repos/SiteMapper/SiteMapper.cpp

// Site Mapper
//
// Douglas Thrift
//
// $Id$

#include "SiteMapper.hpp"
#include "Matcher.hpp"
#include "Page.hpp"

#include <menes-api/exename.hpp>
#include <menes-app/application.hpp>

ext::String program;
bool debug(false);

struct SiteMapperCommand : public app::Application
{
        virtual int Run(const app::ArgumentList& args)
        {
                program = api::GetExecutablePath().GetName();

                ext::String siteIndex, siteMap;

                _foreach (app::ArgumentList, arg, args)
                {
                        Matcher matcher;

                        if (*arg == matcher("^-index=(.*)$"))
                        {
                                siteIndex = matcher[1];
                        }
                        else if (*arg == matcher("^-map=(.*)$"))
                        {
                                siteMap = matcher[1];
                        }
                        else if (*arg == "-D")
                        {
                                if (!debug) debug = true;
                        }
                }

                if (!siteIndex.IsEmpty() && !siteMap.IsEmpty())
                {
                        SiteMapper mapper(siteIndex, siteMap);
                }
                else
                {
                        api::Cout << "Usage: " << program << " -index=index -map=map [-D]\n";
                }

                return 0;
        }
} mapper;

SiteMapper::SiteMapper(const ext::String& siteIndex, const ext::String& siteMap)
{
        oldMap(siteMap);
        newIndex(siteIndex);
        newMap(siteMap);
}

void SiteMapper::oldMap(const ext::String& siteMap)
{
        ext::Handle<xml::Document> document(xml::Parse(siteMap));
        ext::Handle<xml::Node> list(*document/"page"/"section"/"list");

        comment = *document/"comment()";

        if (debug) api::Cerr << "comment = " << comment << ios::NewLine;

        oldMap(pages, list);
}

void SiteMapper::oldMap(ext::Vector<Page>& pages, xml::Node* list)
{
        xml::NodeSet nodes(*list/"item");

        _foreach (xml::NodeSet, node, nodes)
        {
                ext::String url(**node/"link"/"@address"), title(**node/"link");
                Page page(url, title);
                ext::Handle<xml::Node> list(**node/"list");

                if (!list.IsEmpty()) oldMap(page.getChildren(), list);

                pages.InsertLast(page);
        }
}

void SiteMapper::newIndex(const ext::String& siteIndex)
{
        ext::Handle<xml::Document> document(xml::Parse(siteIndex));
        xml::NodeSet nodes(*document/"index"/"page");

        _foreach (xml::NodeSet, node, nodes)
        {
                ios::String address(**node/"address");
                ext::String port(**node/"port");

                if (!port.IsEmpty()) address << ":" << port;

                ext::String path(**node/"path"), title(**node/"title");
                Page page(address, path, title);
                Matcher matcher;

                if (page == matcher("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDouglas\\sThrift's\\sBlog:\\s(.+)$"))
                {
                        if (Matcher("^\\w+\\s\\d{4}\\sArchives$") == matcher[1])
                        {
                                page.setTitle(matcher[1]);

                                if (newIndex(pages, page)) continue;
                        }
                        else continue;
                }
                else if (page == matcher("^Douglas\\sThrift's.+Website\\s\\|\\s(.+)$"))
                {
                        page.setTitle(matcher[1]);

                        if (newIndex(pages, page)) continue;
                }
                else continue;

                std::multimap<std::string, Page> items;

                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));
        }
}

bool SiteMapper::newIndex(ext::Vector<Page>& pages, Page& page)
{
        _mforeach (ext::Vector<Page>, page_, pages)
        {
                if (*page_ == page.getAddress())
                {
                        Matcher matcher;

                        if (*page_ == page)
                        {
                                page.setChildren(page_->getChildren());

                                *page_ = page;

                                api::Cout << "Updated: " << page.getUrl() << ios::NewLine;

                                return true;
                        }
                        else if (matcher("^" + page_->getPath()) == page)
                        {
                                page.setChildOf(matcher[0]);

                                if (matcher("^" + page_->getTitle() + "\\s\\|\\s(.+)$") == page) page.setTitle(matcher[1]);

                                return newIndex(page_->getChildren(), page);
                        }
                }
        }

        return false;
}

void SiteMapper::newMap(const ext::String& siteMap)
{
        api::FileWriter file(siteMap);
        ios::FormatWriter fout(file);
        xml::TextWriter xml(file);

        fout << ios::NewLine << "<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl\"?>" << ios::NewLine << "<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">";

        xml.OutputComment(comment);

        xml::ScopeElement page(xml, "page");

        xml.OpenElement("title");
        xml.OutputText("Sitemap");
        xml.CloseElement();

        xml::ScopeElement section(xml, "section"), list(xml, "list");

        _mforeach (ext::Vector<Page>, page, pages)
        {
                if (newPages.find(page->getAddress()) != newPages.end()) newMap(page->getChildren(), page->getPath(), newPages.find(page->getAddress())->second);

                xml << *page;
        }
}

void SiteMapper::newMap(ext::Vector<Page>& pages, const ext::String& childOf, std::multimap<std::string, Page>& newPages)
{
        _mforeach (ext::Vector<Page>, page, pages) newMap(page->getChildren(), page->getPath(), newPages);

        typedef std::multimap<std::string, Page> MultiMap;

        _for (MultiMap::const_iterator, itor, newPages.lower_bound(childOf), newPages.upper_bound(childOf))
        {
                api::Cout << "Added: " << itor->second.getUrl() << ios::NewLine;

                pages.InsertLast(itor->second);
        }

        newPages.erase(childOf);
}
Revision:	303
Committed:	2004-12-12T17:58:03-08:00 (20 years, 6 months ago) by douglas
File size:	4750 byte(s)
Log Message:	Seems to almost work.
#	User	Rev	Content
1	Douglas Thrift	126	// Site Mapper
2			//
3			// Douglas Thrift
4			//
5			// $Id$
6
7			#include "SiteMapper.hpp"
8			#include "Matcher.hpp"
9	Douglas Thrift	128	#include "Page.hpp"
10	Douglas Thrift	126
11	Douglas Thrift	249	#include <menes-api/exename.hpp>
12			#include <menes-app/application.hpp>
13
14	douglas	285	ext::String program;
15	Douglas Thrift	152	bool debug(false);
16	Douglas Thrift	128
17	Douglas Thrift	249	struct SiteMapperCommand : public app::Application
18	Douglas Thrift	126	{
19	Douglas Thrift	249	virtual int Run(const app::ArgumentList& args)
20			{
21	Douglas Thrift	259	program = api::GetExecutablePath().GetName();
22	Douglas Thrift	132
23	douglas	285	ext::String siteIndex, siteMap;
24	Douglas Thrift	126
25	douglas	285	_foreach (app::ArgumentList, arg, args)
26	Douglas Thrift	249	{
27			Matcher matcher;
28	Douglas Thrift	126
29	douglas	285	if (arg == matcher("^-index=(.)$"))
30	Douglas Thrift	249	{
31			siteIndex = matcher[1];
32			}
33	douglas	285	else if (arg == matcher("^-map=(.)$"))
34	Douglas Thrift	249	{
35			siteMap = matcher[1];
36			}
37	douglas	285	else if (*arg == "-D")
38	Douglas Thrift	249	{
39			if (!debug) debug = true;
40			}
41	Douglas Thrift	128	}
42	Douglas Thrift	249
43	douglas	285	if (!siteIndex.IsEmpty() && !siteMap.IsEmpty())
44	Douglas Thrift	128	{
45	Douglas Thrift	249	SiteMapper mapper(siteIndex, siteMap);
46	Douglas Thrift	128	}
47	Douglas Thrift	249	else
48	Douglas Thrift	133	{
49	douglas	285	api::Cout << "Usage: " << program << " -index=index -map=map [-D]\n";
50	Douglas Thrift	133	}
51	Douglas Thrift	126
52	Douglas Thrift	249	return 0;
53	Douglas Thrift	128	}
54	Douglas Thrift	249	} mapper;
55	Douglas Thrift	129
56	douglas	285	SiteMapper::SiteMapper(const ext::String& siteIndex, const ext::String& siteMap)
57	Douglas Thrift	126	{
58	Douglas Thrift	133	oldMap(siteMap);
59	Douglas Thrift	142	newIndex(siteIndex);
60	Douglas Thrift	134	newMap(siteMap);
61	Douglas Thrift	133	}
62
63	douglas	285	void SiteMapper::oldMap(const ext::String& siteMap)
64	Douglas Thrift	133	{
65	Douglas Thrift	154	ext::Handle<xml::Document> document(xml::Parse(siteMap));
66			ext::Handle<xml::Node> list(*document/"page"/"section"/"list");
67	Douglas Thrift	133
68	douglas	285	comment = *document/"comment()";
69	Douglas Thrift	133
70	douglas	303	if (debug) api::Cerr << "comment = " << comment << ios::NewLine;
71	Douglas Thrift	249
72	Douglas Thrift	137	oldMap(pages, list);
73	Douglas Thrift	133	}
74
75	douglas	285	void SiteMapper::oldMap(ext::Vector<Page>& pages, xml::Node* list)
76	Douglas Thrift	135	{
77	Douglas Thrift	154	xml::NodeSet nodes(*list/"item");
78	Douglas Thrift	135
79	douglas	285	_foreach (xml::NodeSet, node, nodes)
80	Douglas Thrift	135	{
81	douglas	285	ext::String url(node/"link"/"@address"), title(node/"link");
82	Douglas Thrift	154	Page page(url, title);
83			ext::Handle<xml::Node> list(**node/"list");
84	Douglas Thrift	141
85	Douglas Thrift	154	if (!list.IsEmpty()) oldMap(page.getChildren(), list);
86	Douglas Thrift	135
87	douglas	285	pages.InsertLast(page);
88	Douglas Thrift	135	}
89			}
90
91	douglas	285	void SiteMapper::newIndex(const ext::String& siteIndex)
92	Douglas Thrift	133	{
93	Douglas Thrift	154	ext::Handle<xml::Document> document(xml::Parse(siteIndex));
94			xml::NodeSet nodes(*document/"index"/"page");
95	Douglas Thrift	138
96	douglas	285	_foreach (xml::NodeSet, node, nodes)
97	Douglas Thrift	138	{
98	douglas	285	ios::String address(**node/"address");
99			ext::String port(**node/"port");
100	Douglas Thrift	139
101	douglas	285	if (!port.IsEmpty()) address << ":" << port;
102	Douglas Thrift	139
103	douglas	285	ext::String path(node/"path"), title(node/"title");
104	Douglas Thrift	154	Page page(address, path, title);
105	Douglas Thrift	142	Matcher matcher;
106	Douglas Thrift	140
107	douglas	285	if (page == matcher("^Douglas\\sThrift's\\sWebsite\\s\\\|\\sDouglas\\sThrift's\\sBlog:\\s(.+)$"))
108	Douglas Thrift	140	{
109	Douglas Thrift	156	if (Matcher("^\\w+\\s\\d{4}\\sArchives$") == matcher[1])
110	Douglas Thrift	142	{
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(.+)$"))
118			{
119	Douglas Thrift	140	page.setTitle(matcher[1]);
120
121	Douglas Thrift	142	if (newIndex(pages, page)) continue;
122	Douglas Thrift	140	}
123	Douglas Thrift	142	else continue;
124
125	douglas	285	std::multimap<std::string, Page> items;
126	Douglas Thrift	143
127	douglas	285	newPages.insert(std::pair<std::string, std::multimap<std::string, Page> >(page.getAddress(), items)).first->second.insert(std::pair<std::string, Page>(page.getChildOf(), page));
128	Douglas Thrift	138	}
129	Douglas Thrift	126	}
130	Douglas Thrift	133
131	douglas	285	bool SiteMapper::newIndex(ext::Vector<Page>& pages, Page& page)
132	Douglas Thrift	142	{
133	douglas	285	_mforeach (ext::Vector<Page>, page_, pages)
134	Douglas Thrift	142	{
135	douglas	285	if (*page_ == page.getAddress())
136	Douglas Thrift	142	{
137			Matcher matcher;
138
139	douglas	285	if (*page_ == page)
140	Douglas Thrift	142	{
141	douglas	285	page.setChildren(page_->getChildren());
142	Douglas Thrift	142
143	douglas	285	*page_ = page;
144	Douglas Thrift	142
145	douglas	303	api::Cout << "Updated: " << page.getUrl() << ios::NewLine;
146	Douglas Thrift	156
147	Douglas Thrift	142	return true;
148			}
149	douglas	285	else if (matcher("^" + page_->getPath()) == page)
150	Douglas Thrift	142	{
151			page.setChildOf(matcher[0]);
152
153	douglas	285	if (matcher("^" + page_->getTitle() + "\\s\\\|\\s(.+)$") == page) page.setTitle(matcher[1]);
154	Douglas Thrift	142
155	douglas	285	return newIndex(page_->getChildren(), page);
156	Douglas Thrift	142	}
157			}
158			}
159
160			return false;
161			}
162
163	douglas	285	void SiteMapper::newMap(const ext::String& siteMap)
164	Douglas Thrift	133	{
165	douglas	285	api::FileWriter file(siteMap);
166			ios::FormatWriter fout(file);
167			xml::TextWriter xml(file);
168	Douglas Thrift	144
169	douglas	285	fout << ios::NewLine << "<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl\"?>" << ios::NewLine << "<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">";
170	Douglas Thrift	144
171	douglas	285	xml.OutputComment(comment);
172	Douglas Thrift	143
173	douglas	285	xml::ScopeElement page(xml, "page");
174	Douglas Thrift	144
175	douglas	285	xml.OpenElement("title");
176			xml.OutputText("Sitemap");
177			xml.CloseElement();
178	Douglas Thrift	144
179	douglas	285	xml::ScopeElement section(xml, "section"), list(xml, "list");
180	Douglas Thrift	143
181	douglas	285	_mforeach (ext::Vector<Page>, page, pages)
182	Douglas Thrift	143	{
183	douglas	285	if (newPages.find(page->getAddress()) != newPages.end()) newMap(page->getChildren(), page->getPath(), newPages.find(page->getAddress())->second);
184
185			xml << *page;
186	Douglas Thrift	143	}
187	douglas	285	}
188	Douglas Thrift	143
189	douglas	285	void SiteMapper::newMap(ext::Vector<Page>& pages, const ext::String& childOf, std::multimap<std::string, Page>& newPages)
190			{
191			_mforeach (ext::Vector<Page>, page, pages) newMap(page->getChildren(), page->getPath(), newPages);
192
193			typedef std::multimap<std::string, Page> MultiMap;
194
195			_for (MultiMap::const_iterator, itor, newPages.lower_bound(childOf), newPages.upper_bound(childOf))
196	Douglas Thrift	143	{
197	douglas	303	api::Cout << "Added: " << itor->second.getUrl() << ios::NewLine;
198	Douglas Thrift	156
199	douglas	285	pages.InsertLast(itor->second);
200	Douglas Thrift	143	}
201
202			newPages.erase(childOf);
203			}