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 137 by Douglas Thrift, 2004-03-25T02:08:18-08:00 vs.
Revision 303 by douglas, 2004-12-12T17:58:03-08:00

#	Line 8 | Line 8
8		#include "Matcher.hpp"
9		#include "Page.hpp"
10
11	<	string program;
12	<	bool debug = false;
11	>	#include <menes-api/exename.hpp>
12	>	#include <menes-app/application.hpp>
13
14	<	int main(int argc, char* argv[])
15	<	{
16	<	program = argv[0];
17	<
18	<	string siteIndex, siteMap;
14	>	ext::String program;
15	>	bool debug(false);
16
17	<	for (int index = 1; index < argc; index++)
17	>	struct SiteMapperCommand : public app::Application
18	>	{
19	>	virtual int Run(const app::ArgumentList& args)
20		{
21	<	string arg(argv[index]);
23	<	Matcher matcher;
21	>	program = api::GetExecutablePath().GetName();
22
23	<	if (arg == matcher("^-index=(.*)$"))
23	>	ext::String siteIndex, siteMap;
24	>
25	>	_foreach (app::ArgumentList, arg, args)
26		{
27	<	siteIndex = matcher[1];
27	>	Matcher matcher;
28	>
29	>	if (*arg == matcher("^-index=(.*)$"))
30	>	{
31	>	siteIndex = matcher[1];
32	>	}
33	>	else if (*arg == matcher("^-map=(.*)$"))
34	>	{
35	>	siteMap = matcher[1];
36	>	}
37	>	else if (*arg == "-D")
38	>	{
39	>	if (!debug) debug = true;
40	>	}
41		}
42	<	else if (arg == matcher("^-map=(.*)$"))
42	>
43	>	if (!siteIndex.IsEmpty() && !siteMap.IsEmpty())
44		{
45	<	siteMap = matcher[1];
45	>	SiteMapper mapper(siteIndex, siteMap);
46		}
47	<	else if (arg == "-D")
47	>	else
48		{
49	<	if (!debug) debug = true;
49	>	api::Cout << "Usage: " << program << " -index=index -map=map [-D]\n";
50		}
37	–	}
38	–
39	–	if (siteIndex != "" && siteMap != "")
40	–	{
41	–	XMLPlatformUtils::Initialize();
42	–	XPathEvaluator::initialize();
51
52	<	SiteMapper mapper(siteIndex, siteMap);
45	<
46	<	XPathEvaluator::terminate();
47	<	XMLPlatformUtils::Terminate();
48	<	}
49	<	else
50	<	{
51	<	cout << "Usage: " << program << " -index=index -map=map [-D]\n";
52	>	return 0;
53		}
54	+	} mapper;
55
56	<	return 0;
55	<	}
56	<
57	<	SiteMapper::SiteMapper(const string& siteIndex, const string& siteMap)
56	>	SiteMapper::SiteMapper(const ext::String& siteIndex, const ext::String& siteMap)
57		{
58		oldMap(siteMap);
59	<	index(siteIndex);
59	>	newIndex(siteIndex);
60		newMap(siteMap);
61		}
62
63	<	void SiteMapper::oldMap(const string& siteMap)
63	>	void SiteMapper::oldMap(const ext::String& siteMap)
64		{
65	<	support.setParserLiaison(&liaison);
65	>	ext::Handle<xml::Document> document(xml::Parse(siteMap));
66	>	ext::Handle<xml::Node> list(*document/"page"/"section"/"list");
67
68	<	XalanDOMString file(siteMap.c_str());
69	<	LocalFileInputSource source(file.c_str());
68	>	comment = *document/"comment()";
69
70	<	XalanDocument* document = liaison.parseXMLStream(source);
70	>	if (debug) api::Cerr << "comment = " << comment << ios::NewLine;
71
72	<	if (document == 0) return;
72	>	oldMap(pages, list);
73	>	}
74
75	<	XalanNode* list = evaluator.selectSingleNode(support, document,
76	<	XalanDOMString("/page/section/list").c_str());
75	>	void SiteMapper::oldMap(ext::Vector<Page>& pages, xml::Node* list)
76	>	{
77	>	xml::NodeSet nodes(*list/"item");
78
79	<	if (list == 0) return;
79	>	_foreach (xml::NodeSet, node, nodes)
80	>	{
81	>	ext::String url(**node/"link"/"@address"), title(**node/"link");
82	>	Page page(url, title);
83	>	ext::Handle<xml::Node> list(**node/"list");
84
85	<	item = evaluator.createXPath(XalanDOMString("item").c_str());
81	<	address = evaluator.createXPath(XalanDOMString("link/@address").c_str());
82	<	link = evaluator.createXPath(XalanDOMString("link").c_str());
83	<	this->list = evaluator.createXPath(XalanDOMString("list").c_str());
85	>	if (!list.IsEmpty()) oldMap(page.getChildren(), list);
86
87	<	oldMap(pages, list);
87	>	pages.InsertLast(page);
88	>	}
89		}
90
91	<	void SiteMapper::oldMap(vector<Page>& pages, XalanNode* list)
91	>	void SiteMapper::newIndex(const ext::String& siteIndex)
92		{
93	<	NodeRefList nodes = evaluator.selectNodeList(support, list, *item);
93	>	ext::Handle<xml::Document> document(xml::Parse(siteIndex));
94	>	xml::NodeSet nodes(*document/"index"/"page");
95
96	<	for (int index = 0; index < nodes.getLength(); index++)
96	>	_foreach (xml::NodeSet, node, nodes)
97		{
98	<	XalanNode* node = nodes.item(index);
99	<	XObjectPtr address = evaluator.evaluate(support, node,
100	<	*(this->address));
101	<	ostringstream url;
98	>	ios::String address(**node/"address");
99	>	ext::String port(**node/"port");
100	>
101	>	if (!port.IsEmpty()) address << ":" << port;
102
103	<	if (!address.null())
103	>	ext::String path(**node/"path"), title(**node/"title");
104	>	Page page(address, path, title);
105	>	Matcher matcher;
106	>
107	>	if (page == matcher("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDouglas\\sThrift's\\sBlog:\\s(.+)$"))
108	>	{
109	>	if (Matcher("^\\w+\\s\\d{4}\\sArchives$") == matcher[1])
110	>	{
111	>	page.setTitle(matcher[1]);
112	>
113	>	if (newIndex(pages, page)) continue;
114	>	}
115	>	else continue;
116	>	}
117	>	else if (page == matcher("^Douglas\\sThrift's.+Website\\s\\|\\s(.+)$"))
118		{
119	<	url << address->str();
119	>	page.setTitle(matcher[1]);
120	>
121	>	if (newIndex(pages, page)) continue;
122		}
123	+	else continue;
124
125	<	XObjectPtr link = evaluator.evaluate(support, node, *(this->link));
105	<	ostringstream title;
125	>	std::multimap<std::string, Page> items;
126
127	<	if (!link.null())
127	>	newPages.insert(std::pair<std::string, std::multimap<std::string, Page> >(page.getAddress(), items)).first->second.insert(std::pair<std::string, Page>(page.getChildOf(), page));
128	>	}
129	>	}
130	>
131	>	bool SiteMapper::newIndex(ext::Vector<Page>& pages, Page& page)
132	>	{
133	>	_mforeach (ext::Vector<Page>, page_, pages)
134	>	{
135	>	if (*page_ == page.getAddress())
136		{
137	<	title << link->str();
138	<	}
137	>	Matcher matcher;
138	>
139	>	if (*page_ == page)
140	>	{
141	>	page.setChildren(page_->getChildren());
142	>
143	>	*page_ = page;
144
145	<	Page page(url.str(), title.str());
113	<	XalanNode* list = evaluator.selectSingleNode(support, node,
114	<	*(this->list));
145	>	api::Cout << "Updated: " << page.getUrl() << ios::NewLine;
146
147	<	if (list != 0) oldMap(page.getChildren(), list);
147	>	return true;
148	>	}
149	>	else if (matcher("^" + page_->getPath()) == page)
150	>	{
151	>	page.setChildOf(matcher[0]);
152
153	<	pages.push_back(page);
153	>	if (matcher("^" + page_->getTitle() + "\\s\\|\\s(.+)$") == page) page.setTitle(matcher[1]);
154	>
155	>	return newIndex(page_->getChildren(), page);
156	>	}
157	>	}
158		}
159	+
160	+	return false;
161		}
162
163	<	void SiteMapper::index(const string& siteIndex)
163	>	void SiteMapper::newMap(const ext::String& siteMap)
164		{
165	<	//
165	>	api::FileWriter file(siteMap);
166	>	ios::FormatWriter fout(file);
167	>	xml::TextWriter xml(file);
168	>
169	>	fout << ios::NewLine << "<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl\"?>" << ios::NewLine << "<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">";
170	>
171	>	xml.OutputComment(comment);
172	>
173	>	xml::ScopeElement page(xml, "page");
174	>
175	>	xml.OpenElement("title");
176	>	xml.OutputText("Sitemap");
177	>	xml.CloseElement();
178	>
179	>	xml::ScopeElement section(xml, "section"), list(xml, "list");
180	>
181	>	_mforeach (ext::Vector<Page>, page, pages)
182	>	{
183	>	if (newPages.find(page->getAddress()) != newPages.end()) newMap(page->getChildren(), page->getPath(), newPages.find(page->getAddress())->second);
184	>
185	>	xml << *page;
186	>	}
187		}
188
189	<	void SiteMapper::newMap(const string& siteMap)
189	>	void SiteMapper::newMap(ext::Vector<Page>& pages, const ext::String& childOf, std::multimap<std::string, Page>& newPages)
190		{
191	<	//
191	>	_mforeach (ext::Vector<Page>, page, pages) newMap(page->getChildren(), page->getPath(), newPages);
192
193	<	for (int index = 0; index < pages.size(); index++)
193	>	typedef std::multimap<std::string, Page> MultiMap;
194	>
195	>	_for (MultiMap::const_iterator, itor, newPages.lower_bound(childOf), newPages.upper_bound(childOf))
196		{
197	<	cout << pages[index] << '\n';
197	>	api::Cout << "Added: " << itor->second.getUrl() << ios::NewLine;
198	>
199	>	pages.InsertLast(itor->second);
200		}
201	+
202	+	newPages.erase(childOf);
203		}

Diff Legend

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