ViewVC Help

View File | Revision Log | Show Annotations | Download File | View Changeset | Root Listing

root/proj/trunk/Search/HttpHandler.cpp

Comparing trunk/Search/HttpHandler.cpp (file contents):
Revision 1 by douglas, 2002-12-04T20:22:59-08:00 vs.
Revision 28 by douglas, 2003-01-02T19:42:33-08:00

#	Line 1 | Line 1
1		/* ============================================================================
2		* Douglas Thrift's Search Engine License
3		*
4	<	* Copyright (C) 2002, Douglas Thrift. All Rights Reserved.
4	>	* Copyright (C) 2002-2003, 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		*
#	Line 50 | Line 50
50
51		#include "HttpHandler.h"
52
53	–	#ifdef _WIN32
54	–	#define PATH_SEPARATOR "\\"
55	–	#define ENV_SEPARATOR ';'
56	–	#else
57	–	#define PATH_SEPARATOR "/"
58	–	#define ENV_SEPARATOR ':'
59	–	#endif
60	–
53		HttpHandler::HttpHandler()
54		{
55	<	string arguments = "(Ljava/lang/String;ILjava/lang/String;Ljava/lang/Strin"
64	<	+ string("g;Ljava/lang/String;)Ljava/lang/String;");
55	>	buffer = new char[BUFSIZ + 1];
56
57	<	begin = 0;
58	<
59	<	setJarPath();
57	>	#ifdef _WIN32
58	>	if (WSAStartup(MAKEWORD(2, 0), &data) != 0)
59	>	{
60	>	error(program + ": WSAStartup");
61	>	exit(1);
62	>	}
63	>	#endif // _WIN32
64
65	<	//      options[0].optionString = jarPath;
66	<	//      memset(&vm_args, 0, sizeof (vm_args));
72	<	//      vm_args.version = JNI_VERSION_1_4;
73	<	//      vm_args.nOptions = 1;
74	<	//      vm_args.options = options;
75	<
76	<	//      status = JNI_CreateJavaVM(&jvm, (void**)&env, &vm_args);
77	<
78	<	//      if (status != JNI_ERR)
79	<	//      {
80	<	//              cls = env->FindClass("HttpConnector");
81	<
82	<	//              if (cls != 0)
83	<	//              {
84	<	//                      mid = env->GetStaticMethodID(cls, "connect", arguments.c_str());
85	<	//              }
86	<	//      }
65	>	length = 0;
66	>	chunked = false;
67		}
68
69		HttpHandler::~HttpHandler()
70		{
71	<	//      if (status != JNI_ERR)
92	<	//      {
93	<	//              jvm->DestroyJavaVM();
94	<	//      }
71	>	delete [] buffer;
72
73	<	//      delete [] jarPath;
73	>	#ifdef _WIN32
74	>	WSACleanup();
75	>	#endif // _WIN32
76		}
77
78	<	bool HttpHandler::connect(URL& url)
78	>	bool HttpHandler::handle(URL &url, const string referer, bool head)
79		{
80		bool answer = false;
81
82	<	//      if (status != JNI_ERR)
83	<	//      {
84	<	//              if (cls != 0)
85	<	//              {
86	<	//                      if (mid != 0)
108	<	//                      {
109	<	//                              jstring addressJ = env->NewStringUTF(url.getAddress().c_str());
110	<	//                              jint portJ = url.getPort();
111	<	//                              jstring pathJ = env->NewStringUTF(url.getPath().c_str());
112	<	//                              jstring programNameJ = env->NewStringUTF(programName.c_str());
113	<	//                              jstring programVersionJ =
114	<	//                                      env->NewStringUTF(programVersion.c_str());
115	<
116	<	//                              jstring pageJ = (jstring)env->CallStaticObjectMethod(cls, mid,
117	<	//                                      addressJ, portJ, pathJ, programNameJ, programVersionJ);
118	<
119	<	//                              const char* pageC = env->GetStringUTFChars(pageJ, 0);
120	<	//                              page = pageC;
121	<	//                              env->ReleaseStringUTFChars(pageJ, pageC);
122	<
123	<	//                              if (page != "unknown host\n" && page != "io exception\n" &&
124	<	//                                      page != "bad headers\n") answer = true;
125	<	//                      }
126	<	//              }
127	<	//      }
82	>	if ((http = socket(PF_INET, SOCK_STREAM, 0)) == INVALID_SOCKET)
83	>	{
84	>	error(program + ": Socket");
85	>	exit(1);
86	>	}
87
88	<	return answer;
89	<	}
88	>	sockaddr_in address;
89	>	hostent* host;
90
91	<	HttpHandler& HttpHandler::getline(string& line, char endline)
133	<	{
134	<	int end = page.find(endline, begin);
135	<	int newline = page.find('\n', begin);
91	>	address.sin_family = AF_INET;
92
93	<	if (newline < end || end == string::npos)
93	>	if ((host = gethostbyname(url.getAddress().c_str())) == NULL)
94		{
95	<	end = newline;
95	>	error(program + ": Host: " + url.getAddress(), true);
96	>	return answer;
97	>	}
98	>
99	>	address.sin_addr = *((in_addr*)*host->h_addr_list);
100	>	address.sin_port = htons(url.getPort());
101	>
102	>	if (connect(http, (sockaddr*)&address, sizeof(sockaddr_in)) ==
103	>	SOCKET_ERROR)
104	>	{
105	>	error(program + ": Connect");
106	>	return answer;
107	>	}
108	>
109	>	if (head)
110	>	{
111	>	putline("HEAD " + url.getPath() + " HTTP/1.1");
112	>	}
113	>	else
114	>	{
115	>	putline("GET " + url.getPath() + " HTTP/1.1");
116		}
117
118	<	line = page.substr(begin, end - begin);
118	>	putline("Accept: text/html; text/plain");
119	>	putline("User-Agent: " + agent(true) + ' ' + platform());
120
121	<	if (end == string::npos)
121	>	if (url.getPort() == 80)
122		{
123	<	begin = end;
123	>	putline("Host: " + url.getAddress());
124		}
125		else
126		{
127	<	begin = end + 1;
127	>	char* port = new char[1024];
128	>	sprintf(port, "%u", url.getPort());
129	>
130	>	putline("Host: " + url.getAddress() + ':' + port);
131	>
132	>	delete [] port;
133		}
134
135	<	return *this;
136	<	}
135	>	if (referer != "")
136	>	{
137	>	putline("Referer: " + referer);
138	>	}
139
140	<	bool HttpHandler::good()
141	<	{
142	<	bool answer = true;
140	>	putline("Connection: close");
141	>	putline();
142	>
143	>	code response;
144	>	string line;
145
146	<	if (begin >= page.length())
146	>	do
147		{
148	<	answer = false;
148	>	line = getline();
149	>
150	>	if (line.find("HTTP/") != 0)
151	>	{
152	>	return answer;
153	>	}
154	>
155	>	unsigned dot = line.find('.');
156	>	unsigned space = line.find(' ');
157	>
158	>	unsigned major = strtoul(line.substr(5, dot - 5).c_str(), 0, 10);
159	>	unsigned minor = strtoul(line.substr(dot + 1, space - dot - 1).c_str(),
160	>	0, 10);
161	>
162	>	if (major > 1)
163	>	{
164	>	cerr << program << ": Potentially Incompatible Server: HTTP/" <<
165	>	major << "." << minor << "\n";
166	>
167	>	return answer;
168	>	}
169	>
170	>	response = code(strtoul(line.substr(space + 1).c_str(), 0, 10));
171	>
172	>	if (response < ok) do line = getline(); while (line != "");
173	>	}
174	>	while (response < ok);
175	>
176	>	do
177	>	{
178	>	line = getline();
179	>
180	>	if (line != "")
181	>	{
182	>	unsigned colon = line.find(':');
183	>
184	>	string field = line.substr(0, colon);
185	>	string value = line.substr(colon + 1);
186	>
187	>	while (isspace(value[0])) value.erase(0, 1);
188	>
189	>	if (field == "Content-Type")
190	>	{
191	>	type = value;
192	>	}
193	>	else if (field == "Content-Length")
194	>	{
195	>	length = strtoul(value.c_str(), 0, 10);
196	>	}
197	>	else if (field == "Location")
198	>	{
199	>	location = value;
200	>	}
201	>	else if (field == "Transfer-Encoding")
202	>	{
203	>	chunked = value == "chunked";
204	>	}
205	>	}
206		}
207	<	else if (begin == string::npos)
207	>	while (line != "");
208	>
209	>	switch (response)
210		{
211	<	answer = false;
211	>	case ok:
212	>	if (debug) cerr << "response = " << response << "\n";
213	>	answer = true;
214	>	break;
215	>	case choices:
216	>	case moved:
217	>	case found:
218	>	if (debug) cerr << "response = " << response << "\n"
219	>	<< "location = " << location << "\n";
220	>	location = getLink(location, url);
221	>	break;
222	>	case notfound:
223	>	case internal:
224	>	if (debug) cerr << "response = " << response << "\n";
225	>	break;
226	>	default:
227	>	if (debug) cerr << "response = " << response << "\n";
228	>	if (response <= 299)
229	>	{
230	>	answer = true;
231	>	}
232	>	else if (response <= 399)
233	>	{
234	>	location = getLink(location, url);
235	>	}
236	>	break;
237		}
238
239	+	if (!head && answer) populate();
240	+
241		return answer;
242		}
243
244	+	HttpHandler& HttpHandler::getline(string& line, char endline)
245	+	{
246	+	unsigned end = page.find(endline);
247	+	unsigned newline = page.find('\n');
248	+
249	+	if (newline < end || end == string::npos)
250	+	{
251	+	end = newline;
252	+	}
253	+
254	+	line = page.substr(0, end);
255	+	page.erase(0, (end == string::npos ? end : end + 1));
256	+
257	+	return *this;
258	+	}
259	+
260		void HttpHandler::clear()
261		{
262	<	begin = 0;
262	>	closesocket(http);
263	>
264	>	type = "";
265	>	length = 0;
266	>	location = "";
267		page = "";
268	+	chunked = false;
269		}
270
271	<	void HttpHandler::setJarPath()
271	>	void HttpHandler::populate()
272		{
273	<	const string jarFile = "HttpConnector.jar";
181	<	string jarFilePath = jarFile;
182	<
183	<	ifstream fin(jarFilePath.c_str());
184	<	if (!fin.is_open())
273	>	if (!chunked)
274		{
275	<	unsigned end = program.rfind(PATH_SEPARATOR);
276	<	if (end != string::npos)
275	>	unsigned left = length;
276	>
277	>	while (left > 0)
278		{
279	<	string path = program.substr(0, end);
279	>	memset(buffer, 0, BUFSIZ + 1);
280
281	<	jarFilePath = path + PATH_SEPARATOR + jarFile;
281	>	unsigned bytes = left > BUFSIZ ? BUFSIZ : left;
282	>	unsigned received;
283
284	<	fin.open(jarFilePath.c_str());
194	<	if (!fin.is_open())
284	>	if ((received = recv(http, buffer, bytes, 0)) == SOCKET_ERROR)
285		{
286	<	cerr << program << ": Could not find required file: "
197	<	<< jarFile << "\n";
286	>	error(program + ": Recv");
287		exit(1);
288		}
289	<	else
289	>	else if (received != bytes)
290		{
291	<	fin.close();
291	>	left -= received;
292	>	page += buffer;
293	>
294	>	memset(buffer, 0, BUFSIZ + 1);
295	>
296	>	bytes -= received;
297	>	if (recv(http, buffer, bytes, 0) == SOCKET_ERROR)
298	>	{
299	>	error(program + ": Recv");
300	>	exit(1);
301	>	}
302		}
303	+
304	+	page += buffer;
305	+	left -= bytes;
306		}
307	<	else
307	>	}
308	>	else
309	>	{
310	>	unsigned chunk;
311	>
312	>	do
313		{
314	<	string envPath = getenv("PATH");
314	>	chunk = strtoul(getline().c_str(), 0, 16);
315
316	<	if (debug) cerr << "envPath = " << envPath << "\n";
316	>	unsigned left = chunk;
317
318	<	unsigned begin = 0;
212	<	do
318	>	while (left > 0)
319		{
320	<	unsigned end = envPath.find(ENV_SEPARATOR, begin);
215	<	string path = envPath.substr(begin, end);
320	>	memset(buffer, 0, BUFSIZ + 1);
321
322	<	if (debug) cerr << "path = " << path << "\n";
322	>	unsigned bytes = left > BUFSIZ ? BUFSIZ : left;
323	>	unsigned received;
324
325	<	jarFilePath = path + PATH_SEPARATOR + jarFile;
220	<
221	<	fin.open(jarFilePath.c_str());
222	<	if (fin.is_open())
325	>	if ((received = recv(http, buffer, bytes, 0)) == SOCKET_ERROR)
326		{
327	<	fin.close();
328	<	break;
327	>	error(program + ": Recv");
328	>	exit(1);
329		}
330	+	else if (received != bytes)
331	+	{
332	+	left -= received;
333	+	page += buffer;
334
335	<	begin = end != string::npos ? end + 1 : string::npos;
229	<	}
230	<	while (begin < envPath.length());
335	>	memset(buffer, 0, BUFSIZ + 1);
336
337	<	if (begin == string::npos)
338	<	{
339	<	cerr << program << ": Could not find required file: "
340	<	<< jarFile << "\n";
341	<	exit(1);
337	>	bytes -= received;
338	>	if (recv(http, buffer, bytes, 0) == SOCKET_ERROR)
339	>	{
340	>	error(program + ": Recv");
341	>	exit(1);
342	>	}
343	>	}
344	>
345	>	page += buffer;
346	>	left -= bytes;
347		}
348	+
349	+	getline();
350	+	length += chunk;
351		}
352	+	while (chunk > 0);
353		}
354	<	else
354	>
355	>	for (unsigned index = 0; index < page.length(); index++)
356		{
357	<	fin.close();
357	>	if (page[index] == '\r' && (index + 1 < page.length()) ? page[index +
358	>	1] == '\n' : false)
359	>	{
360	>	page.erase(index, 1);
361	>	}
362	>	else if (page[index] == '\r')
363	>	{
364	>	page[index] = '\n';
365	>	}
366		}
367	+	}
368
369	<	string jarPath = "-Djava.class.path=" + jarFilePath;
369	>	void HttpHandler::putline(const string line)
370	>	{
371	>	sprintf(buffer, "%s\r\n", line.c_str());
372	>	if (send(http, buffer, strlen(buffer), 0) == SOCKET_ERROR)
373	>	{
374	>	error(program + ": Send");
375	>	exit(1);
376	>	}
377	>	}
378
379	<	//      this->jarPath = new char[jarPath.length()];
380	<	//      strcpy(this->jarPath, jarPath.c_str());
379	>	string HttpHandler::getline()
380	>	{
381	>	string line;
382	>	char byte;
383	>
384	>	do
385	>	{
386	>	if (recv(http, &byte, 1, 0) == SOCKET_ERROR)
387	>	{
388	>	error(program + ": Recv");
389	>	}
390	>
391	>	if (byte != '\r' && byte != '\n')
392	>	{
393	>	line += byte;
394	>	}
395	>	}
396	>	while (byte != '\n');
397	>
398	>	return line;
399	>	}
400	>
401	>	void HttpHandler::error(const string& prefix, bool host)
402	>	{
403	>	#ifdef _WIN32
404	>	string error;
405	>
406	>	switch (WSAGetLastError())
407	>	{
408	>	case WSAEACCES:
409	>	error = "Permission denied.";
410	>	break;
411	>	case WSAEADDRINUSE:
412	>	error = "Address already in use.";
413	>	break;
414	>	case WSAEADDRNOTAVAIL:
415	>	error = "Cannot assign requested address.";
416	>	break;
417	>	case WSAEAFNOSUPPORT:
418	>	error = "Address family not supported by protocol family.";
419	>	break;
420	>	case WSAEALREADY:
421	>	error = "Operation already in progress.";
422	>	break;
423	>	case WSAECONNABORTED:
424	>	error = "Software caused connection abort.";
425	>	break;
426	>	case WSAECONNREFUSED:
427	>	error = "Connection refused.";
428	>	break;
429	>	case WSAECONNRESET:
430	>	error = "Connection reset by peer.";
431	>	break;
432	>	case WSAEDESTADDRREQ:
433	>	error = "Destination address required.";
434	>	break;
435	>	case WSAEFAULT:
436	>	error = "Bad address.";
437	>	break;
438	>	case WSAEHOSTDOWN:
439	>	error = "Host is down.";
440	>	break;
441	>	case WSAEHOSTUNREACH:
442	>	error = "No route to host.";
443	>	break;
444	>	case WSAEINPROGRESS:
445	>	error = "Operation now in progress.";
446	>	break;
447	>	case WSAEINTR:
448	>	error = "Interrupted function call.";
449	>	break;
450	>	case WSAEINVAL:
451	>	error = "Invalid argument.";
452	>	break;
453	>	case WSAEISCONN:
454	>	error = "Socket is already connected.";
455	>	break;
456	>	case WSAEMFILE:
457	>	error = "Too many open files.";
458	>	break;
459	>	case WSAEMSGSIZE:
460	>	error = "Message too long.";
461	>	break;
462	>	case WSAENETDOWN:
463	>	error = "Network is down.";
464	>	break;
465	>	case WSAENETRESET:
466	>	error = "Network dropped connection on reset.";
467	>	break;
468	>	case WSAENETUNREACH:
469	>	error = "Network is unreachable.";
470	>	break;
471	>	case WSAENOBUFS:
472	>	error = "No buffer space available.";
473	>	break;
474	>	case WSAENOPROTOOPT:
475	>	error = "Bad protocol option.";
476	>	break;
477	>	case WSAENOTCONN:
478	>	error = "Socket is not connected.";
479	>	break;
480	>	case WSAENOTSOCK:
481	>	error = "Socket operation on non-socket.";
482	>	break;
483	>	case WSAEOPNOTSUPP:
484	>	error = "Operation not supported.";
485	>	break;
486	>	case WSAEPFNOSUPPORT:
487	>	error = "Protocol family not supported.";
488	>	break;
489	>	case WSAEPROCLIM:
490	>	error = "Too many processes.";
491	>	break;
492	>	case WSAEPROTONOSUPPORT:
493	>	error = "Protocol not supported.";
494	>	break;
495	>	case WSAEPROTOTYPE:
496	>	error = "Protocol wrong type for socket.";
497	>	break;
498	>	case WSAESHUTDOWN:
499	>	error = "Cannot send after socket shutdown.";
500	>	break;
501	>	case WSAESOCKTNOSUPPORT:
502	>	error = "Socket type not supported.";
503	>	break;
504	>	case WSAETIMEDOUT:
505	>	error = "Connection timed out.";
506	>	break;
507	>	case WSATYPE_NOT_FOUND:
508	>	error = "Class type not found.";
509	>	break;
510	>	case WSAEWOULDBLOCK:
511	>	error = "Resource temporarily unavailable.";
512	>	break;
513	>	case WSAHOST_NOT_FOUND:
514	>	error = "Host not found.";
515	>	break;
516	>	case WSA_INVALID_HANDLE:
517	>	error = "Specified event object handle is invalid.";
518	>	break;
519	>	case WSA_INVALID_PARAMETER:
520	>	error = "One or more parameters are invalid.";
521	>	break;
522	>	//      case WSAINVALIDPROCTABLE:
523	>	//              error = "Invalid procedure table from service provider.";
524	>	//              break;
525	>	//      case WSAINVALIDPROVIDER:
526	>	//              error = "Invalid service provider version number.";
527	>	//              break;
528	>	case WSA_IO_INCOMPLETE:
529	>	error = "Overlapped I/O event object not in signaled state.";
530	>	break;
531	>	case WSA_IO_PENDING:
532	>	error = "Overlapped operations will complete later.";
533	>	break;
534	>	case WSA_NOT_ENOUGH_MEMORY:
535	>	error = "Insufficient memory available.";
536	>	break;
537	>	case WSANOTINITIALISED:
538	>	error = "Successful WSAStartup not yet performed.";
539	>	break;
540	>	case WSANO_DATA:
541	>	error = "Valid name, no data record of requested type.";
542	>	break;
543	>	case WSANO_RECOVERY:
544	>	error = "This is a non-recoverable error.";
545	>	break;
546	>	//      case WSAPROVIDERFAILEDINIT:
547	>	//              error = "Unable to initialize a service provider.";
548	>	//              break;
549	>	case WSASYSCALLFAILURE:
550	>	error = "System call failure.";
551	>	break;
552	>	case WSASYSNOTREADY:
553	>	error = "Network subsystem is unavailable.";
554	>	break;
555	>	case WSATRY_AGAIN:
556	>	error = "Non-authoritative host not found.";
557	>	break;
558	>	case WSAVERNOTSUPPORTED:
559	>	error = "WINSOCK.DLL version out of range.";
560	>	break;
561	>	case WSAEDISCON:
562	>	error = "Graceful shutdown in progress.";
563	>	break;
564	>	case WSA_OPERATION_ABORTED:
565	>	error = "Overlapped operation aborted.";
566	>	break;
567	>	default:
568	>	error = "Unknown error.";
569	>	break;
570	>	}
571	>
572	>	cerr << prefix << ": " << error << "\n";
573	>	#else
574	>	if (host)
575	>	{
576	>	herror(prefix.c_str());
577	>	}
578	>	else
579	>	{
580	>	perror(prefix.c_str());
581	>	}
582	>	#endif // _WIN32
583		}

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines