repos/SiteMapper/SiteMapper.cpp

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

#ifdef _WIN32
#pragma warning(disable:4503)
#endif

#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 != "" && siteMap != "")
        {
                XMLPlatformUtils::Initialize();
                XPathEvaluator::initialize();

                SiteMapper mapper(siteIndex, siteMap);

                XPathEvaluator::terminate();
                XMLPlatformUtils::Terminate();
        }
        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)
{
        support.setParserLiaison(&liaison);

        XalanDOMString file(siteMap.c_str());
        LocalFileInputSource source(file.c_str());

        XalanDocument* document = liaison.parseXMLStream(source);

        if (document == 0) return;

        XalanNode* list = evaluator.selectSingleNode(support, document,
                XalanDOMString("/page/section/list").c_str());

        if (list == 0) return;

        item = evaluator.createXPath(XalanDOMString("item").c_str());
        address = evaluator.createXPath(XalanDOMString("link/@address").c_str());
        link = evaluator.createXPath(XalanDOMString("link").c_str());
        this->list = evaluator.createXPath(XalanDOMString("list").c_str());

        oldMap(pages, list);

        evaluator.destroyXPath(item);
        evaluator.destroyXPath(address);
        evaluator.destroyXPath(link);
        evaluator.destroyXPath(this->list);
}

void SiteMapper::oldMap(vector<Page>& pages, XalanNode* list)
{
        NodeRefList nodes = evaluator.selectNodeList(support, list, *item);

        for (int index = 0; index < nodes.getLength(); index++)
        {
                XalanNode* node = nodes.item(index);
                ostringstream url, title;

                url << evaluator.evaluate(support, node, *address)->str();
                title << evaluator.evaluate(support, node, *link)->str();

                Page page(url.str(), title.str());
                XalanNode* list = evaluator.selectSingleNode(support, node,
                        *(this->list));

                if (list != 0) oldMap(page.getChildren(), list);

                pages.push_back(page);
        }
}

void SiteMapper::newIndex(const string& siteIndex)
{
        XalanDOMString file(siteIndex.c_str());
        LocalFileInputSource source(file.c_str());

        XalanDocument* document = liaison.parseXMLStream(source);

        if (document == 0) return;

        address = evaluator.createXPath(XalanDOMString("address").c_str());
        port = evaluator.createXPath(XalanDOMString("port").c_str());
        path = evaluator.createXPath(XalanDOMString("path").c_str());
        title = evaluator.createXPath(XalanDOMString("title").c_str());

        NodeRefList nodes = evaluator.selectNodeList(support, document,
                XalanDOMString("/index/page").c_str());

        for (int index = 0; index < nodes.getLength(); index++)
        {
                XalanNode* node = nodes.item(index);
                ostringstream address;

                address << evaluator.evaluate(support, node, *(this->address))->str();

                double port = evaluator.evaluate(support, node, *(this->port))->num();

                if (port >= 0 && port <= 65535)
                {
                        address << ':' << int(port);
                }

                ostringstream path, title;

                path << evaluator.evaluate(support, node, *(this->path))->str();
                title << evaluator.evaluate(support, node, *(this->title))->str();

                Page page(address.str(), path.str(), title.str());
                Matcher matcher;

                if (page == matcher(string("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDou")
                        + "glas\\sThrift's\\sBlog:\\s(.+)$"))
                {
                        if (Matcher("^\\w+\\s\\d\\d\\d\\d\\sArchives$") == 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));
        }

        evaluator.destroyXPath(address);
        evaluator.destroyXPath(port);
        evaluator.destroyXPath(path);
        evaluator.destroyXPath(title);
}

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;

                                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)
{
        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);
                }

                cout << pages[index] << '\n';
        }
}

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++)
        {
                pages.push_back(itor->second);
        }

        newPages.erase(childOf);
}
Revision:	143
Committed:	2004-03-31T22:55:16-08:00 (21 years, 2 months ago) by Douglas Thrift
File size:	5835 byte(s)
Log Message:	Almost there!
#	User	Rev	Content
1	Douglas Thrift	126	// Site Mapper
2			//
3			// Douglas Thrift
4			//
5			// $Id$
6
7	Douglas Thrift	143	#ifdef _WIN32
8			#pragma warning(disable:4503)
9			#endif
10
11	Douglas Thrift	126	#include "SiteMapper.hpp"
12			#include "Matcher.hpp"
13	Douglas Thrift	128	#include "Page.hpp"
14	Douglas Thrift	126
15	Douglas Thrift	128	string program;
16			bool debug = false;
17
18	Douglas Thrift	126	int main(int argc, char* argv[])
19			{
20	Douglas Thrift	128	program = argv[0];
21	Douglas Thrift	132
22	Douglas Thrift	126	string siteIndex, siteMap;
23
24	Douglas Thrift	132	for (int index = 1; index < argc; index++)
25	Douglas Thrift	126	{
26	Douglas Thrift	128	string arg(argv[index]);
27			Matcher matcher;
28	Douglas Thrift	126
29	Douglas Thrift	128	if (arg == matcher("^-index=(.*)$"))
30			{
31			siteIndex = matcher[1];
32			}
33			else if (arg == matcher("^-map=(.*)$"))
34			{
35			siteMap = matcher[1];
36			}
37	Douglas Thrift	133	else if (arg == "-D")
38			{
39			if (!debug) debug = true;
40			}
41	Douglas Thrift	126	}
42
43	Douglas Thrift	128	if (siteIndex != "" && siteMap != "")
44	Douglas Thrift	132	{
45	Douglas Thrift	128	XMLPlatformUtils::Initialize();
46			XPathEvaluator::initialize();
47	Douglas Thrift	126
48	Douglas Thrift	128	SiteMapper mapper(siteIndex, siteMap);
49
50			XPathEvaluator::terminate();
51			XMLPlatformUtils::Terminate();
52			}
53			else
54			{
55	Douglas Thrift	133	cout << "Usage: " << program << " -index=index -map=map [-D]\n";
56	Douglas Thrift	128	}
57	Douglas Thrift	129
58	Douglas Thrift	126	return 0;
59			}
60
61			SiteMapper::SiteMapper(const string& siteIndex, const string& siteMap)
62			{
63	Douglas Thrift	133	oldMap(siteMap);
64	Douglas Thrift	142	newIndex(siteIndex);
65	Douglas Thrift	134	newMap(siteMap);
66	Douglas Thrift	133	}
67
68			void SiteMapper::oldMap(const string& siteMap)
69			{
70			support.setParserLiaison(&liaison);
71
72	Douglas Thrift	134	XalanDOMString file(siteMap.c_str());
73			LocalFileInputSource source(file.c_str());
74	Douglas Thrift	133
75	Douglas Thrift	134	XalanDocument* document = liaison.parseXMLStream(source);
76	Douglas Thrift	133
77	Douglas Thrift	134	if (document == 0) return;
78	Douglas Thrift	133
79	Douglas Thrift	134	XalanNode* list = evaluator.selectSingleNode(support, document,
80			XalanDOMString("/page/section/list").c_str());
81	Douglas Thrift	133
82	Douglas Thrift	134	if (list == 0) return;
83	Douglas Thrift	135
84	Douglas Thrift	137	item = evaluator.createXPath(XalanDOMString("item").c_str());
85			address = evaluator.createXPath(XalanDOMString("link/@address").c_str());
86			link = evaluator.createXPath(XalanDOMString("link").c_str());
87	Douglas Thrift	143	this->list = evaluator.createXPath(XalanDOMString("list").c_str());
88	Douglas Thrift	137
89			oldMap(pages, list);
90	Douglas Thrift	138
91			evaluator.destroyXPath(item);
92			evaluator.destroyXPath(address);
93			evaluator.destroyXPath(link);
94	Douglas Thrift	143	evaluator.destroyXPath(this->list);
95	Douglas Thrift	133	}
96
97	Douglas Thrift	143	void SiteMapper::oldMap(vector<Page>& pages, XalanNode* list)
98	Douglas Thrift	135	{
99	Douglas Thrift	143	NodeRefList nodes = evaluator.selectNodeList(support, list, *item);
100	Douglas Thrift	135
101			for (int index = 0; index < nodes.getLength(); index++)
102			{
103			XalanNode* node = nodes.item(index);
104	Douglas Thrift	138	ostringstream url, title;
105	Douglas Thrift	141
106	Douglas Thrift	138	url << evaluator.evaluate(support, node, *address)->str();
107			title << evaluator.evaluate(support, node, *link)->str();
108	Douglas Thrift	135
109			Page page(url.str(), title.str());
110	Douglas Thrift	141	XalanNode* list = evaluator.selectSingleNode(support, node,
111	Douglas Thrift	143	*(this->list));
112	Douglas Thrift	135
113	Douglas Thrift	137	if (list != 0) oldMap(page.getChildren(), list);
114	Douglas Thrift	135
115			pages.push_back(page);
116			}
117			}
118
119	Douglas Thrift	142	void SiteMapper::newIndex(const string& siteIndex)
120	Douglas Thrift	133	{
121	Douglas Thrift	138	XalanDOMString file(siteIndex.c_str());
122			LocalFileInputSource source(file.c_str());
123
124			XalanDocument* document = liaison.parseXMLStream(source);
125
126			if (document == 0) return;
127
128			address = evaluator.createXPath(XalanDOMString("address").c_str());
129	Douglas Thrift	139	port = evaluator.createXPath(XalanDOMString("port").c_str());
130	Douglas Thrift	138	path = evaluator.createXPath(XalanDOMString("path").c_str());
131			title = evaluator.createXPath(XalanDOMString("title").c_str());
132	Douglas Thrift	141
133	Douglas Thrift	138	NodeRefList nodes = evaluator.selectNodeList(support, document,
134			XalanDOMString("/index/page").c_str());
135
136			for (int index = 0; index < nodes.getLength(); index++)
137			{
138			XalanNode* node = nodes.item(index);
139	Douglas Thrift	139	ostringstream address;
140
141			address << evaluator.evaluate(support, node, *(this->address))->str();
142
143			double port = evaluator.evaluate(support, node, *(this->port))->num();
144
145			if (port >= 0 && port <= 65535)
146			{
147			address << ':' << int(port);
148			}
149
150	Douglas Thrift	140	ostringstream path, title;
151
152			path << evaluator.evaluate(support, node, *(this->path))->str();
153			title << evaluator.evaluate(support, node, *(this->title))->str();
154
155			Page page(address.str(), path.str(), title.str());
156	Douglas Thrift	142	Matcher matcher;
157	Douglas Thrift	140
158	Douglas Thrift	142	if (page == matcher(string("^Douglas\\sThrift's\\sWebsite\\s\\\|\\sDou")
159			+ "glas\\sThrift's\\sBlog:\\s(.+)$"))
160	Douglas Thrift	140	{
161	Douglas Thrift	142	if (Matcher("^\\w+\\s\\d\\d\\d\\d\\sArchives$") == matcher[1])
162			{
163			page.setTitle(matcher[1]);
164
165			if (newIndex(pages, page)) continue;
166			}
167			else continue;
168			}
169			else if (page == matcher("^Douglas\\sThrift's.+Website\\s\\\|\\s(.+)$"))
170			{
171	Douglas Thrift	140	page.setTitle(matcher[1]);
172
173	Douglas Thrift	142	if (newIndex(pages, page)) continue;
174	Douglas Thrift	140	}
175	Douglas Thrift	142	else continue;
176
177	Douglas Thrift	143	multimap<string, Page> items;
178
179			newPages.insert(pair<string, multimap<string, Page>
180			>(page.getAddress(), items)).first->second.insert(pair<string,
181			Page>(page.getChildOf(), page));
182	Douglas Thrift	138	}
183
184			evaluator.destroyXPath(address);
185	Douglas Thrift	139	evaluator.destroyXPath(port);
186	Douglas Thrift	138	evaluator.destroyXPath(path);
187			evaluator.destroyXPath(title);
188	Douglas Thrift	126	}
189	Douglas Thrift	133
190	Douglas Thrift	142	bool SiteMapper::newIndex(vector<Page>& pages, Page& page)
191			{
192			for (unsigned index = 0; index < pages.size(); index++)
193			{
194			if (pages[index] == page.getAddress())
195			{
196			Matcher matcher;
197
198			if (pages[index] == page)
199			{
200			page.setChildren(pages[index].getChildren());
201
202			pages[index] = page;
203
204			return true;
205			}
206			else if (matcher('^' + pages[index].getPath()) == page)
207			{
208			page.setChildOf(matcher[0]);
209
210			if (matcher('^' + pages[index].getTitle() + "\\s\\\|\\s(.+)$")
211			== page)
212			{
213			page.setTitle(matcher[1]);
214			}
215
216			return newIndex(pages[index].getChildren(), page);
217			}
218			}
219			}
220
221			return false;
222			}
223
224	Douglas Thrift	133	void SiteMapper::newMap(const string& siteMap)
225			{
226	Douglas Thrift	141	for (unsigned index = 0; index < pages.size(); index++)
227	Douglas Thrift	140	{
228	Douglas Thrift	143	if (newPages.find(pages[index].getAddress()) != newPages.end())
229			{
230			newMap(pages[index].getChildren(), pages[index].getPath(),
231			newPages.find(pages[index].getAddress())->second);
232			}
233
234	Douglas Thrift	141	cout << pages[index] << '\n';
235	Douglas Thrift	140	}
236	Douglas Thrift	133	}
237	Douglas Thrift	143
238			void SiteMapper::newMap(vector<Page>& pages, const string& childOf,
239			multimap<string, Page>& newPages)
240			{
241			for (unsigned index = 0; index < pages.size(); index++)
242			{
243			newMap(pages[index].getChildren(), pages[index].getPath(), newPages);
244			}
245
246			for (multimap<string, Page>::iterator itor = newPages.lower_bound(childOf);
247			itor != newPages.upper_bound(childOf); itor++)
248			{
249			pages.push_back(itor->second);
250			}
251
252			newPages.erase(childOf);
253			}