| 1 |
douglas |
672 |
// Google Tron |
| 2 |
|
|
// |
| 3 |
|
|
// Douglas Thrift |
| 4 |
|
|
// |
| 5 |
|
|
// $Id$ |
| 6 |
|
|
|
| 7 |
|
|
#include <cxx/standard.hh> |
| 8 |
|
|
|
| 9 |
douglas |
679 |
#include <api/exename.hpp> |
| 10 |
|
|
#include <api/pcre/regex.hpp> |
| 11 |
douglas |
673 |
#include <api/process.hpp> |
| 12 |
douglas |
672 |
#include <app/simple.hpp> |
| 13 |
douglas |
680 |
#include <hop/bind.hpp> |
| 14 |
douglas |
673 |
#include <xml/textwriter.hpp> |
| 15 |
douglas |
672 |
|
| 16 |
douglas |
676 |
#include "GoogleTron.hpp" |
| 17 |
douglas |
677 |
#include "Sleep.hpp" |
| 18 |
douglas |
680 |
#include "Zlib/GzipWriter.hpp" |
| 19 |
douglas |
674 |
|
| 20 |
douglas |
672 |
int Main(const app::Options &options) |
| 21 |
|
|
{ |
| 22 |
douglas |
679 |
cse::String sitemap(_B("sitemap.gz")); |
| 23 |
douglas |
676 |
|
| 24 |
douglas |
679 |
api::Pcre::RegEx sitemap_(_B("^-sitemap=(.+)$")); |
| 25 |
|
|
|
| 26 |
|
|
_foreach (const app::ArgumentList, arg, app::GetArguments()) |
| 27 |
|
|
{ |
| 28 |
|
|
api::Pcre::RegEx::Match match; |
| 29 |
|
|
|
| 30 |
|
|
if (match = sitemap_(*arg)) |
| 31 |
|
|
sitemap = match[1]; |
| 32 |
|
|
else |
| 33 |
|
|
{ |
| 34 |
|
|
api::Cout << _B("Usage: ") << api::GetExecutablePath().GetName() << _B(" [-sitemap=sitemap]") << ios::NewLine; |
| 35 |
|
|
|
| 36 |
|
|
return 1; |
| 37 |
|
|
} |
| 38 |
|
|
} |
| 39 |
|
|
|
| 40 |
|
|
GoogleTron tron(sitemap); |
| 41 |
|
|
|
| 42 |
douglas |
676 |
return 0; |
| 43 |
|
|
} |
| 44 |
|
|
|
| 45 |
douglas |
681 |
GoogleTron::GoogleTron(const cse::String &sitemap) : working(true), sitemap(sitemap), output(hop::BindAll(&GoogleTron::Output, this)) |
| 46 |
douglas |
676 |
{ |
| 47 |
douglas |
677 |
// XXX: start some threads |
| 48 |
douglas |
676 |
} |
| 49 |
|
|
|
| 50 |
|
|
GoogleTron::~GoogleTron() |
| 51 |
|
|
{ |
| 52 |
douglas |
681 |
threads.Join(); |
| 53 |
|
|
|
| 54 |
douglas |
677 |
working = false; |
| 55 |
|
|
|
| 56 |
douglas |
681 |
output.Join(); |
| 57 |
douglas |
676 |
} |
| 58 |
|
|
|
| 59 |
|
|
int GoogleTron::Output() |
| 60 |
|
|
{ |
| 61 |
douglas |
680 |
_S<Zlib::GzipWriter> gzip(sitemap); |
| 62 |
douglas |
674 |
_S<xml::TextWriter> writer(gzip); |
| 63 |
douglas |
673 |
xml::ScopeElement urlset(writer, _B("urlset")); |
| 64 |
|
|
|
| 65 |
|
|
writer.SetAttribute(_B("xmlns"), _B("http://www.google.com/schemas/sitemap/0.84")); |
| 66 |
douglas |
679 |
writer.SetAttribute(_B("xmlns:xsi"), _B("http://www.w3.org/2001/XMLSchema-instance")); |
| 67 |
|
|
writer.SetAttribute(_B("xsi:schemaLocation"), _B("http://www.google.com/schemas/sitemap/0.84 http://www.google.com/schemas/sitemap/0.84/sitemap.xsd")); |
| 68 |
douglas |
673 |
|
| 69 |
douglas |
677 |
do |
| 70 |
|
|
{ |
| 71 |
|
|
Sleep(); |
| 72 |
|
|
|
| 73 |
|
|
_synchronized (queueLock) |
| 74 |
|
|
if (queue.size()) |
| 75 |
|
|
{ |
| 76 |
|
|
xml::ScopeElement url_(writer, _B("url")); |
| 77 |
|
|
const Url &url(queue.front()); |
| 78 |
|
|
|
| 79 |
|
|
{ |
| 80 |
|
|
xml::ScopeElement loc(writer, _B("loc")); |
| 81 |
|
|
|
| 82 |
|
|
writer.OutputText(url.GetLocation()); |
| 83 |
|
|
} |
| 84 |
|
|
|
| 85 |
|
|
{ |
| 86 |
|
|
xml::ScopeElement lastmod(writer, _B("lastmod")); |
| 87 |
|
|
|
| 88 |
|
|
writer.OutputText(url.GetModified()); |
| 89 |
|
|
} |
| 90 |
|
|
|
| 91 |
|
|
{ |
| 92 |
|
|
xml::ScopeElement changefreq(writer, _B("changefreq")); |
| 93 |
|
|
|
| 94 |
|
|
writer.OutputText(url.GetFrequency()); |
| 95 |
|
|
} |
| 96 |
|
|
|
| 97 |
|
|
{ |
| 98 |
|
|
xml::ScopeElement priority(writer, _B("priority")); |
| 99 |
|
|
|
| 100 |
|
|
writer.OutputText(url.GetPriority()); |
| 101 |
|
|
} |
| 102 |
|
|
|
| 103 |
|
|
queue.pop(); |
| 104 |
|
|
} |
| 105 |
|
|
} |
| 106 |
|
|
while (working); |
| 107 |
|
|
|
| 108 |
douglas |
672 |
return 0; |
| 109 |
|
|
} |
