| 4 |
|
// |
| 5 |
|
// $Id$ |
| 6 |
|
|
| 7 |
– |
#ifdef _WIN32 |
| 8 |
– |
#pragma warning(disable:4503) |
| 9 |
– |
#endif |
| 10 |
– |
|
| 7 |
|
#include "SiteMapper.hpp" |
| 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); |
| 16 |
|
|
| 17 |
< |
int main(int argc, char* argv[]) |
| 17 |
> |
struct SiteMapperCommand : public app::Application |
| 18 |
|
{ |
| 19 |
< |
program = argv[0]; |
| 21 |
< |
|
| 22 |
< |
string siteIndex, siteMap; |
| 23 |
< |
|
| 24 |
< |
for (int index(1); index < argc; index++) |
| 19 |
> |
virtual int Run(const app::ArgumentList& args) |
| 20 |
|
{ |
| 21 |
< |
string arg(argv[index]); |
| 22 |
< |
Matcher matcher; |
| 21 |
> |
program = api::GetExecutablePath().GetName(); |
| 22 |
> |
|
| 23 |
> |
string siteIndex, siteMap; |
| 24 |
|
|
| 25 |
< |
if (arg == matcher("^-index=(.*)$")) |
| 25 |
> |
for (size_t index(0); index < args.GetSize(); index++) |
| 26 |
|
{ |
| 27 |
< |
siteIndex = matcher[1]; |
| 27 |
> |
string arg(args[index]); |
| 28 |
> |
Matcher matcher; |
| 29 |
> |
|
| 30 |
> |
if (arg == matcher("^-index=(.*)$")) |
| 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 |
|
} |
| 41 |
– |
} |
| 52 |
|
|
| 53 |
< |
if (siteIndex != "" && siteMap != "") |
| 44 |
< |
{ |
| 45 |
< |
SiteMapper mapper(siteIndex, siteMap); |
| 46 |
< |
} |
| 47 |
< |
else |
| 48 |
< |
{ |
| 49 |
< |
cout << "Usage: " << program << " -index=index -map=map [-D]\n"; |
| 53 |
> |
return 0; |
| 54 |
|
} |
| 55 |
< |
|
| 52 |
< |
return 0; |
| 53 |
< |
} |
| 55 |
> |
} mapper; |
| 56 |
|
|
| 57 |
|
SiteMapper::SiteMapper(const string& siteIndex, const string& siteMap) |
| 58 |
|
{ |
| 66 |
|
ext::Handle<xml::Document> document(xml::Parse(siteMap)); |
| 67 |
|
ext::Handle<xml::Node> list(*document/"page"/"section"/"list"); |
| 68 |
|
|
| 69 |
< |
comment = *document/"comment()"; |
| 69 |
> |
comment = ext::String(*document/"comment()"); |
| 70 |
> |
|
| 71 |
> |
if (debug) cerr << "comment = " << comment << '\n'; |
| 72 |
|
|
| 73 |
|
oldMap(pages, list); |
| 74 |
|
} |
| 80 |
|
for (xml::NodeSet::Iterator node(nodes.Begin()); node != nodes.End(); |
| 81 |
|
++node) |
| 82 |
|
{ |
| 83 |
< |
string url(**node/"link"/"@address"), title(**node/"link"); |
| 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 |
|
|
| 99 |
|
for (xml::NodeSet::Iterator node(nodes.Begin()); node != nodes.End(); |
| 100 |
|
++node) |
| 101 |
|
{ |
| 102 |
< |
string address(**node/"address"); |
| 103 |
< |
string port(**node/"port"); |
| 102 |
> |
string address(ext::String(**node/"address")), |
| 103 |
> |
port(ext::String(**node/"port")); |
| 104 |
|
|
| 105 |
|
if (!port.empty()) |
| 106 |
|
{ |
| 107 |
|
address += ':' + port; |
| 108 |
|
} |
| 109 |
|
|
| 110 |
< |
string path(**node/"path"), title(**node/"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 (page == matcher(string("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDou") |
| 116 |
|
+ "glas\\sThrift's\\sBlog:\\s(.+)$")) |
| 117 |
|
{ |
| 118 |
< |
if (Matcher("^\\w+\\s\\d\\d\\d\\d\\sArchives$") == matcher[1]) |
| 118 |
> |
if (Matcher("^\\w+\\s\\d{4}\\sArchives$") == matcher[1]) |
| 119 |
|
{ |
| 120 |
|
page.setTitle(matcher[1]); |
| 121 |
|
|
| 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) |
| 218 |
|
for (multimap<string, Page>::iterator itor(newPages.lower_bound(childOf)); |
| 219 |
|
itor != newPages.upper_bound(childOf); itor++) |
| 220 |
|
{ |
| 221 |
+ |
cout << "Added: " << itor->second.getUrl() << '\n'; |
| 222 |
+ |
|
| 223 |
|
pages.push_back(itor->second); |
| 224 |
|
} |
| 225 |
|
|
