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 13 by douglas, 2002-12-06T19:56:56-08:00 vs.
Revision 25 by douglas, 2002-12-22T23:32:58-08:00

#	Line 52 | Line 52
52
53		HttpHandler::HttpHandler()
54		{
55	+	buffer = new char[BUFSIZ + 1];
56	+
57		#ifdef _WIN32
58	<	WSADATA data;
57	<	if (WSAStartup(MAKEWORD(2, 0) != 0, &data))
58	>	if (WSAStartup(MAKEWORD(2, 0), &data) != 0)
59		{
60	<	error(program + ": WSAStartup()");
60	>	error(program + ": WSAStartup");
61		exit(1);
62		}
63		#endif // _WIN32
64
65	<	begin = 0;
65	>	length = 0;
66	>	chunked = false;
67		}
68
69		HttpHandler::~HttpHandler()
70		{
71	+	delete [] buffer;
72	+
73		#ifdef _WIN32
74		WSACleanup();
75		#endif // _WIN32
76		}
77
78	<	bool HttpHandler::connect(URL &url, bool head)
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)
87	<	//                      {
88	<	//                              jstring addressJ = env->NewStringUTF(url.getAddress().c_str());
89	<	//                              jint portJ = url.getPort();
90	<	//                              jstring pathJ = env->NewStringUTF(url.getPath().c_str());
91	<	//                              jstring programNameJ = env->NewStringUTF(programName.c_str());
92	<	//                              jstring programVersionJ =
93	<	//                                      env->NewStringUTF(programVersion.c_str());
94	<
95	<	//                              jstring pageJ = (jstring)env->CallStaticObjectMethod(cls, mid,
96	<	//                                      addressJ, portJ, pathJ, programNameJ, programVersionJ);
97	<
98	<	//                              const char* pageC = env->GetStringUTFChars(pageJ, 0);
99	<	//                              page = pageC;
100	<	//                              env->ReleaseStringUTFChars(pageJ, pageC);
101	<
102	<	//                              if (page != "unknown host\n" && page != "io exception\n" &&
103	<	//                                      page != "bad headers\n") answer = true;
104	<	//                      }
105	<	//              }
106	<	//      }
82	>	if ((http = socket(PF_INET, SOCK_STREAM, 0)) == INVALID_SOCKET)
83	>	{
84	>	error(program + ": Socket");
85	>	exit(1);
86	>	}
87	>
88	>	sockaddr_in address;
89	>	hostent* host;
90	>
91	>	address.sin_family = AF_INET;
92	>
93	>	if ((host = gethostbyname(url.getAddress().c_str())) == NULL)
94	>	{
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	>	putline("Accept: text/html; text/plain");
119	>	putline("User-Agent: " + agent(true) + ' ' + platform());
120	>
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	>	delete [] port;
133	>	}
134	>
135	>	if (referer != "")
136	>	{
137	>	putline("Referer: " + referer);
138	>	}
139	>
140	>	putline("Connection: close");
141	>	putline();
142	>
143	>	code response;
144	>	string line;
145	>
146	>	do
147	>	{
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	<	int end = page.find(endline, begin);
247	<	int newline = page.find('\n', begin);
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(begin, end - begin);
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	>	closesocket(http);
263
264	<	if (end == string::npos)
264	>	type = "";
265	>	length = 0;
266	>	location = "";
267	>	page = "";
268	>	chunked = false;
269	>	}
270	>
271	>	void HttpHandler::populate()
272	>	{
273	>	if (!chunked)
274		{
275	<	begin = end;
275	>	unsigned left = length;
276	>
277	>	while (left > 0)
278	>	{
279	>	memset(buffer, 0, BUFSIZ + 1);
280	>
281	>	unsigned bytes = left > BUFSIZ ? BUFSIZ : left;
282	>	unsigned received;
283	>
284	>	if ((received = recv(http, buffer, bytes, 0)) == SOCKET_ERROR)
285	>	{
286	>	error(program + ": Recv");
287	>	exit(1);
288	>	}
289	>	else if (received != bytes)
290	>	{
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		}
308		else
309		{
310	<	begin = end + 1;
310	>	unsigned chunk;
311	>
312	>	do
313	>	{
314	>	chunk = strtoul(getline().c_str(), 0, 16);
315	>
316	>	unsigned left = chunk;
317	>
318	>	while (left > 0)
319	>	{
320	>	memset(buffer, 0, BUFSIZ + 1);
321	>
322	>	unsigned bytes = left > BUFSIZ ? BUFSIZ : left;
323	>	unsigned received;
324	>
325	>	if ((received = recv(http, buffer, bytes, 0)) == SOCKET_ERROR)
326	>	{
327	>	error(program + ": Recv");
328	>	exit(1);
329	>	}
330	>	else if (received != bytes)
331	>	{
332	>	left -= received;
333	>	page += buffer;
334	>
335	>	memset(buffer, 0, BUFSIZ + 1);
336	>
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
355	<	return *this;
355	>	for (unsigned index = 0; index < page.length(); index++)
356	>	{
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	<	bool HttpHandler::good()
369	>	void HttpHandler::putline(const string line)
370		{
371	<	bool answer = true;
372	<
135	<	if (begin >= page.length())
136	<	{
137	<	answer = false;
138	<	}
139	<	else if (begin == string::npos)
371	>	sprintf(buffer, "%s\r\n", line.c_str());
372	>	if (send(http, buffer, strlen(buffer), 0) == SOCKET_ERROR)
373		{
374	<	answer = false;
374	>	error(program + ": Send");
375	>	exit(1);
376		}
143	–
144	–	return answer;
377		}
378
379	<	void HttpHandler::clear()
379	>	string HttpHandler::getline()
380		{
381	<	begin = 0;
382	<	page = "";
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)
401	>	void HttpHandler::error(const string& prefix, bool host)
402		{
403		#ifdef _WIN32
404		string error;

Diff Legend

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