ViewVC Help

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

root/repos/SiteMapper/SiteMapper.cpp

Comparing SiteMapper/SiteMapper.cpp (file contents):
Revision 129 by Douglas Thrift, 2004-03-23T22:49:41-08:00 vs.
Revision 154 by Douglas Thrift, 2004-06-04T02:40:33-07:00

#	Line 4 | Line 4
4		//
5		// $Id$
6
7	+	#ifdef _WIN32
8	+	#pragma warning(disable:4503)
9	+	#endif
10	+
11		#include "SiteMapper.hpp"
12		#include "Matcher.hpp"
13		#include "Page.hpp"
14
11	–	#include <xalanc/Include/PlatformDefinitions.hpp>
12	–	#include <xercesc/util/PlatformUtils.hpp>
13	–	#include <xalanc/XPath/XPathEvaluator.hpp>
14	–
15	–	XALAN_USING_XERCES(XMLPlatformUtils)
16	–	XALAN_USING_XALAN(XPathEvaluator)
17	–
15		string program;
16	<	bool debug = false;
16	>	bool debug(false);
17
18		int main(int argc, char* argv[])
19		{
20		program = argv[0];
21	<
21	>
22		string siteIndex, siteMap;
23
24	<	for (int index = 1; index < argc; argc++)
24	>	for (int index(1); index < argc; index++)
25		{
26		string arg(argv[index]);
27		Matcher matcher;
#	Line 37 | Line 34 | int main(int argc, char* argv[])
34		{
35		siteMap = matcher[1];
36		}
37	+	else if (arg == "-D")
38	+	{
39	+	if (!debug) debug = true;
40	+	}
41		}
42
43		if (siteIndex != "" && siteMap != "")
44	<	{
44	<	XMLPlatformUtils::Initialize();
45	<	XPathEvaluator::initialize();
46	<
44	>	{
45		SiteMapper mapper(siteIndex, siteMap);
48	–
49	–	XPathEvaluator::terminate();
50	–	XMLPlatformUtils::Terminate();
46		}
47		else
48		{
49	<	cout << "Usage: " << program << " -index=index -map=map\n";
49	>	cout << "Usage: " << program << " -index=index -map=map [-D]\n";
50		}
51
52		return 0;
#	Line 59 | Line 54 | int main(int argc, char* argv[])
54
55		SiteMapper::SiteMapper(const string& siteIndex, const string& siteMap)
56		{
57	<	//
57	>	oldMap(siteMap);
58	>	newIndex(siteIndex);
59	>	newMap(siteMap);
60	>	}
61	>
62	>	void SiteMapper::oldMap(const string& siteMap)
63	>	{
64	>	ext::Handle<xml::Document> document(xml::Parse(siteMap));
65	>	ext::Handle<xml::Node> list(*document/"page"/"section"/"list");
66	>
67	>	comment = *document/"comment()";
68	>
69	>	oldMap(pages, list);
70	>	}
71	>
72	>	void SiteMapper::oldMap(vector<Page>& pages, xml::Node* list)
73	>	{
74	>	xml::NodeSet nodes(*list/"item");
75	>
76	>	for (xml::NodeSet::Iterator node(nodes.Begin()); node != nodes.End();
77	>	++node)
78	>	{
79	>	string url(**node/"link"/"@address"), title(**node/"link");
80	>	Page page(url, title);
81	>	ext::Handle<xml::Node> list(**node/"list");
82	>
83	>	if (!list.IsEmpty()) oldMap(page.getChildren(), list);
84	>
85	>	pages.push_back(page);
86	>	}
87	>	}
88	>
89	>	void SiteMapper::newIndex(const string& siteIndex)
90	>	{
91	>	ext::Handle<xml::Document> document(xml::Parse(siteIndex));
92	>	xml::NodeSet nodes(*document/"index"/"page");
93	>
94	>	for (xml::NodeSet::Iterator node(nodes.Begin()); node != nodes.End();
95	>	++node)
96	>	{
97	>	string address(**node/"address");
98	>	string port(**node/"port");
99	>
100	>	if (!port.empty())
101	>	{
102	>	address += ':' + port;
103	>	}
104	>
105	>	string path(**node/"path"), title(**node/"title");
106	>	Page page(address, path, title);
107	>	Matcher matcher;
108	>
109	>	if (page == matcher(string("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDou")
110	>	+ "glas\\sThrift's\\sBlog:\\s(.+)$"))
111	>	{
112	>	if (Matcher("^\\w+\\s\\d\\d\\d\\d\\sArchives$") == matcher[1])
113	>	{
114	>	page.setTitle(matcher[1]);
115	>
116	>	if (newIndex(pages, page)) continue;
117	>	}
118	>	else continue;
119	>	}
120	>	else if (page == matcher("^Douglas\\sThrift's.+Website\\s\\|\\s(.+)$"))
121	>	{
122	>	page.setTitle(matcher[1]);
123	>
124	>	if (newIndex(pages, page)) continue;
125	>	}
126	>	else continue;
127	>
128	>	multimap<string, Page> items;
129	>
130	>	newPages.insert(pair<string, multimap<string, Page>
131	>	>(page.getAddress(), items)).first->second.insert(pair<string,
132	>	Page>(page.getChildOf(), page));
133	>	}
134	>	}
135	>
136	>	bool SiteMapper::newIndex(vector<Page>& pages, Page& page)
137	>	{
138	>	for (unsigned index(0); index < pages.size(); ++index)
139	>	{
140	>	if (pages[index] == page.getAddress())
141	>	{
142	>	Matcher matcher;
143	>
144	>	if (pages[index] == page)
145	>	{
146	>	page.setChildren(pages[index].getChildren());
147	>
148	>	pages[index] = page;
149	>
150	>	return true;
151	>	}
152	>	else if (matcher('^' + pages[index].getPath()) == page)
153	>	{
154	>	page.setChildOf(matcher[0]);
155	>
156	>	if (matcher('^' + pages[index].getTitle() + "\\s\\|\\s(.+)$")
157	>	== page)
158	>	{
159	>	page.setTitle(matcher[1]);
160	>	}
161	>
162	>	return newIndex(pages[index].getChildren(), page);
163	>	}
164	>	}
165	>	}
166	>
167	>	return false;
168	>	}
169	>
170	>	void SiteMapper::newMap(const string& siteMap)
171	>	{
172	>	ofstream fout(siteMap.c_str());
173	>
174	>	fout << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
175	>	<< "<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl"
176	>	<< "\"?>\n"
177	>	<< "<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">\n"
178	>	<< "<!--" << comment << "-->\n"
179	>	<< "<page>\n"
180	>	<< "\t<title>Sitemap</title>\n"
181	>	<< "\t<section>\n"
182	>	<< "\t\t<list>\n";
183	>
184	>	for (unsigned index(0); index < pages.size(); ++index)
185	>	{
186	>	if (newPages.find(pages[index].getAddress()) != newPages.end())
187	>	{
188	>	newMap(pages[index].getChildren(), pages[index].getPath(),
189	>	newPages.find(pages[index].getAddress())->second);
190	>	}
191	>
192	>	fout << pages[index](3) << '\n';
193	>	}
194	>
195	>	fout << "\t\t</list>\n"
196	>	<< "\t</section>\n"
197	>	<< "</page>\n";
198	>
199	>	fout.close();
200	>	}
201	>
202	>	void SiteMapper::newMap(vector<Page>& pages, const string& childOf,
203	>	multimap<string, Page>& newPages)
204	>	{
205	>	for (unsigned index(0); index < pages.size(); ++index)
206	>	{
207	>	newMap(pages[index].getChildren(), pages[index].getPath(), newPages);
208	>	}
209	>
210	>	for (multimap<string, Page>::iterator itor(newPages.lower_bound(childOf));
211	>	itor != newPages.upper_bound(childOf); itor++)
212	>	{
213	>	pages.push_back(itor->second);
214	>	}
215	>
216	>	newPages.erase(childOf);
217		}

Diff Legend

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