1 |
|
/* ============================================================================ |
2 |
|
* Douglas Thrift's Search Engine License |
3 |
|
* |
4 |
< |
* Copyright (C) 2002-2003, Douglas Thrift. All Rights Reserved. |
4 |
> |
* Copyright (C) 2002-2004, Douglas Thrift. All Rights Reserved. |
5 |
|
* Redistribution and use in source and binary forms, with or without |
6 |
|
* modification, are permitted provided that the following conditions are met: |
7 |
|
* |
46 |
|
// |
47 |
|
// Douglas Thrift |
48 |
|
// |
49 |
< |
// $Id: HttpHandler.cpp,v 1.21 2003/07/18 07:26:41 douglas Exp $ |
49 |
> |
// $Id$ |
50 |
|
|
51 |
< |
#include "HttpHandler.h" |
51 |
> |
#include "HttpHandler.hpp" |
52 |
|
|
53 |
|
// Lovely C Sockets! |
54 |
|
#ifndef _WIN32 |
59 |
|
#include <netinet/in.h> |
60 |
|
#include <netdb.h> |
61 |
|
|
62 |
– |
#define INVALID_SOCKET -1 |
63 |
– |
#define SOCKET_ERROR -1 |
64 |
– |
|
62 |
|
inline int closesocket(SOCKET s) { return close(s); } |
63 |
|
#endif |
64 |
|
|
65 |
< |
HttpHandler::HttpHandler() |
65 |
> |
#ifndef _OpenSSL_ |
66 |
> |
HttpHandler::HttpHandler() : binary(false), length(0), chunked(false) |
67 |
> |
#else |
68 |
> |
HttpHandler::HttpHandler() : binary(false), length(0), chunked(false), |
69 |
> |
tls(false) |
70 |
> |
#endif |
71 |
|
{ |
70 |
– |
buffer = new char[BUFSIZ + 1]; |
71 |
– |
|
72 |
|
#ifdef _WIN32 |
73 |
|
if (WSAStartup(MAKEWORD(2, 0), &data) != 0) |
74 |
|
{ |
76 |
|
exit(1); |
77 |
|
} |
78 |
|
#endif // _WIN32 |
79 |
– |
|
80 |
– |
binary = false; |
81 |
– |
length = 0; |
82 |
– |
chunked = false; |
83 |
– |
#ifdef _OpenSSL_ |
84 |
– |
tls = false; |
85 |
– |
#endif |
79 |
|
} |
80 |
|
|
81 |
|
HttpHandler::~HttpHandler() |
82 |
|
{ |
90 |
– |
delete [] buffer; |
91 |
– |
|
83 |
|
#ifdef _WIN32 |
84 |
|
WSACleanup(); |
85 |
|
#endif // _WIN32 |
87 |
|
|
88 |
|
bool HttpHandler::handle(URL &url, const string referer, bool head) |
89 |
|
{ |
90 |
< |
bool answer = false; |
90 |
> |
bool answer(false); |
91 |
|
|
92 |
|
if ((http = socket(PF_INET, SOCK_STREAM, 0)) == INVALID_SOCKET) |
93 |
|
{ |
103 |
|
if ((host = gethostbyname(url.getAddress().c_str())) == NULL) |
104 |
|
{ |
105 |
|
error(program + ": Host: " + url.getAddress(), true); |
106 |
+ |
|
107 |
|
return answer; |
108 |
|
} |
109 |
|
|
114 |
|
SOCKET_ERROR) |
115 |
|
{ |
116 |
|
error(program + ": Connect"); |
117 |
+ |
|
118 |
|
return answer; |
119 |
|
} |
120 |
|
|
174 |
|
{ |
175 |
|
line = getline(); |
176 |
|
|
177 |
< |
if (line.find("HTTP/") != 0) |
185 |
< |
{ |
186 |
< |
return answer; |
187 |
< |
} |
177 |
> |
if (line.find("HTTP/") != 0) return answer; |
178 |
|
|
179 |
< |
unsigned dot = line.find('.'); |
180 |
< |
unsigned space = line.find(' '); |
179 |
> |
unsigned dot(line.find('.')), space(line.find(' ')), major, minor; |
180 |
> |
istringstream number(line.substr(5, dot - 5) + " " + line.substr(dot |
181 |
> |
+ 1, space - dot - 1)); |
182 |
|
|
183 |
< |
unsigned major = strtoul(line.substr(5, dot - 5).c_str(), 0, 10); |
184 |
< |
unsigned minor = strtoul(line.substr(dot + 1, space - dot - 1).c_str(), |
194 |
< |
0, 10); |
183 |
> |
number >> major; |
184 |
> |
number >> minor; |
185 |
|
|
186 |
|
if (major > 1) |
187 |
|
{ |
191 |
|
return answer; |
192 |
|
} |
193 |
|
|
194 |
< |
response = code(strtoul(line.substr(space + 1).c_str(), 0, 10)); |
194 |
> |
number.clear(); |
195 |
> |
number.str(line.substr(space + 1, 3)); |
196 |
> |
|
197 |
> |
number >> response; |
198 |
|
|
199 |
|
if (response < ok) do line = getline(); while (line != ""); |
200 |
|
} |
206 |
|
|
207 |
|
if (line != "") |
208 |
|
{ |
209 |
< |
unsigned colon = line.find(':'); |
210 |
< |
|
218 |
< |
string field = line.substr(0, colon); |
219 |
< |
string value = line.substr(colon + 1); |
209 |
> |
unsigned colon(line.find(':')); |
210 |
> |
string field(line.substr(0, colon)), value(line.substr(colon + 1)); |
211 |
|
|
212 |
|
while (isspace(value[0])) value.erase(0, 1); |
213 |
|
|
217 |
|
} |
218 |
|
else if (field == "Content-Length") |
219 |
|
{ |
220 |
< |
length = strtoul(value.c_str(), 0, 10); |
220 |
> |
istringstream number(value); |
221 |
> |
|
222 |
> |
number >> length; |
223 |
|
} |
224 |
|
else if (field == "Location") |
225 |
|
{ |
237 |
|
{ |
238 |
|
case ok: |
239 |
|
if (debug) cerr << "response = " << response << "\n"; |
240 |
+ |
|
241 |
|
answer = true; |
242 |
|
break; |
243 |
|
case choices: |
245 |
|
case found: |
246 |
|
if (debug) cerr << "response = " << response << "\n" |
247 |
|
<< "location = " << location << "\n"; |
248 |
+ |
|
249 |
|
location = getLink(location, url); |
250 |
|
break; |
251 |
|
case notfound: |
254 |
|
break; |
255 |
|
default: |
256 |
|
if (debug) cerr << "response = " << response << "\n"; |
257 |
< |
if (response <= 299) |
258 |
< |
{ |
264 |
< |
answer = true; |
265 |
< |
} |
266 |
< |
else if (response <= 399) |
257 |
> |
|
258 |
> |
if (response <= 299) answer = true; else if (response <= 399) |
259 |
|
{ |
260 |
|
location = getLink(location, url); |
261 |
|
} |
269 |
|
|
270 |
|
void HttpHandler::clear() |
271 |
|
{ |
272 |
+ |
#ifdef _OpenSSL_ |
273 |
|
if (tls) |
274 |
|
{ |
275 |
|
SSL_shutdown(ssl); |
276 |
|
SSL_free(ssl); |
277 |
|
SSL_CTX_free(ctx); |
278 |
|
} |
279 |
+ |
#endif |
280 |
|
|
281 |
|
closesocket(http); |
282 |
|
|
283 |
|
type = ""; |
284 |
|
length = 0; |
285 |
|
location = ""; |
286 |
+ |
|
287 |
|
page.clear(); |
288 |
|
page.str(""); |
289 |
+ |
|
290 |
|
chunked = false; |
291 |
|
#ifdef _OpenSSL_ |
292 |
|
tls = false; |
297 |
|
{ |
298 |
|
if (!chunked) |
299 |
|
{ |
300 |
< |
unsigned left = length; |
300 |
> |
unsigned left(length); |
301 |
|
|
302 |
|
while (left > 0) |
303 |
|
{ |
304 |
|
memset(buffer, 0, BUFSIZ + 1); |
305 |
|
|
306 |
< |
unsigned bytes = left > BUFSIZ ? BUFSIZ : left; |
306 |
> |
unsigned bytes(left > BUFSIZ ? BUFSIZ : left); |
307 |
|
long received; |
308 |
|
|
309 |
|
while (true) |
325 |
|
else if (received != bytes) |
326 |
|
{ |
327 |
|
left -= received; |
328 |
+ |
|
329 |
|
page << buffer; |
330 |
|
|
331 |
|
memset(buffer, 0, BUFSIZ + 1); |
332 |
|
|
333 |
|
bytes -= received; |
334 |
|
} |
335 |
< |
else |
339 |
< |
{ |
340 |
< |
break; |
341 |
< |
} |
335 |
> |
else break; |
336 |
|
} |
337 |
|
|
338 |
|
page << buffer; |
339 |
+ |
|
340 |
|
left -= bytes; |
341 |
|
} |
342 |
|
} |
346 |
|
|
347 |
|
do |
348 |
|
{ |
349 |
< |
chunk = strtoul(getline().c_str(), 0, 16); |
349 |
> |
istringstream number(getline()); |
350 |
> |
|
351 |
> |
number.setf(ios_base::hex, ios_base::basefield); |
352 |
|
|
353 |
< |
unsigned left = chunk; |
353 |
> |
number >> chunk; |
354 |
> |
|
355 |
> |
unsigned left(chunk); |
356 |
|
|
357 |
|
while (left > 0) |
358 |
|
{ |
359 |
|
memset(buffer, 0, BUFSIZ + 1); |
360 |
|
|
361 |
< |
unsigned bytes = left > BUFSIZ ? BUFSIZ : left; |
361 |
> |
unsigned bytes(left > BUFSIZ ? BUFSIZ : left); |
362 |
|
long received; |
363 |
|
|
364 |
|
while (true) |
376 |
|
{ |
377 |
|
!tls ? error(program + ": Recv") : error(program + |
378 |
|
": SSL Read", int(received)); |
379 |
+ |
|
380 |
|
exit(1); |
381 |
|
} |
382 |
|
#endif |
389 |
|
|
390 |
|
bytes -= received; |
391 |
|
} |
392 |
< |
else |
393 |
< |
{ |
394 |
< |
break; |
395 |
< |
} |
392 |
> |
else break; |
393 |
|
} |
394 |
|
|
395 |
|
page << buffer; |
396 |
+ |
|
397 |
|
left -= bytes; |
398 |
|
} |
399 |
|
|
400 |
|
getline(); |
401 |
+ |
|
402 |
|
length += chunk; |
403 |
|
} |
404 |
|
while (chunk > 0); |
406 |
|
|
407 |
|
if (!binary) |
408 |
|
{ |
409 |
< |
string page = this->page.str(); |
409 |
> |
string page(this->page.str()); |
410 |
|
|
411 |
< |
for (unsigned index = 0; index < page.length(); index++) |
411 |
> |
for (unsigned index(0); index < page.length(); index++) |
412 |
|
{ |
413 |
< |
if (page[index] == '\r' && (index + 1 < page.length()) ? page[index + |
414 |
< |
1] == '\n' : false) |
413 |
> |
if (page[index] == '\r' && (index + 1 < page.length()) ? |
414 |
> |
page[index + 1] == '\n' : false) |
415 |
|
{ |
416 |
|
page.erase(index, 1); |
417 |
|
} |
760 |
|
SSL_load_error_strings(); |
761 |
|
SSL_library_init(); |
762 |
|
|
763 |
< |
// |
763 |
> |
#ifndef _urandomdev_ |
764 |
> |
int pid(getpid()), now(time(NULL)); |
765 |
> |
unsigned seed(now > pid ? now - pid : pid - now); |
766 |
> |
char* junk = new char[seed % 30 + 2]; |
767 |
> |
|
768 |
> |
junk[0] = pid; |
769 |
> |
junk[seed % 30 + 1] = now; |
770 |
> |
|
771 |
> |
srand(seed); |
772 |
> |
|
773 |
> |
for (size_t index = 1; index < seed % 30 + 1; index++) |
774 |
> |
{ |
775 |
> |
junk[index] = rand(); |
776 |
> |
} |
777 |
> |
|
778 |
> |
if (debug) |
779 |
> |
{ |
780 |
> |
cerr << "junk = {\n"; |
781 |
> |
|
782 |
> |
for (size_t index = 1; index < seed % 30 + 2; index++) |
783 |
> |
{ |
784 |
> |
cerr << " [" << index << "] = " << unsigned(junk[index]) << "\n"; |
785 |
> |
} |
786 |
> |
|
787 |
> |
cerr << "}\n"; |
788 |
> |
} |
789 |
> |
|
790 |
> |
RAND_seed(junk, seed % 30 + 2); |
791 |
> |
|
792 |
> |
delete junk; |
793 |
> |
#else |
794 |
> |
if (debug) cerr << "junk = /dev/urandom\n"; |
795 |
> |
#endif |
796 |
> |
|
797 |
> |
ctx = SSL_CTX_new(TLSv1_client_method()); |
798 |
> |
|
799 |
> |
if (ctx == NULL) |
800 |
> |
{ |
801 |
> |
cerr << program << ": SSL CTX New: " |
802 |
> |
<< ERR_reason_error_string(ERR_get_error()) << "\n"; |
803 |
> |
|
804 |
> |
return false; |
805 |
> |
} |
806 |
> |
|
807 |
> |
ssl = SSL_new(ctx); |
808 |
> |
|
809 |
> |
if (SSL_set_fd(ssl, http) == 0) |
810 |
> |
{ |
811 |
> |
cerr << program << ": SSL Set FD: " |
812 |
> |
<< ERR_reason_error_string(ERR_get_error()) << "\n"; |
813 |
> |
|
814 |
> |
return false; |
815 |
> |
} |
816 |
> |
|
817 |
> |
int number; |
818 |
> |
|
819 |
> |
if ((number = SSL_connect(ssl)) <= 0) |
820 |
> |
{ |
821 |
> |
error(program + ": SSL Connect", number); |
822 |
> |
|
823 |
> |
return false; |
824 |
> |
} |
825 |
|
|
826 |
|
return true; |
827 |
|
} |