| 1 |
// Site Mapper |
| 2 |
// |
| 3 |
// Douglas Thrift |
| 4 |
// |
| 5 |
// $Id$ |
| 6 |
|
| 7 |
#ifdef _WIN32 |
| 8 |
#pragma warning(disable:4503) |
| 9 |
#endif |
| 10 |
|
| 11 |
#include "SiteMapper.hpp" |
| 12 |
#include "Matcher.hpp" |
| 13 |
#include "Page.hpp" |
| 14 |
|
| 15 |
string program; |
| 16 |
bool debug(false); |
| 17 |
|
| 18 |
int main(int argc, char* argv[]) |
| 19 |
{ |
| 20 |
program = argv[0]; |
| 21 |
|
| 22 |
string siteIndex, siteMap; |
| 23 |
|
| 24 |
for (int index(1); index < argc; index++) |
| 25 |
{ |
| 26 |
string arg(argv[index]); |
| 27 |
Matcher matcher; |
| 28 |
|
| 29 |
if (arg == matcher("^-index=(.*)$")) |
| 30 |
{ |
| 31 |
siteIndex = matcher[1]; |
| 32 |
} |
| 33 |
else if (arg == matcher("^-map=(.*)$")) |
| 34 |
{ |
| 35 |
siteMap = matcher[1]; |
| 36 |
} |
| 37 |
else if (arg == "-D") |
| 38 |
{ |
| 39 |
if (!debug) debug = true; |
| 40 |
} |
| 41 |
} |
| 42 |
|
| 43 |
if (!siteIndex.empty() && !siteMap.empty()) |
| 44 |
{ |
| 45 |
XMLPlatformUtils::Initialize(); |
| 46 |
XPathEvaluator::initialize(); |
| 47 |
|
| 48 |
SiteMapper mapper(siteIndex, siteMap); |
| 49 |
|
| 50 |
XPathEvaluator::terminate(); |
| 51 |
XMLPlatformUtils::Terminate(); |
| 52 |
} |
| 53 |
else |
| 54 |
{ |
| 55 |
cout << "Usage: " << program << " -index=index -map=map [-D]\n"; |
| 56 |
} |
| 57 |
|
| 58 |
return 0; |
| 59 |
} |
| 60 |
|
| 61 |
SiteMapper::SiteMapper(const string& siteIndex, const string& siteMap) |
| 62 |
{ |
| 63 |
oldMap(siteMap); |
| 64 |
newIndex(siteIndex); |
| 65 |
newMap(siteMap); |
| 66 |
} |
| 67 |
|
| 68 |
void SiteMapper::oldMap(const string& siteMap) |
| 69 |
{ |
| 70 |
support.setParserLiaison(&liaison); |
| 71 |
|
| 72 |
XalanDOMString file(siteMap.c_str()); |
| 73 |
LocalFileInputSource source(file.c_str()); |
| 74 |
|
| 75 |
XalanDocument* document = liaison.parseXMLStream(source); |
| 76 |
|
| 77 |
if (document == 0) return; |
| 78 |
|
| 79 |
XalanNode* list = evaluator.selectSingleNode(support, document, |
| 80 |
XalanDOMString("/page/section/list").c_str()); |
| 81 |
|
| 82 |
if (list == 0) return; |
| 83 |
|
| 84 |
comment << evaluator.evaluate(support, document, |
| 85 |
XalanDOMString("comment()").c_str())->str(); |
| 86 |
|
| 87 |
item = evaluator.createXPath(XalanDOMString("item").c_str()); |
| 88 |
address = evaluator.createXPath(XalanDOMString("link/@address").c_str()); |
| 89 |
link = evaluator.createXPath(XalanDOMString("link").c_str()); |
| 90 |
this->list = evaluator.createXPath(XalanDOMString("list").c_str()); |
| 91 |
|
| 92 |
oldMap(pages, list); |
| 93 |
|
| 94 |
evaluator.destroyXPath(item); |
| 95 |
evaluator.destroyXPath(address); |
| 96 |
evaluator.destroyXPath(link); |
| 97 |
evaluator.destroyXPath(this->list); |
| 98 |
} |
| 99 |
|
| 100 |
void SiteMapper::oldMap(vector<Page>& pages, XalanNode* list) |
| 101 |
{ |
| 102 |
NodeRefList nodes = evaluator.selectNodeList(support, list, *item); |
| 103 |
|
| 104 |
for (int index(0); index < nodes.getLength(); ++index) |
| 105 |
{ |
| 106 |
XalanNode* node = nodes.item(index); |
| 107 |
ostringstream url, title; |
| 108 |
|
| 109 |
url << evaluator.evaluate(support, node, *address)->str(); |
| 110 |
title << evaluator.evaluate(support, node, *link)->str(); |
| 111 |
|
| 112 |
Page page(url.str(), title.str()); |
| 113 |
XalanNode* list = evaluator.selectSingleNode(support, node, |
| 114 |
*(this->list)); |
| 115 |
|
| 116 |
if (list != 0) oldMap(page.getChildren(), list); |
| 117 |
|
| 118 |
pages.push_back(page); |
| 119 |
} |
| 120 |
} |
| 121 |
|
| 122 |
void SiteMapper::newIndex(const string& siteIndex) |
| 123 |
{ |
| 124 |
XalanDOMString file(siteIndex.c_str()); |
| 125 |
LocalFileInputSource source(file.c_str()); |
| 126 |
|
| 127 |
XalanDocument* document = liaison.parseXMLStream(source); |
| 128 |
|
| 129 |
if (document == 0) return; |
| 130 |
|
| 131 |
address = evaluator.createXPath(XalanDOMString("address").c_str()); |
| 132 |
port = evaluator.createXPath(XalanDOMString("port").c_str()); |
| 133 |
path = evaluator.createXPath(XalanDOMString("path").c_str()); |
| 134 |
title = evaluator.createXPath(XalanDOMString("title").c_str()); |
| 135 |
|
| 136 |
NodeRefList nodes = evaluator.selectNodeList(support, document, |
| 137 |
XalanDOMString("/index/page").c_str()); |
| 138 |
|
| 139 |
for (int index(0); index < nodes.getLength(); ++index) |
| 140 |
{ |
| 141 |
XalanNode* node = nodes.item(index); |
| 142 |
ostringstream address; |
| 143 |
|
| 144 |
address << evaluator.evaluate(support, node, *(this->address))->str(); |
| 145 |
|
| 146 |
double port = evaluator.evaluate(support, node, *(this->port))->num(); |
| 147 |
|
| 148 |
if (port >= 0 && port <= 65535) |
| 149 |
{ |
| 150 |
address << ':' << int(port); |
| 151 |
} |
| 152 |
|
| 153 |
ostringstream path, title; |
| 154 |
|
| 155 |
path << evaluator.evaluate(support, node, *(this->path))->str(); |
| 156 |
title << evaluator.evaluate(support, node, *(this->title))->str(); |
| 157 |
|
| 158 |
Page page(address.str(), path.str(), title.str()); |
| 159 |
Matcher matcher; |
| 160 |
|
| 161 |
if (page == matcher(string("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDou") |
| 162 |
+ "glas\\sThrift's\\sBlog:\\s(.+)$")) |
| 163 |
{ |
| 164 |
if (Matcher("^\\w+\\s\\d{4}\\sArchives$") == matcher[1]) |
| 165 |
{ |
| 166 |
page.setTitle(matcher[1]); |
| 167 |
|
| 168 |
if (newIndex(pages, page)) continue; |
| 169 |
} |
| 170 |
else continue; |
| 171 |
} |
| 172 |
else if (page == matcher("^Douglas\\sThrift's.+Website\\s\\|\\s(.+)$")) |
| 173 |
{ |
| 174 |
page.setTitle(matcher[1]); |
| 175 |
|
| 176 |
if (newIndex(pages, page)) continue; |
| 177 |
} |
| 178 |
else continue; |
| 179 |
|
| 180 |
multimap<string, Page> items; |
| 181 |
|
| 182 |
newPages.insert(pair<string, multimap<string, Page> |
| 183 |
>(page.getAddress(), items)).first->second.insert(pair<string, |
| 184 |
Page>(page.getChildOf(), page)); |
| 185 |
} |
| 186 |
|
| 187 |
evaluator.destroyXPath(address); |
| 188 |
evaluator.destroyXPath(port); |
| 189 |
evaluator.destroyXPath(path); |
| 190 |
evaluator.destroyXPath(title); |
| 191 |
} |
| 192 |
|
| 193 |
bool SiteMapper::newIndex(vector<Page>& pages, Page& page) |
| 194 |
{ |
| 195 |
for (unsigned index(0); index < pages.size(); ++index) |
| 196 |
{ |
| 197 |
if (pages[index] == page.getAddress()) |
| 198 |
{ |
| 199 |
Matcher matcher; |
| 200 |
|
| 201 |
if (pages[index] == page) |
| 202 |
{ |
| 203 |
page.setChildren(pages[index].getChildren()); |
| 204 |
|
| 205 |
pages[index] = page; |
| 206 |
|
| 207 |
cout << "Updated: " << page.getUrl() << '\n'; |
| 208 |
|
| 209 |
return true; |
| 210 |
} |
| 211 |
else if (matcher('^' + pages[index].getPath()) == page) |
| 212 |
{ |
| 213 |
page.setChildOf(matcher[0]); |
| 214 |
|
| 215 |
if (matcher('^' + pages[index].getTitle() + "\\s\\|\\s(.+)$") |
| 216 |
== page) |
| 217 |
{ |
| 218 |
page.setTitle(matcher[1]); |
| 219 |
} |
| 220 |
|
| 221 |
return newIndex(pages[index].getChildren(), page); |
| 222 |
} |
| 223 |
} |
| 224 |
} |
| 225 |
|
| 226 |
return false; |
| 227 |
} |
| 228 |
|
| 229 |
void SiteMapper::newMap(const string& siteMap) |
| 230 |
{ |
| 231 |
ofstream fout(siteMap.c_str()); |
| 232 |
|
| 233 |
fout << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" |
| 234 |
<< "<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl" |
| 235 |
<< "\"?>\n" |
| 236 |
<< "<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">\n" |
| 237 |
<< "<!--" << comment.str() << "-->\n" |
| 238 |
<< "<page>\n" |
| 239 |
<< "\t<title>Sitemap</title>\n" |
| 240 |
<< "\t<section>\n" |
| 241 |
<< "\t\t<list>\n"; |
| 242 |
|
| 243 |
for (unsigned index(0); index < pages.size(); ++index) |
| 244 |
{ |
| 245 |
if (newPages.find(pages[index].getAddress()) != newPages.end()) |
| 246 |
{ |
| 247 |
newMap(pages[index].getChildren(), pages[index].getPath(), |
| 248 |
newPages.find(pages[index].getAddress())->second); |
| 249 |
} |
| 250 |
|
| 251 |
fout << pages[index](3) << '\n'; |
| 252 |
} |
| 253 |
|
| 254 |
fout << "\t\t</list>\n" |
| 255 |
<< "\t</section>\n" |
| 256 |
<< "</page>\n"; |
| 257 |
|
| 258 |
fout.close(); |
| 259 |
} |
| 260 |
|
| 261 |
void SiteMapper::newMap(vector<Page>& pages, const string& childOf, |
| 262 |
multimap<string, Page>& newPages) |
| 263 |
{ |
| 264 |
for (unsigned index(0); index < pages.size(); ++index) |
| 265 |
{ |
| 266 |
newMap(pages[index].getChildren(), pages[index].getPath(), newPages); |
| 267 |
} |
| 268 |
|
| 269 |
for (multimap<string, Page>::iterator itor(newPages.lower_bound(childOf)); |
| 270 |
itor != newPages.upper_bound(childOf); itor++) |
| 271 |
{ |
| 272 |
cout << "Added: " << itor->second.getUrl() << '\n'; |
| 273 |
|
| 274 |
pages.push_back(itor->second); |
| 275 |
} |
| 276 |
|
| 277 |
newPages.erase(childOf); |
| 278 |
} |
