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 145 by douglas, 2003-05-31T14:31:50-07:00

#	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	<	line = page.substr(begin, end - begin);
99	>	address.sin_addr = *((in_addr*)*host->h_addr_list);
100	>	address.sin_port = htons(url.getPort());
101
102	<	if (end == string::npos)
102	>	if (connect(http, (sockaddr*)&address, sizeof(sockaddr_in)) ==
103	>	SOCKET_ERROR)
104		{
105	<	begin = end;
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	<	begin = end + 1;
115	>	putline("GET " + url.getPath() + " HTTP/1.1");
116		}
117
118	<	return *this;
119	<	}
118	>	putline("Accept: text/html; text/plain");
119	>	putline("User-Agent: " + agent(true) + ' ' + platform());
120
121	<	bool HttpHandler::good()
122	<	{
123	<	bool answer = true;
121	>	if (url.getPort() == 80)
122	>	{
123	>	putline("Host: " + url.getAddress());
124	>	}
125	>	else
126	>	{
127	>	char* port = new char[1024];
128	>	sprintf(port, "%u", url.getPort());
129	>
130	>	putline("Host: " + url.getAddress() + ':' + port);
131
132	<	if (begin >= page.length())
132	>	delete [] port;
133	>	}
134	>
135	>	if (referer != "")
136		{
137	<	answer = false;
137	>	putline("Referer: " + referer);
138		}
139	<	else if (begin == string::npos)
139	>
140	>	putline("Connection: close");
141	>	putline();
142	>
143	>	code response;
144	>	string line;
145	>
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	+	while (line != "");
208	+
209	+	switch (response)
210	+	{
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	>	while (true)
285		{
286	<	cerr << program << ": Could not find required file: "
287	<	<< jarFile << "\n";
288	<	exit(1);
289	<	}
290	<	else
291	<	{
292	<	fin.close();
286	>	if ((received = recv(http, buffer, bytes, 0)) == SOCKET_ERROR)
287	>	{
288	>	error(program + ": Recv");
289	>	exit(1);
290	>	}
291	>	else if (received != bytes)
292	>	{
293	>	left -= received;
294	>	page += buffer;
295	>
296	>	memset(buffer, 0, BUFSIZ + 1);
297	>
298	>	bytes -= received;
299	>	}
300	>	else
301	>	{
302	>	break;
303	>	}
304		}
305	+
306	+	page += buffer;
307	+	left -= bytes;
308		}
309	<	else
309	>	}
310	>	else
311	>	{
312	>	unsigned chunk;
313	>
314	>	do
315		{
316	<	string envPath = getenv("PATH");
316	>	chunk = strtoul(getline().c_str(), 0, 16);
317
318	<	if (debug) cerr << "envPath = " << envPath << "\n";
318	>	unsigned left = chunk;
319
320	<	unsigned begin = 0;
212	<	do
320	>	while (left > 0)
321		{
322	<	unsigned end = envPath.find(ENV_SEPARATOR, begin);
215	<	string path = envPath.substr(begin, end);
216	<
217	<	if (debug) cerr << "path = " << path << "\n";
322	>	memset(buffer, 0, BUFSIZ + 1);
323
324	<	jarFilePath = path + PATH_SEPARATOR + jarFile;
324	>	unsigned bytes = left > BUFSIZ ? BUFSIZ : left;
325	>	unsigned received;
326
327	<	fin.open(jarFilePath.c_str());
222	<	if (fin.is_open())
327	>	while (true)
328		{
329	<	fin.close();
330	<	break;
329	>	if ((received = recv(http, buffer, bytes, 0)) ==
330	>	SOCKET_ERROR)
331	>	{
332	>	error(program + ": Recv");
333	>	exit(1);
334	>	}
335	>	else if (received != bytes)
336	>	{
337	>	left -= received;
338	>	page += buffer;
339	>
340	>	memset(buffer, 0, BUFSIZ + 1);
341	>
342	>	bytes -= received;
343	>	}
344	>	else
345	>	{
346	>	break;
347	>	}
348		}
349
350	<	begin = end != string::npos ? end + 1 : string::npos;
350	>	page += buffer;
351	>	left -= bytes;
352		}
230	–	while (begin < envPath.length());
353
354	<	if (begin == string::npos)
355	<	{
234	<	cerr << program << ": Could not find required file: "
235	<	<< jarFile << "\n";
236	<	exit(1);
237	<	}
354	>	getline();
355	>	length += chunk;
356		}
357	+	while (chunk > 0);
358		}
359	<	else
359	>
360	>	for (unsigned index = 0; index < page.length(); index++)
361		{
362	<	fin.close();
362	>	if (page[index] == '\r' && (index + 1 < page.length()) ? page[index +
363	>	1] == '\n' : false)
364	>	{
365	>	page.erase(index, 1);
366	>	}
367	>	else if (page[index] == '\r')
368	>	{
369	>	page[index] = '\n';
370	>	}
371	>	}
372	>	}
373	>
374	>	void HttpHandler::putline(const string line)
375	>	{
376	>	sprintf(buffer, "%s\r\n", line.c_str());
377	>	if (send(http, buffer, strlen(buffer), 0) == SOCKET_ERROR)
378	>	{
379	>	error(program + ": Send");
380	>	exit(1);
381	>	}
382	>	}
383	>
384	>	string HttpHandler::getline()
385	>	{
386	>	string line;
387	>	char byte;
388	>
389	>	do
390	>	{
391	>	if (recv(http, &byte, 1, 0) == SOCKET_ERROR)
392	>	{
393	>	error(program + ": Recv");
394	>	}
395	>
396	>	if (byte != '\r' && byte != '\n')
397	>	{
398	>	line += byte;
399	>	}
400		}
401	+	while (byte != '\n');
402
403	<	string jarPath = "-Djava.class.path=" + jarFilePath;
403	>	return line;
404	>	}
405
406	<	//      this->jarPath = new char[jarPath.length()];
407	<	//      strcpy(this->jarPath, jarPath.c_str());
406	>	void HttpHandler::error(const string& prefix, bool host)
407	>	{
408	>	#ifdef _WIN32
409	>	string error;
410	>
411	>	switch (WSAGetLastError())
412	>	{
413	>	case WSAEACCES:
414	>	error = "Permission denied.";
415	>	break;
416	>	case WSAEADDRINUSE:
417	>	error = "Address already in use.";
418	>	break;
419	>	case WSAEADDRNOTAVAIL:
420	>	error = "Cannot assign requested address.";
421	>	break;
422	>	case WSAEAFNOSUPPORT:
423	>	error = "Address family not supported by protocol family.";
424	>	break;
425	>	case WSAEALREADY:
426	>	error = "Operation already in progress.";
427	>	break;
428	>	case WSAECONNABORTED:
429	>	error = "Software caused connection abort.";
430	>	break;
431	>	case WSAECONNREFUSED:
432	>	error = "Connection refused.";
433	>	break;
434	>	case WSAECONNRESET:
435	>	error = "Connection reset by peer.";
436	>	break;
437	>	case WSAEDESTADDRREQ:
438	>	error = "Destination address required.";
439	>	break;
440	>	case WSAEFAULT:
441	>	error = "Bad address.";
442	>	break;
443	>	case WSAEHOSTDOWN:
444	>	error = "Host is down.";
445	>	break;
446	>	case WSAEHOSTUNREACH:
447	>	error = "No route to host.";
448	>	break;
449	>	case WSAEINPROGRESS:
450	>	error = "Operation now in progress.";
451	>	break;
452	>	case WSAEINTR:
453	>	error = "Interrupted function call.";
454	>	break;
455	>	case WSAEINVAL:
456	>	error = "Invalid argument.";
457	>	break;
458	>	case WSAEISCONN:
459	>	error = "Socket is already connected.";
460	>	break;
461	>	case WSAEMFILE:
462	>	error = "Too many open files.";
463	>	break;
464	>	case WSAEMSGSIZE:
465	>	error = "Message too long.";
466	>	break;
467	>	case WSAENETDOWN:
468	>	error = "Network is down.";
469	>	break;
470	>	case WSAENETRESET:
471	>	error = "Network dropped connection on reset.";
472	>	break;
473	>	case WSAENETUNREACH:
474	>	error = "Network is unreachable.";
475	>	break;
476	>	case WSAENOBUFS:
477	>	error = "No buffer space available.";
478	>	break;
479	>	case WSAENOPROTOOPT:
480	>	error = "Bad protocol option.";
481	>	break;
482	>	case WSAENOTCONN:
483	>	error = "Socket is not connected.";
484	>	break;
485	>	case WSAENOTSOCK:
486	>	error = "Socket operation on non-socket.";
487	>	break;
488	>	case WSAEOPNOTSUPP:
489	>	error = "Operation not supported.";
490	>	break;
491	>	case WSAEPFNOSUPPORT:
492	>	error = "Protocol family not supported.";
493	>	break;
494	>	case WSAEPROCLIM:
495	>	error = "Too many processes.";
496	>	break;
497	>	case WSAEPROTONOSUPPORT:
498	>	error = "Protocol not supported.";
499	>	break;
500	>	case WSAEPROTOTYPE:
501	>	error = "Protocol wrong type for socket.";
502	>	break;
503	>	case WSAESHUTDOWN:
504	>	error = "Cannot send after socket shutdown.";
505	>	break;
506	>	case WSAESOCKTNOSUPPORT:
507	>	error = "Socket type not supported.";
508	>	break;
509	>	case WSAETIMEDOUT:
510	>	error = "Connection timed out.";
511	>	break;
512	>	case WSATYPE_NOT_FOUND:
513	>	error = "Class type not found.";
514	>	break;
515	>	case WSAEWOULDBLOCK:
516	>	error = "Resource temporarily unavailable.";
517	>	break;
518	>	case WSAHOST_NOT_FOUND:
519	>	error = "Host not found.";
520	>	break;
521	>	case WSA_INVALID_HANDLE:
522	>	error = "Specified event object handle is invalid.";
523	>	break;
524	>	case WSA_INVALID_PARAMETER:
525	>	error = "One or more parameters are invalid.";
526	>	break;
527	>	//      case WSAINVALIDPROCTABLE:
528	>	//              error = "Invalid procedure table from service provider.";
529	>	//              break;
530	>	//      case WSAINVALIDPROVIDER:
531	>	//              error = "Invalid service provider version number.";
532	>	//              break;
533	>	case WSA_IO_INCOMPLETE:
534	>	error = "Overlapped I/O event object not in signaled state.";
535	>	break;
536	>	case WSA_IO_PENDING:
537	>	error = "Overlapped operations will complete later.";
538	>	break;
539	>	case WSA_NOT_ENOUGH_MEMORY:
540	>	error = "Insufficient memory available.";
541	>	break;
542	>	case WSANOTINITIALISED:
543	>	error = "Successful WSAStartup not yet performed.";
544	>	break;
545	>	case WSANO_DATA:
546	>	error = "Valid name, no data record of requested type.";
547	>	break;
548	>	case WSANO_RECOVERY:
549	>	error = "This is a non-recoverable error.";
550	>	break;
551	>	//      case WSAPROVIDERFAILEDINIT:
552	>	//              error = "Unable to initialize a service provider.";
553	>	//              break;
554	>	case WSASYSCALLFAILURE:
555	>	error = "System call failure.";
556	>	break;
557	>	case WSASYSNOTREADY:
558	>	error = "Network subsystem is unavailable.";
559	>	break;
560	>	case WSATRY_AGAIN:
561	>	error = "Non-authoritative host not found.";
562	>	break;
563	>	case WSAVERNOTSUPPORTED:
564	>	error = "WINSOCK.DLL version out of range.";
565	>	break;
566	>	case WSAEDISCON:
567	>	error = "Graceful shutdown in progress.";
568	>	break;
569	>	case WSA_OPERATION_ABORTED:
570	>	error = "Overlapped operation aborted.";
571	>	break;
572	>	default:
573	>	error = "Unknown error.";
574	>	break;
575	>	}
576	>
577	>	cerr << prefix << ": " << error << "\n";
578	>	#else
579	>	if (host)
580	>	{
581	>	herror(prefix.c_str());
582	>	}
583	>	else
584	>	{
585	>	perror(prefix.c_str());
586	>	}
587	>	#endif // _WIN32
588		}

Diff Legend

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