| 1 |
// Google Tron |
| 2 |
// |
| 3 |
// Douglas Thrift |
| 4 |
// |
| 5 |
// $Id$ |
| 6 |
|
| 7 |
#include <cxx/standard.hh> |
| 8 |
|
| 9 |
#include <api/exename.hpp> |
| 10 |
#include <api/pcre/regex.hpp> |
| 11 |
#include <api/process.hpp> |
| 12 |
#include <app/simple.hpp> |
| 13 |
#include <hop/bind.hpp> |
| 14 |
#include <xml/textwriter.hpp> |
| 15 |
|
| 16 |
#include "GoogleTron.hpp" |
| 17 |
#include "Sleep.hpp" |
| 18 |
#include "Zlib/GzipWriter.hpp" |
| 19 |
|
| 20 |
int Usage() |
| 21 |
{ |
| 22 |
api::Cout << _B("Usage: ") << api::GetExecutablePath().GetName() << _B(" [-sitemap=.+] -base=.+ [-fs=.*] [-blog=.*]") << ios::NewLine; |
| 23 |
|
| 24 |
return 1; |
| 25 |
} |
| 26 |
|
| 27 |
int Main(const app::Options &options) |
| 28 |
{ |
| 29 |
cse::String sitemap(_B("sitemap.gz")), base; |
| 30 |
_L<_R<Command> > commands; |
| 31 |
|
| 32 |
api::Pcre::RegEx sitemap_(_B("^-sitemap=(.+)$")), base_(_B("^-base=(.+)$")), fs_(_B("^-fs=(.*)$")), blog_(_B("^-blog=(.*)$")), exclude(_B("^-exclude=(.+)$")); |
| 33 |
|
| 34 |
_foreach (const app::ArgumentList, arg, app::GetArguments()) |
| 35 |
{ |
| 36 |
api::Pcre::RegEx::Match match; |
| 37 |
_R<FileSystemCommand> fs; |
| 38 |
_R<BlogCommand> blog; |
| 39 |
|
| 40 |
if (match = sitemap_(*arg)) |
| 41 |
sitemap = match[1]; |
| 42 |
else if (match = base_(*arg)) |
| 43 |
base = match[1]; |
| 44 |
else if (match = fs_(*arg)) |
| 45 |
commands.InsertLast(new _H<FileSystemCommand>(match[1])); |
| 46 |
else if (match = blog_(*arg)) |
| 47 |
commands.InsertLast(new _H<BlogCommand>(match[1])); |
| 48 |
else |
| 49 |
return Usage(); |
| 50 |
} |
| 51 |
|
| 52 |
if (base.IsEmpty() || commands.IsEmpty()) |
| 53 |
return Usage(); |
| 54 |
|
| 55 |
GoogleTron tron(sitemap, base, commands); |
| 56 |
|
| 57 |
return 0; |
| 58 |
} |
| 59 |
|
| 60 |
GoogleTron::GoogleTron(const cse::String &sitemap, const cse::String &base, const _L<_R<Command> > &commands) : working(true), sitemap(sitemap), base(base), output(hop::BindAll(&GoogleTron::Output, this)) |
| 61 |
{ |
| 62 |
_foreach (const _L<_R<Command> >, command_, commands) |
| 63 |
if (_R<FileSystemCommand> command = dynamic_cast<FileSystemCommand *>(command_->GetValue())) |
| 64 |
threads.Add(hop::BindAll(&GoogleTron::FileSystem_, this, command)); |
| 65 |
else if (_R<BlogCommand> command = dynamic_cast<BlogCommand *>(command_->GetValue())) |
| 66 |
threads.Add(hop::BindAll(&GoogleTron::Blog_, this, command)); |
| 67 |
} |
| 68 |
|
| 69 |
GoogleTron::~GoogleTron() |
| 70 |
{ |
| 71 |
threads.Join(); |
| 72 |
|
| 73 |
working = false; |
| 74 |
|
| 75 |
output.Join(); |
| 76 |
} |
| 77 |
|
| 78 |
int GoogleTron::Output() |
| 79 |
{ |
| 80 |
_S<Zlib::GzipWriter> gzip(sitemap); |
| 81 |
_S<xml::TextWriter> writer(gzip); |
| 82 |
xml::ScopeElement urlset(writer, _B("urlset")); |
| 83 |
|
| 84 |
writer.SetAttribute(_B("xmlns"), _B("http://www.google.com/schemas/sitemap/0.84")); |
| 85 |
writer.SetAttribute(_B("xmlns:xsi"), _B("http://www.w3.org/2001/XMLSchema-instance")); |
| 86 |
writer.SetAttribute(_B("xsi:schemaLocation"), _B("http://www.google.com/schemas/sitemap/0.84 http://www.google.com/schemas/sitemap/0.84/sitemap.xsd")); |
| 87 |
|
| 88 |
do |
| 89 |
{ |
| 90 |
Sleep(); |
| 91 |
|
| 92 |
_synchronized (queueLock) |
| 93 |
if (queue.size()) |
| 94 |
{ |
| 95 |
xml::ScopeElement url_(writer, _B("url")); |
| 96 |
const Url &url(queue.front()); |
| 97 |
|
| 98 |
{ |
| 99 |
xml::ScopeElement loc(writer, _B("loc")); |
| 100 |
|
| 101 |
writer.OutputText(url.GetLocation()); |
| 102 |
} |
| 103 |
|
| 104 |
{ |
| 105 |
xml::ScopeElement lastmod(writer, _B("lastmod")); |
| 106 |
|
| 107 |
writer.OutputText(url.GetModified()); |
| 108 |
} |
| 109 |
|
| 110 |
{ |
| 111 |
xml::ScopeElement changefreq(writer, _B("changefreq")); |
| 112 |
|
| 113 |
writer.OutputText(url.GetFrequency()); |
| 114 |
} |
| 115 |
|
| 116 |
{ |
| 117 |
xml::ScopeElement priority(writer, _B("priority")); |
| 118 |
|
| 119 |
writer.OutputText(url.GetPriority()); |
| 120 |
} |
| 121 |
|
| 122 |
queue.pop(); |
| 123 |
} |
| 124 |
} |
| 125 |
while (working); |
| 126 |
|
| 127 |
return 0; |
| 128 |
} |
| 129 |
|
| 130 |
int GoogleTron::FileSystem_(const _R<FileSystemCommand> &command) |
| 131 |
{ |
| 132 |
FileSystem fs(sitemap, base, queue, queueLock, command); |
| 133 |
|
| 134 |
return 0; |
| 135 |
} |
| 136 |
|
| 137 |
int GoogleTron::Blog_(const _R<BlogCommand> &command) |
| 138 |
{ |
| 139 |
Blog blog(sitemap, base, queue, queueLock, command); |
| 140 |
|
| 141 |
return 0; |
| 142 |
} |
