| 4 |
|
// |
| 5 |
|
// $Id$ |
| 6 |
|
|
| 7 |
< |
#include "SiteMapper.hpp" |
| 8 |
< |
#include "Matcher.hpp" |
| 9 |
< |
#include "Page.hpp" |
| 7 |
> |
#include <menes/cxx/standard.hh> |
| 8 |
> |
|
| 9 |
> |
#include <menes/api/pcre/regex.hpp> |
| 10 |
> |
#include <menes/api/exename.hpp> |
| 11 |
> |
#include <menes/app/simple.hpp> |
| 12 |
|
|
| 13 |
< |
string program; |
| 12 |
< |
bool debug = false; |
| 13 |
> |
#include "SiteMapper.hpp" |
| 14 |
|
|
| 15 |
< |
int main(int argc, char* argv[]) |
| 15 |
> |
int Main(const app::Options& options) |
| 16 |
|
{ |
| 17 |
< |
program = argv[0]; |
| 17 |
> |
SiteMapper::program = api::GetExecutablePath().GetName(); |
| 18 |
|
|
| 19 |
< |
string siteIndex, siteMap; |
| 19 |
> |
cse::String siteIndex, siteMap; |
| 20 |
> |
api::Pcre::RegEx index(_B("^-index=(.+)$")), map(_B("^-map=(.+)$")); |
| 21 |
|
|
| 22 |
< |
for (int index = 1; index < argc; index++) |
| 22 |
> |
_foreach (const app::ArgumentList, arg, app::GetArguments()) |
| 23 |
|
{ |
| 24 |
< |
string arg(argv[index]); |
| 23 |
< |
Matcher matcher; |
| 24 |
> |
api::Pcre::RegEx::Match match; |
| 25 |
|
|
| 26 |
< |
if (arg == matcher("^-index=(.*)$")) |
| 26 |
> |
if (match = index(*arg)) |
| 27 |
|
{ |
| 28 |
< |
siteIndex = matcher[1]; |
| 28 |
> |
siteIndex = match[1]; |
| 29 |
|
} |
| 30 |
< |
else if (arg == matcher("^-map=(.*)$")) |
| 30 |
> |
else if (match = map(*arg)) |
| 31 |
|
{ |
| 32 |
< |
siteMap = matcher[1]; |
| 32 |
> |
siteMap = match[1]; |
| 33 |
|
} |
| 34 |
< |
else if (arg == "-D") |
| 34 |
> |
else if (*arg == _B("-D")) |
| 35 |
|
{ |
| 36 |
< |
if (!debug) debug = true; |
| 36 |
> |
if (!SiteMapper::debug) SiteMapper::debug = true; |
| 37 |
|
} |
| 38 |
|
} |
| 39 |
|
|
| 40 |
< |
if (siteIndex != "" && siteMap != "") |
| 40 |
< |
{ |
| 41 |
< |
XMLPlatformUtils::Initialize(); |
| 42 |
< |
XPathEvaluator::initialize(); |
| 43 |
< |
|
| 40 |
> |
if (!siteIndex.IsEmpty() && !siteMap.IsEmpty()) |
| 41 |
|
SiteMapper mapper(siteIndex, siteMap); |
| 45 |
– |
|
| 46 |
– |
XPathEvaluator::terminate(); |
| 47 |
– |
XMLPlatformUtils::Terminate(); |
| 48 |
– |
} |
| 42 |
|
else |
| 43 |
< |
{ |
| 51 |
< |
cout << "Usage: " << program << " -index=index -map=map [-D]\n"; |
| 52 |
< |
} |
| 43 |
> |
api::Cout << _B("Usage: ") << SiteMapper::program << _B(" -index=index -map=map [-D]") << ios::NewLine; |
| 44 |
|
|
| 45 |
|
return 0; |
| 46 |
|
} |
| 47 |
|
|
| 48 |
< |
SiteMapper::SiteMapper(const string& siteIndex, const string& siteMap) |
| 48 |
> |
SiteMapper::SiteMapper(const cse::String& siteIndex, const cse::String& siteMap) |
| 49 |
|
{ |
| 50 |
|
oldMap(siteMap); |
| 51 |
< |
index(siteIndex); |
| 51 |
> |
newIndex(siteIndex); |
| 52 |
|
newMap(siteMap); |
| 53 |
|
} |
| 54 |
|
|
| 55 |
< |
void SiteMapper::oldMap(const string& siteMap) |
| 56 |
< |
{ |
| 66 |
< |
support.setParserLiaison(&liaison); |
| 55 |
> |
cse::String SiteMapper::program; |
| 56 |
> |
bool SiteMapper::debug(false); |
| 57 |
|
|
| 58 |
< |
XalanDOMString file(siteMap.c_str()); |
| 59 |
< |
LocalFileInputSource source(file.c_str()); |
| 58 |
> |
void SiteMapper::oldMap(const cse::String& siteMap) |
| 59 |
> |
{ |
| 60 |
> |
ext::Handle<xml::Document> document(xml::Parse(siteMap)); |
| 61 |
> |
ext::Handle<xml::Node> list(*document/_B("page")/_B("section")/_B("list")); |
| 62 |
|
|
| 63 |
< |
XalanDocument* document = liaison.parseXMLStream(source); |
| 63 |
> |
comment = *document/_B("comment()"); |
| 64 |
|
|
| 65 |
< |
if (document == 0) return; |
| 65 |
> |
if (debug) api::Cerr << _B("comment = ") << comment << ios::NewLine; |
| 66 |
|
|
| 67 |
< |
XalanNode* list = evaluator.selectSingleNode(support, document, |
| 68 |
< |
XalanDOMString("/page/section/list").c_str()); |
| 67 |
> |
oldMap(pages, list); |
| 68 |
> |
} |
| 69 |
|
|
| 70 |
< |
if (list == 0) return; |
| 70 |
> |
void SiteMapper::oldMap(ext::Vector<Page>& pages, xml::Node* list) |
| 71 |
> |
{ |
| 72 |
> |
xml::NodeSet nodes(*list/_B("item")); |
| 73 |
|
|
| 74 |
< |
item = evaluator.createXPath(XalanDOMString("item").c_str()); |
| 75 |
< |
address = evaluator.createXPath(XalanDOMString("link/@address").c_str()); |
| 76 |
< |
link = evaluator.createXPath(XalanDOMString("link").c_str()); |
| 77 |
< |
this->list = evaluator.createXPath(XalanDOMString("list").c_str()); |
| 74 |
> |
_foreach (xml::NodeSet, node, nodes) |
| 75 |
> |
{ |
| 76 |
> |
cse::String url(**node/_B("link")/_B("@address")), title(**node/_B("link")); |
| 77 |
> |
Page page(url, title); |
| 78 |
> |
ext::Handle<xml::Node> list(**node/_B("list")); |
| 79 |
|
|
| 80 |
< |
oldMap(pages, list); |
| 80 |
> |
if (!list.IsEmpty()) oldMap(page.GetChildren(), list); |
| 81 |
|
|
| 82 |
< |
evaluator.destroyXPath(item); |
| 83 |
< |
evaluator.destroyXPath(address); |
| 89 |
< |
evaluator.destroyXPath(link); |
| 90 |
< |
evaluator.destroyXPath(this->list); |
| 82 |
> |
pages.InsertLast(page); |
| 83 |
> |
} |
| 84 |
|
} |
| 85 |
|
|
| 86 |
< |
void SiteMapper::oldMap(vector<Page>& pages, XalanNode* list) |
| 86 |
> |
void SiteMapper::newIndex(const cse::String& siteIndex) |
| 87 |
|
{ |
| 88 |
< |
NodeRefList nodes = evaluator.selectNodeList(support, list, *item); |
| 88 |
> |
ext::Handle<xml::Document> document(xml::Parse(siteIndex)); |
| 89 |
> |
xml::NodeSet nodes(*document/_B("index")/_B("page")); |
| 90 |
|
|
| 91 |
< |
for (int index = 0; index < nodes.getLength(); index++) |
| 91 |
> |
_foreach (xml::NodeSet, node, nodes) |
| 92 |
|
{ |
| 93 |
< |
XalanNode* node = nodes.item(index); |
| 94 |
< |
ostringstream url, title; |
| 95 |
< |
|
| 96 |
< |
url << evaluator.evaluate(support, node, *address)->str(); |
| 97 |
< |
title << evaluator.evaluate(support, node, *link)->str(); |
| 93 |
> |
_S<ios::String> address(**node/_B("address")); |
| 94 |
> |
cse::String port(**node/_B("port")); |
| 95 |
> |
|
| 96 |
> |
if (!port.IsEmpty()) address << _B(":") << port; |
| 97 |
> |
|
| 98 |
> |
cse::String path(**node/_B("path")), title(**node/_B("title")); |
| 99 |
> |
Page page(address, path, title); |
| 100 |
> |
static api::Pcre::RegEx blog(_B("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDouglas\\sThrift's\\sBlog:\\s(.+)$")), page_(_B("^Douglas\\sThrift's.+Website\\s\\|\\s(.+)$")); |
| 101 |
> |
|
| 102 |
> |
if (api::Pcre::RegEx::Match match = blog(page.GetTitle())) |
| 103 |
> |
{ |
| 104 |
> |
static api::Pcre::RegEx archives(_B("^\\w+\\s\\d{4}\\sArchives$")); |
| 105 |
> |
|
| 106 |
> |
if (archives(match[1])) |
| 107 |
> |
{ |
| 108 |
> |
page.SetTitle(match[1]); |
| 109 |
> |
|
| 110 |
> |
if (newIndex(pages, page)) |
| 111 |
> |
continue; |
| 112 |
> |
} |
| 113 |
> |
else |
| 114 |
> |
continue; |
| 115 |
> |
} |
| 116 |
> |
else if (api::Pcre::RegEx::Match match = page_(page.GetTitle())) |
| 117 |
> |
{ |
| 118 |
> |
page.SetTitle(match[1]); |
| 119 |
|
|
| 120 |
< |
Page page(url.str(), title.str()); |
| 121 |
< |
XalanNode* list = evaluator.selectSingleNode(support, node, |
| 122 |
< |
*(this->list)); |
| 120 |
> |
if (newIndex(pages, page)) |
| 121 |
> |
continue; |
| 122 |
> |
} |
| 123 |
> |
else |
| 124 |
> |
continue; |
| 125 |
|
|
| 126 |
< |
if (list != 0) oldMap(page.getChildren(), list); |
| 126 |
> |
std::multimap<std::string, Page> items; |
| 127 |
|
|
| 128 |
< |
pages.push_back(page); |
| 128 |
> |
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)); |
| 129 |
|
} |
| 130 |
|
} |
| 131 |
|
|
| 132 |
< |
void SiteMapper::index(const string& siteIndex) |
| 132 |
> |
bool SiteMapper::newIndex(ext::Vector<Page>& pages, Page& page) |
| 133 |
|
{ |
| 134 |
< |
XalanDOMString file(siteIndex.c_str()); |
| 135 |
< |
LocalFileInputSource source(file.c_str()); |
| 136 |
< |
|
| 137 |
< |
XalanDocument* document = liaison.parseXMLStream(source); |
| 134 |
> |
_foreach (ext::Vector<Page>, page_, pages) |
| 135 |
> |
{ |
| 136 |
> |
if (*page_ == page.GetAddress()) |
| 137 |
> |
{ |
| 138 |
> |
if (*page_ == page) |
| 139 |
> |
{ |
| 140 |
> |
page.SetChildren(page_->GetChildren()); |
| 141 |
|
|
| 142 |
< |
if (document == 0) return; |
| 142 |
> |
*page_ = page; |
| 143 |
|
|
| 144 |
< |
address = evaluator.createXPath(XalanDOMString("address").c_str()); |
| 125 |
< |
port = evaluator.createXPath(XalanDOMString("port").c_str()); |
| 126 |
< |
path = evaluator.createXPath(XalanDOMString("path").c_str()); |
| 127 |
< |
title = evaluator.createXPath(XalanDOMString("title").c_str()); |
| 128 |
< |
|
| 129 |
< |
NodeRefList nodes = evaluator.selectNodeList(support, document, |
| 130 |
< |
XalanDOMString("/index/page").c_str()); |
| 144 |
> |
api::Cout << _B("Updated: ") << page.GetUrl() << ios::NewLine; |
| 145 |
|
|
| 146 |
< |
for (int index = 0; index < nodes.getLength(); index++) |
| 147 |
< |
{ |
| 148 |
< |
XalanNode* node = nodes.item(index); |
| 149 |
< |
ostringstream address; |
| 146 |
> |
return true; |
| 147 |
> |
} |
| 148 |
> |
else if (page.GetPath().StartsWithAll(page_->GetPath())) |
| 149 |
> |
{ |
| 150 |
> |
page.SetChildOf(page_->GetPath()); |
| 151 |
|
|
| 152 |
< |
address << evaluator.evaluate(support, node, *(this->address))->str(); |
| 152 |
> |
api::Pcre::RegEx title(_S<ios::String>() << _B("^") << page_->GetTitle() << "\\s\\|\\s(.+)$"); |
| 153 |
|
|
| 154 |
< |
double port = evaluator.evaluate(support, node, *(this->port))->num(); |
| 154 |
> |
if (api::Pcre::RegEx::Match match = title(page.GetTitle())) |
| 155 |
> |
page.SetTitle(match[1]); |
| 156 |
|
|
| 157 |
< |
if (port >= 0 && port <= 65535) |
| 158 |
< |
{ |
| 143 |
< |
address << ':' << int(port); |
| 157 |
> |
return newIndex(page_->GetChildren(), page); |
| 158 |
> |
} |
| 159 |
|
} |
| 145 |
– |
|
| 146 |
– |
cerr << address.str() << '\n'; |
| 160 |
|
} |
| 161 |
|
|
| 162 |
< |
evaluator.destroyXPath(address); |
| 163 |
< |
evaluator.destroyXPath(port); |
| 164 |
< |
evaluator.destroyXPath(path); |
| 165 |
< |
evaluator.destroyXPath(title); |
| 162 |
> |
return false; |
| 163 |
> |
} |
| 164 |
> |
|
| 165 |
> |
void SiteMapper::newMap(const cse::String& siteMap) |
| 166 |
> |
{ |
| 167 |
> |
_S<api::FileWriter> file(siteMap); |
| 168 |
> |
_S<ios::FormatWriter> fout(file); |
| 169 |
> |
_S<xml::TextWriter> xml(file); |
| 170 |
> |
|
| 171 |
> |
fout << ios::NewLine << _B("<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl\"?>") << ios::NewLine << _B("<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">"); |
| 172 |
> |
|
| 173 |
> |
xml.OutputComment(comment); |
| 174 |
> |
|
| 175 |
> |
xml::ScopeElement page(xml, _B("page")); |
| 176 |
> |
|
| 177 |
> |
xml.OpenElement(_B("title")); |
| 178 |
> |
xml.OutputText(_B("Sitemap")); |
| 179 |
> |
xml.CloseElement(); |
| 180 |
> |
|
| 181 |
> |
xml::ScopeElement section(xml, _B("section")), list(xml, _B("list")); |
| 182 |
> |
|
| 183 |
> |
_foreach (ext::Vector<Page>, page, pages) |
| 184 |
> |
{ |
| 185 |
> |
if (newPages.find(page->GetAddress()) != newPages.end()) newMap(page->GetChildren(), page->GetPath(), newPages.find(page->GetAddress())->second); |
| 186 |
> |
|
| 187 |
> |
xml << *page; |
| 188 |
> |
} |
| 189 |
|
} |
| 190 |
|
|
| 191 |
< |
void SiteMapper::newMap(const string& siteMap) |
| 191 |
> |
void SiteMapper::newMap(ext::Vector<Page>& pages, const cse::String& childOf, std::multimap<std::string, Page>& newPages) |
| 192 |
|
{ |
| 193 |
< |
// |
| 193 |
> |
_foreach (ext::Vector<Page>, page, pages) newMap(page->GetChildren(), page->GetPath(), newPages); |
| 194 |
> |
|
| 195 |
> |
typedef std::multimap<std::string, Page> MultiMap; |
| 196 |
> |
|
| 197 |
> |
_forall (MultiMap::const_iterator, itor, newPages.lower_bound(childOf), newPages.upper_bound(childOf)) |
| 198 |
> |
{ |
| 199 |
> |
api::Cout << _B("Added: ") << itor->second.GetUrl() << ios::NewLine; |
| 200 |
> |
|
| 201 |
> |
pages.InsertLast(itor->second); |
| 202 |
> |
} |
| 203 |
|
|
| 204 |
< |
// for (int index = 0; index < pages.size(); index++) |
| 160 |
< |
// { |
| 161 |
< |
// cout << pages[index] << '\n'; |
| 162 |
< |
// } |
| 204 |
> |
newPages.erase(childOf); |
| 205 |
|
} |
