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>

string program;
bool debug(false);

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

                string siteIndex, siteMap;

                for (size_t index(0); index < args.GetSize(); index++)
                {
                        string arg(args[index]);
                        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.empty() && !siteMap.empty())
                {
                        SiteMapper mapper(siteIndex, siteMap);
                }
                else
                {
                        cout << "Usage: " << program << " -index=index -map=map [-D]\n";
                }

                return 0;
        }
} mapper;

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

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

        comment = ext::String(*document/"comment()");

        if (debug) cerr << "comment = " << comment << '\n';

        oldMap(pages, list);
}

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

        for (xml::NodeSet::Iterator node(nodes.Begin()); node != nodes.End();
                ++node)
        {
                string url(ext::String(**node/"link"/"@address")),
                        title(ext::String(**node/"link"));
                Page page(url, title);
                ext::Handle<xml::Node> list(**node/"list");

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

                pages.push_back(page);
        }
}

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

        for (xml::NodeSet::Iterator node(nodes.Begin()); node != nodes.End();
                ++node)
        {
                string address(ext::String(**node/"address")),
                        port(ext::String(**node/"port"));

                if (!port.empty())
                {
                        address += ':' + port;
                }

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

                if (page == matcher(string("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDou")
                        + "glas\\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;

                multimap<string, Page> items;

                newPages.insert(pair<string, multimap<string, Page>
                        >(page.getAddress(), items)).first->second.insert(pair<string,
                        Page>(page.getChildOf(), page));
        }
}

bool SiteMapper::newIndex(vector<Page>& pages, Page& page)
{
        for (unsigned index(0); index < pages.size(); ++index)
        {
                if (pages[index] == page.getAddress())
                {
                        Matcher matcher;

                        if (pages[index] == page)
                        {
                                page.setChildren(pages[index].getChildren());

                                pages[index] = page;

                                cout << "Updated: " << page.getUrl() << '\n';

                                return true;
                        }
                        else if (matcher('^' + pages[index].getPath()) == page)
                        {
                                page.setChildOf(matcher[0]);

                                if (matcher('^' + pages[index].getTitle() + "\\s\\|\\s(.+)$")
                                        == page)
                                {
                                        page.setTitle(matcher[1]);
                                }

                                return newIndex(pages[index].getChildren(), page);
                        }
                }
        }

        return false;
}

void SiteMapper::newMap(const string& siteMap)
{
        ofstream fout(siteMap.c_str());

        fout << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
                << "<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl"
                << "\"?>\n"
                << "<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">\n"
                << "<!--" << comment << "-->\n"
                << "<page>\n"
                << "\t<title>Sitemap</title>\n"
                << "\t<section>\n"
                << "\t\t<list>\n";

        for (unsigned index(0); index < pages.size(); ++index)
        {
                if (newPages.find(pages[index].getAddress()) != newPages.end())
                {
                        newMap(pages[index].getChildren(), pages[index].getPath(),
                                newPages.find(pages[index].getAddress())->second);
                }

                fout << pages[index](3) << '\n';
        }

        fout << "\t\t</list>\n"
                << "\t</section>\n"
                << "</page>\n";

        fout.close();
}

void SiteMapper::newMap(vector<Page>& pages, const string& childOf,
        multimap<string, Page>& newPages)
{
        for (unsigned index(0); index < pages.size(); ++index)
        {
                newMap(pages[index].getChildren(), pages[index].getPath(), newPages);
        }

        for (multimap<string, Page>::iterator itor(newPages.lower_bound(childOf));
                itor != newPages.upper_bound(childOf); itor++)
        {
                cout << "Added: " << itor->second.getUrl() << '\n';

                pages.push_back(itor->second);
        }

        newPages.erase(childOf);
}
Revision:	259
Committed:	2004-10-01T17:19:10-07:00 (20 years, 8 months ago) by Douglas Thrift
File size:	4949 byte(s)
Log Message:	Worky, worky!
#	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 Thrift	128	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 Thrift	249	string siteIndex, siteMap;
24	Douglas Thrift	126
25	Douglas Thrift	249	for (size_t index(0); index < args.GetSize(); index++)
26			{
27			string arg(args[index]);
28			Matcher matcher;
29	Douglas Thrift	126
30	Douglas Thrift	249	if (arg == matcher("^-index=(.*)$"))
31			{
32			siteIndex = matcher[1];
33			}
34			else if (arg == matcher("^-map=(.*)$"))
35			{
36			siteMap = matcher[1];
37			}
38			else if (arg == "-D")
39			{
40			if (!debug) debug = true;
41			}
42	Douglas Thrift	128	}
43	Douglas Thrift	249
44			if (!siteIndex.empty() && !siteMap.empty())
45	Douglas Thrift	128	{
46	Douglas Thrift	249	SiteMapper mapper(siteIndex, siteMap);
47	Douglas Thrift	128	}
48	Douglas Thrift	249	else
49	Douglas Thrift	133	{
50	Douglas Thrift	249	cout << "Usage: " << program << " -index=index -map=map [-D]\n";
51	Douglas Thrift	133	}
52	Douglas Thrift	126
53	Douglas Thrift	249	return 0;
54	Douglas Thrift	128	}
55	Douglas Thrift	249	} mapper;
56	Douglas Thrift	129
57	Douglas Thrift	126	SiteMapper::SiteMapper(const string& siteIndex, const string& siteMap)
58			{
59	Douglas Thrift	133	oldMap(siteMap);
60	Douglas Thrift	142	newIndex(siteIndex);
61	Douglas Thrift	134	newMap(siteMap);
62	Douglas Thrift	133	}
63
64			void SiteMapper::oldMap(const string& siteMap)
65			{
66	Douglas Thrift	154	ext::Handle<xml::Document> document(xml::Parse(siteMap));
67			ext::Handle<xml::Node> list(*document/"page"/"section"/"list");
68	Douglas Thrift	133
69	Douglas Thrift	237	comment = ext::String(*document/"comment()");
70	Douglas Thrift	133
71	Douglas Thrift	249	if (debug) cerr << "comment = " << comment << '\n';
72
73	Douglas Thrift	137	oldMap(pages, list);
74	Douglas Thrift	133	}
75
76	Douglas Thrift	154	void SiteMapper::oldMap(vector<Page>& pages, xml::Node* list)
77	Douglas Thrift	135	{
78	Douglas Thrift	154	xml::NodeSet nodes(*list/"item");
79	Douglas Thrift	135
80	Douglas Thrift	154	for (xml::NodeSet::Iterator node(nodes.Begin()); node != nodes.End();
81			++node)
82	Douglas Thrift	135	{
83	Douglas Thrift	237	string url(ext::String(**node/"link"/"@address")),
84			title(ext::String(**node/"link"));
85	Douglas Thrift	154	Page page(url, title);
86			ext::Handle<xml::Node> list(**node/"list");
87	Douglas Thrift	141
88	Douglas Thrift	154	if (!list.IsEmpty()) oldMap(page.getChildren(), list);
89	Douglas Thrift	135
90			pages.push_back(page);
91			}
92			}
93
94	Douglas Thrift	142	void SiteMapper::newIndex(const string& siteIndex)
95	Douglas Thrift	133	{
96	Douglas Thrift	154	ext::Handle<xml::Document> document(xml::Parse(siteIndex));
97			xml::NodeSet nodes(*document/"index"/"page");
98	Douglas Thrift	138
99	Douglas Thrift	154	for (xml::NodeSet::Iterator node(nodes.Begin()); node != nodes.End();
100			++node)
101	Douglas Thrift	138	{
102	Douglas Thrift	249	string address(ext::String(**node/"address")),
103			port(ext::String(**node/"port"));
104	Douglas Thrift	139
105	Douglas Thrift	154	if (!port.empty())
106	Douglas Thrift	139	{
107	Douglas Thrift	154	address += ':' + port;
108	Douglas Thrift	139	}
109
110	Douglas Thrift	237	string path(ext::String(**node/"path")),
111			title(ext::String(**node/"title"));
112	Douglas Thrift	154	Page page(address, path, title);
113	Douglas Thrift	142	Matcher matcher;
114	Douglas Thrift	140
115	Douglas Thrift	142	if (page == matcher(string("^Douglas\\sThrift's\\sWebsite\\s\\\|\\sDou")
116			+ "glas\\sThrift's\\sBlog:\\s(.+)$"))
117	Douglas Thrift	140	{
118	Douglas Thrift	156	if (Matcher("^\\w+\\s\\d{4}\\sArchives$") == matcher[1])
119	Douglas Thrift	142	{
120			page.setTitle(matcher[1]);
121
122			if (newIndex(pages, page)) continue;
123			}
124			else continue;
125			}
126			else if (page == matcher("^Douglas\\sThrift's.+Website\\s\\\|\\s(.+)$"))
127			{
128	Douglas Thrift	140	page.setTitle(matcher[1]);
129
130	Douglas Thrift	142	if (newIndex(pages, page)) continue;
131	Douglas Thrift	140	}
132	Douglas Thrift	142	else continue;
133
134	Douglas Thrift	143	multimap<string, Page> items;
135
136			newPages.insert(pair<string, multimap<string, Page>
137			>(page.getAddress(), items)).first->second.insert(pair<string,
138			Page>(page.getChildOf(), page));
139	Douglas Thrift	138	}
140	Douglas Thrift	126	}
141	Douglas Thrift	133
142	Douglas Thrift	142	bool SiteMapper::newIndex(vector<Page>& pages, Page& page)
143			{
144	Douglas Thrift	153	for (unsigned index(0); index < pages.size(); ++index)
145	Douglas Thrift	142	{
146			if (pages[index] == page.getAddress())
147			{
148			Matcher matcher;
149
150			if (pages[index] == page)
151			{
152			page.setChildren(pages[index].getChildren());
153
154			pages[index] = page;
155
156	Douglas Thrift	156	cout << "Updated: " << page.getUrl() << '\n';
157
158	Douglas Thrift	142	return true;
159			}
160			else if (matcher('^' + pages[index].getPath()) == page)
161			{
162			page.setChildOf(matcher[0]);
163
164			if (matcher('^' + pages[index].getTitle() + "\\s\\\|\\s(.+)$")
165			== page)
166			{
167			page.setTitle(matcher[1]);
168			}
169
170			return newIndex(pages[index].getChildren(), page);
171			}
172			}
173			}
174
175			return false;
176			}
177
178	Douglas Thrift	133	void SiteMapper::newMap(const string& siteMap)
179			{
180	Douglas Thrift	144	ofstream fout(siteMap.c_str());
181
182			fout << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
183			<< "<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl"
184			<< "\"?>\n"
185			<< "<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">\n"
186	Douglas Thrift	154	<< "<!--" << comment << "-->\n"
187	Douglas Thrift	144	<< "<page>\n"
188			<< "\t<title>Sitemap</title>\n"
189			<< "\t<section>\n"
190			<< "\t\t<list>\n";
191
192	Douglas Thrift	153	for (unsigned index(0); index < pages.size(); ++index)
193	Douglas Thrift	140	{
194	Douglas Thrift	143	if (newPages.find(pages[index].getAddress()) != newPages.end())
195			{
196			newMap(pages[index].getChildren(), pages[index].getPath(),
197			newPages.find(pages[index].getAddress())->second);
198			}
199
200	Douglas Thrift	144	fout << pages[index](3) << '\n';
201	Douglas Thrift	140	}
202	Douglas Thrift	144
203			fout << "\t\t</list>\n"
204			<< "\t</section>\n"
205			<< "</page>\n";
206
207			fout.close();
208	Douglas Thrift	133	}
209	Douglas Thrift	143
210			void SiteMapper::newMap(vector<Page>& pages, const string& childOf,
211			multimap<string, Page>& newPages)
212			{
213	Douglas Thrift	153	for (unsigned index(0); index < pages.size(); ++index)
214	Douglas Thrift	143	{
215			newMap(pages[index].getChildren(), pages[index].getPath(), newPages);
216			}
217
218	Douglas Thrift	153	for (multimap<string, Page>::iterator itor(newPages.lower_bound(childOf));
219	Douglas Thrift	143	itor != newPages.upper_bound(childOf); itor++)
220			{
221	Douglas Thrift	156	cout << "Added: " << itor->second.getUrl() << '\n';
222
223	Douglas Thrift	143	pages.push_back(itor->second);
224			}
225
226			newPages.erase(childOf);
227			}