Annotation of parser3/src/main/pa_http.C, revision 1.148
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:
1.148 ! moko 21: volatile const char * IDENT_PA_HTTP_C="$Id: pa_http.C,v 1.147 2026/10/29 19:51:53 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
1.148 ! moko 49: #ifdef FD_CLOEXEC
! 50: // FreeBSD < 10: a program executed by another thread between the call and fcntl still inherits the socket
! 51: static SOCKET pa_cloexec(SOCKET sock) {
! 52: if(sock != INVALID_SOCKET)
! 53: fcntl(sock, F_SETFD, FD_CLOEXEC);
! 54: return sock;
! 55: }
! 56: #define PA_ACCEPT(sock, addr, len) pa_cloexec(::accept(sock, addr, len))
! 57: #define PA_SOCKET(domain, type, protocol) pa_cloexec(::socket(domain, type, protocol))
! 58: #else
1.146 moko 59: #define PA_ACCEPT(sock, addr, len) ::accept(sock, addr, len)
60: #define PA_SOCKET(domain, type, protocol) ::socket(domain, type, protocol)
61: #endif
1.148 ! moko 62: #endif
1.146 moko 63:
1.1 paf 64: #undef CRLF
65: #define CRLF "\r\n"
66:
1.138 moko 67: #define HTTP_TIMEOUT_COMMENT "timeout occurred while retrieving document"
68:
1.54 misha 69: // helpers
1.56 misha 70:
1.130 moko 71: #ifdef HTTPD_DEBUG
72: void pa_log(const char* fmt, ...);
73: #define LOG(action) action
74: #else
75: #define LOG(action)
76: #endif
77:
78: ssize_t pa_send(int sock, const char *buffer, size_t len){
79: size_t total_sent = 0;
80: while (total_sent < len) {
1.141 moko 81: ssize_t bytes_sent=pa_sock_send(sock, buffer + total_sent, len - total_sent);
1.130 moko 82: if (bytes_sent < 0) {
83: return bytes_sent;
84: } else if (bytes_sent == 0) {
85: // Connection closed by the remote peer?
86: break;
87: }
88: LOG(
89: if(bytes_sent != len - total_sent)
90: pa_log("httpd [%d] partial send %d of (%d)", sock, bytes_sent, len - total_sent)
91: );
92: total_sent += bytes_sent;
93: }
94: return total_sent;
95: }
96:
1.138 moko 97: // socket operation failed on SO_*TIMEO timeout
98: static bool sock_timed_out(int no){
99: #ifdef WIN32
100: return no==WSAETIMEDOUT;
101: #else
102: return no==EAGAIN || no==EWOULDBLOCK || no==EINPROGRESS /* connect on linux */;
103: #endif
104: }
105:
106: static int check_socket_timeout(int no){
107: if(sock_timed_out(no))
108: throw Exception("http.timeout", 0, HTTP_TIMEOUT_COMMENT);
109: return no;
110: }
111:
1.85 moko 112: bool HTTP_Headers::add_header(const char *line){
1.78 moko 113: const char *value=strchr(line, ':');
114:
115: if(value && value != line){ // we need only headers, not the response code
1.134 moko 116: Header header = Header(String::Body(str_upper(line, value-line)), String::Body(value+1).trim(String::TRIM_BOTH, " \t\n\r"));
1.78 moko 117:
118: if(header.name == String::Body(HTTP_CONTENT_TYPE_UPPER) && content_type.is_empty())
119: content_type=header.value;
120:
121: if(header.name == String::Body("CONTENT-LENGTH") && content_length==0)
1.95 moko 122: ALTER_EXCEPTION_COMMENT(content_length=pa_atoul(header.value.cstr()), " for content-length");
1.78 moko 123:
124: headers+=header;
125:
126: return true;
127: }
128: return false;
129: }
130:
1.54 misha 131: class Cookies_table_template_columns: public ArrayString {
132: public:
133: Cookies_table_template_columns() {
134: *this+=new String("name");
135: *this+=new String("value");
136: *this+=new String("expires");
137: *this+=new String("max-age");
138: *this+=new String("domain");
139: *this+=new String("path");
140: *this+=new String("httponly");
141: *this+=new String("secure");
142: }
143: };
144:
145:
1.1 paf 146: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
1.22 misha 147: memset(addr, 0, sizeof(*addr));
148: addr->sin_family=AF_INET;
149: addr->sin_port=htons(port);
150: if(host) {
1.65 moko 151: struct hostent *hostIP=gethostbyname(host);
152: if(hostIP && hostIP->h_addrtype == AF_INET){
153: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
154: return true;
155: }
156: }
157: return false;
1.1 paf 158: }
159:
1.84 moko 160: class HTTP_response : public PA_Allocated {
1.78 moko 161: public:
162: char *buf;
163: size_t length;
164: size_t buf_size;
165: size_t body_offset;
166:
1.85 moko 167: HTTP_Headers headers;
1.78 moko 168:
1.97 moko 169: HTTP_response() : buf(NULL), length(0), buf_size(0), body_offset(0){}
1.78 moko 170:
171: void resize(size_t size){
172: buf_size=size;
173: buf=(char *)pa_realloc(buf, size + 1);
174: }
175:
1.120 moko 176: bool read(SOCKET sock, size_t size){
1.103 moko 177: if(length + size > buf_size)
178: resize(buf_size * 2 + size);
1.141 moko 179: ssize_t received_size=pa_sock_recv(sock, buf + length, size);
1.103 moko 180: if(received_size == 0)
1.78 moko 181: return false;
1.103 moko 182: if(received_size < 0) {
1.138 moko 183: if(int no = check_socket_timeout(pa_socks_errno()))
1.102 moko 184: throw Exception("http.timeout", 0, "error receiving response: %s (%d)", pa_socks_strerr(no), no);
1.78 moko 185: return false;
186: }
187: length+=received_size;
188: buf[length]='\0';
189: return true;
190: }
191:
1.83 moko 192: size_t first_line(){
1.89 moko 193: char *header=strchr(buf, '\n');
194: if(!header)
1.78 moko 195: return false;
196:
1.89 moko 197: return header-buf;
1.78 moko 198: }
199:
200: const char *status_code(char *status_line, int &result){
201: char* status_start = strchr(status_line, ' ');
202:
203: if(!(status_start++))
204: return status_line;
205:
206: char* status_end=strchr(status_start, ' ');
207:
208: if(!status_end)
209: return status_line;
210:
211: if(status_end==status_start)
212: return status_line;
1.1 paf 213:
1.78 moko 214: const char *result_str=pa_strdup(status_start, status_end-status_start);
1.95 moko 215: ALTER_EXCEPTION_COMMENT(result=pa_atoui(result_str), " for HTTP status");
1.78 moko 216: return result_str;
217: }
1.2 paf 218:
1.78 moko 219: bool body_start(){
220: char *p=buf;
221: while((p=strchr(p, '\n'))) {
222: if(p[1]=='\r' && p[2]=='\n'){ // \r\n\r\n
223: *p='\0';
224: body_offset=p-buf+3;
225: return true;
226: }
227: if(p[1]=='\n') { // \n\n
228: *p='\0';
229: body_offset=p-buf+2;
230: return true;
231: }
232: p++;
233: }
234: return false;
1.2 paf 235: }
1.78 moko 236:
237: void parse_headers(){
238: const String header_block(buf, String::L_TAINTED);
239:
240: ArrayString aheaders;
241: header_block.split(aheaders, 0, "\n");
242:
1.124 moko 243: ArrayString::Iterator i(aheaders);
1.78 moko 244: i.next(); // skipping status
1.125 moko 245: for(;i;){
1.78 moko 246: const char *line=i.next()->cstr();
247: if(!headers.add_header(line))
1.97 moko 248: throw Exception("http.response", 0, "bad response from host - bad header \"%s\"", line);
1.78 moko 249: }
1.1 paf 250: }
251:
1.120 moko 252: int read_response(SOCKET sock, bool fail_on_status_ne_200);
1.78 moko 253: };
254:
255: enum HTTP_response_state {
256: HTTP_STATUS_CODE,
257: HTTP_HEADERS,
258: HTTP_BODY
259: };
260:
1.120 moko 261: int HTTP_response::read_response(SOCKET sock, bool fail_on_status_ne_200) {
1.78 moko 262: HTTP_response_state state=HTTP_STATUS_CODE;
263: int result=0;
264:
265: size_t chunk_size=0x400*16;
1.88 moko 266: resize(2*chunk_size);
1.78 moko 267:
1.88 moko 268: while(read(sock, chunk_size)){
1.78 moko 269: switch(state){
270: case HTTP_STATUS_CODE: {
1.88 moko 271: size_t status_size=first_line();
1.78 moko 272: if(!status_size)
273: break;
274:
1.88 moko 275: const char *status=status_code(pa_strdup(buf, status_size), result);
1.78 moko 276:
277: if(!result || fail_on_status_ne_200 && result!=200)
278: throw Exception("http.status", status ? new String(status) : &String::Empty, "invalid HTTP response status");
279:
280: state=HTTP_HEADERS;
281: }
282:
283: case HTTP_HEADERS: {
1.88 moko 284: if(!body_start())
1.78 moko 285: break;
286:
1.88 moko 287: parse_headers();
1.78 moko 288:
1.97 moko 289: size_t content_length=check_file_size(headers.content_length, 0);
1.88 moko 290: if(content_length>0 && (content_length + body_offset) > length){
291: resize(content_length + body_offset + 0x400*64);
1.78 moko 292: }
293:
294: state=HTTP_BODY;
1.1 paf 295: break;
296: }
1.78 moko 297:
298: case HTTP_BODY: {
299: chunk_size=0x400*64;
1.1 paf 300: break;
301: }
302: }
303: }
1.78 moko 304:
305: if(state==HTTP_STATUS_CODE)
1.97 moko 306: throw Exception("http.response", 0, "bad response from host - no status found (size=%u)", length);
1.78 moko 307:
308: if(state==HTTP_HEADERS){
1.88 moko 309: parse_headers();
310: body_offset=length;
1.1 paf 311: }
1.78 moko 312:
313: return result;
1.1 paf 314: }
315:
316: /* ********************** request *************************** */
317:
318: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
1.138 moko 319: # define PA_HAVE_ALARM
1.1 paf 320: #endif
321:
1.138 moko 322: #ifdef PA_HAVE_ALARM
1.1 paf 323: static sigjmp_buf timeout_env;
324: static void timeout_handler(int /*sig*/){
1.101 moko 325: siglongjmp(timeout_env, 1);
1.1 paf 326: }
1.118 moko 327:
1.138 moko 328: // alarm is per process, threads use SO_*TIMEO
329: #define PA_USE_ALARM (HTTPD_Server::mode != HTTPD_Server::MULTITHREADED)
1.118 moko 330:
1.138 moko 331: #define ALARM(value) if(PA_USE_ALARM) alarm(value)
1.101 moko 332: #else
1.138 moko 333: #define PA_USE_ALARM false
1.101 moko 334: #define ALARM(value)
1.1 paf 335: #endif
336:
1.138 moko 337: // SO_*TIMEO can be defined in .h but not implemented in protocol, so errors are ignored
338: static void set_socket_timeout(SOCKET sock, int option, int timeout_secs){
339: #ifdef WIN32
340: int timeout=timeout_secs < INT_MAX/1000 ? timeout_secs*1000 : 0 /* infinite */;
341: #else
342: struct timeval timeout={timeout_secs, 0};
343: #endif
344: setsockopt(sock, SOL_SOCKET, option, (const char*)&timeout, sizeof(timeout));
345: }
346:
1.78 moko 347: 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 348: if(!host)
1.73 moko 349: throw Exception("http.host", 0, "zero hostname"); //never
1.1 paf 350:
1.120 moko 351: volatile SOCKET sock=INVALID_SOCKET; // to prevent makeing it register variable, because it will be clobbered by longjmp [thanks gcc warning]
1.101 moko 352:
1.138 moko 353: #ifdef PA_HAVE_ALARM
354: if(PA_USE_ALARM) signal(SIGALRM, timeout_handler);
355: if(PA_USE_ALARM && sigsetjmp(timeout_env, 1)) {
1.101 moko 356: // duplicating closesocket to make code more simple for old compilers
1.120 moko 357: if(sock != INVALID_SOCKET)
1.101 moko 358: closesocket(sock);
1.138 moko 359: throw Exception("http.timeout", 0, HTTP_TIMEOUT_COMMENT);
1.1 paf 360: return 0; // never
1.101 moko 361: } else
1.1 paf 362: #endif
1.101 moko 363: {
364: ALARM(timeout_secs);
1.1 paf 365: try {
366: int result;
367: struct sockaddr_in dest;
368:
369: if(!set_addr(&dest, host, port))
1.127 moko 370: throw Exception("http.host", 0, "cannot resolve hostname \"%s\"", host);
1.1 paf 371:
1.120 moko 372: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/)) == INVALID_SOCKET) {
1.1 paf 373: int no=pa_socks_errno();
1.127 moko 374: throw Exception("http.connect", 0, "cannot make socket: %s (%d)", pa_socks_strerr(no), no);
1.1 paf 375: }
376:
377: // To enable SO_DONTLINGER (that is, disable SO_LINGER)
378: // l_onoff should be set to zero and setsockopt should be called
379: linger dont_linger={0,0};
380: setsockopt(sock, SOL_SOCKET, SO_LINGER, (const char *)&dont_linger, sizeof(dont_linger));
381:
1.138 moko 382: if(!PA_USE_ALARM){
383: set_socket_timeout(sock, SO_SNDTIMEO, timeout_secs);
384: set_socket_timeout(sock, SO_RCVTIMEO, timeout_secs);
385: }
1.1 paf 386:
1.147 moko 387: if(pa_sock_connect(sock, (struct sockaddr *)&dest, sizeof(dest))) {
1.138 moko 388: int no=check_socket_timeout(pa_socks_errno());
1.127 moko 389: throw Exception("http.connect", 0, "cannot connect to host \"%s\": %s (%d)", host, pa_socks_strerr(no), no);
1.1 paf 390: }
1.22 misha 391:
1.130 moko 392: if(pa_send(sock, request, request_size) < 0) {
1.138 moko 393: int no=check_socket_timeout(pa_socks_errno());
1.78 moko 394: throw Exception("http.timeout", 0, "error sending request: %s (%d)", pa_socks_strerr(no), no);
1.1 paf 395: }
396:
1.88 moko 397: result=response.read_response(sock, fail_on_status_ne_200);
1.78 moko 398: closesocket(sock);
1.101 moko 399: ALARM(0);
1.1 paf 400: return result;
401: } catch(...) {
1.101 moko 402: ALARM(0);
1.120 moko 403: if(sock != INVALID_SOCKET)
1.78 moko 404: closesocket(sock);
1.1 paf 405: rethrow;
406: }
407: }
408: }
409:
410: #ifndef DOXYGEN
411: struct Http_pass_header_info {
412: Request_charsets* charsets;
413: String* request;
1.35 misha 414: bool* user_agent_specified;
415: bool* content_type_specified;
416: bool* content_type_url_encoded;
1.1 paf 417: };
418: #endif
1.50 moko 419:
420: char *pa_http_safe_header_name(const char *name) {
421: char *result=pa_strdup(name);
422: char *n=result;
1.52 misha 423: if(!pa_isalpha((unsigned char)*n))
1.50 moko 424: *n++ = '_';
425: for(; *n; ++n) {
1.52 misha 426: if (!pa_isalnum((unsigned char)*n) && *n != '-' && *n != '_')
1.50 moko 427: *n = '_';
428: }
429: return result;
430: }
431:
1.101 moko 432: static void http_pass_header(HashStringValue::key_type aname, HashStringValue::value_type avalue, Http_pass_header_info *info) {
1.9 misha 433:
1.41 misha 434: const char* name_cstr=aname.cstr();
435:
1.38 misha 436: if(strcasecmp(name_cstr, HTTP_CONTENT_LENGTH)==0)
437: return;
438:
1.50 moko 439: String name=String(pa_http_safe_header_name(capitalize(name_cstr)), String::L_AS_IS);
440: String value=attributed_meaning_to_string(*avalue, String::L_HTTP_HEADER, true);
1.9 misha 441:
1.35 misha 442: *info->request << name << ": " << value << CRLF;
1.1 paf 443:
1.38 misha 444: if(strcasecmp(name_cstr, HTTP_USER_AGENT)==0)
1.35 misha 445: *info->user_agent_specified=true;
1.38 misha 446: if(strcasecmp(name_cstr, HTTP_CONTENT_TYPE)==0){
1.35 misha 447: *info->content_type_specified=true;
1.62 moko 448: *info->content_type_url_encoded=pa_strncasecmp(value.cstr(), HTTP_CONTENT_TYPE_FORM_URLENCODED)==0;
1.35 misha 449: }
1.1 paf 450: }
451:
1.10 misha 452: static void http_pass_cookie(HashStringValue::key_type name,
1.20 misha 453: HashStringValue::value_type value,
454: Http_pass_header_info *info) {
1.10 misha 455:
1.17 misha 456: *info->request << String(name, String::L_HTTP_COOKIE) << "="
1.31 misha 457: << attributed_meaning_to_string(*value, String::L_HTTP_COOKIE, true)
1.10 misha 458: << "; ";
459:
460: }
1.1 paf 461:
462: static const String* basic_authorization_field(const char* user, const char* pass) {
463: if(!user&& !pass)
464: return 0;
465:
466: String combined;
467: if(user)
468: combined<<user;
469: combined<<":";
470: if(pass)
471: combined<<pass;
472:
1.20 misha 473: String* result=new String("Basic ");
1.82 moko 474: *result<<pa_base64_encode(combined.cstr(), combined.length(), Base64Options(false /*no wrap*/));
1.1 paf 475: return result;
476: }
477:
1.73 moko 478: static void form_string_value2string(HashStringValue::key_type key, const String& value, String& result) {
1.30 misha 479: result << String(key, String::L_URI) << "=" << String(value, String::L_URI) << "&";
1.1 paf 480: }
1.20 misha 481:
1.1 paf 482: #ifndef DOXYGEN
483: struct Form_table_value2string_info {
484: HashStringValue::key_type key;
485: String& result;
486:
487: Form_table_value2string_info(HashStringValue::key_type akey, String& aresult):
488: key(akey), result(aresult) {}
489: };
490: #endif
491: static void form_table_value2string(Table::element_type row, Form_table_value2string_info* info) {
492: form_string_value2string(info->key, *row->get(0), info->result);
493: }
1.73 moko 494:
495: static void form_value2string(HashStringValue::key_type key, HashStringValue::value_type value, String* result) {
1.1 paf 496: if(const String* svalue=value->get_string())
497: form_string_value2string(key, *svalue, *result);
498: else if(Table* tvalue=value->get_table()) {
499: Form_table_value2string_info info(key, *result);
500: tvalue->for_each(form_table_value2string, &info);
501: } else
1.73 moko 502: throw Exception(PARSER_RUNTIME, new String(key, String::L_TAINTED),
1.63 moko 503: "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 504: }
1.20 misha 505:
1.5 misha 506: const char* pa_form2string(HashStringValue& form, Request_charsets& charsets) {
1.1 paf 507: String string;
1.3 paf 508: form.for_each<String*>(form_value2string, &string);
1.44 misha 509: return string.untaint_and_transcode_cstr(String::L_URI, &charsets);
1.1 paf 510: }
1.22 misha 511:
512: struct FormPart {
513: Request* r;
514: const char* boundary;
1.48 moko 515: String* string;
1.22 misha 516: Form_table_value2string_info* info;
1.48 moko 517:
518: struct BinaryBlock{
519: const char* ptr;
520: size_t length;
521:
522: BinaryBlock(String* astring, Request* r): ptr(astring->untaint_and_transcode_cstr(String::L_AS_IS, &r->charsets)), length(strlen(ptr)){}
523: BinaryBlock(const char* aptr, size_t alength): ptr(aptr), length(alength){}
524: };
525:
526: Array<BinaryBlock> blocks;
527:
528: FormPart(Request* ar, const char* aboundary): r(ar), boundary(aboundary), string(new String()){}
529:
530: const char *post(size_t &length){
531: if(blocks.count()){
532: blocks+=BinaryBlock(string, r);
533:
534: length=0;
535: for(size_t i=0; i<blocks.count(); i++)
536: length+=blocks[i].length;
537:
538: char *result=(char *)pa_malloc_atomic(length);
539: char *ptr=result;
540:
541: for(size_t i=0; i<blocks.count(); i++){
542: memcpy(ptr, blocks[i].ptr, blocks[i].length);
543: ptr+=blocks[i].length;
544: }
545:
546: return result;
547: } else {
548: BinaryBlock result(string, r);
549: length=result.length;
550: return result.ptr;
551: }
552: }
1.22 misha 553: };
554:
1.73 moko 555: static void form_part_boundary_header(FormPart& part, String::Body name, const char* file_name=0) {
556: *part.string << "--" << part.boundary << CRLF CONTENT_DISPOSITION_CAPITALIZED ": form-data; name=\"" << name << "\"";
1.22 misha 557: if(file_name){
558: if(strcmp(file_name, NONAME_DAT)!=0)
1.48 moko 559: *part.string << "; filename=\"" << file_name << "\"";
560: *part.string << CRLF HTTP_CONTENT_TYPE_CAPITALIZED ": " << part.r->mime_type_of(file_name);
1.22 misha 561: }
1.48 moko 562: *part.string << CRLF CRLF;
1.22 misha 563: }
564:
1.73 moko 565: static void form_string_value2part(HashStringValue::key_type key, const String& value, FormPart& part) {
1.28 misha 566: form_part_boundary_header(part, key);
1.48 moko 567: *part.string << value << CRLF;
1.22 misha 568: }
569:
1.73 moko 570: static void form_file_value2part(HashStringValue::key_type key, VFile& vfile, FormPart& part) {
1.28 misha 571: form_part_boundary_header(part, key, vfile.fields().get(name_name)->as_string().cstr());
1.48 moko 572: part.blocks+=FormPart::BinaryBlock(part.string, part.r);
573: part.blocks+=FormPart::BinaryBlock(vfile.value_ptr(), vfile.value_size());
574: part.string=new String();
575: *part.string << CRLF;
1.22 misha 576: }
577:
578: static void form_table_value2part(Table::element_type row, FormPart* part) {
579: form_string_value2part(part->info->key, *row->get(0), *part);
580: }
581:
1.136 moko 582: static bool form_element_value2part(HashStringValue::key_type key, HashStringValue::value_type value, FormPart& part) {
583: if(const String* svalue=value->get_string()){
584: form_string_value2part(key, *svalue, part);
585: } else if(VFile* vfile=dynamic_cast<VFile *>(value)){
586: form_file_value2part(key, *vfile, part);
587: } else {
588: return false;
589: }
590: return true;
591: }
592:
1.73 moko 593: static void form_value2part(HashStringValue::key_type key, HashStringValue::value_type value, FormPart& part) {
1.136 moko 594: if(form_element_value2part(key, value, part)) {
595: // string or file
596: } else if(Table* tvalue=value->get_table()) {
1.48 moko 597: Form_table_value2string_info info(key, *part.string);
1.22 misha 598: part.info = &info;
599: tvalue->for_each(form_table_value2part, &part);
1.136 moko 600: } else if(HashStringValue* hvalue=value->get_hash()){
601: // hash or array: several strings/files under the same field name
602: for(HashStringValue::Iterator j(*hvalue); j; j.next())
603: if(!form_element_value2part(key, j.value(), part))
604: 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 605: } else
1.136 moko 606: 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 607: }
608:
609: const char* pa_form2string_multipart(HashStringValue& form, Request& r, const char* boundary, size_t& post_size){
1.48 moko 610: FormPart formpart(&r, boundary);
1.22 misha 611: form.for_each<FormPart&>(form_value2part, formpart);
1.48 moko 612: *formpart.string << "--" << boundary << "--";
613: // @todo: return binary blocks here to save memory in pa_internal_file_read_http
614: return formpart.post(post_size);
1.22 misha 615: }
616:
1.54 misha 617: // Set-Cookie: name=value; Domain=docs.foo.com; Path=/accounts; Expires=Wed, 13-Jan-2021 22:23:01 GMT; Secure; HttpOnly
618: static ArrayString* parse_cookie(Request& r, const String& cookie) {
1.64 moko 619: char *current=pa_strdup(cookie.cstr());
1.54 misha 620:
621: const String* name=0;
1.55 moko 622: const String* value=&String::Empty;
623: const String* expires=&String::Empty;
624: const String* max_age=&String::Empty;
625: const String* path=&String::Empty;
626: const String* domain=&String::Empty;
627: const String* httponly=&String::Empty;
628: const String* secure=&String::Empty;
1.54 misha 629:
630: bool first_pair=true;
631:
632: do {
633: if(char *meaning=search_stop(current, ';'))
634: if(char *attribute=search_stop(meaning, '=')) {
635: const String* sname=new String(unescape_chars(attribute, strlen(attribute), &r.charsets.source(), true/*don't convert '"' to space*/), String::L_TAINTED);
636: const String* smeaning=0;
637: if(meaning)
638: smeaning=new String(unescape_chars(meaning, strlen(meaning), &r.charsets.source(), true/*don't convert '"' to space*/), String::L_TAINTED);
639:
640: if(first_pair) {
641: // name + value
642: name=sname;
1.122 moko 643: if(smeaning)
644: value=smeaning;
1.54 misha 645: first_pair=false;
646: } else {
647: const String& slower=sname->change_case(r.charsets.source(), String::CC_LOWER);
648:
649: if(slower == "expires")
650: expires=smeaning;
651: else if(slower == "max-age")
652: max_age=smeaning;
653: else if(slower == "domain")
654: domain=smeaning;
655: else if(slower == "path")
656: path=smeaning;
657: else if(slower == "httponly")
658: httponly=new String("1", String::L_CLEAN);
659: else if(slower == "secure")
660: secure=new String("1", String::L_CLEAN);
661: else {
662: // todo@ ?
663: }
664: }
665: }
666: } while(current);
667:
668: if(!name)
669: return 0;
670:
671: ArrayString* result=new ArrayString(8);
672: *result+=name;
673: *result+=value;
674: *result+=expires;
675: *result+=max_age;
676: *result+=domain;
677: *result+=path;
678: *result+=httponly;
679: *result+=secure;
680:
681: return result;
682: }
683:
1.56 misha 684: Table* parse_cookies(Request& r, Table *cookies){
685: Table& result=*new Table(new Cookies_table_template_columns);
686:
1.125 moko 687: for(Array_iterator<Table::element_type> i(*cookies); i; )
1.56 misha 688: if(ArrayString* row=parse_cookie(r, *i.next()->get(0)))
689: result+=row;
690:
691: return &result;
692: }
693:
1.75 moko 694: void tables_update(HashStringValue& tables, const String::Body name, const String& value){
1.72 moko 695: Table *table;
696: if(Value *valready=tables.get(name)) {
697: // second+ appearence
698: table=valready->get_table();
699: } else {
700: // first appearence
701: Table::columns_type columns=new ArrayString(1);
702: *columns+=new String("value");
703: table=new Table(columns);
704: tables.put(name, new VTable(table));
705: }
706: // this string becomes next row
707: ArrayString& row=*new ArrayString(1);
708: row+=&value;
709: *table+=&row;
710: }
711:
1.1 paf 712: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.72 moko 713: 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 714: File_read_http_result result;
1.20 misha 715: char host[MAX_STRING];
1.66 moko 716: const char *idna_host;
1.1 paf 717: const char* uri;
1.49 moko 718: short port=80;
1.10 misha 719: const char* method="GET";
1.21 misha 720: bool method_is_get=true;
1.1 paf 721: HashStringValue* form=0;
722: int timeout_secs=2;
723: bool fail_on_status_ne_200=true;
1.12 misha 724: bool omit_post_charset=false;
1.1 paf 725: Value* vheaders=0;
1.10 misha 726: Value* vcookies=0;
1.11 misha 727: Value* vbody=0;
1.72 moko 728: Charset* asked_remote_charset=0;
1.58 moko 729: Charset* real_remote_charset=0;
1.1 paf 730: const char* user_cstr=0;
731: const char* password_cstr=0;
1.22 misha 732: const char* encode=0;
733: bool multipart=false;
1.1 paf 734:
735: if(options) {
736: int valid_options=pa_get_valid_file_options_count(*options);
737:
738: if(Value* vmethod=options->get(HTTP_METHOD_NAME)) {
739: valid_options++;
1.21 misha 740: method=vmethod->as_string().change_case(r.charsets.source(), String::CC_UPPER).cstr();
741: method_is_get=strcmp(method, "GET")==0;
1.1 paf 742: }
1.22 misha 743: if(Value* vencode=options->get(HTTP_FORM_ENCTYPE_NAME)) {
744: valid_options++;
745: encode=vencode->as_string().cstr();
746: }
1.1 paf 747: if(Value* vform=options->get(HTTP_FORM_NAME)) {
748: valid_options++;
749: form=vform->get_hash();
750: }
1.11 misha 751: if(vbody=options->get(HTTP_BODY_NAME)) {
1.1 paf 752: valid_options++;
753: }
754: if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) {
755: valid_options++;
756: timeout_secs=vtimeout->as_int();
757: }
1.11 misha 758: if(vheaders=options->get(HTTP_HEADERS_NAME)) {
1.1 paf 759: valid_options++;
760: }
1.11 misha 761: if(vcookies=options->get(HTTP_COOKIES_NAME)) {
1.10 misha 762: valid_options++;
763: }
1.1 paf 764: if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) {
765: valid_options++;
766: fail_on_status_ne_200=!vany_status->as_bool();
1.12 misha 767: }
1.20 misha 768: if(Value* vomit_post_charset=options->get(HTTP_OMIT_POST_CHARSET_NAME)){
1.12 misha 769: valid_options++;
770: omit_post_charset=vomit_post_charset->as_bool();
771: }
1.6 misha 772: if(Value* vcharset_name=options->get(PA_CHARSET_NAME)) {
1.77 moko 773: asked_remote_charset=&pa_charsets.get(vcharset_name->as_string());
1.58 moko 774: }
775: if(Value* vresponse_charset_name=options->get(PA_RESPONSE_CHARSET_NAME)) {
1.61 moko 776: valid_options++;
1.77 moko 777: real_remote_charset=&pa_charsets.get(vresponse_charset_name->as_string());
1.1 paf 778: }
779: if(Value* vuser=options->get(HTTP_USER)) {
780: valid_options++;
781: user_cstr=vuser->as_string().cstr();
782: }
783: if(Value* vpassword=options->get(HTTP_PASSWORD)) {
784: valid_options++;
785: password_cstr=vpassword->as_string().cstr();
786: }
787:
788: if(valid_options!=options->count())
1.46 misha 789: throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION);
1.1 paf 790: }
791: if(!asked_remote_charset) // defaulting to $request:charset
1.22 misha 792: asked_remote_charset=&(r.charsets).source();
793:
794: if(encode){
795: if(method_is_get)
1.127 moko 796: throw Exception(PARSER_RUNTIME, 0, "you cannot use $." HTTP_FORM_ENCTYPE_NAME " option with method GET");
1.22 misha 797:
798: multipart=strcasecmp(encode, HTTP_CONTENT_TYPE_MULTIPART_FORMDATA)==0;
799:
800: if(!multipart && strcasecmp(encode, HTTP_CONTENT_TYPE_FORM_URLENCODED)!=0)
1.72 moko 801: throw Exception(PARSER_RUNTIME, 0, "$." HTTP_FORM_ENCTYPE_NAME " option value can be " HTTP_CONTENT_TYPE_FORM_URLENCODED " or " HTTP_CONTENT_TYPE_MULTIPART_FORMDATA " only");
1.22 misha 802: }
1.1 paf 803:
1.11 misha 804: if(vbody){
805: if(method_is_get)
1.127 moko 806: throw Exception(PARSER_RUNTIME, 0, "you cannot use $." HTTP_BODY_NAME " option with method GET");
1.11 misha 807:
808: if(form)
1.127 moko 809: throw Exception(PARSER_RUNTIME, 0, "you cannot use options $." HTTP_BODY_NAME " and $." HTTP_FORM_NAME " together");
1.11 misha 810: }
1.1 paf 811:
812: //preparing request
1.29 misha 813: String& connect_string=*new String(file_spec);
1.1 paf 814:
1.48 moko 815: const char* request;
816: size_t request_size;
1.1 paf 817: {
818: // influence URLencoding of tainted pieces to String::L_URI lang
1.22 misha 819: Temp_client_charset temp(r.charsets, *asked_remote_charset);
1.1 paf 820:
1.44 misha 821: const char* connect_string_cstr=connect_string.untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.1 paf 822:
823: const char* current=connect_string_cstr;
824: if(strncmp(current, "http://", 7)!=0)
1.72 moko 825: throw Exception(PARSER_RUNTIME, &connect_string, "does not start with http://"); //never
1.1 paf 826: current+=7;
827:
1.119 moko 828: pa_strncpy(host, current, sizeof(host));
1.34 misha 829: char* host_uri=lsplit(host, '/');
830: uri=host_uri?current+(host_uri-1-host):"/";
831: char* port_cstr=lsplit(host, ':');
1.49 moko 832:
833: if (port_cstr){
834: char* error_pos=0;
835: port=(short)strtol(port_cstr, &error_pos, 10);
836: if(port==0 || *error_pos)
837: throw Exception(PARSER_RUNTIME, &connect_string, "invalid port number '%s'", port_cstr);
838: }
1.1 paf 839:
1.66 moko 840: idna_host=pa_idna_encode(host, r.charsets.source());
841:
1.11 misha 842: // making request head
1.1 paf 843: String head;
1.11 misha 844: head << method << " " << uri;
1.28 misha 845: if(method_is_get && form)
846: head << (strchr(uri, '?')!=0?"&":"?") << pa_form2string(*form, r.charsets);
1.11 misha 847:
1.66 moko 848: head <<" HTTP/1.0" CRLF "Host: "<< idna_host;
1.49 moko 849: if (port != 80)
850: head << ":" << port_cstr;
851: head << CRLF;
1.11 misha 852:
1.71 moko 853: char* boundary= multipart ? get_uuid_boundary() : 0;
1.22 misha 854:
1.35 misha 855: String user_headers;
856: bool user_agent_specified=false;
857: bool content_type_specified=false;
858: bool content_type_url_encoded=false;
859: if(vheaders && !vheaders->is_string()) { // allow empty
860: if(HashStringValue *headers=vheaders->get_hash()) {
861: Http_pass_header_info info={
862: &(r.charsets),
863: &user_headers,
864: &user_agent_specified,
865: &content_type_specified,
866: &content_type_url_encoded};
867: headers->for_each<Http_pass_header_info*>(http_pass_header, &info);
868: } else
1.72 moko 869: throw Exception(PARSER_RUNTIME, 0, "headers param must be hash");
1.35 misha 870: };
871:
1.48 moko 872: const char* request_body=0;
1.22 misha 873: size_t post_size=0;
874: if(form && !method_is_get) {
1.38 misha 875: head << "Content-Type: " << (multipart ? HTTP_CONTENT_TYPE_MULTIPART_FORMDATA : HTTP_CONTENT_TYPE_FORM_URLENCODED);
1.28 misha 876:
877: if(!omit_post_charset)
878: head << "; charset=" << asked_remote_charset->NAME_CSTR();
879:
1.22 misha 880: if(multipart) {
1.28 misha 881: head << "; boundary=" << boundary;
1.48 moko 882: request_body=pa_form2string_multipart(*form, r/*charsets & mime_type needed*/, boundary, post_size/*correct post_size returned here*/);
1.22 misha 883: } else {
1.48 moko 884: request_body=pa_form2string(*form, r.charsets);
885: post_size=strlen(request_body);
1.22 misha 886: }
1.28 misha 887: head << CRLF;
1.35 misha 888: } else if(vbody) {
1.38 misha 889: // $.body was specified
1.35 misha 890: if(content_type_url_encoded){
1.36 misha 891: // transcode + url-encode
1.48 moko 892: request_body=vbody->as_string().untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.35 misha 893: } else {
1.36 misha 894: // content-type != application/x-www-form-urlencoded -> transcode only, don't url-encode!
1.72 moko 895: const String &sbody=vbody->as_string();
896: request_body=Charset::transcode(String::C(sbody.cstr(), sbody.length()), r.charsets.source(), *asked_remote_charset).str;
1.35 misha 897: }
1.48 moko 898: post_size=strlen(request_body);
1.1 paf 899: }
900:
901: // http://www.ietf.org/rfc/rfc2617.txt
902: if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr))
1.38 misha 903: head << "Authorization: " << *authorization_field_value << CRLF;
1.1 paf 904:
1.35 misha 905: head << user_headers;
906:
1.1 paf 907: if(!user_agent_specified) // defaulting
1.38 misha 908: head << "User-Agent: " DEFAULT_USER_AGENT CRLF;
1.1 paf 909:
1.12 misha 910: if(form && !method_is_get && content_type_specified) // POST + form + content-type was specified
1.72 moko 911: throw Exception(PARSER_RUNTIME, 0, "$.content-type can't be specified with method POST");
1.12 misha 912:
1.11 misha 913: if(vcookies && !vcookies->is_string()){ // allow empty
1.10 misha 914: if(HashStringValue* cookies=vcookies->get_hash()) {
1.37 misha 915: head << "Cookie: ";
1.35 misha 916: Http_pass_header_info info={&(r.charsets), &head, 0, 0, 0};
1.10 misha 917: cookies->for_each<Http_pass_header_info*>(http_pass_cookie, &info);
918: head << CRLF;
919: } else
1.72 moko 920: throw Exception(PARSER_RUNTIME, 0, "cookies param must be hash");
1.10 misha 921: }
922:
1.48 moko 923: if(request_body)
1.128 moko 924: head << "Content-Length: " << pa_uitoa(post_size) << CRLF;
1.48 moko 925:
926: head << CRLF;
927:
928: const char *request_head=head.untaint_and_transcode_cstr(String::L_URI, &(r.charsets));
1.1 paf 929:
1.48 moko 930: if(request_body){
931: size_t head_size = strlen(request_head);
932: request_size=post_size + head_size;
933: char *ptr=(char *)pa_malloc_atomic(request_size);
934: memcpy(ptr, request_head, head_size);
935: memcpy(ptr+head_size, request_body, post_size);
936: request=ptr;
937: } else {
938: request_size=strlen(request_head);
939: request=request_head;
940: }
1.1 paf 941: }
942:
1.78 moko 943:
1.97 moko 944: HTTP_response response;
1.22 misha 945:
1.28 misha 946: // sending request
1.95 moko 947: int status_code;
948: 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 949:
1.72 moko 950: // processing results
1.78 moko 951: char* raw_body=response.buf + response.body_offset;
952: size_t raw_body_size=response.length - response.body_offset;
953:
1.1 paf 954: result.headers=new HashStringValue;
955: VHash* vtables=new VHash;
1.133 moko 956: HASH_PUT_CSTR(*result.headers, "tables", vtables);
1.72 moko 957:
1.78 moko 958: if (!real_remote_charset && !response.headers.content_type.is_empty())
959: real_remote_charset=detect_charset(response.headers.content_type.cstr());
1.1 paf 960:
1.72 moko 961: if(as_text)
1.77 moko 962: real_remote_charset=pa_charsets.checkBOM(raw_body, raw_body_size, real_remote_charset);
1.72 moko 963:
964: if (!real_remote_charset)
965: real_remote_charset=asked_remote_charset; // never null
966:
1.125 moko 967: for(Array_iterator<HTTP_Headers::Header> i(response.headers.headers); i; ){
1.85 moko 968: HTTP_Headers::Header header=i.next();
1.72 moko 969:
970: header.transcode(*real_remote_charset, r.charsets.source());
971:
972: String &header_value=*new String(header.value, String::L_TAINTED);
973:
974: tables_update(vtables->hash(), header.name, header_value);
975: result.headers->put(header.name, new VString(header_value));
1.16 misha 976: }
977:
1.72 moko 978: // filling $.cookies
1.89 moko 979: if(vcookies=vtables->hash().get("SET-COOKIE"))
1.133 moko 980: HASH_PUT_CSTR(*result.headers, HTTP_COOKIES_NAME, new VTable(parse_cookies(r, vcookies->get_table())));
1.72 moko 981:
1.1 paf 982: // output response
983: String::C real_body=String::C(raw_body, raw_body_size);
1.16 misha 984:
985: 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 986: real_body=Charset::transcode(real_body, *real_remote_charset, r.charsets.source());
1.1 paf 987: }
988:
989: result.str=const_cast<char *>(real_body.str); // hacking a little
990: result.length=real_body.length;
1.16 misha 991:
1.22 misha 992: if(as_text && result.length)
993: fix_line_breaks(result.str, result.length);
994:
1.1 paf 995: result.headers->put(file_status_name, new VInt(status_code));
1.16 misha 996:
1.1 paf 997: return result;
998: }
1.84 moko 999:
1000: /* ********************** httpd *************************** */
1001:
1.100 moko 1002: enum EscapeState {
1003: Initial,
1004: Default,
1005: EscapeFirst,
1006: EscapeSecond
1007: };
1008:
1009: static bool check_uri(const char *uri){
1010: EscapeState state=Initial;
1.120 moko 1011: uint escapedValue=0;
1.100 moko 1012:
1013: const char *pattern="/../";
1014: const char *pos=pattern;
1015:
1016: while(*uri){
1017: uchar c=(uchar)*(uri++);
1018: switch(state) {
1019: case Initial:
1020: if(c!='/')
1021: return false;
1022: state=Default;
1023: break;
1024: case Default:
1025: if(c=='%'){
1026: state=EscapeFirst;
1027: continue;
1028: }
1029: if(c=='?')
1030: return true;
1031: break;
1032: case EscapeFirst:
1033: if(isxdigit(c)){
1034: state=EscapeSecond;
1035: escapedValue=hex_value[c] << 4;
1036: continue;
1037: }
1038: return false;
1039: case EscapeSecond:
1040: if(isxdigit(c)){
1041: state=Default;
1.105 moko 1042: c=(uchar)(escapedValue + hex_value[c]);
1.100 moko 1043:
1044: // implementing Apache AllowEncodedSlashes Off just in case
1045: if(c=='/' || c=='\\')
1046: return false;
1047:
1048: break;
1049: }
1050: return false;
1051: }
1052:
1053: if(c==*pos || c=='\\' && *pos=='/'){
1054: if(!*(++pos))
1055: return false;
1056: } else {
1057: pos=pattern;
1058: }
1059: }
1060: return true;
1061: }
1062:
1.84 moko 1063: class HTTPD_request : public HTTP_response {
1064: public:
1065: const char *method;
1066: const char *uri;
1067:
1.97 moko 1068: HTTPD_request() : HTTP_response(), method(NULL), uri(NULL){};
1.84 moko 1069:
1.120 moko 1070: ssize_t pa_recv(SOCKET sockfd, char *buf, size_t len);
1.103 moko 1071:
1.120 moko 1072: bool read(SOCKET sock, size_t size){
1.103 moko 1073: if(length + size > buf_size)
1074: resize(buf_size * 2 + size);
1075: ssize_t received_size=pa_recv(sock, buf + length, size);
1076: if(received_size == 0)
1077: return false;
1078: if(received_size < 0) {
1079: if(int no = pa_socks_errno())
1.111 moko 1080: throw Exception("httpd.read", 0, "error receiving request: %s (%d)", pa_socks_strerr(no), no);
1.103 moko 1081: return false;
1082: }
1083: length+=received_size;
1084: buf[length]='\0';
1085: return true;
1086: }
1087:
1.84 moko 1088: const char *extract_method(char *method_line){
1089: char* uri_start = strchr(method_line, ' ');
1090:
1091: if(!uri_start || uri_start == method_line)
1092: return NULL;
1093:
1094: char* uri_end=strchr(uri_start+1, ' ');
1095:
1096: if(!uri_end || uri_end == uri_start+1)
1097: return NULL;
1098:
1099: uri=pa_strdup(uri_start+1, uri_end-uri_start-1);
1.100 moko 1100: if(!check_uri(uri))
1101: throw Exception("httpd.request", 0, "invalid uri '%s'", uri);
1102:
1.84 moko 1103: return str_upper(method_line, uri_start-method_line);
1104: }
1105:
1.103 moko 1106:
1.120 moko 1107: bool read_header(SOCKET);
1108: size_t read_post(SOCKET, char *, size_t);
1.84 moko 1109: };
1110:
1111: enum HTTPD_request_state {
1112: HTTPD_METHOD,
1113: HTTPD_HEADERS
1114: };
1115:
1.138 moko 1116: // timeout on "void" connection is normal
1117: #define PA_RECV_TIMEOUT \
1118: LOG(pa_log("httpd [%d] recv got %d sec timeout", sockfd, pa_httpd_timeout)); \
1119: if(length) \
1120: throw Exception("httpd.timeout", 0, "timeout occurred while receiving request"); \
1121: return 0;
1122:
1.120 moko 1123: ssize_t HTTPD_request::pa_recv(SOCKET sockfd, char *buffer, size_t len){
1.111 moko 1124: LOG(pa_log("httpd [%d] recv %d appending to %d ...", sockfd, len, length));
1.107 moko 1125:
1.138 moko 1126: #ifdef PA_HAVE_ALARM
1127: if(PA_USE_ALARM) signal(SIGALRM, timeout_handler);
1128: if(PA_USE_ALARM && sigsetjmp(timeout_env, 1)) {
1129: PA_RECV_TIMEOUT
1.103 moko 1130: } else
1.107 moko 1131: #endif
1.103 moko 1132: {
1.107 moko 1133: ALARM(pa_httpd_timeout);
1.141 moko 1134: ssize_t result=pa_sock_recv(sockfd, buffer, len);
1.107 moko 1135: ALARM(0);
1.138 moko 1136: if(result < 0 && sock_timed_out(pa_socks_errno())) {
1137: PA_RECV_TIMEOUT
1138: }
1.111 moko 1139: LOG(pa_log("httpd [%d] recv got %d bytes", sockfd, result));
1.112 moko 1140: LOG(pa_log("httpd [%d] %s", sockfd, buffer));
1.103 moko 1141: return result;
1142: }
1143: }
1144:
1.115 moko 1145: static bool valid_http_method(const char * method){
1146: return method && (
1147: !strcmp(method, "GET") ||
1148: !strcmp(method, "HEAD") ||
1149: !strcmp(method, "POST") ||
1150: !strcmp(method, "PUT") ||
1151: !strcmp(method, "DELETE") ||
1152: !strcmp(method, "CONNECT") ||
1153: !strcmp(method, "OPTIONS") ||
1154: !strcmp(method, "TRACE") ||
1155: !strcmp(method, "PATCH")
1156: );
1157: }
1158:
1.120 moko 1159: bool HTTPD_request::read_header(SOCKET sock) {
1.84 moko 1160: enum HTTPD_request_state state = HTTPD_METHOD;
1161:
1162: size_t chunk_size = 0x400*4;
1163: resize(chunk_size);
1164:
1165: while(read(sock, chunk_size)){
1166: switch(state){
1167: case HTTPD_METHOD: {
1168: size_t method_size = first_line();
1169: if(!method_size)
1170: break;
1171:
1172: char *method_line = pa_strdup(buf, method_size);
1173: method = extract_method(method_line);
1174:
1.115 moko 1175: if(!valid_http_method(method))
1.84 moko 1176: throw Exception("httpd.method", new String(method ? method : method_line), "invalid request method");
1177: state = HTTPD_HEADERS;
1178: }
1179:
1180: case HTTPD_HEADERS: {
1181: if(!body_start())
1182: break;
1183:
1184: parse_headers();
1.144 moko 1185:
1186: if(length==body_offset){ // nothing but the headers (GET, websocket handshake), the buffer is no longer needed
1187: buf=NULL;
1188: length=body_offset=buf_size=0;
1189: }
1.110 moko 1190: return true;
1.84 moko 1191: }
1192: }
1193: }
1194:
1.111 moko 1195: if(!length){ // browsers open connections in advance and they will be empty unless user requests more pages
1196: LOG(pa_log("httpd [%d] void request", sock));
1.110 moko 1197: return false;
1.111 moko 1198: }
1.110 moko 1199:
1.84 moko 1200: if(state == HTTPD_METHOD)
1201: throw Exception("httpd.request", 0, "bad request from host - no method found (size=%u)", length);
1202:
1203: if(state == HTTPD_HEADERS){
1204: parse_headers();
1205: body_offset=length;
1206: }
1.110 moko 1207:
1208: return true;
1.84 moko 1209: }
1210:
1.120 moko 1211: size_t HTTPD_request::read_post(SOCKET sock, char *body, size_t max_bytes) {
1.87 moko 1212: size_t total_read = min(length - body_offset, max_bytes);
1.144 moko 1213: if(total_read)
1214: memcpy(body, buf + body_offset, total_read);
1.87 moko 1215:
1216: while (total_read < max_bytes){
1.103 moko 1217: ssize_t received_size = pa_recv(sock, body + total_read, max_bytes - total_read);
1.87 moko 1218: if(received_size == 0)
1219: return total_read;
1220: if(received_size < 0) {
1221: if(int no = pa_socks_errno())
1.111 moko 1222: throw Exception("httpd.read", new String(uri), "error receiving request body: %s (%d)", pa_socks_strerr(no), no);
1.87 moko 1223: return total_read;
1224: }
1225: total_read += received_size;
1226: }
1227: return total_read;
1228: }
1229:
1.84 moko 1230: /* ********************************************************** */
1231:
1.85 moko 1232: Array<HTTP_Headers::Header> &HTTPD_Connection::headers() {
1.84 moko 1233: return request->headers.headers;
1234: }
1235:
1236: const char *HTTPD_Connection::method() {
1237: return request->method;
1238: }
1239:
1240: const char *HTTPD_Connection::uri() {
1241: return request->uri;
1242: }
1243:
1244: const char *HTTPD_Connection::content_type() {
1245: return request->headers.content_type.cstr();
1246: }
1247:
1248: uint64_t HTTPD_Connection::content_length(){
1249: return request->headers.content_length;
1250: }
1251:
1.110 moko 1252: bool HTTPD_Connection::read_header(){
1.84 moko 1253: request = new HTTPD_request();
1.111 moko 1254: bool result = request->read_header(sock);
1255: LOG(if(result){
1256: pa_log("httpd [%d] got %s \"%s\"", sock, method(), uri());
1257: })
1258: return result;
1.84 moko 1259: }
1260:
1.87 moko 1261: size_t HTTPD_Connection::read_post(char *body, size_t max_bytes) {
1262: return request->read_post(sock, body, max_bytes);
1263: }
1264:
1.90 moko 1265: size_t HTTPD_Connection::send_body(const void *buf, size_t size) {
1.112 moko 1266: LOG(pa_log("httpd [%d] response %d bytes", sock, size));
1267: LOG(pa_log("httpd [%d] %s", sock, buf));
1.131 moko 1268: ssize_t result=pa_send(sock, (const char*)buf, size);
1269: if(result < 0) {
1.90 moko 1270: int no=pa_socks_errno();
1.111 moko 1271: throw Exception("httpd.write", 0, "error sending response: %s (%d)", pa_socks_strerr(no), no);
1.90 moko 1272: }
1.131 moko 1273: return result;
1.90 moko 1274: }
1275:
1.140 moko 1276: String::C HTTPD_Connection::unprocessed() {
1277: return String::C(request->buf + request->body_offset, request->length - request->body_offset);
1278: }
1279:
1280: size_t HTTPD_Connection::receive(char *dest, size_t max_bytes) {
1.141 moko 1281: ssize_t received_size=pa_sock_recv(sock, dest, max_bytes);
1.140 moko 1282: if(received_size < 0) {
1283: int no=pa_socks_errno();
1284: if(pa_socks_is_reset(no))
1285: return 0;
1286: throw Exception("httpd.read", 0, "error receiving data: %s (%d)", pa_socks_strerr(no), no);
1287: }
1288: return received_size;
1289: }
1290:
1291: void HTTPD_Connection::set_nodelay() {
1292: int nodelay=1;
1293: setsockopt(sock, IPPROTO_TCP, TCP_NODELAY, (const char *)&nodelay, sizeof(nodelay));
1294: }
1295:
1296: void HTTPD_Connection::set_send_timeout(int timeout_secs) {
1297: set_socket_timeout(sock, SO_SNDTIMEO, timeout_secs);
1298: }
1299:
1.93 moko 1300: HTTPD_Connection::~HTTPD_Connection(){
1.120 moko 1301: if(sock != INVALID_SOCKET){
1.111 moko 1302: LOG(pa_log("httpd [%d] closed", sock));
1.93 moko 1303: closesocket(sock);
1.145 moko 1304: HTTPD_Server::track_connection(-1);
1.111 moko 1305: }
1.93 moko 1306: }
1307:
1.120 moko 1308: bool HTTPD_Connection::accept(SOCKET server_sock, int timeout_value) {
1.139 moko 1309: int ready = pa_sock_ready(server_sock, false, timeout_value);
1.93 moko 1310: if (ready < 0) {
1311: int no=pa_socks_errno();
1312: throw Exception("httpd.accept", 0, "error waiting for connection: %s (%d)", pa_socks_strerr(no), no);
1313: }
1314: if (ready == 0)
1315: return false; /* Timeout */
1316:
1317: struct sockaddr_in addr;
1318: socklen_t sock_addr_len = sizeof(struct sockaddr_in);
1319: memset(&addr, 0, sock_addr_len);
1320:
1.146 moko 1321: sock = PA_ACCEPT(server_sock, (struct sockaddr *)&addr, &sock_addr_len);
1.120 moko 1322: if(sock == INVALID_SOCKET){
1.93 moko 1323: int no=pa_socks_errno();
1324: throw Exception("httpd.accept", 0, "error accepting connection: %s (%d)", pa_socks_strerr(no), no);
1325: }
1.145 moko 1326: HTTPD_Server::track_connection(1); // taken back by the owner
1.93 moko 1327:
1.123 moko 1328: // Has no positive performance effect, requires include <netinet/tcp.h>
1329: // static int sock_on=1;
1330: // setsockopt(sock, IPPROTO_TCP, TCP_NODELAY, (char *)&sock_on, sizeof(sock_on));
1331:
1.138 moko 1332: if(!PA_USE_ALARM)
1333: set_socket_timeout(sock, SO_RCVTIMEO, pa_httpd_timeout);
1334:
1.93 moko 1335: remote_addr = pa_strdup(inet_ntoa(addr.sin_addr));
1.111 moko 1336: LOG(pa_log("httpd [%d] accepted from %s", sock, remote_addr));
1.93 moko 1337: return true;
1338: }
1.84 moko 1339:
1.107 moko 1340: HTTPD_Server::HTTPD_MODE HTTPD_Server::mode = HTTPD_Server::SEQUENTIAL;
1.114 moko 1341: const char *HTTPD_Server::port=NULL;
1.145 moko 1342: int HTTPD_Server::max_connections=INT_MAX;
1343: int HTTPD_Server::connections=0;
1344:
1345: void HTTPD_Server::track_connection(int delta){
1346: SYNCHRONIZED;
1347: connections+=delta;
1348: }
1.106 moko 1349:
1.108 moko 1350: void HTTPD_Server::set_mode(const String &value){
1351: if(value == "sequental") mode = SEQUENTIAL;
1.113 moko 1352: #ifdef HAVE_TLS
1.108 moko 1353: else if (value == "threaded") mode = MULTITHREADED;
1.113 moko 1354: #endif
1.108 moko 1355: #ifdef _MSC_VER
1.117 moko 1356: else throw Exception("httpd.mode", &value, "$MAIN:HTTPD.mode must be 'sequental' or 'threaded'");
1.108 moko 1357: #else
1358: else if (value == "parallel") mode = PARALLEL;
1.117 moko 1359: else throw Exception("httpd.mode", &value, "$MAIN:HTTPD.mode must be 'sequental', 'parallel' or 'threaded'");
1.108 moko 1360: #endif
1361: }
1362:
1.120 moko 1363: SOCKET HTTPD_Server::bind(const char *host_port){
1.84 moko 1364: struct sockaddr_in me;
1365:
1.114 moko 1366: port = strchr(host_port, ':');
1.86 moko 1367: const char *host = NULL;
1.116 moko 1368: if(port){
1.114 moko 1369: if(port > host_port)
1370: host = pa_strdup(host_port, port - host_port);
1.86 moko 1371: port += 1;
1372: } else {
1373: port = host_port;
1374: }
1375:
1.105 moko 1376: if(!set_addr(&me, host, (short)pa_atoui(port))){
1.84 moko 1377: if (host)
1.127 moko 1378: throw Exception("httpd.bind", 0, "cannot resolve hostname \"%s\"", host);
1.84 moko 1379: me.sin_addr.s_addr=INADDR_ANY;
1380: }
1381:
1.146 moko 1382: SOCKET sock = PA_SOCKET(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/);
1.84 moko 1383:
1.120 moko 1384: if(sock == INVALID_SOCKET){
1.84 moko 1385: int no=pa_socks_errno();
1.127 moko 1386: throw Exception("httpd.bind", 0, "cannot make socket: %s (%d)", pa_socks_strerr(no), no);
1.84 moko 1387: }
1388:
1.93 moko 1389: static int sock_on = 1;
1390:
1.84 moko 1391: if (setsockopt(sock, SOL_SOCKET, SO_REUSEADDR, (char *)&sock_on, sizeof(sock_on)) ||
1392: setsockopt(sock, SOL_SOCKET, SO_KEEPALIVE, (char *)&sock_on, sizeof(sock_on)) ||
1393: ::bind(sock, (struct sockaddr*)&me, sizeof(me)) ||
1.143 moko 1394: listen(sock, 511)) { // same default as in apache/nginx
1.89 moko 1395: closesocket(sock);
1.84 moko 1396: int no = pa_socks_errno();
1.127 moko 1397: throw Exception("httpd.bind", 0, "cannot bind socket: %s (%d)", pa_socks_strerr(no), no);
1.84 moko 1398: }
1399: return sock;
1400: }
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