| 8 |
|
#include "Matcher.hpp" |
| 9 |
|
#include "Page.hpp" |
| 10 |
|
|
| 11 |
+ |
#include <menes-api/exename.hpp> |
| 12 |
+ |
#include <menes-app/application.hpp> |
| 13 |
+ |
|
| 14 |
|
string program; |
| 15 |
< |
bool debug = false; |
| 15 |
> |
bool debug(false); |
| 16 |
|
|
| 17 |
< |
int main(int argc, char* argv[]) |
| 17 |
> |
struct SiteMapperCommand : public app::Application |
| 18 |
|
{ |
| 19 |
< |
program = argv[0]; |
| 17 |
< |
|
| 18 |
< |
string siteIndex, siteMap; |
| 19 |
< |
|
| 20 |
< |
for (int index = 1; index < argc; index++) |
| 19 |
> |
virtual int Run(const app::ArgumentList& args) |
| 20 |
|
{ |
| 21 |
< |
string arg(argv[index]); |
| 23 |
< |
Matcher matcher; |
| 21 |
> |
program = api::GetExecutablePath().GetName(); |
| 22 |
|
|
| 23 |
< |
if (arg == matcher("^-index=(.*)$")) |
| 23 |
> |
string siteIndex, siteMap; |
| 24 |
> |
|
| 25 |
> |
for (size_t index(0); index < args.GetSize(); index++) |
| 26 |
|
{ |
| 27 |
< |
siteIndex = matcher[1]; |
| 27 |
> |
string arg(args[index]); |
| 28 |
> |
Matcher matcher; |
| 29 |
> |
|
| 30 |
> |
if (arg == matcher("^-index=(.*)$")) |
| 31 |
> |
{ |
| 32 |
> |
siteIndex = matcher[1]; |
| 33 |
> |
} |
| 34 |
> |
else if (arg == matcher("^-map=(.*)$")) |
| 35 |
> |
{ |
| 36 |
> |
siteMap = matcher[1]; |
| 37 |
> |
} |
| 38 |
> |
else if (arg == "-D") |
| 39 |
> |
{ |
| 40 |
> |
if (!debug) debug = true; |
| 41 |
> |
} |
| 42 |
|
} |
| 43 |
< |
else if (arg == matcher("^-map=(.*)$")) |
| 43 |
> |
|
| 44 |
> |
if (!siteIndex.empty() && !siteMap.empty()) |
| 45 |
|
{ |
| 46 |
< |
siteMap = matcher[1]; |
| 46 |
> |
SiteMapper mapper(siteIndex, siteMap); |
| 47 |
|
} |
| 48 |
< |
else if (arg == "-D") |
| 48 |
> |
else |
| 49 |
|
{ |
| 50 |
< |
if (!debug) debug = true; |
| 50 |
> |
cout << "Usage: " << program << " -index=index -map=map [-D]\n"; |
| 51 |
|
} |
| 37 |
– |
} |
| 38 |
– |
|
| 39 |
– |
if (siteIndex != "" && siteMap != "") |
| 40 |
– |
{ |
| 41 |
– |
XMLPlatformUtils::Initialize(); |
| 42 |
– |
XPathEvaluator::initialize(); |
| 52 |
|
|
| 53 |
< |
SiteMapper mapper(siteIndex, siteMap); |
| 45 |
< |
|
| 46 |
< |
XPathEvaluator::terminate(); |
| 47 |
< |
XMLPlatformUtils::Terminate(); |
| 48 |
< |
} |
| 49 |
< |
else |
| 50 |
< |
{ |
| 51 |
< |
cout << "Usage: " << program << " -index=index -map=map [-D]\n"; |
| 53 |
> |
return 0; |
| 54 |
|
} |
| 55 |
< |
|
| 54 |
< |
return 0; |
| 55 |
< |
} |
| 55 |
> |
} mapper; |
| 56 |
|
|
| 57 |
|
SiteMapper::SiteMapper(const string& siteIndex, const string& siteMap) |
| 58 |
|
{ |
| 59 |
|
oldMap(siteMap); |
| 60 |
< |
index(siteIndex); |
| 60 |
> |
newIndex(siteIndex); |
| 61 |
|
newMap(siteMap); |
| 62 |
|
} |
| 63 |
|
|
| 64 |
|
void SiteMapper::oldMap(const string& siteMap) |
| 65 |
|
{ |
| 66 |
< |
support.setParserLiaison(&liaison); |
| 66 |
> |
ext::Handle<xml::Document> document(xml::Parse(siteMap)); |
| 67 |
> |
ext::Handle<xml::Node> list(*document/"page"/"section"/"list"); |
| 68 |
|
|
| 69 |
< |
XalanDOMString file(siteMap.c_str()); |
| 69 |
< |
LocalFileInputSource source(file.c_str()); |
| 69 |
> |
comment = ext::String(*document/"comment()"); |
| 70 |
|
|
| 71 |
< |
XalanDocument* document = liaison.parseXMLStream(source); |
| 71 |
> |
if (debug) cerr << "comment = " << comment << '\n'; |
| 72 |
|
|
| 73 |
< |
if (document == 0) return; |
| 73 |
> |
oldMap(pages, list); |
| 74 |
> |
} |
| 75 |
|
|
| 76 |
< |
XalanNode* list = evaluator.selectSingleNode(support, document, |
| 77 |
< |
XalanDOMString("/page/section/list").c_str()); |
| 76 |
> |
void SiteMapper::oldMap(vector<Page>& pages, xml::Node* list) |
| 77 |
> |
{ |
| 78 |
> |
xml::NodeSet nodes(*list/"item"); |
| 79 |
|
|
| 80 |
< |
if (list == 0) return; |
| 80 |
> |
for (xml::NodeSet::Iterator node(nodes.Begin()); node != nodes.End(); |
| 81 |
> |
++node) |
| 82 |
> |
{ |
| 83 |
> |
string url(ext::String(**node/"link"/"@address")), |
| 84 |
> |
title(ext::String(**node/"link")); |
| 85 |
> |
Page page(url, title); |
| 86 |
> |
ext::Handle<xml::Node> list(**node/"list"); |
| 87 |
|
|
| 88 |
< |
item = evaluator.createXPath(XalanDOMString("item").c_str()); |
| 81 |
< |
address = evaluator.createXPath(XalanDOMString("link/@address").c_str()); |
| 82 |
< |
link = evaluator.createXPath(XalanDOMString("link").c_str()); |
| 83 |
< |
this->list = evaluator.createXPath(XalanDOMString("list").c_str()); |
| 88 |
> |
if (!list.IsEmpty()) oldMap(page.getChildren(), list); |
| 89 |
|
|
| 90 |
< |
oldMap(pages, list); |
| 90 |
> |
pages.push_back(page); |
| 91 |
> |
} |
| 92 |
|
} |
| 93 |
|
|
| 94 |
< |
void SiteMapper::oldMap(vector<Page>& pages, XalanNode* list) |
| 94 |
> |
void SiteMapper::newIndex(const string& siteIndex) |
| 95 |
|
{ |
| 96 |
< |
NodeRefList nodes = evaluator.selectNodeList(support, list, *item); |
| 96 |
> |
ext::Handle<xml::Document> document(xml::Parse(siteIndex)); |
| 97 |
> |
xml::NodeSet nodes(*document/"index"/"page"); |
| 98 |
|
|
| 99 |
< |
for (int index = 0; index < nodes.getLength(); index++) |
| 99 |
> |
for (xml::NodeSet::Iterator node(nodes.Begin()); node != nodes.End(); |
| 100 |
> |
++node) |
| 101 |
|
{ |
| 102 |
< |
XalanNode* node = nodes.item(index); |
| 103 |
< |
XObjectPtr address = evaluator.evaluate(support, node, |
| 96 |
< |
*(this->address)); |
| 97 |
< |
ostringstream url; |
| 102 |
> |
string address(ext::String(**node/"address")), |
| 103 |
> |
port(ext::String(**node/"port")); |
| 104 |
|
|
| 105 |
< |
if (!address.null()) |
| 105 |
> |
if (!port.empty()) |
| 106 |
|
{ |
| 107 |
< |
url << address->str(); |
| 107 |
> |
address += ':' + port; |
| 108 |
|
} |
| 109 |
|
|
| 110 |
< |
XObjectPtr link = evaluator.evaluate(support, node, *(this->link)); |
| 111 |
< |
ostringstream title; |
| 110 |
> |
string path(ext::String(**node/"path")), |
| 111 |
> |
title(ext::String(**node/"title")); |
| 112 |
> |
Page page(address, path, title); |
| 113 |
> |
Matcher matcher; |
| 114 |
|
|
| 115 |
< |
if (!link.null()) |
| 115 |
> |
if (page == matcher(string("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDou") |
| 116 |
> |
+ "glas\\sThrift's\\sBlog:\\s(.+)$")) |
| 117 |
|
{ |
| 118 |
< |
title << link->str(); |
| 118 |
> |
if (Matcher("^\\w+\\s\\d{4}\\sArchives$") == matcher[1]) |
| 119 |
> |
{ |
| 120 |
> |
page.setTitle(matcher[1]); |
| 121 |
> |
|
| 122 |
> |
if (newIndex(pages, page)) continue; |
| 123 |
> |
} |
| 124 |
> |
else continue; |
| 125 |
|
} |
| 126 |
+ |
else if (page == matcher("^Douglas\\sThrift's.+Website\\s\\|\\s(.+)$")) |
| 127 |
+ |
{ |
| 128 |
+ |
page.setTitle(matcher[1]); |
| 129 |
|
|
| 130 |
< |
Page page(url.str(), title.str()); |
| 131 |
< |
XalanNode* list = evaluator.selectSingleNode(support, node, |
| 132 |
< |
*(this->list)); |
| 130 |
> |
if (newIndex(pages, page)) continue; |
| 131 |
> |
} |
| 132 |
> |
else continue; |
| 133 |
|
|
| 134 |
< |
if (list != 0) oldMap(page.getChildren(), list); |
| 134 |
> |
multimap<string, Page> items; |
| 135 |
|
|
| 136 |
< |
pages.push_back(page); |
| 136 |
> |
newPages.insert(pair<string, multimap<string, Page> |
| 137 |
> |
>(page.getAddress(), items)).first->second.insert(pair<string, |
| 138 |
> |
Page>(page.getChildOf(), page)); |
| 139 |
|
} |
| 140 |
|
} |
| 141 |
|
|
| 142 |
< |
void SiteMapper::index(const string& siteIndex) |
| 142 |
> |
bool SiteMapper::newIndex(vector<Page>& pages, Page& page) |
| 143 |
|
{ |
| 144 |
< |
// |
| 144 |
> |
for (unsigned index(0); index < pages.size(); ++index) |
| 145 |
> |
{ |
| 146 |
> |
if (pages[index] == page.getAddress()) |
| 147 |
> |
{ |
| 148 |
> |
Matcher matcher; |
| 149 |
> |
|
| 150 |
> |
if (pages[index] == page) |
| 151 |
> |
{ |
| 152 |
> |
page.setChildren(pages[index].getChildren()); |
| 153 |
> |
|
| 154 |
> |
pages[index] = page; |
| 155 |
> |
|
| 156 |
> |
cout << "Updated: " << page.getUrl() << '\n'; |
| 157 |
> |
|
| 158 |
> |
return true; |
| 159 |
> |
} |
| 160 |
> |
else if (matcher('^' + pages[index].getPath()) == page) |
| 161 |
> |
{ |
| 162 |
> |
page.setChildOf(matcher[0]); |
| 163 |
> |
|
| 164 |
> |
if (matcher('^' + pages[index].getTitle() + "\\s\\|\\s(.+)$") |
| 165 |
> |
== page) |
| 166 |
> |
{ |
| 167 |
> |
page.setTitle(matcher[1]); |
| 168 |
> |
} |
| 169 |
> |
|
| 170 |
> |
return newIndex(pages[index].getChildren(), page); |
| 171 |
> |
} |
| 172 |
> |
} |
| 173 |
> |
} |
| 174 |
> |
|
| 175 |
> |
return false; |
| 176 |
|
} |
| 177 |
|
|
| 178 |
|
void SiteMapper::newMap(const string& siteMap) |
| 179 |
|
{ |
| 180 |
< |
// |
| 180 |
> |
ofstream fout(siteMap.c_str()); |
| 181 |
> |
|
| 182 |
> |
fout << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" |
| 183 |
> |
<< "<?xml-stylesheet type=\"text/xsl\" href=\"stylesheets/sitemap.xsl" |
| 184 |
> |
<< "\"?>\n" |
| 185 |
> |
<< "<!DOCTYPE page SYSTEM \"stylesheets/page.dtd\">\n" |
| 186 |
> |
<< "<!--" << comment << "-->\n" |
| 187 |
> |
<< "<page>\n" |
| 188 |
> |
<< "\t<title>Sitemap</title>\n" |
| 189 |
> |
<< "\t<section>\n" |
| 190 |
> |
<< "\t\t<list>\n"; |
| 191 |
> |
|
| 192 |
> |
for (unsigned index(0); index < pages.size(); ++index) |
| 193 |
> |
{ |
| 194 |
> |
if (newPages.find(pages[index].getAddress()) != newPages.end()) |
| 195 |
> |
{ |
| 196 |
> |
newMap(pages[index].getChildren(), pages[index].getPath(), |
| 197 |
> |
newPages.find(pages[index].getAddress())->second); |
| 198 |
> |
} |
| 199 |
> |
|
| 200 |
> |
fout << pages[index](3) << '\n'; |
| 201 |
> |
} |
| 202 |
> |
|
| 203 |
> |
fout << "\t\t</list>\n" |
| 204 |
> |
<< "\t</section>\n" |
| 205 |
> |
<< "</page>\n"; |
| 206 |
> |
|
| 207 |
> |
fout.close(); |
| 208 |
> |
} |
| 209 |
> |
|
| 210 |
> |
void SiteMapper::newMap(vector<Page>& pages, const string& childOf, |
| 211 |
> |
multimap<string, Page>& newPages) |
| 212 |
> |
{ |
| 213 |
> |
for (unsigned index(0); index < pages.size(); ++index) |
| 214 |
> |
{ |
| 215 |
> |
newMap(pages[index].getChildren(), pages[index].getPath(), newPages); |
| 216 |
> |
} |
| 217 |
|
|
| 218 |
< |
for (int index = 0; index < pages.size(); index++) |
| 218 |
> |
for (multimap<string, Page>::iterator itor(newPages.lower_bound(childOf)); |
| 219 |
> |
itor != newPages.upper_bound(childOf); itor++) |
| 220 |
|
{ |
| 221 |
< |
cout << pages[index] << '\n'; |
| 221 |
> |
cout << "Added: " << itor->second.getUrl() << '\n'; |
| 222 |
> |
|
| 223 |
> |
pages.push_back(itor->second); |
| 224 |
|
} |
| 225 |
+ |
|
| 226 |
+ |
newPages.erase(childOf); |
| 227 |
|
} |
