Annotation of parser3/src/main/pa_http.C, revision 1.146
1.1 paf 1: /** @file
2: Parser: http support functions.
3:
1.135 moko 4: Copyright (c) 2001-2026 Art. Lebedev Studio (https://www.artlebedev.com)
1.121 moko 5: Authors: Konstantin Morshnev <moko@design.ru>, Alexandr Petrosian <paf@design.ru>
1.1 paf 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_vfile.h"
14: #include "pa_random.h"
1.145 moko 15: #include "pa_threads.h"
1.1 paf 16:
1.140 moko 17: #ifndef WIN32
18: #include <netinet/tcp.h> // for TCP_NODELAY
19: #endif
20:
1.146 ! moko 21: volatile const char * IDENT_PA_HTTP_C="$Id: pa_http.C,v 1.145 2026/10/26 19:06:51 moko Exp $" IDENT_PA_HTTP_H;
1.53 moko 22:
1.1 paf 23: // defines
24:
1.19 misha 25: #define HTTP_METHOD_NAME "method"
26: #define HTTP_FORM_NAME "form"
27: #define HTTP_BODY_NAME "body"
28: #define HTTP_TIMEOUT_NAME "timeout"
29: #define HTTP_HEADERS_NAME "headers"
1.22 misha 30: #define HTTP_FORM_ENCTYPE_NAME "enctype"
1.19 misha 31: #define HTTP_ANY_STATUS_NAME "any-status"
1.59 moko 32: #define HTTP_OMIT_POST_CHARSET_NAME "omit-post-charset" // ^file::load[...;http://...;$.method[post]] by default adds charset to content-type
1.12 misha 33:
1.1 paf 34: #define HTTP_USER "user"
35: #define HTTP_PASSWORD "password"
36:
1.70 moko 37: #define HTTP_USER_AGENT "user-agent"
1.1 paf 38: #define DEFAULT_USER_AGENT "parser3"
39:
1.59 moko 40: #ifndef INADDR_NONE
41: #define INADDR_NONE ((ulong) -1)
42: #endif
1.1 paf 43:
1.146 ! moko 44: #ifdef SOCK_CLOEXEC
! 45: // avoid inheritance by the executed programs
! 46: #define PA_ACCEPT(sock, addr, len) ::accept4(sock, addr, len, SOCK_CLOEXEC)
! 47: #define PA_SOCKET(domain, type, protocol) ::socket(domain, (type) | SOCK_CLOEXEC, protocol)
! 48: #else
! 49: #define PA_ACCEPT(sock, addr, len) ::accept(sock, addr, len)
! 50: #define PA_SOCKET(domain, type, protocol) ::socket(domain, type, protocol)
! 51: #endif
! 52:
1.1 paf 53: #undef CRLF
54: #define CRLF "\r\n"
55:
1.138 moko 56: #define HTTP_TIMEOUT_COMMENT "timeout occurred while retrieving document"
57:
1.54 misha 58: // helpers
1.56 misha 59:
1.130 moko 60: #ifdef HTTPD_DEBUG
61: void pa_log(const char* fmt, ...);
62: #define LOG(action) action
63: #else
64: #define LOG(action)
65: #endif
66:
67: ssize_t pa_send(int sock, const char *buffer, size_t len){
68: size_t total_sent = 0;
69: while (total_sent < len) {
1.141 moko 70: ssize_t bytes_sent=pa_sock_send(sock, buffer + total_sent, len - total_sent);
1.130 moko 71: if (bytes_sent < 0) {
72: return bytes_sent;
73: } else if (bytes_sent == 0) {
74: // Connection closed by the remote peer?
75: break;
76: }
77: LOG(
78: if(bytes_sent != len - total_sent)
79: pa_log("httpd [%d] partial send %d of (%d)", sock, bytes_sent, len - total_sent)
80: );
81: total_sent += bytes_sent;
82: }
83: return total_sent;
84: }
85:
1.138 moko 86: // socket operation failed on SO_*TIMEO timeout
87: static bool sock_timed_out(int no){
88: #ifdef WIN32
89: return no==WSAETIMEDOUT;
90: #else
91: return no==EAGAIN || no==EWOULDBLOCK || no==EINPROGRESS /* connect on linux */;
92: #endif
93: }
94:
95: static int check_socket_timeout(int no){
96: if(sock_timed_out(no))
97: throw Exception("http.timeout", 0, HTTP_TIMEOUT_COMMENT);
98: return no;
99: }
100:
1.85 moko 101: bool HTTP_Headers::add_header(const char *line){
1.78 moko 102: const char *value=strchr(line, ':');
103:
104: if(value && value != line){ // we need only headers, not the response code
1.134 moko 105: Header header = Header(String::Body(str_upper(line, value-line)), String::Body(value+1).trim(String::TRIM_BOTH, " \t\n\r"));
1.78 moko 106:
107: if(header.name == String::Body(HTTP_CONTENT_TYPE_UPPER) && content_type.is_empty())
108: content_type=header.value;
109:
110: if(header.name == String::Body("CONTENT-LENGTH") && content_length==0)
1.95 moko 111: ALTER_EXCEPTION_COMMENT(content_length=pa_atoul(header.value.cstr()), " for content-length");
1.78 moko 112:
113: headers+=header;
114:
115: return true;
116: }
117: return false;
118: }
119:
1.54 misha 120: class Cookies_table_template_columns: public ArrayString {
121: public:
122: Cookies_table_template_columns() {
123: *this+=new String("name");
124: *this+=new String("value");
125: *this+=new String("expires");
126: *this+=new String("max-age");
127: *this+=new String("domain");
128: *this+=new String("path");
129: *this+=new String("httponly");
130: *this+=new String("secure");
131: }
132: };
133:
134:
1.1 paf 135: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
1.22 misha 136: memset(addr, 0, sizeof(*addr));
137: addr->sin_family=AF_INET;
138: addr->sin_port=htons(port);
139: if(host) {
1.65 moko 140: struct hostent *hostIP=gethostbyname(host);
141: if(hostIP && hostIP->h_addrtype == AF_INET){
142: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
143: return true;
144: }
145: }
146: return false;
1.1 paf 147: }
148:
1.84 moko 149: class HTTP_response : public PA_Allocated {
1.78 moko 150: public:
151: char *buf;
152: size_t length;
153: size_t buf_size;
154: size_t body_offset;
155:
1.85 moko 156: HTTP_Headers headers;
1.78 moko 157:
1.97 moko 158: HTTP_response() : buf(NULL), length(0), buf_size(0), body_offset(0){}
1.78 moko 159:
160: void resize(size_t size){
161: buf_size=size;
162: buf=(char *)pa_realloc(buf, size + 1);
163: }
164:
1.120 moko 165: bool read(SOCKET sock, size_t size){
1.103 moko 166: if(length + size > buf_size)
167: resize(buf_size * 2 + size);
1.141 moko 168: ssize_t received_size=pa_sock_recv(sock, buf + length, size);
1.103 moko 169: if(received_size == 0)
1.78 moko 170: return false;
1.103 moko 171: if(received_size < 0) {
1.138 moko 172: if(int no = check_socket_timeout(pa_socks_errno()))
1.102 moko 173: throw Exception("http.timeout", 0, "error receiving response: %s (%d)", pa_socks_strerr(no), no);
1.78 moko 174: return false;
175: }
176: length+=received_size;
177: buf[length]='\0';
178: return true;
179: }
180:
1.83 moko 181: size_t first_line(){
1.89 moko 182: char *header=strchr(buf, '\n');
183: if(!header)
1.78 moko 184: return false;
185:
1.89 moko 186: return header-buf;
1.78 moko 187: }
188:
189: const char *status_code(char *status_line, int &result){
190: char* status_start = strchr(status_line, ' ');
191:
192: if(!(status_start++))
193: return status_line;
194:
195: char* status_end=strchr(status_start, ' ');
196:
197: if(!status_end)
198: return status_line;
199:
200: if(status_end==status_start)
201: return status_line;
1.1 paf 202:
1.78 moko 203: const char *result_str=pa_strdup(status_start, status_end-status_start);
1.95 moko 204: ALTER_EXCEPTION_COMMENT(result=pa_atoui(result_str), " for HTTP status");
1.78 moko 205: return result_str;
206: }
1.2 paf 207:
1.78 moko 208: bool body_start(){
209: char *p=buf;
210: while((p=strchr(p, '\n'))) {
211: if(p[1]=='\r' && p[2]=='\n'){ // \r\n\r\n
212: *p='\0';
213: body_offset=p-buf+3;
214: return true;
215: }
216: if(p[1]=='\n') { // \n\n
217: *p='\0';
218: body_offset=p-buf+2;
219: return true;
220: }
221: p++;
222: }
223: return false;
1.2 paf 224: }
1.78 moko 225:
226: void parse_headers(){
227: const String header_block(buf, String::L_TAINTED);
228:
229: ArrayString aheaders;
230: header_block.split(aheaders, 0, "\n");
231:
1.124 moko 232: ArrayString::Iterator i(aheaders);
1.78 moko 233: i.next(); // skipping status
1.125 moko 234: for(;i;){
1.78 moko 235: const char *line=i.next()->cstr();
236: if(!headers.add_header(line))
1.97 moko 237: throw Exception("http.response", 0, "bad response from host - bad header \"%s\"", line);
1.78 moko 238: }
1.1 paf 239: }
240:
1.120 moko 241: int read_response(SOCKET sock, bool fail_on_status_ne_200);
1.78 moko 242: };
243:
244: enum HTTP_response_state {
245: HTTP_STATUS_CODE,
246: HTTP_HEADERS,
247: HTTP_BODY
248: };
249:
1.120 moko 250: int HTTP_response::read_response(SOCKET sock, bool fail_on_status_ne_200) {
1.78 moko 251: HTTP_response_state state=HTTP_STATUS_CODE;
252: int result=0;
253:
254: size_t chunk_size=0x400*16;
1.88 moko 255: resize(2*chunk_size);
1.78 moko 256:
1.88 moko 257: while(read(sock, chunk_size)){
1.78 moko 258: switch(state){
259: case HTTP_STATUS_CODE: {
1.88 moko 260: size_t status_size=first_line();
1.78 moko 261: if(!status_size)
262: break;
263:
1.88 moko 264: const char *status=status_code(pa_strdup(buf, status_size), result);
1.78 moko 265:
266: if(!result || fail_on_status_ne_200 && result!=200)
267: throw Exception("http.status", status ? new String(status) : &String::Empty, "invalid HTTP response status");
268:
269: state=HTTP_HEADERS;
270: }
271:
272: case HTTP_HEADERS: {
1.88 moko 273: if(!body_start())
1.78 moko 274: break;
275:
1.88 moko 276: parse_headers();
1.78 moko 277:
1.97 moko 278: size_t content_length=check_file_size(headers.content_length, 0);
1.88 moko 279: if(content_length>0 && (content_length + body_offset) > length){
280: resize(content_length + body_offset + 0x400*64);
1.78 moko 281: }
282:
283: state=HTTP_BODY;
1.1 paf 284: break;
285: }
1.78 moko 286:
287: case HTTP_BODY: {
288: chunk_size=0x400*64;
1.1 paf 289: break;
290: }
291: }
292: }
1.78 moko 293:
294: if(state==HTTP_STATUS_CODE)
1.97 moko 295: throw Exception("http.response", 0, "bad response from host - no status found (size=%u)", length);
1.78 moko 296:
297: if(state==HTTP_HEADERS){
1.88 moko 298: parse_headers();
299: body_offset=length;
1.1 paf 300: }
1.78 moko 301:
302: return result;
1.1 paf 303: }
304:
305: /* ********************** request *************************** */
306:
307: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
1.138 moko 308: # define PA_HAVE_ALARM
1.1 paf 309: #endif
310:
1.138 moko 311: #ifdef PA_HAVE_ALARM
1.1 paf 312: static sigjmp_buf timeout_env;
313: static void timeout_handler(int /*sig*/){
1.101 moko 314: siglongjmp(timeout_env, 1);
1.1 paf 315: }
1.118 moko 316:
1.138 moko 317: // alarm is per process, threads use SO_*TIMEO
318: #define PA_USE_ALARM (HTTPD_Server::mode != HTTPD_Server::MULTITHREADED)
1.118 moko 319:
1.138 moko 320: #define ALARM(value) if(PA_USE_ALARM) alarm(value)
1.101 moko 321: #else
1.138 moko 322: #define PA_USE_ALARM false
1.101 moko 323: #define ALARM(value)
1.1 paf 324: #endif
325:
1.138 moko 326: // SO_*TIMEO can be defined in .h but not implemented in protocol, so errors are ignored
327: static void set_socket_timeout(SOCKET sock, int option, int timeout_secs){
328: #ifdef WIN32
329: int timeout=timeout_secs < INT_MAX/1000 ? timeout_secs*1000 : 0 /* infinite */;
330: #else
331: struct timeval timeout={timeout_secs, 0};
332: #endif
333: setsockopt(sock, SOL_SOCKET, option, (const char*)&timeout, sizeof(timeout));
334: }
335:
1.78 moko 336: 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 337: if(!host)
1.73 moko 338: throw Exception("http.host", 0, "zero hostname"); //never
1.1 paf 339:
1.120 moko 340: volatile SOCKET sock=INVALID_SOCKET; // to prevent makeing it register variable, because it will be clobbered by longjmp [thanks gcc warning]
1.101 moko 341:
1.138 moko 342: #ifdef PA_HAVE_ALARM
343: if(PA_USE_ALARM) signal(SIGALRM, timeout_handler);
344: if(PA_USE_ALARM && sigsetjmp(timeout_env, 1)) {
1.101 moko 345: // duplicating closesocket to make code more simple for old compilers
1.120 moko 346: if(sock != INVALID_SOCKET)
1.101 moko 347: closesocket(sock);
1.138 moko 348: throw Exception("http.timeout", 0, HTTP_TIMEOUT_COMMENT);
1.1 paf 349: return 0; // never
1.101 moko 350: } else
1.1 paf 351: #endif
1.101 moko 352: {
353: ALARM(timeout_secs);
1.1 paf 354: try {
355: int result;
356: struct sockaddr_in dest;
357:
358: if(!set_addr(&dest, host, port))
1.127 moko 359: throw Exception("http.host", 0, "cannot resolve hostname \"%s\"", host);
1.1 paf 360:
1.120 moko 361: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/)) == INVALID_SOCKET) {
1.1 paf 362: int no=pa_socks_errno();
1.127 moko 363: throw Exception("http.connect", 0, "cannot make socket: %s (%d)", pa_socks_strerr(no), no);
1.1 paf 364: }
365:
366: // To enable SO_DONTLINGER (that is, disable SO_LINGER)
367: // l_onoff should be set to zero and setsockopt should be called
368: linger dont_linger={0,0};
369: setsockopt(sock, SOL_SOCKET, SO_LINGER, (const char *)&dont_linger, sizeof(dont_linger));
370:
1.138 moko 371: if(!PA_USE_ALARM){
372: set_socket_timeout(sock, SO_SNDTIMEO, timeout_secs);
373: set_socket_timeout(sock, SO_RCVTIMEO, timeout_secs);
374: }
1.1 paf 375:
376: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) {
1.138 moko 377: int no=check_socket_timeout(pa_socks_errno());
1.127 moko 378: throw Exception("http.connect", 0, "cannot connect to host \"%s\": %s (%d)", host, pa_socks_strerr(no), no);
1.1 paf 379: }
1.22 misha 380:
1.130 moko 381: if(pa_send(sock, request, request_size) < 0) {
1.138 moko 382: int no=check_socket_timeout(pa_socks_errno());
1.78 moko 383: throw Exception("http.timeout", 0, "error sending request: %s (%d)", pa_socks_strerr(no), no);
1.1 paf 384: }
385:
1.88 moko 386: result=response.read_response(sock, fail_on_status_ne_200);
1.78 moko 387: closesocket(sock);
1.101 moko 388: ALARM(0);
1.1 paf 389: return result;
390: } catch(...) {
1.101 moko 391: ALARM(0);
1.120 moko 392: if(sock != INVALID_SOCKET)
1.78 moko 393: closesocket(sock);
1.1 paf 394: rethrow;
395: }
396: }
397: }
398:
399: #ifndef DOXYGEN
400: struct Http_pass_header_info {
401: Request_charsets* charsets;
402: String* request;
1.35 misha 403: bool* user_agent_specified;
404: bool* content_type_specified;
405: bool* content_type_url_encoded;
1.1 paf 406: };
407: #endif
1.50 moko 408:
409: char *pa_http_safe_header_name(const char *name) {
410: char *result=pa_strdup(name);
411: char *n=result;
1.52 misha 412: if(!pa_isalpha((unsigned char)*n))
1.50 moko 413: *n++ = '_';
414: for(; *n; ++n) {
1.52 misha 415: if (!pa_isalnum((unsigned char)*n) && *n != '-' && *n != '_')
1.50 moko 416: *n = '_';
417: }
418: return result;
419: }
420:
1.101 moko 421: static void http_pass_header(HashStringValue::key_type aname, HashStringValue::value_type avalue, Http_pass_header_info *info) {
1.9 misha 422:
1.41 misha 423: const char* name_cstr=aname.cstr();
424:
1.38 misha 425: if(strcasecmp(name_cstr, HTTP_CONTENT_LENGTH)==0)
426: return;
427:
1.50 moko 428: String name=String(pa_http_safe_header_name(capitalize(name_cstr)), String::L_AS_IS);
429: String value=attributed_meaning_to_string(*avalue, String::L_HTTP_HEADER, true);
1.9 misha 430:
1.35 misha 431: *info->request << name << ": " << value << CRLF;
1.1 paf 432:
1.38 misha 433: if(strcasecmp(name_cstr, HTTP_USER_AGENT)==0)
1.35 misha 434: *info->user_agent_specified=true;
1.38 misha 435: if(strcasecmp(name_cstr, HTTP_CONTENT_TYPE)==0){
1.35 misha 436: *info->content_type_specified=true;
1.62 moko 437: *info->content_type_url_encoded=pa_strncasecmp(value.cstr(), HTTP_CONTENT_TYPE_FORM_URLENCODED)==0;
1.35 misha 438: }
1.1 paf 439: }
440:
1.10 misha 441: static void http_pass_cookie(HashStringValue::key_type name,
1.20 misha 442: HashStringValue::value_type value,
443: Http_pass_header_info *info) {
1.10 misha 444:
1.17 misha 445: *info->request << String(name, String::L_HTTP_COOKIE) << "="
1.31 misha 446: << attributed_meaning_to_string(*value, String::L_HTTP_COOKIE, true)
1.10 misha 447: << "; ";
448:
449: }
1.1 paf 450:
451: static const String* basic_authorization_field(const char* user, const char* pass) {
452: if(!user&& !pass)
453: return 0;
454:
455: String combined;
456: if(user)
457: combined<<user;
458: combined<<":";
459: if(pass)
460: combined<<pass;
461:
1.20 misha 462: String* result=new String("Basic ");
1.82 moko 463: *result<<pa_base64_encode(combined.cstr(), combined.length(), Base64Options(false /*no wrap*/));
1.1 paf 464: return result;
465: }
466:
1.73 moko 467: static void form_string_value2string(HashStringValue::key_type key, const String& value, String& result) {
1.30 misha 468: result << String(key, String::L_URI) << "=" << String(value, String::L_URI) << "&";
1.1 paf 469: }
1.20 misha 470:
1.1 paf 471: #ifndef DOXYGEN
472: struct Form_table_value2string_info {
473: HashStringValue::key_type key;
474: String& result;
475:
476: Form_table_value2string_info(HashStringValue::key_type akey, String& aresult):
477: key(akey), result(aresult) {}
478: };
479: #endif
480: static void form_table_value2string(Table::element_type row, Form_table_value2string_info* info) {
481: form_string_value2string(info->key, *row->get(0), info->result);
482: }
1.73 moko 483:
484: static void form_value2string(HashStringValue::key_type key, HashStringValue::value_type value, String* result) {
1.1 paf 485: if(const String* svalue=value->get_string())
486: form_string_value2string(key, *svalue, *result);
487: else if(Table* tvalue=value->get_table()) {
488: Form_table_value2string_info info(key, *result);
489: tvalue->for_each(form_table_value2string, &info);
490: } else
1.73 moko 491: throw Exception(PARSER_RUNTIME, new String(key, String::L_TAINTED),
1.63 moko 492: "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 493: }
1.20 misha 494:
1.5 misha 495: const char* pa_form2string(HashStringValue& form, Request_charsets& charsets) {
1.1 paf 496: String string;
1.3 paf 497: form.for_each<String*>(form_value2string, &string);
1.44 misha 498: return string.untaint_and_transcode_cstr(String::L_URI, &charsets);
1.1 paf 499: }
1.22 misha 500:
501: struct FormPart {
502: Request* r;
503: const char* boundary;
1.48 moko 504: String* string;
1.22 misha 505: Form_table_value2string_info* info;
1.48 moko 506:
507: struct BinaryBlock{
508: const char* ptr;
509: size_t length;
510:
511: BinaryBlock(String* astring, Request* r): ptr(astring->untaint_and_transcode_cstr(String::L_AS_IS, &r->charsets)), length(strlen(ptr)){}
512: BinaryBlock(const char* aptr, size_t alength): ptr(aptr), length(alength){}
513: };
514:
515: Array<BinaryBlock> blocks;
516:
517: FormPart(Request* ar, const char* aboundary): r(ar), boundary(aboundary), string(new String()){}
518:
519: const char *post(size_t &length){
520: if(blocks.count()){
521: blocks+=BinaryBlock(string, r);
522:
523: length=0;
524: for(size_t i=0; i<blocks.count(); i++)
525: length+=blocks[i].length;
526:
527: char *result=(char *)pa_malloc_atomic(length);
528: char *ptr=result;
529:
530: for(size_t i=0; i<blocks.count(); i++){
531: memcpy(ptr, blocks[i].ptr, blocks[i].length);
532: ptr+=blocks[i].length;
533: }
534:
535: return result;
536: } else {
537: BinaryBlock result(string, r);
538: length=result.length;
539: return result.ptr;
540: }
541: }
1.22 misha 542: };
543:
1.73 moko 544: static void form_part_boundary_header(FormPart& part, String::Body name, const char* file_name=0) {
545: *part.string << "--" << part.boundary << CRLF CONTENT_DISPOSITION_CAPITALIZED ": form-data; name=\"" << name << "\"";
1.22 misha 546: if(file_name){
547: if(strcmp(file_name, NONAME_DAT)!=0)
1.48 moko 548: *part.string << "; filename=\"" << file_name << "\"";
549: *part.string << CRLF HTTP_CONTENT_TYPE_CAPITALIZED ": " << part.r->mime_type_of(file_name);
1.22 misha 550: }
1.48 moko 551: *part.string << CRLF CRLF;
1.22 misha 552: }
553:
1.73 moko 554: static void form_string_value2part(HashStringValue::key_type key, const String& value, FormPart& part) {
1.28 misha 555: form_part_boundary_header(part, key);
1.48 moko 556: *part.string << value << CRLF;
1.22 misha 557: }
558:
1.73 moko 559: static void form_file_value2part(HashStringValue::key_type key, VFile& vfile, FormPart& part) {
1.28 misha 560: form_part_boundary_header(part, key, vfile.fields().get(name_name)->as_string().cstr());
1.48 moko 561: part.blocks+=FormPart::BinaryBlock(part.string, part.r);
562: part.blocks+=FormPart::BinaryBlock(vfile.value_ptr(), vfile.value_size());
563: part.string=new String();
564: *part.string << CRLF;
1.22 misha 565: }
566:
567: static void form_table_value2part(Table::element_type row, FormPart* part) {
568: form_string_value2part(part->info->key, *row->get(0), *part);
569: }
570:
1.136 moko 571: static bool form_element_value2part(HashStringValue::key_type key, HashStringValue::value_type value, FormPart& part) {
572: if(const String* svalue=value->get_string()){
573: form_string_value2part(key, *svalue, part);
574: } else if(VFile* vfile=dynamic_cast<VFile *>(value)){
575: form_file_value2part(key, *vfile, part);
576: } else {
577: return false;
578: }
579: return true;
580: }
581:
1.73 moko 582: static void form_value2part(HashStringValue::key_type key, HashStringValue::value_type value, FormPart& part) {
1.136 moko 583: if(form_element_value2part(key, value, part)) {
584: // string or file
585: } else if(Table* tvalue=value->get_table()) {
1.48 moko 586: Form_table_value2string_info info(key, *part.string);
1.22 misha 587: part.info = &info;
588: tvalue->for_each(form_table_value2part, &part);
1.136 moko 589: } else if(HashStringValue* hvalue=value->get_hash()){
590: // hash or array: several strings/files under the same field name
591: for(HashStringValue::Iterator j(*hvalue); j; j.next())
592: if(!form_element_value2part(key, j.value(), part))
593: throw Exception(PARSER_RUNTIME, new String(j.key(), String::L_TAINTED), "is %s, multiple values of " HTTP_FORM_NAME " field '%s' must be strings or files", j.value()->type(), key.cstr());
1.22 misha 594: } else
1.136 moko 595: throw Exception(PARSER_RUNTIME, new String(key, String::L_TAINTED), "is %s, " HTTP_FORM_NAME " field value can be string, table, file, hash or array", value->type());
1.22 misha 596: }
597:
598: const char* pa_form2string_multipart(HashStringValue& form, Request& r, const char* boundary, size_t& post_size){
1.48 moko 599: FormPart formpart(&r, boundary);
1.22 misha 600: form.for_each<FormPart&>(form_value2part, formpart);
1.48 moko 601: *formpart.string << "--" << boundary << "--";
602: // @todo: return binary blocks here to save memory in pa_internal_file_read_http
603: return formpart.post(post_size);
1.22 misha 604: }
605:
1.54 misha 606: // Set-Cookie: name=value; Domain=docs.foo.com; Path=/accounts; Expires=Wed, 13-Jan-2021 22:23:01 GMT; Secure; HttpOnly
607: static ArrayString* parse_cookie(Request& r, const String& cookie) {
1.64 moko 608: char *current=pa_strdup(cookie.cstr());
1.54 misha 609:
610: const String* name=0;
1.55 moko 611: const String* value=&String::Empty;
612: const String* expires=&String::Empty;
613: const String* max_age=&String::Empty;
614: const String* path=&String::Empty;
615: const String* domain=&String::Empty;
616: const String* httponly=&String::Empty;
617: const String* secure=&String::Empty;
1.54 misha 618:
619: bool first_pair=true;
620:
621: do {
622: if(char *meaning=search_stop(current, ';'))
623: if(char *attribute=search_stop(meaning, '=')) {
624: const String* sname=new String(unescape_chars(attribute, strlen(attribute), &r.charsets.source(), true/*don't convert '"' to space*/), String::L_TAINTED);
625: const String* smeaning=0;
626: if(meaning)
627: smeaning=new String(unescape_chars(meaning, strlen(meaning), &r.charsets.source(), true/*don't convert '"' to space*/), String::L_TAINTED);
628:
629: if(first_pair) {
630: // name + value
631: name=sname;
1.122 moko 632: if(smeaning)
633: value=smeaning;
1.54 misha 634: first_pair=false;
635: } else {
636: const String& slower=sname->change_case(r.charsets.source(), String::CC_LOWER);
637:
638: if(slower == "expires")
639: expires=smeaning;
640: else if(slower == "max-age")
641: max_age=smeaning;
642: else if(slower == "domain")
643: domain=smeaning;
644: else if(slower == "path")
645: path=smeaning;
646: else if(slower == "httponly")
647: httponly=new String("1", String::L_CLEAN);
648: else if(slower == "secure")
649: secure=new String("1", String::L_CLEAN);
650: else {
651: // todo@ ?
652: }
653: }
654: }
655: } while(current);
656:
657: if(!name)
658: return 0;
659:
660: ArrayString* result=new ArrayString(8);
661: *result+=name;
662: *result+=value;
663: *result+=expires;
664: *result+=max_age;
665: *result+=domain;
666: *result+=path;
667: *result+=httponly;
668: *result+=secure;
669:
670: return result;
671: }
672:
1.56 misha 673: Table* parse_cookies(Request& r, Table *cookies){
674: Table& result=*new Table(new Cookies_table_template_columns);
675:
1.125 moko 676: for(Array_iterator<Table::element_type> i(*cookies); i; )
1.56 misha 677: if(ArrayString* row=parse_cookie(r, *i.next()->get(0)))
678: result+=row;
679:
680: return &result;
681: }
682:
1.75 moko 683: void tables_update(HashStringValue& tables, const String::Body name, const String& value){
1.72 moko 684: Table *table;
685: if(Value *valready=tables.get(name)) {
686: // second+ appearence
687: table=valready->get_table();
688: } else {
689: // first appearence
690: Table::columns_type columns=new ArrayString(1);
691: *columns+=new String("value");
692: table=new Table(columns);
693: tables.put(name, new VTable(table));
694: }
695: // this string becomes next row
696: ArrayString& row=*new ArrayString(1);
697: row+=&value;
698: *table+=&row;
699: }
700:
1.1 paf 701: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.72 moko 702: 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 703: File_read_http_result result;
1.20 misha 704: char host[MAX_STRING];
1.66 moko 705: const char *idna_host;
1.1 paf 706: const char* uri;
1.49 moko 707: short port=80;
1.10 misha 708: const char* method="GET";
1.21 misha 709: bool method_is_get=true;
1.1 paf 710: HashStringValue* form=0;
711: int timeout_secs=2;
712: bool fail_on_status_ne_200=true;
1.12 misha 713: bool omit_post_charset=false;
1.1 paf 714: Value* vheaders=0;
1.10 misha 715: Value* vcookies=0;
1.11 misha 716: Value* vbody=0;
1.72 moko 717: Charset* asked_remote_charset=0;
1.58 moko 718: Charset* real_remote_charset=0;
1.1 paf 719: const char* user_cstr=0;
720: const char* password_cstr=0;
1.22 misha 721: const char* encode=0;
722: bool multipart=false;
1.1 paf 723:
724: if(options) {
725: int valid_options=pa_get_valid_file_options_count(*options);
726:
727: if(Value* vmethod=options->get(HTTP_METHOD_NAME)) {
728: valid_options++;
1.21 misha 729: method=vmethod->as_string().change_case(r.charsets.source(), String::CC_UPPER).cstr();
730: method_is_get=strcmp(method, "GET")==0;
1.1 paf 731: }
1.22 misha 732: if(Value* vencode=options->get(HTTP_FORM_ENCTYPE_NAME)) {
733: valid_options++;
734: encode=vencode->as_string().cstr();
735: }
1.1 paf 736: if(Value* vform=options->get(HTTP_FORM_NAME)) {
737: valid_options++;
738: form=vform->get_hash();
739: }
1.11 misha 740: if(vbody=options->get(HTTP_BODY_NAME)) {
1.1 paf 741: valid_options++;
742: }
743: if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) {
744: valid_options++;
745: timeout_secs=vtimeout->as_int();
746: }
1.11 misha 747: if(vheaders=options->get(HTTP_HEADERS_NAME)) {
1.1 paf 748: valid_options++;
749: }
1.11 misha 750: if(vcookies=options->get(HTTP_COOKIES_NAME)) {
1.10 misha 751: valid_options++;
752: }
1.1 paf 753: if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) {
754: valid_options++;
755: fail_on_status_ne_200=!vany_status->as_bool();
1.12 misha 756: }
1.20 misha 757: if(Value* vomit_post_charset=options->get(HTTP_OMIT_POST_CHARSET_NAME)){
1.12 misha 758: valid_options++;
759: omit_post_charset=vomit_post_charset->as_bool();
760: }
1.6 misha 761: if(Value* vcharset_name=options->get(PA_CHARSET_NAME)) {
1.77 moko 762: asked_remote_charset=&pa_charsets.get(vcharset_name->as_string());
1.58 moko 763: }
764: if(Value* vresponse_charset_name=options->get(PA_RESPONSE_CHARSET_NAME)) {
1.61 moko 765: valid_options++;
1.77 moko 766: real_remote_charset=&pa_charsets.get(vresponse_charset_name->as_string());
1.1 paf 767: }
768: if(Value* vuser=options->get(HTTP_USER)) {
769: valid_options++;
770: user_cstr=vuser->as_string().cstr();
771: }
772: if(Value* vpassword=options->get(HTTP_PASSWORD)) {
773: valid_options++;
774: password_cstr=vpassword->as_string().cstr();
775: }
776:
777: if(valid_options!=options->count())
1.46 misha 778: throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION);
1.1 paf 779: }
780: if(!asked_remote_charset) // defaulting to $request:charset
1.22 misha 781: asked_remote_charset=&(r.charsets).source();
782:
783: if(encode){
784: if(method_is_get)
1.127 moko 785: throw Exception(PARSER_RUNTIME, 0, "you cannot use $." HTTP_FORM_ENCTYPE_NAME " option with method GET");
1.22 misha 786:
787: multipart=strcasecmp(encode, HTTP_CONTENT_TYPE_MULTIPART_FORMDATA)==0;
788:
789: if(!multipart && strcasecmp(encode, HTTP_CONTENT_TYPE_FORM_URLENCODED)!=0)
1.72 moko 790: throw Exception(PARSER_RUNTIME, 0, "$." HTTP_FORM_ENCTYPE_NAME " option value can be " HTTP_CONTENT_TYPE_FORM_URLENCODED " or " HTTP_CONTENT_TYPE_MULTIPART_FORMDATA " only");
1.22 misha 791: }
1.1 paf 792:
1.11 misha 793: if(vbody){
794: if(method_is_get)
1.127 moko 795: throw Exception(PARSER_RUNTIME, 0, "you cannot use $." HTTP_BODY_NAME " option with method GET");
1.11 misha 796:
797: if(form)
1.127 moko 798: throw Exception(PARSER_RUNTIME, 0, "you cannot use options $." HTTP_BODY_NAME " and $." HTTP_FORM_NAME " together");
1.11 misha 799: }
1.1 paf 800:
801: //preparing request
1.29 misha 802: String& connect_string=*new String(file_spec);
1.1 paf 803:
1.48 moko 804: const char* request;
805: size_t request_size;
1.1 paf 806: {
807: // influence URLencoding of tainted pieces to String::L_URI lang
1.22 misha 808: Temp_client_charset temp(r.charsets, *asked_remote_charset);
1.1 paf 809:
1.44 misha 810: const char* connect_string_cstr=connect_string.untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.1 paf 811:
812: const char* current=connect_string_cstr;
813: if(strncmp(current, "http://", 7)!=0)
1.72 moko 814: throw Exception(PARSER_RUNTIME, &connect_string, "does not start with http://"); //never
1.1 paf 815: current+=7;
816:
1.119 moko 817: pa_strncpy(host, current, sizeof(host));
1.34 misha 818: char* host_uri=lsplit(host, '/');
819: uri=host_uri?current+(host_uri-1-host):"/";
820: char* port_cstr=lsplit(host, ':');
1.49 moko 821:
822: if (port_cstr){
823: char* error_pos=0;
824: port=(short)strtol(port_cstr, &error_pos, 10);
825: if(port==0 || *error_pos)
826: throw Exception(PARSER_RUNTIME, &connect_string, "invalid port number '%s'", port_cstr);
827: }
1.1 paf 828:
1.66 moko 829: idna_host=pa_idna_encode(host, r.charsets.source());
830:
1.11 misha 831: // making request head
1.1 paf 832: String head;
1.11 misha 833: head << method << " " << uri;
1.28 misha 834: if(method_is_get && form)
835: head << (strchr(uri, '?')!=0?"&":"?") << pa_form2string(*form, r.charsets);
1.11 misha 836:
1.66 moko 837: head <<" HTTP/1.0" CRLF "Host: "<< idna_host;
1.49 moko 838: if (port != 80)
839: head << ":" << port_cstr;
840: head << CRLF;
1.11 misha 841:
1.71 moko 842: char* boundary= multipart ? get_uuid_boundary() : 0;
1.22 misha 843:
1.35 misha 844: String user_headers;
845: bool user_agent_specified=false;
846: bool content_type_specified=false;
847: bool content_type_url_encoded=false;
848: if(vheaders && !vheaders->is_string()) { // allow empty
849: if(HashStringValue *headers=vheaders->get_hash()) {
850: Http_pass_header_info info={
851: &(r.charsets),
852: &user_headers,
853: &user_agent_specified,
854: &content_type_specified,
855: &content_type_url_encoded};
856: headers->for_each<Http_pass_header_info*>(http_pass_header, &info);
857: } else
1.72 moko 858: throw Exception(PARSER_RUNTIME, 0, "headers param must be hash");
1.35 misha 859: };
860:
1.48 moko 861: const char* request_body=0;
1.22 misha 862: size_t post_size=0;
863: if(form && !method_is_get) {
1.38 misha 864: head << "Content-Type: " << (multipart ? HTTP_CONTENT_TYPE_MULTIPART_FORMDATA : HTTP_CONTENT_TYPE_FORM_URLENCODED);
1.28 misha 865:
866: if(!omit_post_charset)
867: head << "; charset=" << asked_remote_charset->NAME_CSTR();
868:
1.22 misha 869: if(multipart) {
1.28 misha 870: head << "; boundary=" << boundary;
1.48 moko 871: request_body=pa_form2string_multipart(*form, r/*charsets & mime_type needed*/, boundary, post_size/*correct post_size returned here*/);
1.22 misha 872: } else {
1.48 moko 873: request_body=pa_form2string(*form, r.charsets);
874: post_size=strlen(request_body);
1.22 misha 875: }
1.28 misha 876: head << CRLF;
1.35 misha 877: } else if(vbody) {
1.38 misha 878: // $.body was specified
1.35 misha 879: if(content_type_url_encoded){
1.36 misha 880: // transcode + url-encode
1.48 moko 881: request_body=vbody->as_string().untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.35 misha 882: } else {
1.36 misha 883: // content-type != application/x-www-form-urlencoded -> transcode only, don't url-encode!
1.72 moko 884: const String &sbody=vbody->as_string();
885: request_body=Charset::transcode(String::C(sbody.cstr(), sbody.length()), r.charsets.source(), *asked_remote_charset).str;
1.35 misha 886: }
1.48 moko 887: post_size=strlen(request_body);
1.1 paf 888: }
889:
890: // http://www.ietf.org/rfc/rfc2617.txt
891: if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr))
1.38 misha 892: head << "Authorization: " << *authorization_field_value << CRLF;
1.1 paf 893:
1.35 misha 894: head << user_headers;
895:
1.1 paf 896: if(!user_agent_specified) // defaulting
1.38 misha 897: head << "User-Agent: " DEFAULT_USER_AGENT CRLF;
1.1 paf 898:
1.12 misha 899: if(form && !method_is_get && content_type_specified) // POST + form + content-type was specified
1.72 moko 900: throw Exception(PARSER_RUNTIME, 0, "$.content-type can't be specified with method POST");
1.12 misha 901:
1.11 misha 902: if(vcookies && !vcookies->is_string()){ // allow empty
1.10 misha 903: if(HashStringValue* cookies=vcookies->get_hash()) {
1.37 misha 904: head << "Cookie: ";
1.35 misha 905: Http_pass_header_info info={&(r.charsets), &head, 0, 0, 0};
1.10 misha 906: cookies->for_each<Http_pass_header_info*>(http_pass_cookie, &info);
907: head << CRLF;
908: } else
1.72 moko 909: throw Exception(PARSER_RUNTIME, 0, "cookies param must be hash");
1.10 misha 910: }
911:
1.48 moko 912: if(request_body)
1.128 moko 913: head << "Content-Length: " << pa_uitoa(post_size) << CRLF;
1.48 moko 914:
915: head << CRLF;
916:
917: const char *request_head=head.untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.1 paf 918:
1.48 moko 919: if(request_body){
920: size_t head_size = strlen(request_head);
921: request_size=post_size + head_size;
922: char *ptr=(char *)pa_malloc_atomic(request_size);
923: memcpy(ptr, request_head, head_size);
924: memcpy(ptr+head_size, request_body, post_size);
925: request=ptr;
926: } else {
927: request_size=strlen(request_head);
928: request=request_head;
929: }
1.1 paf 930: }
931:
1.78 moko 932:
1.97 moko 933: HTTP_response response;
1.22 misha 934:
1.28 misha 935: // sending request
1.95 moko 936: int status_code;
937: ALTER_EXCEPTION_SOURCE(status_code=http_request(response, idna_host, port, request, request_size, timeout_secs, fail_on_status_ne_200), &connect_string);
1.78 moko 938:
1.72 moko 939: // processing results
1.78 moko 940: char* raw_body=response.buf + response.body_offset;
941: size_t raw_body_size=response.length - response.body_offset;
942:
1.1 paf 943: result.headers=new HashStringValue;
944: VHash* vtables=new VHash;
1.133 moko 945: HASH_PUT_CSTR(*result.headers, "tables", vtables);
1.72 moko 946:
1.78 moko 947: if (!real_remote_charset && !response.headers.content_type.is_empty())
948: real_remote_charset=detect_charset(response.headers.content_type.cstr());
1.1 paf 949:
1.72 moko 950: if(as_text)
1.77 moko 951: real_remote_charset=pa_charsets.checkBOM(raw_body, raw_body_size, real_remote_charset);
1.72 moko 952:
953: if (!real_remote_charset)
954: real_remote_charset=asked_remote_charset; // never null
955:
1.125 moko 956: for(Array_iterator<HTTP_Headers::Header> i(response.headers.headers); i; ){
1.85 moko 957: HTTP_Headers::Header header=i.next();
1.72 moko 958:
959: header.transcode(*real_remote_charset, r.charsets.source());
960:
961: String &header_value=*new String(header.value, String::L_TAINTED);
962:
963: tables_update(vtables->hash(), header.name, header_value);
964: result.headers->put(header.name, new VString(header_value));
1.16 misha 965: }
966:
1.72 moko 967: // filling $.cookies
1.89 moko 968: if(vcookies=vtables->hash().get("SET-COOKIE"))
1.133 moko 969: HASH_PUT_CSTR(*result.headers, HTTP_COOKIES_NAME, new VTable(parse_cookies(r, vcookies->get_table())));
1.72 moko 970:
1.1 paf 971: // output response
972: String::C real_body=String::C(raw_body, raw_body_size);
1.16 misha 973:
974: 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 975: real_body=Charset::transcode(real_body, *real_remote_charset, r.charsets.source());
1.1 paf 976: }
977:
978: result.str=const_cast<char *>(real_body.str); // hacking a little
979: result.length=real_body.length;
1.16 misha 980:
1.22 misha 981: if(as_text && result.length)
982: fix_line_breaks(result.str, result.length);
983:
1.1 paf 984: result.headers->put(file_status_name, new VInt(status_code));
1.16 misha 985:
1.1 paf 986: return result;
987: }
1.84 moko 988:
989: /* ********************** httpd *************************** */
990:
1.100 moko 991: enum EscapeState {
992: Initial,
993: Default,
994: EscapeFirst,
995: EscapeSecond
996: };
997:
998: static bool check_uri(const char *uri){
999: EscapeState state=Initial;
1.120 moko 1000: uint escapedValue=0;
1.100 moko 1001:
1002: const char *pattern="/../";
1003: const char *pos=pattern;
1004:
1005: while(*uri){
1006: uchar c=(uchar)*(uri++);
1007: switch(state) {
1008: case Initial:
1009: if(c!='/')
1010: return false;
1011: state=Default;
1012: break;
1013: case Default:
1014: if(c=='%'){
1015: state=EscapeFirst;
1016: continue;
1017: }
1018: if(c=='?')
1019: return true;
1020: break;
1021: case EscapeFirst:
1022: if(isxdigit(c)){
1023: state=EscapeSecond;
1024: escapedValue=hex_value[c] << 4;
1025: continue;
1026: }
1027: return false;
1028: case EscapeSecond:
1029: if(isxdigit(c)){
1030: state=Default;
1.105 moko 1031: c=(uchar)(escapedValue + hex_value[c]);
1.100 moko 1032:
1033: // implementing Apache AllowEncodedSlashes Off just in case
1034: if(c=='/' || c=='\\')
1035: return false;
1036:
1037: break;
1038: }
1039: return false;
1040: }
1041:
1042: if(c==*pos || c=='\\' && *pos=='/'){
1043: if(!*(++pos))
1044: return false;
1045: } else {
1046: pos=pattern;
1047: }
1048: }
1049: return true;
1050: }
1051:
1.84 moko 1052: class HTTPD_request : public HTTP_response {
1053: public:
1054: const char *method;
1055: const char *uri;
1056:
1.97 moko 1057: HTTPD_request() : HTTP_response(), method(NULL), uri(NULL){};
1.84 moko 1058:
1.120 moko 1059: ssize_t pa_recv(SOCKET sockfd, char *buf, size_t len);
1.103 moko 1060:
1.120 moko 1061: bool read(SOCKET sock, size_t size){
1.103 moko 1062: if(length + size > buf_size)
1063: resize(buf_size * 2 + size);
1064: ssize_t received_size=pa_recv(sock, buf + length, size);
1065: if(received_size == 0)
1066: return false;
1067: if(received_size < 0) {
1068: if(int no = pa_socks_errno())
1.111 moko 1069: throw Exception("httpd.read", 0, "error receiving request: %s (%d)", pa_socks_strerr(no), no);
1.103 moko 1070: return false;
1071: }
1072: length+=received_size;
1073: buf[length]='\0';
1074: return true;
1075: }
1076:
1.84 moko 1077: const char *extract_method(char *method_line){
1078: char* uri_start = strchr(method_line, ' ');
1079:
1080: if(!uri_start || uri_start == method_line)
1081: return NULL;
1082:
1083: char* uri_end=strchr(uri_start+1, ' ');
1084:
1085: if(!uri_end || uri_end == uri_start+1)
1086: return NULL;
1087:
1088: uri=pa_strdup(uri_start+1, uri_end-uri_start-1);
1.100 moko 1089: if(!check_uri(uri))
1090: throw Exception("httpd.request", 0, "invalid uri '%s'", uri);
1091:
1.84 moko 1092: return str_upper(method_line, uri_start-method_line);
1093: }
1094:
1.103 moko 1095:
1.120 moko 1096: bool read_header(SOCKET);
1097: size_t read_post(SOCKET, char *, size_t);
1.84 moko 1098: };
1099:
1100: enum HTTPD_request_state {
1101: HTTPD_METHOD,
1102: HTTPD_HEADERS
1103: };
1104:
1.138 moko 1105: // timeout on "void" connection is normal
1106: #define PA_RECV_TIMEOUT \
1107: LOG(pa_log("httpd [%d] recv got %d sec timeout", sockfd, pa_httpd_timeout)); \
1108: if(length) \
1109: throw Exception("httpd.timeout", 0, "timeout occurred while receiving request"); \
1110: return 0;
1111:
1.120 moko 1112: ssize_t HTTPD_request::pa_recv(SOCKET sockfd, char *buffer, size_t len){
1.111 moko 1113: LOG(pa_log("httpd [%d] recv %d appending to %d ...", sockfd, len, length));
1.107 moko 1114:
1.138 moko 1115: #ifdef PA_HAVE_ALARM
1116: if(PA_USE_ALARM) signal(SIGALRM, timeout_handler);
1117: if(PA_USE_ALARM && sigsetjmp(timeout_env, 1)) {
1118: PA_RECV_TIMEOUT
1.103 moko 1119: } else
1.107 moko 1120: #endif
1.103 moko 1121: {
1.107 moko 1122: ALARM(pa_httpd_timeout);
1.141 moko 1123: ssize_t result=pa_sock_recv(sockfd, buffer, len);
1.107 moko 1124: ALARM(0);
1.138 moko 1125: if(result < 0 && sock_timed_out(pa_socks_errno())) {
1126: PA_RECV_TIMEOUT
1127: }
1.111 moko 1128: LOG(pa_log("httpd [%d] recv got %d bytes", sockfd, result));
1.112 moko 1129: LOG(pa_log("httpd [%d] %s", sockfd, buffer));
1.103 moko 1130: return result;
1131: }
1132: }
1133:
1.115 moko 1134: static bool valid_http_method(const char * method){
1135: return method && (
1136: !strcmp(method, "GET") ||
1137: !strcmp(method, "HEAD") ||
1138: !strcmp(method, "POST") ||
1139: !strcmp(method, "PUT") ||
1140: !strcmp(method, "DELETE") ||
1141: !strcmp(method, "CONNECT") ||
1142: !strcmp(method, "OPTIONS") ||
1143: !strcmp(method, "TRACE") ||
1144: !strcmp(method, "PATCH")
1145: );
1146: }
1147:
1.120 moko 1148: bool HTTPD_request::read_header(SOCKET sock) {
1.84 moko 1149: enum HTTPD_request_state state = HTTPD_METHOD;
1150:
1151: size_t chunk_size = 0x400*4;
1152: resize(chunk_size);
1153:
1154: while(read(sock, chunk_size)){
1155: switch(state){
1156: case HTTPD_METHOD: {
1157: size_t method_size = first_line();
1158: if(!method_size)
1159: break;
1160:
1161: char *method_line = pa_strdup(buf, method_size);
1162: method = extract_method(method_line);
1163:
1.115 moko 1164: if(!valid_http_method(method))
1.84 moko 1165: throw Exception("httpd.method", new String(method ? method : method_line), "invalid request method");
1166: state = HTTPD_HEADERS;
1167: }
1168:
1169: case HTTPD_HEADERS: {
1170: if(!body_start())
1171: break;
1172:
1173: parse_headers();
1.144 moko 1174:
1175: if(length==body_offset){ // nothing but the headers (GET, websocket handshake), the buffer is no longer needed
1176: buf=NULL;
1177: length=body_offset=buf_size=0;
1178: }
1.110 moko 1179: return true;
1.84 moko 1180: }
1181: }
1182: }
1183:
1.111 moko 1184: if(!length){ // browsers open connections in advance and they will be empty unless user requests more pages
1185: LOG(pa_log("httpd [%d] void request", sock));
1.110 moko 1186: return false;
1.111 moko 1187: }
1.110 moko 1188:
1.84 moko 1189: if(state == HTTPD_METHOD)
1190: throw Exception("httpd.request", 0, "bad request from host - no method found (size=%u)", length);
1191:
1192: if(state == HTTPD_HEADERS){
1193: parse_headers();
1194: body_offset=length;
1195: }
1.110 moko 1196:
1197: return true;
1.84 moko 1198: }
1199:
1.120 moko 1200: size_t HTTPD_request::read_post(SOCKET sock, char *body, size_t max_bytes) {
1.87 moko 1201: size_t total_read = min(length - body_offset, max_bytes);
1.144 moko 1202: if(total_read)
1203: memcpy(body, buf + body_offset, total_read);
1.87 moko 1204:
1205: while (total_read < max_bytes){
1.103 moko 1206: ssize_t received_size = pa_recv(sock, body + total_read, max_bytes - total_read);
1.87 moko 1207: if(received_size == 0)
1208: return total_read;
1209: if(received_size < 0) {
1210: if(int no = pa_socks_errno())
1.111 moko 1211: throw Exception("httpd.read", new String(uri), "error receiving request body: %s (%d)", pa_socks_strerr(no), no);
1.87 moko 1212: return total_read;
1213: }
1214: total_read += received_size;
1215: }
1216: return total_read;
1217: }
1218:
1.84 moko 1219: /* ********************************************************** */
1220:
1.85 moko 1221: Array<HTTP_Headers::Header> &HTTPD_Connection::headers() {
1.84 moko 1222: return request->headers.headers;
1223: }
1224:
1225: const char *HTTPD_Connection::method() {
1226: return request->method;
1227: }
1228:
1229: const char *HTTPD_Connection::uri() {
1230: return request->uri;
1231: }
1232:
1233: const char *HTTPD_Connection::content_type() {
1234: return request->headers.content_type.cstr();
1235: }
1236:
1237: uint64_t HTTPD_Connection::content_length(){
1238: return request->headers.content_length;
1239: }
1240:
1.110 moko 1241: bool HTTPD_Connection::read_header(){
1.84 moko 1242: request = new HTTPD_request();
1.111 moko 1243: bool result = request->read_header(sock);
1244: LOG(if(result){
1245: pa_log("httpd [%d] got %s \"%s\"", sock, method(), uri());
1246: })
1247: return result;
1.84 moko 1248: }
1249:
1.87 moko 1250: size_t HTTPD_Connection::read_post(char *body, size_t max_bytes) {
1251: return request->read_post(sock, body, max_bytes);
1252: }
1253:
1.90 moko 1254: size_t HTTPD_Connection::send_body(const void *buf, size_t size) {
1.112 moko 1255: LOG(pa_log("httpd [%d] response %d bytes", sock, size));
1256: LOG(pa_log("httpd [%d] %s", sock, buf));
1.131 moko 1257: ssize_t result=pa_send(sock, (const char*)buf, size);
1258: if(result < 0) {
1.90 moko 1259: int no=pa_socks_errno();
1.111 moko 1260: throw Exception("httpd.write", 0, "error sending response: %s (%d)", pa_socks_strerr(no), no);
1.90 moko 1261: }
1.131 moko 1262: return result;
1.90 moko 1263: }
1264:
1.140 moko 1265: String::C HTTPD_Connection::unprocessed() {
1266: return String::C(request->buf + request->body_offset, request->length - request->body_offset);
1267: }
1268:
1269: size_t HTTPD_Connection::receive(char *dest, size_t max_bytes) {
1.141 moko 1270: ssize_t received_size=pa_sock_recv(sock, dest, max_bytes);
1.140 moko 1271: if(received_size < 0) {
1272: int no=pa_socks_errno();
1273: if(pa_socks_is_reset(no))
1274: return 0;
1275: throw Exception("httpd.read", 0, "error receiving data: %s (%d)", pa_socks_strerr(no), no);
1276: }
1277: return received_size;
1278: }
1279:
1280: void HTTPD_Connection::set_nodelay() {
1281: int nodelay=1;
1282: setsockopt(sock, IPPROTO_TCP, TCP_NODELAY, (const char *)&nodelay, sizeof(nodelay));
1283: }
1284:
1285: void HTTPD_Connection::set_send_timeout(int timeout_secs) {
1286: set_socket_timeout(sock, SO_SNDTIMEO, timeout_secs);
1287: }
1288:
1.93 moko 1289: HTTPD_Connection::~HTTPD_Connection(){
1.120 moko 1290: if(sock != INVALID_SOCKET){
1.111 moko 1291: LOG(pa_log("httpd [%d] closed", sock));
1.93 moko 1292: closesocket(sock);
1.145 moko 1293: HTTPD_Server::track_connection(-1);
1.111 moko 1294: }
1.93 moko 1295: }
1296:
1.120 moko 1297: bool HTTPD_Connection::accept(SOCKET server_sock, int timeout_value) {
1.139 moko 1298: int ready = pa_sock_ready(server_sock, false, timeout_value);
1.93 moko 1299: if (ready < 0) {
1300: int no=pa_socks_errno();
1301: throw Exception("httpd.accept", 0, "error waiting for connection: %s (%d)", pa_socks_strerr(no), no);
1302: }
1303: if (ready == 0)
1304: return false; /* Timeout */
1305:
1306: struct sockaddr_in addr;
1307: socklen_t sock_addr_len = sizeof(struct sockaddr_in);
1308: memset(&addr, 0, sock_addr_len);
1309:
1.146 ! moko 1310: sock = PA_ACCEPT(server_sock, (struct sockaddr *)&addr, &sock_addr_len);
1.120 moko 1311: if(sock == INVALID_SOCKET){
1.93 moko 1312: int no=pa_socks_errno();
1313: throw Exception("httpd.accept", 0, "error accepting connection: %s (%d)", pa_socks_strerr(no), no);
1314: }
1.145 moko 1315: HTTPD_Server::track_connection(1); // taken back by the owner
1.93 moko 1316:
1.123 moko 1317: // Has no positive performance effect, requires include <netinet/tcp.h>
1318: // static int sock_on=1;
1319: // setsockopt(sock, IPPROTO_TCP, TCP_NODELAY, (char *)&sock_on, sizeof(sock_on));
1320:
1.138 moko 1321: if(!PA_USE_ALARM)
1322: set_socket_timeout(sock, SO_RCVTIMEO, pa_httpd_timeout);
1323:
1.93 moko 1324: remote_addr = pa_strdup(inet_ntoa(addr.sin_addr));
1.111 moko 1325: LOG(pa_log("httpd [%d] accepted from %s", sock, remote_addr));
1.93 moko 1326: return true;
1327: }
1.84 moko 1328:
1.107 moko 1329: HTTPD_Server::HTTPD_MODE HTTPD_Server::mode = HTTPD_Server::SEQUENTIAL;
1.114 moko 1330: const char *HTTPD_Server::port=NULL;
1.145 moko 1331: int HTTPD_Server::max_connections=INT_MAX;
1332: int HTTPD_Server::connections=0;
1333:
1334: void HTTPD_Server::track_connection(int delta){
1335: SYNCHRONIZED;
1336: connections+=delta;
1337: }
1.106 moko 1338:
1.108 moko 1339: void HTTPD_Server::set_mode(const String &value){
1340: if(value == "sequental") mode = SEQUENTIAL;
1.113 moko 1341: #ifdef HAVE_TLS
1.108 moko 1342: else if (value == "threaded") mode = MULTITHREADED;
1.113 moko 1343: #endif
1.108 moko 1344: #ifdef _MSC_VER
1.117 moko 1345: else throw Exception("httpd.mode", &value, "$MAIN:HTTPD.mode must be 'sequental' or 'threaded'");
1.108 moko 1346: #else
1347: else if (value == "parallel") mode = PARALLEL;
1.117 moko 1348: else throw Exception("httpd.mode", &value, "$MAIN:HTTPD.mode must be 'sequental', 'parallel' or 'threaded'");
1.108 moko 1349: #endif
1350: }
1351:
1.120 moko 1352: SOCKET HTTPD_Server::bind(const char *host_port){
1.84 moko 1353: struct sockaddr_in me;
1354:
1.114 moko 1355: port = strchr(host_port, ':');
1.86 moko 1356: const char *host = NULL;
1.116 moko 1357: if(port){
1.114 moko 1358: if(port > host_port)
1359: host = pa_strdup(host_port, port - host_port);
1.86 moko 1360: port += 1;
1361: } else {
1362: port = host_port;
1363: }
1364:
1.105 moko 1365: if(!set_addr(&me, host, (short)pa_atoui(port))){
1.84 moko 1366: if (host)
1.127 moko 1367: throw Exception("httpd.bind", 0, "cannot resolve hostname \"%s\"", host);
1.84 moko 1368: me.sin_addr.s_addr=INADDR_ANY;
1369: }
1370:
1.146 ! moko 1371: SOCKET sock = PA_SOCKET(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/);
1.84 moko 1372:
1.120 moko 1373: if(sock == INVALID_SOCKET){
1.84 moko 1374: int no=pa_socks_errno();
1.127 moko 1375: throw Exception("httpd.bind", 0, "cannot make socket: %s (%d)", pa_socks_strerr(no), no);
1.84 moko 1376: }
1377:
1.93 moko 1378: static int sock_on = 1;
1379:
1.84 moko 1380: if (setsockopt(sock, SOL_SOCKET, SO_REUSEADDR, (char *)&sock_on, sizeof(sock_on)) ||
1381: setsockopt(sock, SOL_SOCKET, SO_KEEPALIVE, (char *)&sock_on, sizeof(sock_on)) ||
1382: ::bind(sock, (struct sockaddr*)&me, sizeof(me)) ||
1.143 moko 1383: listen(sock, 511)) { // same default as in apache/nginx
1.89 moko 1384: closesocket(sock);
1.84 moko 1385: int no = pa_socks_errno();
1.127 moko 1386: throw Exception("httpd.bind", 0, "cannot bind socket: %s (%d)", pa_socks_strerr(no), no);
1.84 moko 1387: }
1388: return sock;
1389: }
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