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 134 by Douglas Thrift, 2004-03-24T22:23:19-08:00 vs.
Revision 156 by Douglas Thrift, 2004-06-10T01:24:32-07:00

#	Line 8 | Line 8
8		#include "Matcher.hpp"
9		#include "Page.hpp"
10
11	–	#include <xalanc/Include/PlatformDefinitions.hpp>
12	–	#include <xercesc/util/PlatformUtils.hpp>
13	–	#include <xercesc/framework/LocalFileInputSource.hpp>
14	–	#include <xalanc/XPath/XObject.hpp>
15	–	#include <xalanc/XPath/XPathEvaluator.hpp>
16	–	#include <xalanc/DOMSupport/XalanDocumentPrefixResolver.hpp>
17	–	#include <xalanc/XalanSourceTree/XalanSourceTreeDOMSupport.hpp>
18	–	#include <xalanc/XalanSourceTree/XalanSourceTreeInit.hpp>
19	–	#include <xalanc/XalanSourceTree/XalanSourceTreeParserLiaison.hpp>
20	–
21	–	XALAN_USING_XERCES(XMLPlatformUtils)
22	–	XALAN_USING_XERCES(LocalFileInputSource)
23	–	XALAN_USING_XALAN(XObjectPtr)
24	–	XALAN_USING_XALAN(XPathEvaluator)
25	–	XALAN_USING_XALAN(XalanDocument)
26	–	XALAN_USING_XALAN(XalanDocumentPrefixResolver)
27	–	XALAN_USING_XALAN(XalanDOMString)
28	–	XALAN_USING_XALAN(XalanNode)
29	–	XALAN_USING_XALAN(XalanSourceTreeInit)
30	–	XALAN_USING_XALAN(XalanSourceTreeDOMSupport)
31	–	XALAN_USING_XALAN(XalanSourceTreeParserLiaison)
32	–
11		string program;
12	<	bool debug = false;
12	>	bool debug(false);
13
14		int main(int argc, char* argv[])
15		{
#	Line 39 | Line 17 | int main(int argc, char* argv[])
17
18		string siteIndex, siteMap;
19
20	<	for (int index = 1; index < argc; index++)
20	>	for (int index(1); index < argc; index++)
21		{
22		string arg(argv[index]);
23		Matcher matcher;
#	Line 58 | Line 36 | int main(int argc, char* argv[])
36		}
37		}
38
39	<	if (siteIndex != "" && siteMap != "")
39	>	if (!siteIndex.empty() && !siteMap.empty())
40		{
63	–	XMLPlatformUtils::Initialize();
64	–	XPathEvaluator::initialize();
65	–
41		SiteMapper mapper(siteIndex, siteMap);
67	–
68	–	XPathEvaluator::terminate();
69	–	XMLPlatformUtils::Terminate();
42		}
43		else
44		{
#	Line 78 | Line 50 | int main(int argc, char* argv[])
50
51		SiteMapper::SiteMapper(const string& siteIndex, const string& siteMap)
52		{
81	–	XalanSourceTreeInit init;
82	–
53		oldMap(siteMap);
54	<	index(siteIndex);
54	>	newIndex(siteIndex);
55		newMap(siteMap);
56		}
57
58		void SiteMapper::oldMap(const string& siteMap)
59		{
60	<	XalanSourceTreeDOMSupport support;
61	<	XalanSourceTreeParserLiaison liaison(support);
60	>	ext::Handle<xml::Document> document(xml::Parse(siteMap));
61	>	ext::Handle<xml::Node> list(*document/"page"/"section"/"list");
62
63	<	support.setParserLiaison(&liaison);
63	>	comment = *document/"comment()";
64
65	<	XalanDOMString file(siteMap.c_str());
66	<	LocalFileInputSource source(file.c_str());
65	>	oldMap(pages, list);
66	>	}
67
68	<	XalanDocument* document = liaison.parseXMLStream(source);
68	>	void SiteMapper::oldMap(vector<Page>& pages, xml::Node* list)
69	>	{
70	>	xml::NodeSet nodes(*list/"item");
71
72	<	if (document == 0) return;
72	>	for (xml::NodeSet::Iterator node(nodes.Begin()); node != nodes.End();
73	>	++node)
74	>	{
75	>	string url(**node/"link"/"@address"), title(**node/"link");
76	>	Page page(url, title);
77	>	ext::Handle<xml::Node> list(**node/"list");
78
79	<	XPathEvaluator evaluator;
103	<	XalanNode* list = evaluator.selectSingleNode(support, document,
104	<	XalanDOMString("/page/section/list").c_str());
79	>	if (!list.IsEmpty()) oldMap(page.getChildren(), list);
80
81	<	if (list == 0) return;
81	>	pages.push_back(page);
82	>	}
83		}
84
85	<	void SiteMapper::index(const string& siteIndex)
85	>	void SiteMapper::newIndex(const string& siteIndex)
86		{
87	<	//
87	>	ext::Handle<xml::Document> document(xml::Parse(siteIndex));
88	>	xml::NodeSet nodes(*document/"index"/"page");
89	>
90	>	for (xml::NodeSet::Iterator node(nodes.Begin()); node != nodes.End();
91	>	++node)
92	>	{
93	>	string address(**node/"address");
94	>	string port(**node/"port");
95	>
96	>	if (!port.empty())
97	>	{
98	>	address += ':' + port;
99	>	}
100	>
101	>	string path(**node/"path"), title(**node/"title");
102	>	Page page(address, path, title);
103	>	Matcher matcher;
104	>
105	>	if (page == matcher(string("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDou")
106	>	+ "glas\\sThrift's\\sBlog:\\s(.+)$"))
107	>	{
108	>	if (Matcher("^\\w+\\s\\d{4}\\sArchives$") == matcher[1])
109	>	{
110	>	page.setTitle(matcher[1]);
111	>
112	>	if (newIndex(pages, page)) continue;
113	>	}
114	>	else continue;
115	>	}
116	>	else if (page == matcher("^Douglas\\sThrift's.+Website\\s\\|\\s(.+)$"))
117	>	{
118	>	page.setTitle(matcher[1]);
119	>
120	>	if (newIndex(pages, page)) continue;
121	>	}
122	>	else continue;
123	>
124	>	multimap<string, Page> items;
125	>
126	>	newPages.insert(pair<string, multimap<string, Page>
127	>	>(page.getAddress(), items)).first->second.insert(pair<string,
128	>	Page>(page.getChildOf(), page));
129	>	}
130	>	}
131	>
132	>	bool SiteMapper::newIndex(vector<Page>& pages, Page& page)
133	>	{
134	>	for (unsigned index(0); index < pages.size(); ++index)
135	>	{
136	>	if (pages[index] == page.getAddress())
137	>	{
138	>	Matcher matcher;
139	>
140	>	if (pages[index] == page)
141	>	{
142	>	page.setChildren(pages[index].getChildren());
143	>
144	>	pages[index] = page;
145	>
146	>	cout << "Updated: " << page.getUrl() << '\n';
147	>
148	>	return true;
149	>	}
150	>	else if (matcher('^' + pages[index].getPath()) == page)
151	>	{
152	>	page.setChildOf(matcher[0]);
153	>
154	>	if (matcher('^' + pages[index].getTitle() + "\\s\\|\\s(.+)$")
155	>	== page)
156	>	{
157	>	page.setTitle(matcher[1]);
158	>	}
159	>
160	>	return newIndex(pages[index].getChildren(), page);
161	>	}
162	>	}
163	>	}
164	>
165	>	return false;
166		}
167
168		void SiteMapper::newMap(const string& siteMap)
169		{
170	<	//
170	>	ofstream fout(siteMap.c_str());
171	>
172	>	fout << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
173	>	<< "<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl"
174	>	<< "\"?>\n"
175	>	<< "<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">\n"
176	>	<< "<!--" << comment << "-->\n"
177	>	<< "<page>\n"
178	>	<< "\t<title>Sitemap</title>\n"
179	>	<< "\t<section>\n"
180	>	<< "\t\t<list>\n";
181	>
182	>	for (unsigned index(0); index < pages.size(); ++index)
183	>	{
184	>	if (newPages.find(pages[index].getAddress()) != newPages.end())
185	>	{
186	>	newMap(pages[index].getChildren(), pages[index].getPath(),
187	>	newPages.find(pages[index].getAddress())->second);
188	>	}
189	>
190	>	fout << pages[index](3) << '\n';
191	>	}
192	>
193	>	fout << "\t\t</list>\n"
194	>	<< "\t</section>\n"
195	>	<< "</page>\n";
196	>
197	>	fout.close();
198	>	}
199	>
200	>	void SiteMapper::newMap(vector<Page>& pages, const string& childOf,
201	>	multimap<string, Page>& newPages)
202	>	{
203	>	for (unsigned index(0); index < pages.size(); ++index)
204	>	{
205	>	newMap(pages[index].getChildren(), pages[index].getPath(), newPages);
206	>	}
207	>
208	>	for (multimap<string, Page>::iterator itor(newPages.lower_bound(childOf));
209	>	itor != newPages.upper_bound(childOf); itor++)
210	>	{
211	>	cout << "Added: " << itor->second.getUrl() << '\n';
212	>
213	>	pages.push_back(itor->second);
214	>	}
215	>
216	>	newPages.erase(childOf);
217		}

Diff Legend

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