| 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; |
| 16 |
> |
bool debug(false); |
| 17 |
|
|
| 18 |
|
int main(int argc, char* argv[]) |
| 19 |
|
{ |
| 21 |
|
|
| 22 |
|
string siteIndex, siteMap; |
| 23 |
|
|
| 24 |
< |
for (int index = 1; index < argc; index++) |
| 24 |
> |
for (int index(1); index < argc; index++) |
| 25 |
|
{ |
| 26 |
|
string arg(argv[index]); |
| 27 |
|
Matcher matcher; |
| 40 |
|
} |
| 41 |
|
} |
| 42 |
|
|
| 43 |
< |
if (siteIndex != "" && siteMap != "") |
| 43 |
> |
if (!siteIndex.empty() && !siteMap.empty()) |
| 44 |
|
{ |
| 45 |
|
XMLPlatformUtils::Initialize(); |
| 46 |
|
XPathEvaluator::initialize(); |
| 81 |
|
|
| 82 |
|
if (list == 0) return; |
| 83 |
|
|
| 84 |
+ |
comment << evaluator.evaluate(support, document, |
| 85 |
+ |
XalanDOMString("comment()").c_str())->str(); |
| 86 |
+ |
|
| 87 |
+ |
if (debug) cerr << "comment = " << comment.str() << '\n'; |
| 88 |
+ |
|
| 89 |
|
item = evaluator.createXPath(XalanDOMString("item").c_str()); |
| 90 |
|
address = evaluator.createXPath(XalanDOMString("link/@address").c_str()); |
| 91 |
|
link = evaluator.createXPath(XalanDOMString("link").c_str()); |
| 92 |
< |
list_ = evaluator.createXPath(XalanDOMString("list").c_str()); |
| 92 |
> |
this->list = evaluator.createXPath(XalanDOMString("list").c_str()); |
| 93 |
|
|
| 94 |
|
oldMap(pages, list); |
| 95 |
|
|
| 96 |
|
evaluator.destroyXPath(item); |
| 97 |
|
evaluator.destroyXPath(address); |
| 98 |
|
evaluator.destroyXPath(link); |
| 99 |
< |
evaluator.destroyXPath(list_); |
| 99 |
> |
evaluator.destroyXPath(this->list); |
| 100 |
|
} |
| 101 |
|
|
| 102 |
< |
void SiteMapper::oldMap(vector<Page>& pages, XalanNode* list_) |
| 102 |
> |
void SiteMapper::oldMap(vector<Page>& pages, XalanNode* list) |
| 103 |
|
{ |
| 104 |
< |
NodeRefList nodes = evaluator.selectNodeList(support, list_, *item); |
| 104 |
> |
NodeRefList nodes = evaluator.selectNodeList(support, list, *item); |
| 105 |
|
|
| 106 |
< |
for (int index = 0; index < nodes.getLength(); index++) |
| 106 |
> |
for (int index(0); index < nodes.getLength(); ++index) |
| 107 |
|
{ |
| 108 |
|
XalanNode* node = nodes.item(index); |
| 109 |
|
ostringstream url, title; |
| 113 |
|
|
| 114 |
|
Page page(url.str(), title.str()); |
| 115 |
|
XalanNode* list = evaluator.selectSingleNode(support, node, |
| 116 |
< |
*(this->list_)); |
| 116 |
> |
*(this->list)); |
| 117 |
|
|
| 118 |
|
if (list != 0) oldMap(page.getChildren(), list); |
| 119 |
|
|
| 138 |
|
NodeRefList nodes = evaluator.selectNodeList(support, document, |
| 139 |
|
XalanDOMString("/index/page").c_str()); |
| 140 |
|
|
| 141 |
< |
for (int index = 0; index < nodes.getLength(); index++) |
| 141 |
> |
for (int index(0); index < nodes.getLength(); ++index) |
| 142 |
|
{ |
| 143 |
|
XalanNode* node = nodes.item(index); |
| 144 |
|
ostringstream address; |
| 163 |
|
if (page == matcher(string("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDou") |
| 164 |
|
+ "glas\\sThrift's\\sBlog:\\s(.+)$")) |
| 165 |
|
{ |
| 166 |
< |
if (Matcher("^\\w+\\s\\d\\d\\d\\d\\sArchives$") == matcher[1]) |
| 166 |
> |
if (Matcher("^\\w+\\s\\d{4}\\sArchives$") == matcher[1]) |
| 167 |
|
{ |
| 168 |
|
page.setTitle(matcher[1]); |
| 169 |
|
|
| 179 |
|
} |
| 180 |
|
else continue; |
| 181 |
|
|
| 182 |
< |
cerr << page.getTitle() << ' ' << page.getChildOf() << '\n'; |
| 183 |
< |
// newPages.insert(page); |
| 182 |
> |
multimap<string, Page> items; |
| 183 |
> |
|
| 184 |
> |
newPages.insert(pair<string, multimap<string, Page> |
| 185 |
> |
>(page.getAddress(), items)).first->second.insert(pair<string, |
| 186 |
> |
Page>(page.getChildOf(), page)); |
| 187 |
|
} |
| 188 |
|
|
| 189 |
|
evaluator.destroyXPath(address); |
| 194 |
|
|
| 195 |
|
bool SiteMapper::newIndex(vector<Page>& pages, Page& page) |
| 196 |
|
{ |
| 197 |
< |
for (unsigned index = 0; index < pages.size(); index++) |
| 197 |
> |
for (unsigned index(0); index < pages.size(); ++index) |
| 198 |
|
{ |
| 199 |
|
if (pages[index] == page.getAddress()) |
| 200 |
|
{ |
| 206 |
|
|
| 207 |
|
pages[index] = page; |
| 208 |
|
|
| 209 |
+ |
cout << "Updated: " << page.getUrl() << '\n'; |
| 210 |
+ |
|
| 211 |
|
return true; |
| 212 |
|
} |
| 213 |
|
else if (matcher('^' + pages[index].getPath()) == page) |
| 225 |
|
} |
| 226 |
|
} |
| 227 |
|
|
| 214 |
– |
cerr << "Here!\n"; |
| 215 |
– |
|
| 228 |
|
return false; |
| 229 |
|
} |
| 230 |
|
|
| 231 |
|
void SiteMapper::newMap(const string& siteMap) |
| 232 |
|
{ |
| 233 |
< |
// |
| 233 |
> |
ofstream fout(siteMap.c_str()); |
| 234 |
> |
|
| 235 |
> |
fout << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" |
| 236 |
> |
<< "<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl" |
| 237 |
> |
<< "\"?>\n" |
| 238 |
> |
<< "<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">\n" |
| 239 |
> |
<< "<!--" << comment.str() << "-->\n" |
| 240 |
> |
<< "<page>\n" |
| 241 |
> |
<< "\t<title>Sitemap</title>\n" |
| 242 |
> |
<< "\t<section>\n" |
| 243 |
> |
<< "\t\t<list>\n"; |
| 244 |
> |
|
| 245 |
> |
for (unsigned index(0); index < pages.size(); ++index) |
| 246 |
> |
{ |
| 247 |
> |
if (newPages.find(pages[index].getAddress()) != newPages.end()) |
| 248 |
> |
{ |
| 249 |
> |
newMap(pages[index].getChildren(), pages[index].getPath(), |
| 250 |
> |
newPages.find(pages[index].getAddress())->second); |
| 251 |
> |
} |
| 252 |
> |
|
| 253 |
> |
fout << pages[index](3) << '\n'; |
| 254 |
> |
} |
| 255 |
> |
|
| 256 |
> |
fout << "\t\t</list>\n" |
| 257 |
> |
<< "\t</section>\n" |
| 258 |
> |
<< "</page>\n"; |
| 259 |
> |
|
| 260 |
> |
fout.close(); |
| 261 |
> |
} |
| 262 |
> |
|
| 263 |
> |
void SiteMapper::newMap(vector<Page>& pages, const string& childOf, |
| 264 |
> |
multimap<string, Page>& newPages) |
| 265 |
> |
{ |
| 266 |
> |
for (unsigned index(0); index < pages.size(); ++index) |
| 267 |
> |
{ |
| 268 |
> |
newMap(pages[index].getChildren(), pages[index].getPath(), newPages); |
| 269 |
> |
} |
| 270 |
|
|
| 271 |
< |
for (unsigned index = 0; index < pages.size(); index++) |
| 271 |
> |
for (multimap<string, Page>::iterator itor(newPages.lower_bound(childOf)); |
| 272 |
> |
itor != newPages.upper_bound(childOf); itor++) |
| 273 |
|
{ |
| 274 |
< |
cout << pages[index] << '\n'; |
| 274 |
> |
cout << "Added: " << itor->second.getUrl() << '\n'; |
| 275 |
> |
|
| 276 |
> |
pages.push_back(itor->second); |
| 277 |
|
} |
| 278 |
+ |
|
| 279 |
+ |
newPages.erase(childOf); |
| 280 |
|
} |
