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