| 1 |
// Site Mapper |
| 2 |
// |
| 3 |
// Douglas Thrift |
| 4 |
// |
| 5 |
// $Id$ |
| 6 |
|
| 7 |
#include "SiteMapper.hpp" |
| 8 |
#include "Matcher/Matcher.hpp" |
| 9 |
|
| 10 |
#include <menes-api/exename.hpp> |
| 11 |
#include <menes-app/application.hpp> |
| 12 |
|
| 13 |
ext::String program; |
| 14 |
bool debug(false); |
| 15 |
|
| 16 |
struct SiteMapperCommand : public app::Application |
| 17 |
{ |
| 18 |
virtual int Run(const app::ArgumentList& args) |
| 19 |
{ |
| 20 |
program = api::GetExecutablePath().GetName(); |
| 21 |
|
| 22 |
ext::String siteIndex, siteMap; |
| 23 |
|
| 24 |
_foreach (app::ArgumentList, arg, args) |
| 25 |
{ |
| 26 |
Matcher matcher; |
| 27 |
|
| 28 |
if (*arg == matcher("^-index=(.*)$")) |
| 29 |
{ |
| 30 |
siteIndex = matcher[1]; |
| 31 |
} |
| 32 |
else if (*arg == matcher("^-map=(.*)$")) |
| 33 |
{ |
| 34 |
siteMap = matcher[1]; |
| 35 |
} |
| 36 |
else if (*arg == "-D") |
| 37 |
{ |
| 38 |
if (!debug) debug = true; |
| 39 |
} |
| 40 |
} |
| 41 |
|
| 42 |
if (!siteIndex.IsEmpty() && !siteMap.IsEmpty()) |
| 43 |
{ |
| 44 |
SiteMapper mapper(siteIndex, siteMap); |
| 45 |
} |
| 46 |
else |
| 47 |
{ |
| 48 |
api::Cout << "Usage: " << program << " -index=index -map=map [-D]\n"; |
| 49 |
} |
| 50 |
|
| 51 |
return 0; |
| 52 |
} |
| 53 |
} mapper; |
| 54 |
|
| 55 |
SiteMapper::SiteMapper(const ext::String& siteIndex, const ext::String& siteMap) |
| 56 |
{ |
| 57 |
oldMap(siteMap); |
| 58 |
newIndex(siteIndex); |
| 59 |
newMap(siteMap); |
| 60 |
} |
| 61 |
|
| 62 |
void SiteMapper::oldMap(const ext::String& siteMap) |
| 63 |
{ |
| 64 |
ext::Handle<xml::Document> document(xml::Parse(siteMap)); |
| 65 |
ext::Handle<xml::Node> list(*document/"page"/"section"/"list"); |
| 66 |
|
| 67 |
comment = *document/"comment()"; |
| 68 |
|
| 69 |
if (debug) api::Cerr << "comment = " << comment << ios::NewLine; |
| 70 |
|
| 71 |
oldMap(pages, list); |
| 72 |
} |
| 73 |
|
| 74 |
void SiteMapper::oldMap(ext::Vector<Page>& pages, xml::Node* list) |
| 75 |
{ |
| 76 |
xml::NodeSet nodes(*list/"item"); |
| 77 |
|
| 78 |
_foreach (xml::NodeSet, node, nodes) |
| 79 |
{ |
| 80 |
ext::String url(**node/"link"/"@address"), title(**node/"link"); |
| 81 |
Page page(url, title); |
| 82 |
ext::Handle<xml::Node> list(**node/"list"); |
| 83 |
|
| 84 |
if (!list.IsEmpty()) oldMap(page.getChildren(), list); |
| 85 |
|
| 86 |
pages.InsertLast(page); |
| 87 |
} |
| 88 |
} |
| 89 |
|
| 90 |
void SiteMapper::newIndex(const ext::String& siteIndex) |
| 91 |
{ |
| 92 |
ext::Handle<xml::Document> document(xml::Parse(siteIndex)); |
| 93 |
xml::NodeSet nodes(*document/"index"/"page"); |
| 94 |
|
| 95 |
_foreach (xml::NodeSet, node, nodes) |
| 96 |
{ |
| 97 |
ios::String address(**node/"address"); |
| 98 |
ext::String port(**node/"port"); |
| 99 |
|
| 100 |
if (!port.IsEmpty()) address << ":" << port; |
| 101 |
|
| 102 |
ext::String path(**node/"path"), title(**node/"title"); |
| 103 |
Page page(address, path, title); |
| 104 |
Matcher matcher(PCRE_UNGREEDY | PCRE_DOTALL); |
| 105 |
|
| 106 |
if (page == matcher("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDouglas\\sThrift's\\sBlog:\\s(.+)$")) |
| 107 |
{ |
| 108 |
if (Matcher("^\\w+\\s\\d{4}\\sArchives$") == matcher[1]) |
| 109 |
{ |
| 110 |
page.setTitle(matcher[1]); |
| 111 |
|
| 112 |
if (newIndex(pages, page)) continue; |
| 113 |
} |
| 114 |
else continue; |
| 115 |
} |
| 116 |
else if (page == matcher("^Douglas\\sThrift's.+Website\\s\\|\\s(.+)$")) |
| 117 |
{ |
| 118 |
page.setTitle(matcher[1]); |
| 119 |
|
| 120 |
if (newIndex(pages, page)) continue; |
| 121 |
} |
| 122 |
else continue; |
| 123 |
|
| 124 |
std::multimap<std::string, Page> items; |
| 125 |
|
| 126 |
newPages.insert(std::pair<std::string, std::multimap<std::string, Page> >(page.getAddress(), items)).first->second.insert(std::pair<std::string, Page>(page.getChildOf(), page)); |
| 127 |
} |
| 128 |
} |
| 129 |
|
| 130 |
bool SiteMapper::newIndex(ext::Vector<Page>& pages, Page& page) |
| 131 |
{ |
| 132 |
_mforeach (ext::Vector<Page>, page_, pages) |
| 133 |
{ |
| 134 |
if (*page_ == page.getAddress()) |
| 135 |
{ |
| 136 |
Matcher matcher; |
| 137 |
|
| 138 |
if (*page_ == page) |
| 139 |
{ |
| 140 |
page.setChildren(page_->getChildren()); |
| 141 |
|
| 142 |
*page_ = page; |
| 143 |
|
| 144 |
api::Cout << "Updated: " << page.getUrl() << ios::NewLine; |
| 145 |
|
| 146 |
return true; |
| 147 |
} |
| 148 |
else if (matcher("^" + page_->getPath()) == page) |
| 149 |
{ |
| 150 |
page.setChildOf(matcher[0]); |
| 151 |
|
| 152 |
if (matcher("^" + page_->getTitle() + "\\s\\|\\s(.+)$") == page) page.setTitle(matcher[1]); |
| 153 |
|
| 154 |
return newIndex(page_->getChildren(), page); |
| 155 |
} |
| 156 |
} |
| 157 |
} |
| 158 |
|
| 159 |
return false; |
| 160 |
} |
| 161 |
|
| 162 |
void SiteMapper::newMap(const ext::String& siteMap) |
| 163 |
{ |
| 164 |
api::FileWriter file(siteMap); |
| 165 |
ios::FormatWriter fout(file); |
| 166 |
xml::TextWriter xml(file); |
| 167 |
|
| 168 |
fout << ios::NewLine << "<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl\"?>" << ios::NewLine << "<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">"; |
| 169 |
|
| 170 |
xml.OutputComment(comment); |
| 171 |
|
| 172 |
xml::ScopeElement page(xml, "page"); |
| 173 |
|
| 174 |
xml.OpenElement("title"); |
| 175 |
xml.OutputText("Sitemap"); |
| 176 |
xml.CloseElement(); |
| 177 |
|
| 178 |
xml::ScopeElement section(xml, "section"), list(xml, "list"); |
| 179 |
|
| 180 |
_mforeach (ext::Vector<Page>, page, pages) |
| 181 |
{ |
| 182 |
if (newPages.find(page->getAddress()) != newPages.end()) newMap(page->getChildren(), page->getPath(), newPages.find(page->getAddress())->second); |
| 183 |
|
| 184 |
xml << *page; |
| 185 |
} |
| 186 |
} |
| 187 |
|
| 188 |
void SiteMapper::newMap(ext::Vector<Page>& pages, const ext::String& childOf, std::multimap<std::string, Page>& newPages) |
| 189 |
{ |
| 190 |
_mforeach (ext::Vector<Page>, page, pages) newMap(page->getChildren(), page->getPath(), newPages); |
| 191 |
|
| 192 |
typedef std::multimap<std::string, Page> MultiMap; |
| 193 |
|
| 194 |
_for (MultiMap::const_iterator, itor, newPages.lower_bound(childOf), newPages.upper_bound(childOf)) |
| 195 |
{ |
| 196 |
api::Cout << "Added: " << itor->second.getUrl() << ios::NewLine; |
| 197 |
|
| 198 |
pages.InsertLast(itor->second); |
| 199 |
} |
| 200 |
|
| 201 |
newPages.erase(childOf); |
| 202 |
} |
