| 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" |
| 61 |
|
SiteMapper::SiteMapper(const string& siteIndex, const string& siteMap) |
| 62 |
|
{ |
| 63 |
|
oldMap(siteMap); |
| 64 |
< |
index(siteIndex); |
| 64 |
> |
newIndex(siteIndex); |
| 65 |
|
newMap(siteMap); |
| 66 |
|
} |
| 67 |
|
|
| 84 |
|
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 |
< |
list_ = evaluator.createXPath(XalanDOMString("list").c_str()); |
| 87 |
> |
this->list = evaluator.createXPath(XalanDOMString("list").c_str()); |
| 88 |
|
|
| 89 |
|
oldMap(pages, list); |
| 90 |
|
|
| 91 |
|
evaluator.destroyXPath(item); |
| 92 |
|
evaluator.destroyXPath(address); |
| 93 |
|
evaluator.destroyXPath(link); |
| 94 |
< |
evaluator.destroyXPath(list_); |
| 94 |
> |
evaluator.destroyXPath(this->list); |
| 95 |
|
} |
| 96 |
|
|
| 97 |
< |
void SiteMapper::oldMap(list<Page>& pages, XalanNode* list) |
| 97 |
> |
void SiteMapper::oldMap(vector<Page>& pages, XalanNode* list) |
| 98 |
|
{ |
| 99 |
|
NodeRefList nodes = evaluator.selectNodeList(support, list, *item); |
| 100 |
|
|
| 102 |
|
{ |
| 103 |
|
XalanNode* node = nodes.item(index); |
| 104 |
|
ostringstream url, title; |
| 105 |
< |
|
| 105 |
> |
|
| 106 |
|
url << evaluator.evaluate(support, node, *address)->str(); |
| 107 |
|
title << evaluator.evaluate(support, node, *link)->str(); |
| 108 |
|
|
| 109 |
|
Page page(url.str(), title.str()); |
| 110 |
< |
XalanNode* list = evaluator.selectSingleNode(support, node, *list_); |
| 110 |
> |
XalanNode* list = evaluator.selectSingleNode(support, node, |
| 111 |
> |
*(this->list)); |
| 112 |
|
|
| 113 |
|
if (list != 0) oldMap(page.getChildren(), list); |
| 114 |
|
|
| 116 |
|
} |
| 117 |
|
} |
| 118 |
|
|
| 119 |
< |
void SiteMapper::index(const string& siteIndex) |
| 119 |
> |
void SiteMapper::newIndex(const string& siteIndex) |
| 120 |
|
{ |
| 121 |
|
XalanDOMString file(siteIndex.c_str()); |
| 122 |
|
LocalFileInputSource source(file.c_str()); |
| 129 |
|
port = evaluator.createXPath(XalanDOMString("port").c_str()); |
| 130 |
|
path = evaluator.createXPath(XalanDOMString("path").c_str()); |
| 131 |
|
title = evaluator.createXPath(XalanDOMString("title").c_str()); |
| 132 |
< |
|
| 132 |
> |
|
| 133 |
|
NodeRefList nodes = evaluator.selectNodeList(support, document, |
| 134 |
|
XalanDOMString("/index/page").c_str()); |
| 135 |
|
|
| 153 |
|
title << evaluator.evaluate(support, node, *(this->title))->str(); |
| 154 |
|
|
| 155 |
|
Page page(address.str(), path.str(), title.str()); |
| 156 |
< |
Matcher matcher("^Douglas Thrift's.+Website \\| (.+)$"); |
| 156 |
> |
Matcher matcher; |
| 157 |
|
|
| 158 |
< |
if (page == matcher) |
| 158 |
> |
if (page == matcher(string("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDou") |
| 159 |
> |
+ "glas\\sThrift's\\sBlog:\\s(.+)$")) |
| 160 |
> |
{ |
| 161 |
> |
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 |
|
page.setTitle(matcher[1]); |
| 172 |
|
|
| 173 |
< |
newPages.insert(page); |
| 173 |
> |
if (newIndex(pages, page)) continue; |
| 174 |
|
} |
| 175 |
+ |
else continue; |
| 176 |
+ |
|
| 177 |
+ |
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 |
|
} |
| 183 |
|
|
| 184 |
|
evaluator.destroyXPath(address); |
| 187 |
|
evaluator.destroyXPath(title); |
| 188 |
|
} |
| 189 |
|
|
| 190 |
+ |
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 |
|
void SiteMapper::newMap(const string& siteMap) |
| 225 |
|
{ |
| 226 |
< |
// |
| 226 |
> |
for (unsigned index = 0; index < pages.size(); index++) |
| 227 |
> |
{ |
| 228 |
> |
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 |
> |
cout << pages[index] << '\n'; |
| 235 |
> |
} |
| 236 |
> |
} |
| 237 |
> |
|
| 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 (list<Page>::iterator itor = pages.begin(); itor != pages.end(); |
| 247 |
< |
itor++) |
| 246 |
> |
for (multimap<string, Page>::iterator itor = newPages.lower_bound(childOf); |
| 247 |
> |
itor != newPages.upper_bound(childOf); itor++) |
| 248 |
|
{ |
| 249 |
< |
cout << *itor << '\n'; |
| 249 |
> |
pages.push_back(itor->second); |
| 250 |
|
} |
| 251 |
+ |
|
| 252 |
+ |
newPages.erase(childOf); |
| 253 |
|
} |
