next(); // skipping status
186: for(;i.has_next();){
187: const char *line=i.next()->cstr();
188: if(!headers.add_header(line))
189: throw Exception("http.response", &url, "bad response from host - bad header \"%s\"", line);
190: }
1.1 paf 191: }
192:
1.88 moko 193: int read_response(int sock, bool fail_on_status_ne_200);
1.78 moko 194: };
195:
196: enum HTTP_response_state {
197: HTTP_STATUS_CODE,
198: HTTP_HEADERS,
199: HTTP_BODY
200: };
201:
1.88 moko 202: int HTTP_response::read_response(int sock, bool fail_on_status_ne_200) {
1.78 moko 203: HTTP_response_state state=HTTP_STATUS_CODE;
204: int result=0;
205:
206: size_t chunk_size=0x400*16;
1.88 moko 207: resize(2*chunk_size);
1.78 moko 208:
1.88 moko 209: while(read(sock, chunk_size)){
1.78 moko 210: switch(state){
211: case HTTP_STATUS_CODE: {
1.88 moko 212: size_t status_size=first_line();
1.78 moko 213: if(!status_size)
214: break;
215:
1.88 moko 216: const char *status=status_code(pa_strdup(buf, status_size), result);
1.78 moko 217:
218: if(!result || fail_on_status_ne_200 && result!=200)
219: throw Exception("http.status", status ? new String(status) : &String::Empty, "invalid HTTP response status");
220:
221: state=HTTP_HEADERS;
222: }
223:
224: case HTTP_HEADERS: {
1.88 moko 225: if(!body_start())
1.78 moko 226: break;
227:
1.88 moko 228: parse_headers();
1.78 moko 229:
1.88 moko 230: size_t content_length=check_file_size(headers.content_length, url);
231: if(content_length>0 && (content_length + body_offset) > length){
232: resize(content_length + body_offset + 0x400*64);
1.78 moko 233: }
234:
235: state=HTTP_BODY;
1.1 paf 236: break;
237: }
1.78 moko 238:
239: case HTTP_BODY: {
240: chunk_size=0x400*64;
1.1 paf 241: break;
242: }
243: }
244: }
1.78 moko 245:
246: if(state==HTTP_STATUS_CODE)
1.88 moko 247: throw Exception("http.response", &url, "bad response from host - no status found (size=%u)", length);
1.78 moko 248:
249: if(state==HTTP_HEADERS){
1.88 moko 250: parse_headers();
251: body_offset=length;
1.1 paf 252: }
1.78 moko 253:
254: return result;
1.1 paf 255: }
256:
257: /* ********************** request *************************** */
258:
259: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
260: # define PA_USE_ALARM
261: #endif
262:
263: #ifdef PA_USE_ALARM
264: static sigjmp_buf timeout_env;
265: static void timeout_handler(int /*sig*/){
1.22 misha 266: siglongjmp(timeout_env, 1);
1.1 paf 267: }
268: #endif
269:
1.78 moko 270: static int http_request(HTTP_response& response, const char* host, short port, const char* request, size_t request_size, int timeout_secs, bool fail_on_status_ne_200) {
1.1 paf 271: if(!host)
1.73 moko 272: throw Exception("http.host", 0, "zero hostname"); //never
1.1 paf 273:
274: volatile // to prevent makeing it register variable, because it will be clobbered by longjmp [thanks gcc warning]
275: int sock=-1;
276: #ifdef PA_USE_ALARM
277: signal(SIGALRM, timeout_handler);
278: #endif
279: #ifdef PA_USE_ALARM
280: if(sigsetjmp(timeout_env, 1)) {
281: // stupid gcc [2.95.4] generated bad code
282: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
283: // rewritten simplier [athough duplicating closesocket code]
284: if(sock>=0)
285: closesocket(sock);
1.80 moko 286: throw Exception("http.timeout", 0, "timeout occurred while retrieving document");
1.1 paf 287: return 0; // never
288: } else {
289: alarm(timeout_secs);
290: #endif
291: try {
292: int result;
293: struct sockaddr_in dest;
294:
295: if(!set_addr(&dest, host, port))
1.73 moko 296: throw Exception("http.host", 0, "can not resolve hostname \"%s\"", host);
1.1 paf 297:
298: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) {
299: int no=pa_socks_errno();
1.73 moko 300: throw Exception("http.connect", 0, "can not make socket: %s (%d)", pa_socks_strerr(no), no);
1.1 paf 301: }
302:
303: // To enable SO_DONTLINGER (that is, disable SO_LINGER)
304: // l_onoff should be set to zero and setsockopt should be called
305: linger dont_linger={0,0};
306: setsockopt(sock, SOL_SOCKET, SO_LINGER, (const char *)&dont_linger, sizeof(dont_linger));
307:
308: #ifdef WIN32
309: // SO_*TIMEO can be defined in .h but not implemlemented in protocol,
310: // failing subsequently with Option not supported by protocol (99) message
311: // could not suppress that, so leaving this only for win32
312: int timeout_ms=timeout_secs*1000;
313: setsockopt(sock, SOL_SOCKET, SO_SNDTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms));
314: setsockopt(sock, SOL_SOCKET, SO_RCVTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms));
315: #endif
316:
317: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) {
318: int no=pa_socks_errno();
1.78 moko 319: throw Exception("http.connect", 0, "can not connect to host \"%s\": %s (%d)", host, pa_socks_strerr(no), no);
1.1 paf 320: }
1.22 misha 321:
1.1 paf 322: if(send(sock, request, request_size, 0)!=(ssize_t)request_size) {
323: int no=pa_socks_errno();
1.78 moko 324: throw Exception("http.timeout", 0, "error sending request: %s (%d)", pa_socks_strerr(no), no);
1.1 paf 325: }
326:
1.88 moko 327: result=response.read_response(sock, fail_on_status_ne_200);
1.78 moko 328: closesocket(sock);
1.1 paf 329: #ifdef PA_USE_ALARM
1.78 moko 330: alarm(0);
1.1 paf 331: #endif
332: return result;
333: } catch(...) {
334: #ifdef PA_USE_ALARM
1.78 moko 335: alarm(0);
1.1 paf 336: #endif
1.78 moko 337: if(sock>=0)
338: closesocket(sock);
1.1 paf 339: rethrow;
340: }
341: #ifdef PA_USE_ALARM
342: }
343: #endif
344: }
345:
346: #ifndef DOXYGEN
347: struct Http_pass_header_info {
348: Request_charsets* charsets;
349: String* request;
1.35 misha 350: bool* user_agent_specified;
351: bool* content_type_specified;
352: bool* content_type_url_encoded;
1.1 paf 353: };
354: #endif
1.50 moko 355:
356: char *pa_http_safe_header_name(const char *name) {
357: char *result=pa_strdup(name);
358: char *n=result;
1.52 misha 359: if(!pa_isalpha((unsigned char)*n))
1.50 moko 360: *n++ = '_';
361: for(; *n; ++n) {
1.52 misha 362: if (!pa_isalnum((unsigned char)*n) && *n != '-' && *n != '_')
1.50 moko 363: *n = '_';
364: }
365: return result;
366: }
367:
1.35 misha 368: static void http_pass_header(HashStringValue::key_type aname,
369: HashStringValue::value_type avalue,
1.22 misha 370: Http_pass_header_info *info) {
1.9 misha 371:
1.41 misha 372: const char* name_cstr=aname.cstr();
373:
1.38 misha 374: if(strcasecmp(name_cstr, HTTP_CONTENT_LENGTH)==0)
375: return;
376:
1.50 moko 377: String name=String(pa_http_safe_header_name(capitalize(name_cstr)), String::L_AS_IS);
378: String value=attributed_meaning_to_string(*avalue, String::L_HTTP_HEADER, true);
1.9 misha 379:
1.35 misha 380: *info->request << name << ": " << value << CRLF;
1.1 paf 381:
1.38 misha 382: if(strcasecmp(name_cstr, HTTP_USER_AGENT)==0)
1.35 misha 383: *info->user_agent_specified=true;
1.38 misha 384: if(strcasecmp(name_cstr, HTTP_CONTENT_TYPE)==0){
1.35 misha 385: *info->content_type_specified=true;
1.62 moko 386: *info->content_type_url_encoded=pa_strncasecmp(value.cstr(), HTTP_CONTENT_TYPE_FORM_URLENCODED)==0;
1.35 misha 387: }
1.1 paf 388: }
389:
1.10 misha 390: static void http_pass_cookie(HashStringValue::key_type name,
1.20 misha 391: HashStringValue::value_type value,
392: Http_pass_header_info *info) {
1.10 misha 393:
1.17 misha 394: *info->request << String(name, String::L_HTTP_COOKIE) << "="
1.31 misha 395: << attributed_meaning_to_string(*value, String::L_HTTP_COOKIE, true)
1.10 misha 396: << "; ";
397:
398: }
1.1 paf 399:
400: static const String* basic_authorization_field(const char* user, const char* pass) {
401: if(!user&& !pass)
402: return 0;
403:
404: String combined;
405: if(user)
406: combined<<user;
407: combined<<":";
408: if(pass)
409: combined<<pass;
410:
1.20 misha 411: String* result=new String("Basic ");
1.82 moko 412: *result<<pa_base64_encode(combined.cstr(), combined.length(), Base64Options(false /*no wrap*/));
1.1 paf 413: return result;
414: }
415:
1.73 moko 416: static void form_string_value2string(HashStringValue::key_type key, const String& value, String& result) {
1.30 misha 417: result << String(key, String::L_URI) << "=" << String(value, String::L_URI) << "&";
1.1 paf 418: }
1.20 misha 419:
1.1 paf 420: #ifndef DOXYGEN
421: struct Form_table_value2string_info {
422: HashStringValue::key_type key;
423: String& result;
424:
425: Form_table_value2string_info(HashStringValue::key_type akey, String& aresult):
426: key(akey), result(aresult) {}
427: };
428: #endif
429: static void form_table_value2string(Table::element_type row, Form_table_value2string_info* info) {
430: form_string_value2string(info->key, *row->get(0), info->result);
431: }
1.73 moko 432:
433: static void form_value2string(HashStringValue::key_type key, HashStringValue::value_type value, String* result) {
1.1 paf 434: if(const String* svalue=value->get_string())
435: form_string_value2string(key, *svalue, *result);
436: else if(Table* tvalue=value->get_table()) {
437: Form_table_value2string_info info(key, *result);
438: tvalue->for_each(form_table_value2string, &info);
439: } else
1.73 moko 440: throw Exception(PARSER_RUNTIME, new String(key, String::L_TAINTED),
1.63 moko 441: "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 442: }
1.20 misha 443:
1.5 misha 444: const char* pa_form2string(HashStringValue& form, Request_charsets& charsets) {
1.1 paf 445: String string;
1.3 paf 446: form.for_each<String*>(form_value2string, &string);
1.44 misha 447: return string.untaint_and_transcode_cstr(String::L_URI, &charsets);
1.1 paf 448: }
1.22 misha 449:
450: struct FormPart {
451: Request* r;
452: const char* boundary;
1.48 moko 453: String* string;
1.22 misha 454: Form_table_value2string_info* info;
1.48 moko 455:
456: struct BinaryBlock{
457: const char* ptr;
458: size_t length;
459:
460: BinaryBlock(String* astring, Request* r): ptr(astring->untaint_and_transcode_cstr(String::L_AS_IS, &r->charsets)), length(strlen(ptr)){}
461: BinaryBlock(const char* aptr, size_t alength): ptr(aptr), length(alength){}
462: };
463:
464: Array<BinaryBlock> blocks;
465:
466: FormPart(Request* ar, const char* aboundary): r(ar), boundary(aboundary), string(new String()){}
467:
468: const char *post(size_t &length){
469: if(blocks.count()){
470: blocks+=BinaryBlock(string, r);
471:
472: length=0;
473: for(size_t i=0; i<blocks.count(); i++)
474: length+=blocks[i].length;
475:
476: char *result=(char *)pa_malloc_atomic(length);
477: char *ptr=result;
478:
479: for(size_t i=0; i<blocks.count(); i++){
480: memcpy(ptr, blocks[i].ptr, blocks[i].length);
481: ptr+=blocks[i].length;
482: }
483:
484: return result;
485: } else {
486: BinaryBlock result(string, r);
487: length=result.length;
488: return result.ptr;
489: }
490: }
1.22 misha 491: };
492:
1.73 moko 493: static void form_part_boundary_header(FormPart& part, String::Body name, const char* file_name=0) {
494: *part.string << "--" << part.boundary << CRLF CONTENT_DISPOSITION_CAPITALIZED ": form-data; name=\"" << name << "\"";
1.22 misha 495: if(file_name){
496: if(strcmp(file_name, NONAME_DAT)!=0)
1.48 moko 497: *part.string << "; filename=\"" << file_name << "\"";
498: *part.string << CRLF HTTP_CONTENT_TYPE_CAPITALIZED ": " << part.r->mime_type_of(file_name);
1.22 misha 499: }
1.48 moko 500: *part.string << CRLF CRLF;
1.22 misha 501: }
502:
1.73 moko 503: static void form_string_value2part(HashStringValue::key_type key, const String& value, FormPart& part) {
1.28 misha 504: form_part_boundary_header(part, key);
1.48 moko 505: *part.string << value << CRLF;
1.22 misha 506: }
507:
1.73 moko 508: static void form_file_value2part(HashStringValue::key_type key, VFile& vfile, FormPart& part) {
1.28 misha 509: form_part_boundary_header(part, key, vfile.fields().get(name_name)->as_string().cstr());
1.48 moko 510: part.blocks+=FormPart::BinaryBlock(part.string, part.r);
511: part.blocks+=FormPart::BinaryBlock(vfile.value_ptr(), vfile.value_size());
512: part.string=new String();
513: *part.string << CRLF;
1.22 misha 514: }
515:
516: static void form_table_value2part(Table::element_type row, FormPart* part) {
517: form_string_value2part(part->info->key, *row->get(0), *part);
518: }
519:
1.73 moko 520: static void form_value2part(HashStringValue::key_type key, HashStringValue::value_type value, FormPart& part) {
1.22 misha 521: if(const String* svalue=value->get_string())
522: form_string_value2part(key, *svalue, part);
523: else if(Table* tvalue=value->get_table()) {
1.48 moko 524: Form_table_value2string_info info(key, *part.string);
1.22 misha 525: part.info = &info;
526: tvalue->for_each(form_table_value2part, &part);
1.33 misha 527: } else if(VFile* vfile=static_cast<VFile *>(value->as("file"))){
1.22 misha 528: form_file_value2part(key, *vfile, part);
529: } else
1.73 moko 530: 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 531: }
532:
533: const char* pa_form2string_multipart(HashStringValue& form, Request& r, const char* boundary, size_t& post_size){
1.48 moko 534: FormPart formpart(&r, boundary);
1.22 misha 535: form.for_each<FormPart&>(form_value2part, formpart);
1.48 moko 536: *formpart.string << "--" << boundary << "--";
537: // @todo: return binary blocks here to save memory in pa_internal_file_read_http
538: return formpart.post(post_size);
1.22 misha 539: }
540:
1.54 misha 541: // Set-Cookie: name=value; Domain=docs.foo.com; Path=/accounts; Expires=Wed, 13-Jan-2021 22:23:01 GMT; Secure; HttpOnly
542: static ArrayString* parse_cookie(Request& r, const String& cookie) {
1.64 moko 543: char *current=pa_strdup(cookie.cstr());
1.54 misha 544:
545: const String* name=0;
1.55 moko 546: const String* value=&String::Empty;
547: const String* expires=&String::Empty;
548: const String* max_age=&String::Empty;
549: const String* path=&String::Empty;
550: const String* domain=&String::Empty;
551: const String* httponly=&String::Empty;
552: const String* secure=&String::Empty;
1.54 misha 553:
554: bool first_pair=true;
555:
556: do {
557: if(char *meaning=search_stop(current, ';'))
558: if(char *attribute=search_stop(meaning, '=')) {
559: const String* sname=new String(unescape_chars(attribute, strlen(attribute), &r.charsets.source(), true/*don't convert '"' to space*/), String::L_TAINTED);
560: const String* smeaning=0;
561: if(meaning)
562: smeaning=new String(unescape_chars(meaning, strlen(meaning), &r.charsets.source(), true/*don't convert '"' to space*/), String::L_TAINTED);
563:
564: if(first_pair) {
565: // name + value
566: name=sname;
567: value=smeaning;
568: first_pair=false;
569: } else {
570: const String& slower=sname->change_case(r.charsets.source(), String::CC_LOWER);
571:
572: if(slower == "expires")
573: expires=smeaning;
574: else if(slower == "max-age")
575: max_age=smeaning;
576: else if(slower == "domain")
577: domain=smeaning;
578: else if(slower == "path")
579: path=smeaning;
580: else if(slower == "httponly")
581: httponly=new String("1", String::L_CLEAN);
582: else if(slower == "secure")
583: secure=new String("1", String::L_CLEAN);
584: else {
585: // todo@ ?
586: }
587: }
588: }
589: } while(current);
590:
591: if(!name)
592: return 0;
593:
594: ArrayString* result=new ArrayString(8);
595: *result+=name;
596: *result+=value;
597: *result+=expires;
598: *result+=max_age;
599: *result+=domain;
600: *result+=path;
601: *result+=httponly;
602: *result+=secure;
603:
604: return result;
605: }
606:
1.56 misha 607: Table* parse_cookies(Request& r, Table *cookies){
608: Table& result=*new Table(new Cookies_table_template_columns);
609:
610: for(Array_iterator<Table::element_type> i(*cookies); i.has_next(); )
611: if(ArrayString* row=parse_cookie(r, *i.next()->get(0)))
612: result+=row;
613:
614: return &result;
615: }
616:
1.75 moko 617: void tables_update(HashStringValue& tables, const String::Body name, const String& value){
1.72 moko 618: Table *table;
619: if(Value *valready=tables.get(name)) {
620: // second+ appearence
621: table=valready->get_table();
622: } else {
623: // first appearence
624: Table::columns_type columns=new ArrayString(1);
625: *columns+=new String("value");
626: table=new Table(columns);
627: tables.put(name, new VTable(table));
628: }
629: // this string becomes next row
630: ArrayString& row=*new ArrayString(1);
631: row+=&value;
632: *table+=&row;
633: }
634:
1.1 paf 635: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.72 moko 636: 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 637: File_read_http_result result;
1.20 misha 638: char host[MAX_STRING];
1.66 moko 639: const char *idna_host;
1.1 paf 640: const char* uri;
1.49 moko 641: short port=80;
1.10 misha 642: const char* method="GET";
1.21 misha 643: bool method_is_get=true;
1.1 paf 644: HashStringValue* form=0;
645: int timeout_secs=2;
646: bool fail_on_status_ne_200=true;
1.12 misha 647: bool omit_post_charset=false;
1.1 paf 648: Value* vheaders=0;
1.10 misha 649: Value* vcookies=0;
1.11 misha 650: Value* vbody=0;
1.72 moko 651: Charset* asked_remote_charset=0;
1.58 moko 652: Charset* real_remote_charset=0;
1.1 paf 653: const char* user_cstr=0;
654: const char* password_cstr=0;
1.22 misha 655: const char* encode=0;
656: bool multipart=false;
1.1 paf 657:
658: if(options) {
659: int valid_options=pa_get_valid_file_options_count(*options);
660:
661: if(Value* vmethod=options->get(HTTP_METHOD_NAME)) {
662: valid_options++;
1.21 misha 663: method=vmethod->as_string().change_case(r.charsets.source(), String::CC_UPPER).cstr();
664: method_is_get=strcmp(method, "GET")==0;
1.1 paf 665: }
1.22 misha 666: if(Value* vencode=options->get(HTTP_FORM_ENCTYPE_NAME)) {
667: valid_options++;
668: encode=vencode->as_string().cstr();
669: }
1.1 paf 670: if(Value* vform=options->get(HTTP_FORM_NAME)) {
671: valid_options++;
672: form=vform->get_hash();
673: }
1.11 misha 674: if(vbody=options->get(HTTP_BODY_NAME)) {
1.1 paf 675: valid_options++;
676: }
677: if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) {
678: valid_options++;
679: timeout_secs=vtimeout->as_int();
680: }
1.11 misha 681: if(vheaders=options->get(HTTP_HEADERS_NAME)) {
1.1 paf 682: valid_options++;
683: }
1.11 misha 684: if(vcookies=options->get(HTTP_COOKIES_NAME)) {
1.10 misha 685: valid_options++;
686: }
1.1 paf 687: if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) {
688: valid_options++;
689: fail_on_status_ne_200=!vany_status->as_bool();
1.12 misha 690: }
1.20 misha 691: if(Value* vomit_post_charset=options->get(HTTP_OMIT_POST_CHARSET_NAME)){
1.12 misha 692: valid_options++;
693: omit_post_charset=vomit_post_charset->as_bool();
694: }
1.6 misha 695: if(Value* vcharset_name=options->get(PA_CHARSET_NAME)) {
1.77 moko 696: asked_remote_charset=&pa_charsets.get(vcharset_name->as_string());
1.58 moko 697: }
698: if(Value* vresponse_charset_name=options->get(PA_RESPONSE_CHARSET_NAME)) {
1.61 moko 699: valid_options++;
1.77 moko 700: real_remote_charset=&pa_charsets.get(vresponse_charset_name->as_string());
1.1 paf 701: }
702: if(Value* vuser=options->get(HTTP_USER)) {
703: valid_options++;
704: user_cstr=vuser->as_string().cstr();
705: }
706: if(Value* vpassword=options->get(HTTP_PASSWORD)) {
707: valid_options++;
708: password_cstr=vpassword->as_string().cstr();
709: }
710:
711: if(valid_options!=options->count())
1.46 misha 712: throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION);
1.1 paf 713: }
714: if(!asked_remote_charset) // defaulting to $request:charset
1.22 misha 715: asked_remote_charset=&(r.charsets).source();
716:
717: if(encode){
718: if(method_is_get)
1.72 moko 719: throw Exception(PARSER_RUNTIME, 0, "you can not use $." HTTP_FORM_ENCTYPE_NAME " option with method GET");
1.22 misha 720:
721: multipart=strcasecmp(encode, HTTP_CONTENT_TYPE_MULTIPART_FORMDATA)==0;
722:
723: if(!multipart && strcasecmp(encode, HTTP_CONTENT_TYPE_FORM_URLENCODED)!=0)
1.72 moko 724: throw Exception(PARSER_RUNTIME, 0, "$." HTTP_FORM_ENCTYPE_NAME " option value can be " HTTP_CONTENT_TYPE_FORM_URLENCODED " or " HTTP_CONTENT_TYPE_MULTIPART_FORMDATA " only");
1.22 misha 725: }
1.1 paf 726:
1.11 misha 727: if(vbody){
728: if(method_is_get)
1.72 moko 729: throw Exception(PARSER_RUNTIME, 0, "you can not use $." HTTP_BODY_NAME " option with method GET");
1.11 misha 730:
731: if(form)
1.72 moko 732: throw Exception(PARSER_RUNTIME, 0, "you can not use options $." HTTP_BODY_NAME " and $." HTTP_FORM_NAME " together");
1.11 misha 733: }
1.1 paf 734:
735: //preparing request
1.29 misha 736: String& connect_string=*new String(file_spec);
1.1 paf 737:
1.48 moko 738: const char* request;
739: size_t request_size;
1.1 paf 740: {
741: // influence URLencoding of tainted pieces to String::L_URI lang
1.22 misha 742: Temp_client_charset temp(r.charsets, *asked_remote_charset);
1.1 paf 743:
1.44 misha 744: const char* connect_string_cstr=connect_string.untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.1 paf 745:
746: const char* current=connect_string_cstr;
747: if(strncmp(current, "http://", 7)!=0)
1.72 moko 748: throw Exception(PARSER_RUNTIME, &connect_string, "does not start with http://"); //never
1.1 paf 749: current+=7;
750:
751: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
1.34 misha 752: char* host_uri=lsplit(host, '/');
753: uri=host_uri?current+(host_uri-1-host):"/";
754: char* port_cstr=lsplit(host, ':');
1.49 moko 755:
756: if (port_cstr){
757: char* error_pos=0;
758: port=(short)strtol(port_cstr, &error_pos, 10);
759: if(port==0 || *error_pos)
760: throw Exception(PARSER_RUNTIME, &connect_string, "invalid port number '%s'", port_cstr);
761: }
1.1 paf 762:
1.66 moko 763: idna_host=pa_idna_encode(host, r.charsets.source());
764:
1.11 misha 765: // making request head
1.1 paf 766: String head;
1.11 misha 767: head << method << " " << uri;
1.28 misha 768: if(method_is_get && form)
769: head << (strchr(uri, '?')!=0?"&":"?") << pa_form2string(*form, r.charsets);
1.11 misha 770:
1.66 moko 771: head <<" HTTP/1.0" CRLF "Host: "<< idna_host;
1.49 moko 772: if (port != 80)
773: head << ":" << port_cstr;
774: head << CRLF;
1.11 misha 775:
1.71 moko 776: char* boundary= multipart ? get_uuid_boundary() : 0;
1.22 misha 777:
1.35 misha 778: String user_headers;
779: bool user_agent_specified=false;
780: bool content_type_specified=false;
781: bool content_type_url_encoded=false;
782: if(vheaders && !vheaders->is_string()) { // allow empty
783: if(HashStringValue *headers=vheaders->get_hash()) {
784: Http_pass_header_info info={
785: &(r.charsets),
786: &user_headers,
787: &user_agent_specified,
788: &content_type_specified,
789: &content_type_url_encoded};
790: headers->for_each<Http_pass_header_info*>(http_pass_header, &info);
791: } else
1.72 moko 792: throw Exception(PARSER_RUNTIME, 0, "headers param must be hash");
1.35 misha 793: };
794:
1.48 moko 795: const char* request_body=0;
1.22 misha 796: size_t post_size=0;
797: if(form && !method_is_get) {
1.38 misha 798: head << "Content-Type: " << (multipart ? HTTP_CONTENT_TYPE_MULTIPART_FORMDATA : HTTP_CONTENT_TYPE_FORM_URLENCODED);
1.28 misha 799:
800: if(!omit_post_charset)
801: head << "; charset=" << asked_remote_charset->NAME_CSTR();
802:
1.22 misha 803: if(multipart) {
1.28 misha 804: head << "; boundary=" << boundary;
1.48 moko 805: request_body=pa_form2string_multipart(*form, r/*charsets & mime_type needed*/, boundary, post_size/*correct post_size returned here*/);
1.22 misha 806: } else {
1.48 moko 807: request_body=pa_form2string(*form, r.charsets);
808: post_size=strlen(request_body);
1.22 misha 809: }
1.28 misha 810: head << CRLF;
1.35 misha 811: } else if(vbody) {
1.38 misha 812: // $.body was specified
1.35 misha 813: if(content_type_url_encoded){
1.36 misha 814: // transcode + url-encode
1.48 moko 815: request_body=vbody->as_string().untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.35 misha 816: } else {
1.36 misha 817: // content-type != application/x-www-form-urlencoded -> transcode only, don't url-encode!
1.72 moko 818: const String &sbody=vbody->as_string();
819: request_body=Charset::transcode(String::C(sbody.cstr(), sbody.length()), r.charsets.source(), *asked_remote_charset).str;
1.35 misha 820: }
1.48 moko 821: post_size=strlen(request_body);
1.1 paf 822: }
823:
824: // http://www.ietf.org/rfc/rfc2617.txt
825: if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr))
1.38 misha 826: head << "Authorization: " << *authorization_field_value << CRLF;
1.1 paf 827:
1.35 misha 828: head << user_headers;
829:
1.1 paf 830: if(!user_agent_specified) // defaulting
1.38 misha 831: head << "User-Agent: " DEFAULT_USER_AGENT CRLF;
1.1 paf 832:
1.12 misha 833: if(form && !method_is_get && content_type_specified) // POST + form + content-type was specified
1.72 moko 834: throw Exception(PARSER_RUNTIME, 0, "$.content-type can't be specified with method POST");
1.12 misha 835:
1.11 misha 836: if(vcookies && !vcookies->is_string()){ // allow empty
1.10 misha 837: if(HashStringValue* cookies=vcookies->get_hash()) {
1.37 misha 838: head << "Cookie: ";
1.35 misha 839: Http_pass_header_info info={&(r.charsets), &head, 0, 0, 0};
1.10 misha 840: cookies->for_each<Http_pass_header_info*>(http_pass_cookie, &info);
841: head << CRLF;
842: } else
1.72 moko 843: throw Exception(PARSER_RUNTIME, 0, "cookies param must be hash");
1.10 misha 844: }
845:
1.48 moko 846: if(request_body)
1.38 misha 847: head << "Content-Length: " << format(post_size, "%u") << CRLF;
1.48 moko 848:
849: head << CRLF;
850:
851: const char *request_head=head.untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.1 paf 852:
1.48 moko 853: if(request_body){
854: size_t head_size = strlen(request_head);
855: request_size=post_size + head_size;
856: char *ptr=(char *)pa_malloc_atomic(request_size);
857: memcpy(ptr, request_head, head_size);
858: memcpy(ptr+head_size, request_body, post_size);
859: request=ptr;
860: } else {
861: request_size=strlen(request_head);
862: request=request_head;
863: }
1.1 paf 864: }
865:
1.78 moko 866:
867: HTTP_response response(connect_string);
1.22 misha 868:
1.28 misha 869: // sending request
1.78 moko 870: int status_code=http_request(response, idna_host, port, request, request_size, timeout_secs, fail_on_status_ne_200);
871:
1.72 moko 872: // processing results
1.78 moko 873: char* raw_body=response.buf + response.body_offset;
874: size_t raw_body_size=response.length - response.body_offset;
875:
1.1 paf 876: result.headers=new HashStringValue;
877: VHash* vtables=new VHash;
1.72 moko 878: result.headers->put("tables", vtables);
879:
1.78 moko 880: if (!real_remote_charset && !response.headers.content_type.is_empty())
881: real_remote_charset=detect_charset(response.headers.content_type.cstr());
1.1 paf 882:
1.72 moko 883: if(as_text)
1.77 moko 884: real_remote_charset=pa_charsets.checkBOM(raw_body, raw_body_size, real_remote_charset);
1.72 moko 885:
886: if (!real_remote_charset)
887: real_remote_charset=asked_remote_charset; // never null
888:
1.85 moko 889: for(Array_iterator<HTTP_Headers::Header> i(response.headers.headers); i.has_next(); ){
890: HTTP_Headers::Header header=i.next();
1.72 moko 891:
892: header.transcode(*real_remote_charset, r.charsets.source());
893:
894: String &header_value=*new String(header.value, String::L_TAINTED);
895:
896: tables_update(vtables->hash(), header.name, header_value);
897: result.headers->put(header.name, new VString(header_value));
1.16 misha 898: }
899:
1.72 moko 900: // filling $.cookies
1.89 moko 901: if(vcookies=vtables->hash().get("SET-COOKIE"))
1.72 moko 902: result.headers->put(HTTP_COOKIES_NAME, new VTable(parse_cookies(r, vcookies->get_table())));
903:
1.1 paf 904: // output response
905: String::C real_body=String::C(raw_body, raw_body_size);
1.16 misha 906:
907: 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 908: real_body=Charset::transcode(real_body, *real_remote_charset, r.charsets.source());
1.1 paf 909: }
910:
911: result.str=const_cast<char *>(real_body.str); // hacking a little
912: result.length=real_body.length;
1.16 misha 913:
1.22 misha 914: if(as_text && result.length)
915: fix_line_breaks(result.str, result.length);
916:
1.1 paf 917: result.headers->put(file_status_name, new VInt(status_code));
1.16 misha 918:
1.1 paf 919: return result;
920: }
1.84 moko 921:
922: /* ********************** httpd *************************** */
923:
924: class HTTPD_request : public HTTP_response {
925: public:
926: const char *method;
927: const char *uri;
928:
929: HTTPD_request() : HTTP_response(String::Empty), method(NULL), uri(NULL){};
930:
931: const char *extract_method(char *method_line){
932: char* uri_start = strchr(method_line, ' ');
933:
934: if(!uri_start || uri_start == method_line)
935: return NULL;
936:
937: char* uri_end=strchr(uri_start+1, ' ');
938:
939: if(!uri_end || uri_end == uri_start+1)
940: return NULL;
941:
942: uri=pa_strdup(uri_start+1, uri_end-uri_start-1);
943: return str_upper(method_line, uri_start-method_line);
944: }
945:
1.87 moko 946: void read_header(int);
947: size_t read_post(int, char *, size_t);
1.84 moko 948: };
949:
950: enum HTTPD_request_state {
951: HTTPD_METHOD,
952: HTTPD_HEADERS
953: };
954:
955: void HTTPD_request::read_header(int sock) {
956: enum HTTPD_request_state state = HTTPD_METHOD;
957:
958: size_t chunk_size = 0x400*4;
959: resize(chunk_size);
960:
961: while(read(sock, chunk_size)){
962: switch(state){
963: case HTTPD_METHOD: {
964: size_t method_size = first_line();
965: if(!method_size)
966: break;
967:
968: char *method_line = pa_strdup(buf, method_size);
969: method = extract_method(method_line);
970:
971: if(!method || strcmp(method, "GET") && strcmp(method, "HEAD") && strcmp(method, "POST") && strcmp(method, "PUT") && strcmp(method, "DELETE"))
972: throw Exception("httpd.method", new String(method ? method : method_line), "invalid request method");
973: state = HTTPD_HEADERS;
974: }
975:
976: case HTTPD_HEADERS: {
977: if(!body_start())
978: break;
979:
980: parse_headers();
981: return;
982: }
983: }
984: }
985:
986: if(state == HTTPD_METHOD)
987: throw Exception("httpd.request", 0, "bad request from host - no method found (size=%u)", length);
988:
989: if(state == HTTPD_HEADERS){
990: parse_headers();
991: body_offset=length;
992: }
993: }
994:
1.87 moko 995: size_t HTTPD_request::read_post(int sock, char *body, size_t max_bytes) {
996: size_t total_read = min(length - body_offset, max_bytes);
997: memcpy(body, buf, total_read);
998:
999: while (total_read < max_bytes){
1000: ssize_t received_size = recv(sock, buf + total_read, max_bytes - total_read, 0);
1001: if(received_size == 0)
1002: return total_read;
1003: if(received_size < 0) {
1004: if(int no = pa_socks_errno())
1005: throw Exception("httpd.timeout", &url, "error receiving request body: %s (%d)", pa_socks_strerr(no), no);
1006: return total_read;
1007: }
1008: total_read += received_size;
1009: }
1010: return total_read;
1011: }
1012:
1.84 moko 1013: /* ********************************************************** */
1014:
1.85 moko 1015: Array<HTTP_Headers::Header> &HTTPD_Connection::headers() {
1.84 moko 1016: return request->headers.headers;
1017: }
1018:
1019: const char *HTTPD_Connection::method() {
1020: return request->method;
1021: }
1022:
1023: const char *HTTPD_Connection::uri() {
1024: return request->uri;
1025: }
1026:
1027: const char *HTTPD_Connection::content_type() {
1028: return request->headers.content_type.cstr();
1029: }
1030:
1031: uint64_t HTTPD_Connection::content_length(){
1032: return request->headers.content_length;
1033: }
1034:
1035: void HTTPD_Connection::read_header(){
1036: request = new HTTPD_request();
1037: request->read_header(sock);
1038: }
1039:
1.87 moko 1040: size_t HTTPD_Connection::read_post(char *body, size_t max_bytes) {
1041: return request->read_post(sock, body, max_bytes);
1042: }
1043:
1.90 ! moko 1044: size_t HTTPD_Connection::send_body(const void *buf, size_t size) {
! 1045: if(send(sock, buf, size, 0) != (ssize_t)size) {
! 1046: int no=pa_socks_errno();
! 1047: throw Exception("httpd.timeout", 0, "error sending response: %s (%d)", pa_socks_strerr(no), no);
! 1048: }
! 1049: return size;
! 1050: }
! 1051:
1.84 moko 1052: static int sock_on = 1;
1053:
1.86 moko 1054: int HTTPD_Server::bind(const char *host_port){
1.84 moko 1055: struct sockaddr_in me;
1056:
1.86 moko 1057: const char *port = strchr(host_port, ':');
1058: const char *host = NULL;
1059: if(port && port > host_port){
1060: host = pa_strdup(host_port, port - host_port);
1061: port += 1;
1062: } else {
1063: port = host_port;
1064: }
1065:
1066: if(!set_addr(&me, host, pa_atoui(port, 10))){
1.84 moko 1067: if (host)
1068: throw Exception("httpd.bind", 0, "can not resolve hostname \"%s\"", host);
1069: me.sin_addr.s_addr=INADDR_ANY;
1070: }
1071:
1072: int sock = socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/);
1073:
1074: if(sock < 0){
1075: int no=pa_socks_errno();
1076: throw Exception("httpd.bind", 0, "can not make socket: %s (%d)", pa_socks_strerr(no), no);
1077: }
1078:
1079: if (setsockopt(sock, SOL_SOCKET, SO_REUSEADDR, (char *)&sock_on, sizeof(sock_on)) ||
1080: setsockopt(sock, SOL_SOCKET, SO_KEEPALIVE, (char *)&sock_on, sizeof(sock_on)) ||
1081: ::bind(sock, (struct sockaddr*)&me, sizeof(me)) ||
1082: listen(sock, 16)) {
1.89 moko 1083: closesocket(sock);
1.84 moko 1084: int no = pa_socks_errno();
1085: throw Exception("httpd.bind", 0, "can not bind socket: %s (%d)", pa_socks_strerr(no), no);
1086: }
1087: return sock;
1088: }
1089:
1.88 moko 1090: static int ready(int fd,int operation,int timeout_value){
1.84 moko 1091: struct timeval timeout = {0, timeout_value * 1000};
1092: fd_set fds;
1093: FD_ZERO(&fds);
1094: FD_SET(fd, &fds);
1095: switch (operation){
1096: case 0: return select(fd + 1, &fds, NULL, NULL, &timeout)>0; /* read */
1097: case 1: return select(fd + 1, NULL, &fds, NULL, &timeout)>0; /* write */
1098: default: return select(fd + 1, &fds, &fds, NULL, &timeout)>0; /* both */
1099: }
1100: }
1101:
1102: HTTPD_Connection *HTTPD_Server::accept(int sock, int timeout_value) {
1103: int ready = ::ready(sock, 0, timeout_value);
1104: if (ready < 0) {
1105: int no=pa_socks_errno();
1106: if(no == EINTR)
1107: return NULL;
1108: throw Exception("httpd.accept", 0, "error waiting for connection: %s (%d)", pa_socks_strerr(no), no);
1109: }
1110: if (ready == 0) {
1111: /* Timeout */
1112: return NULL;
1113: }
1114:
1115: struct sockaddr_in addr;
1.89 moko 1116: socklen_t sock_addr_len = sizeof(struct sockaddr_in);
1.84 moko 1117: memset(&addr, 0, sock_addr_len);
1118:
1119: int csock = ::accept(sock, (struct sockaddr *)&addr, &sock_addr_len);
1120: if(csock == -1){
1121: int no=pa_socks_errno();
1122: throw Exception("httpd.accept", 0, "error accepting connection: %s (%d)", pa_socks_strerr(no), no);
1123: }
1124:
1125: return new HTTPD_Connection(csock, pa_strdup(inet_ntoa(addr.sin_addr)));
1126: }
1127:
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