ViewVC Help

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

root/proj/Search/trunk/Processor.cpp

Comparing:
trunk/Search/Processor.cpp (file contents), Revision 348 by Douglas Thrift, 2004-05-26T17:44:17-07:00 vs.
Search/trunk/Processor.cpp (file contents), Revision 372 by douglas, 2008-08-23T04:00:12-07:00

#	Line 1 | Line 1
1		/* ============================================================================
2		* Douglas Thrift's Search Engine License
3		*
4	<	* Copyright (C) 2002-2004, Douglas Thrift. All Rights Reserved.
4	>	* Copyright (C) 2002-2004, 2008, 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 136 | Line 136 | bool Processor::process(HttpHandler& htt
136		{
137		http.getline(line);
138
139	<	unsigned begin(0), startComment(0);
139	>	size_t begin(0), startComment(0);
140
141		while (begin < line.length())
142		{
143	<	unsigned open(line.find('<', begin)), close(line.find('>', begin));
143	>	size_t open(line.find('<', begin)), close(line.find('>', begin));
144		string next;
145
146		while (close == string::npos && http.good())
#	Line 179 | Line 179 | bool Processor::process(HttpHandler& htt
179		== 0) || (lowerTag.find("meta   ") == 0)) && inHead)
180		{
181		if (lowerTag.find("name=robots") != string::npos ||
182	<	lowerTag.find("name=\"robots\"") != string::npos)
182	>	lowerTag.find("name=\"robots\"") != string::npos ||
183	>	lowerTag.find("name='robots'") != string::npos)
184		{
185	<	unsigned start(lowerTag.find("content=\"") + 9),
185	>	size_t start(lowerTag.find("content=\"") + 9),
186		finish(lowerTag.find('\"', start));
187		string robots(lowerTag.substr(start, finish - start));
188
#	Line 208 | Line 209 | bool Processor::process(HttpHandler& htt
209		}
210		}
211		else if (lowerTag.find("name=description") != string::npos
212	<	|| lowerTag.find("name=\"description\"") !=
213	<	string::npos)
212	>	|| lowerTag.find("name=\"description\"") != string::npos
213	>	|| lowerTag.find("name='description'") != string::npos)
214		{
215	<	unsigned start(lowerTag.find("content=\"") + 9),
215	>	size_t start(lowerTag.find("content=\"") + 9),
216		finish(lowerTag.find('\"', start));
217
218		description = tag.substr(start, finish - start);
#	Line 223 | Line 224 | bool Processor::process(HttpHandler& htt
224		{
225		if (lowerTag.find("href=\"") != string::npos)
226		{
227	<	unsigned start(lowerTag.find("href=\"") + 6),
227	>	size_t start(lowerTag.find("href=\"") + 6),
228		finish(lowerTag.find('\"', start));
229	+	string link(getLink(tag.substr(start, finish -
230	+	start), url));
231
232	<	string link = getLink(tag.substr(start, finish -
233	<	start), url);
232	>	if (!link.empty()) links.insert(link);
233	>	}
234	>	else if (lowerTag.find("href='") != string::npos)
235	>	{
236	>	size_t start(lowerTag.find("href='") + 6),
237	>	finish(lowerTag.find('\'', start));
238	>	string link(getLink(tag.substr(start, finish -
239	>	start), url));
240
241	<	if (link != "") links.insert(link);
241	>	if (!link.empty()) links.insert(link);
242		}
243		else if (lowerTag.find("href=") != string::npos)
244		{
245	<	unsigned start(lowerTag.find("href=") + 5),
245	>	size_t start(lowerTag.find("href=") + 5),
246		finish(lowerTag.find(' ', start));
247
248		if (finish < close)
249		{
250	<	string link = getLink(tag.substr(start, finish -
251	<	start), url);
250	>	string link(getLink(tag.substr(start, finish -
251	>	start), url));
252
253	<	if (link != "") links.insert(link);
253	>	if (!link.empty()) links.insert(link);
254		}
255		else
256		{
257	<	string link = getLink(tag.substr(start, close -
258	<	start), url);
257	>	string link(getLink(tag.substr(start, close -
258	>	start), url));
259
260	<	if (link != "") links.insert(link);
260	>	if (!link.empty()) links.insert(link);
261		}
262		}
263		}
#	Line 258 | Line 267 | bool Processor::process(HttpHandler& htt
267		{
268		if (lowerTag.find("alt=\"") != string::npos)
269		{
270	<	unsigned start(lowerTag.find("alt=\"") + 5),
270	>	size_t start(lowerTag.find("alt=\"") + 5),
271		finish(lowerTag.find('\"', start));
272
273		text += tag.substr(start, finish - start) + ' ';
#	Line 268 | Line 277 | bool Processor::process(HttpHandler& htt
277		}
278		else if (lowerTag.find("alt=") != string::npos)
279		{
280	<	unsigned start(lowerTag.find("alt=") + 4),
280	>	size_t start(lowerTag.find("alt=") + 4),
281		finish(lowerTag.find(' ', start));
282
283		if (finish < close)
#	Line 304 | Line 313 | bool Processor::process(HttpHandler& htt
313		lowerTag.find("h3") == 0 || lowerTag.find("h4") == 0 ||
314		lowerTag.find("h5") == 0 || lowerTag.find("h6") == 0)
315		{
316	<	heading = "";
316	>	heading.erase();
317	>
318		inHeading = true;
319		}
320
#	Line 312 | Line 322 | bool Processor::process(HttpHandler& htt
322		lowerTag.find("/h3") == 0 || lowerTag.find("/h4") == 0 ||
323		lowerTag.find("/h5") == 0 || lowerTag.find("/h6") == 0)
324		{
325	<	if (heading != "") headings.push_back(heading);
325	>	if (!heading.empty()) headings.push_back(heading);
326
327		inHeading = false;
328		}
#	Line 326 | Line 336 | bool Processor::process(HttpHandler& htt
336		if (line.find("-->", begin) >= startComment && line.find("-->",
337		begin) != string::npos)
338		{
339	<	close = line.find("-->", begin) + 3;
339	>	close = line.find("-->", begin) + 2;
340		inComment = false;
341		}
342
#	Line 344 | Line 354 | bool Processor::process(HttpHandler& htt
354		return answer;
355		}
356
357	<	string Processor::getTag(const string& line, unsigned open, unsigned close)
357	>	string Processor::getTag(const string& line, size_t open, size_t close)
358		{
359		return line.substr(open + 1, close - open - 1);
360		}

Diff Legend

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