repos/SiteMapper/SiteMapper.cpp

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

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

string program;
bool debug(false);

int main(int argc, char* argv[])
{
        program = argv[0];

        string siteIndex, siteMap;

        for (int index(1); index < argc; index++)
        {
                string arg(argv[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;
}

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()");

        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"));
                string 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:	237
Committed:	2004-09-10T16:22:07-07:00 (20 years, 9 months ago) by Douglas Thrift
File size:	4682 byte(s)
Log Message:	Yes?
#	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	128	string program;
12	Douglas Thrift	152	bool debug(false);
13	Douglas Thrift	128
14	Douglas Thrift	126	int main(int argc, char* argv[])
15			{
16	Douglas Thrift	128	program = argv[0];
17	Douglas Thrift	132
18	Douglas Thrift	126	string siteIndex, siteMap;
19
20	Douglas Thrift	152	for (int index(1); index < argc; index++)
21	Douglas Thrift	126	{
22	Douglas Thrift	128	string arg(argv[index]);
23			Matcher matcher;
24	Douglas Thrift	126
25	Douglas Thrift	128	if (arg == matcher("^-index=(.*)$"))
26			{
27			siteIndex = matcher[1];
28			}
29			else if (arg == matcher("^-map=(.*)$"))
30			{
31			siteMap = matcher[1];
32			}
33	Douglas Thrift	133	else if (arg == "-D")
34			{
35			if (!debug) debug = true;
36			}
37	Douglas Thrift	126	}
38
39	Douglas Thrift	156	if (!siteIndex.empty() && !siteMap.empty())
40	Douglas Thrift	132	{
41	Douglas Thrift	128	SiteMapper mapper(siteIndex, siteMap);
42			}
43			else
44			{
45	Douglas Thrift	133	cout << "Usage: " << program << " -index=index -map=map [-D]\n";
46	Douglas Thrift	128	}
47	Douglas Thrift	129
48	Douglas Thrift	126	return 0;
49			}
50
51			SiteMapper::SiteMapper(const string& siteIndex, const string& siteMap)
52			{
53	Douglas Thrift	133	oldMap(siteMap);
54	Douglas Thrift	142	newIndex(siteIndex);
55	Douglas Thrift	134	newMap(siteMap);
56	Douglas Thrift	133	}
57
58			void SiteMapper::oldMap(const string& siteMap)
59			{
60	Douglas Thrift	154	ext::Handle<xml::Document> document(xml::Parse(siteMap));
61			ext::Handle<xml::Node> list(*document/"page"/"section"/"list");
62	Douglas Thrift	133
63	Douglas Thrift	237	comment = ext::String(*document/"comment()");
64	Douglas Thrift	133
65	Douglas Thrift	137	oldMap(pages, list);
66	Douglas Thrift	133	}
67
68	Douglas Thrift	154	void SiteMapper::oldMap(vector<Page>& pages, xml::Node* list)
69	Douglas Thrift	135	{
70	Douglas Thrift	154	xml::NodeSet nodes(*list/"item");
71	Douglas Thrift	135
72	Douglas Thrift	154	for (xml::NodeSet::Iterator node(nodes.Begin()); node != nodes.End();
73			++node)
74	Douglas Thrift	135	{
75	Douglas Thrift	237	string url(ext::String(**node/"link"/"@address")),
76			title(ext::String(**node/"link"));
77	Douglas Thrift	154	Page page(url, title);
78			ext::Handle<xml::Node> list(**node/"list");
79	Douglas Thrift	141
80	Douglas Thrift	154	if (!list.IsEmpty()) oldMap(page.getChildren(), list);
81	Douglas Thrift	135
82			pages.push_back(page);
83			}
84			}
85
86	Douglas Thrift	142	void SiteMapper::newIndex(const string& siteIndex)
87	Douglas Thrift	133	{
88	Douglas Thrift	154	ext::Handle<xml::Document> document(xml::Parse(siteIndex));
89			xml::NodeSet nodes(*document/"index"/"page");
90	Douglas Thrift	138
91	Douglas Thrift	154	for (xml::NodeSet::Iterator node(nodes.Begin()); node != nodes.End();
92			++node)
93	Douglas Thrift	138	{
94	Douglas Thrift	237	string address(ext::String(**node/"address"));
95			string port(ext::String(**node/"port"));
96	Douglas Thrift	139
97	Douglas Thrift	154	if (!port.empty())
98	Douglas Thrift	139	{
99	Douglas Thrift	154	address += ':' + port;
100	Douglas Thrift	139	}
101
102	Douglas Thrift	237	string path(ext::String(**node/"path")),
103			title(ext::String(**node/"title"));
104	Douglas Thrift	154	Page page(address, path, title);
105	Douglas Thrift	142	Matcher matcher;
106	Douglas Thrift	140
107	Douglas Thrift	142	if (page == matcher(string("^Douglas\\sThrift's\\sWebsite\\s\\\|\\sDou")
108			+ "glas\\sThrift's\\sBlog:\\s(.+)$"))
109	Douglas Thrift	140	{
110	Douglas Thrift	156	if (Matcher("^\\w+\\s\\d{4}\\sArchives$") == matcher[1])
111	Douglas Thrift	142	{
112			page.setTitle(matcher[1]);
113
114			if (newIndex(pages, page)) continue;
115			}
116			else continue;
117			}
118			else if (page == matcher("^Douglas\\sThrift's.+Website\\s\\\|\\s(.+)$"))
119			{
120	Douglas Thrift	140	page.setTitle(matcher[1]);
121
122	Douglas Thrift	142	if (newIndex(pages, page)) continue;
123	Douglas Thrift	140	}
124	Douglas Thrift	142	else continue;
125
126	Douglas Thrift	143	multimap<string, Page> items;
127
128			newPages.insert(pair<string, multimap<string, Page>
129			>(page.getAddress(), items)).first->second.insert(pair<string,
130			Page>(page.getChildOf(), page));
131	Douglas Thrift	138	}
132	Douglas Thrift	126	}
133	Douglas Thrift	133
134	Douglas Thrift	142	bool SiteMapper::newIndex(vector<Page>& pages, Page& page)
135			{
136	Douglas Thrift	153	for (unsigned index(0); index < pages.size(); ++index)
137	Douglas Thrift	142	{
138			if (pages[index] == page.getAddress())
139			{
140			Matcher matcher;
141
142			if (pages[index] == page)
143			{
144			page.setChildren(pages[index].getChildren());
145
146			pages[index] = page;
147
148	Douglas Thrift	156	cout << "Updated: " << page.getUrl() << '\n';
149
150	Douglas Thrift	142	return true;
151			}
152			else if (matcher('^' + pages[index].getPath()) == page)
153			{
154			page.setChildOf(matcher[0]);
155
156			if (matcher('^' + pages[index].getTitle() + "\\s\\\|\\s(.+)$")
157			== page)
158			{
159			page.setTitle(matcher[1]);
160			}
161
162			return newIndex(pages[index].getChildren(), page);
163			}
164			}
165			}
166
167			return false;
168			}
169
170	Douglas Thrift	133	void SiteMapper::newMap(const string& siteMap)
171			{
172	Douglas Thrift	144	ofstream fout(siteMap.c_str());
173
174			fout << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
175			<< "<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl"
176			<< "\"?>\n"
177			<< "<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">\n"
178	Douglas Thrift	154	<< "<!--" << comment << "-->\n"
179	Douglas Thrift	144	<< "<page>\n"
180			<< "\t<title>Sitemap</title>\n"
181			<< "\t<section>\n"
182			<< "\t\t<list>\n";
183
184	Douglas Thrift	153	for (unsigned index(0); index < pages.size(); ++index)
185	Douglas Thrift	140	{
186	Douglas Thrift	143	if (newPages.find(pages[index].getAddress()) != newPages.end())
187			{
188			newMap(pages[index].getChildren(), pages[index].getPath(),
189			newPages.find(pages[index].getAddress())->second);
190			}
191
192	Douglas Thrift	144	fout << pages[index](3) << '\n';
193	Douglas Thrift	140	}
194	Douglas Thrift	144
195			fout << "\t\t</list>\n"
196			<< "\t</section>\n"
197			<< "</page>\n";
198
199			fout.close();
200	Douglas Thrift	133	}
201	Douglas Thrift	143
202			void SiteMapper::newMap(vector<Page>& pages, const string& childOf,
203			multimap<string, Page>& newPages)
204			{
205	Douglas Thrift	153	for (unsigned index(0); index < pages.size(); ++index)
206	Douglas Thrift	143	{
207			newMap(pages[index].getChildren(), pages[index].getPath(), newPages);
208			}
209
210	Douglas Thrift	153	for (multimap<string, Page>::iterator itor(newPages.lower_bound(childOf));
211	Douglas Thrift	143	itor != newPages.upper_bound(childOf); itor++)
212			{
213	Douglas Thrift	156	cout << "Added: " << itor->second.getUrl() << '\n';
214
215	Douglas Thrift	143	pages.push_back(itor->second);
216			}
217
218			newPages.erase(childOf);
219			}