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 127 by Douglas Thrift, 2004-03-23T19:52:15-08:00 vs.
Revision 637 by douglas, 2005-12-31T21:32:21-08:00

#	Line 4 | Line 4
4		//
5		// $Id$
6
7	<	#include "SiteMapper.hpp"
8	<	#include "Matcher.hpp"
7	>	#include <cxx/standard.hh>
8
9	<	#include <xalanc/Include/PlatformDefinitions.hpp>
10	<	#include <xercesc/util/PlatformUtils.hpp>
11	<	#include <xalanc/XPath/XPathEvaluator.hpp>
9	>	#include <api/pcre/regex.hpp>
10	>	#include <api/exename.hpp>
11	>	#include <app/simple.hpp>
12
13	<	XALAN_USING_XERCES(XMLPlatformUtils)
15	<	XALAN_USING_XALAN(XPathEvaluator)
13	>	#include "SiteMapper.hpp"
14
15	<	int main(int argc, char* argv[])
15	>	int Main(const app::Options &options)
16		{
17	<	string siteIndex, siteMap;
17	>	SiteMapper::program = api::GetExecutablePath().GetName();
18
19	<	for (int index = 1; index < argc; argc++)
19	>	cse::String siteIndex, siteMap;
20	>	api::Pcre::RegEx index(_B("^-index=(.+)$")), map(_B("^-map=(.+)$"));
21	>
22	>	_foreach (const app::ArgumentList, arg, app::GetArguments())
23		{
24	<	string arg = argv[index];
24	>	api::Pcre::RegEx::Match match;
25
26	<	//
26	>	if (match = index(*arg))
27	>	{
28	>	siteIndex = match[1];
29	>	}
30	>	else if (match = map(*arg))
31	>	{
32	>	siteMap = match[1];
33	>	}
34	>	else if (*arg == _B("-D"))
35	>	{
36	>	if (!SiteMapper::debug) SiteMapper::debug = true;
37	>	}
38		}
27	–
28	–	XMLPlatformUtils::Initialize();
29	–	XPathEvaluator::initialize();
39
40	<	SiteMapper mapper(siteIndex, siteMap);
40	>	if (!siteIndex.IsEmpty() && !siteMap.IsEmpty())
41	>	SiteMapper mapper(siteIndex, siteMap);
42	>	else
43	>	api::Cout << _B("Usage: ") << SiteMapper::program << _B(" -index=index -map=map [-D]") << ios::NewLine;
44
33	–	XPathEvaluator::terminate();
34	–	XMLPlatformUtils::Terminate();
35	–
45		return 0;
46		}
47
48	<	SiteMapper::SiteMapper(const string& siteIndex, const string& siteMap)
48	>	SiteMapper::SiteMapper(const cse::String &siteIndex, const cse::String &siteMap)
49	>	{
50	>	oldMap(siteMap);
51	>	newIndex(siteIndex);
52	>	newMap(siteMap);
53	>	}
54	>
55	>	cse::String SiteMapper::program;
56	>	bool SiteMapper::debug(false);
57	>
58	>	void SiteMapper::oldMap(const cse::String &siteMap)
59	>	{
60	>	ext::Handle<xml::Document> document(xml::Parse(siteMap));
61	>	ext::Handle<xml::Node> list(*document/_B("page")/_B("section")/_B("list"));
62	>
63	>	comment = *document/_B("comment()");
64	>
65	>	if (debug) api::Cerr << _B("comment = ") << comment << ios::NewLine;
66	>
67	>	oldMap(pages, list);
68	>	}
69	>
70	>	void SiteMapper::oldMap(ext::Vector<Page> &pages, xml::Node* list)
71		{
72	<	//
72	>	xml::NodeSet nodes(*list/_B("item"));
73	>
74	>	_foreach (xml::NodeSet, node, nodes)
75	>	{
76	>	cse::String url(**node/_B("link")/_B("@address")), title(**node/_B("link"));
77	>	Page page(url, title);
78	>	ext::Handle<xml::Node> list(**node/_B("list"));
79	>
80	>	if (!list.IsEmpty()) oldMap(page.GetChildren(), list);
81	>
82	>	pages.InsertLast(page);
83	>	}
84	>	}
85	>
86	>	void SiteMapper::newIndex(const cse::String &siteIndex)
87	>	{
88	>	ext::Handle<xml::Document> document(xml::Parse(siteIndex));
89	>	xml::NodeSet nodes(*document/_B("index")/_B("page"));
90	>
91	>	_foreach (xml::NodeSet, node, nodes)
92	>	{
93	>	_S<ios::String> address(**node/_B("address"));
94	>	cse::String port(**node/_B("port"));
95	>
96	>	if (!port.IsEmpty()) address << _B(":") << port;
97	>
98	>	cse::String path(**node/_B("path")), title(**node/_B("title"));
99	>	Page page(address, path, title);
100	>	static api::Pcre::RegEx blog(_B("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDouglas\\sThrift's\\sBlog:\\s(.+)$")), page_(_B("^Douglas\\sThrift's.+Website\\s\\|\\s(.+)$"));
101	>
102	>	if (api::Pcre::RegEx::Match match = blog(page.GetTitle()))
103	>	{
104	>	static api::Pcre::RegEx archives(_B("^\\w+\\s\\d{4}\\sArchives$"));
105	>
106	>	if (archives(match[1]))
107	>	{
108	>	page.SetTitle(match[1]);
109	>
110	>	if (newIndex(pages, page))
111	>	continue;
112	>	}
113	>	else
114	>	continue;
115	>	}
116	>	else if (api::Pcre::RegEx::Match match = page_(page.GetTitle()))
117	>	{
118	>	page.SetTitle(match[1]);
119	>
120	>	if (newIndex(pages, page))
121	>	continue;
122	>	}
123	>	else
124	>	continue;
125	>
126	>	newPages[page.GetAddress()][page.GetChildOf()].InsertLast(page);
127	>	}
128	>	}
129	>
130	>	bool SiteMapper::newIndex(ext::Vector<Page> &pages, Page &page)
131	>	{
132	>	_foreach (ext::Vector<Page>, page_, pages)
133	>	{
134	>	if (*page_ == page.GetAddress())
135	>	{
136	>	if (*page_ == page)
137	>	{
138	>	page.SetChildren(page_->GetChildren());
139	>
140	>	*page_ = page;
141	>
142	>	api::Cout << _B("Updated: ") << page.GetUrl() << ios::NewLine;
143	>
144	>	return true;
145	>	}
146	>	else if (page.GetPath().StartsWithAll(page_->GetPath()))
147	>	{
148	>	page.SetChildOf(page_->GetPath());
149	>
150	>	api::Pcre::RegEx title(_S<ios::String>() << _B("^") << page_->GetTitle() << "\\s\\|\\s(.+)$");
151	>
152	>	if (api::Pcre::RegEx::Match match = title(page.GetTitle()))
153	>	page.SetTitle(match[1]);
154	>
155	>	return newIndex(page_->GetChildren(), page);
156	>	}
157	>	}
158	>	}
159	>
160	>	return false;
161	>	}
162	>
163	>	void SiteMapper::newMap(const cse::String &siteMap)
164	>	{
165	>	_S<api::FileWriter> file(siteMap);
166	>	_S<ios::FormatWriter> fout(file);
167	>	_S<xml::TextWriter> xml(file);
168	>
169	>	// XXX: xml::TextWriter should have this kind of stuff, no?
170	>	fout << ios::NewLine << _B("<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl\"?>") << ios::NewLine << _B("<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">");
171	>
172	>	xml.OutputComment(comment);
173	>
174	>	xml::ScopeElement page(xml, _B("page"));
175	>
176	>	xml.OpenElement(_B("title"));
177	>	xml.OutputText(_B("Sitemap"));
178	>	xml.CloseElement();
179	>
180	>	xml::ScopeElement section(xml, _B("section")), list(xml, _B("list"));
181	>
182	>	_foreach (ext::Vector<Page>, page, pages)
183	>	{
184	>	if (newPages.Contains(page->GetAddress()))
185	>	newMap(page->GetChildren(), page->GetPath(), newPages.Find(page->GetAddress())->Second());
186	>
187	>	xml << *page;
188	>	}
189	>	}
190	>
191	>	void SiteMapper::newMap(ext::Vector<Page> &pages, const cse::String &childOf, ext::RedBlackMap<cse::String, ext::Vector<Page>, LessThan> &newPages)
192	>	{
193	>	_foreach (ext::Vector<Page>, page, pages)
194	>	newMap(page->GetChildren(), page->GetPath(), newPages);
195	>
196	>	_foreach (ext::Vector<Page>, page, newPages[childOf])
197	>	{
198	>	api::Cout << _B("Added: ") << page->GetUrl() << ios::NewLine;
199	>
200	>	pages.InsertLast(*page);
201	>	}
202	>
203	>	newPages.Remove(childOf);
204		}

Diff Legend

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