| 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 |
namespace |
| 21 |
{ |
| 22 |
|
| 23 |
static struct Apr |
| 24 |
{ |
| 25 |
Apr() |
| 26 |
{ |
| 27 |
api::Apr::CheckError(::apr_initialize()); |
| 28 |
} |
| 29 |
|
| 30 |
~Apr() |
| 31 |
{ |
| 32 |
::apr_terminate(); |
| 33 |
} |
| 34 |
} |
| 35 |
arp; |
| 36 |
|
| 37 |
} |
| 38 |
|
| 39 |
int Main(const app::Options &options) |
| 40 |
{ |
| 41 |
cse::String sitemap(_B("sitemap.gz")), base; |
| 42 |
_L<_R<Command> > commands; |
| 43 |
|
| 44 |
api::Pcre::RegEx sitemap_(_B("^-sitemap=(.+)$")), base_(_B("^-base=(.+)$")), fs(_B("^-fs=(.*)$")), blog(_B("^-blog=(.*)$")), wiki(_B("^-wiki=(.*)$")); |
| 45 |
|
| 46 |
_foreach (const app::ArgumentList, arg, app::GetArguments()) |
| 47 |
{ |
| 48 |
api::Pcre::RegEx::Match match; |
| 49 |
|
| 50 |
if (match = sitemap_(*arg)) |
| 51 |
sitemap = match[1]; |
| 52 |
else if (match = base_(*arg)) |
| 53 |
base = match[1]; |
| 54 |
else if (match = fs(*arg)) |
| 55 |
commands.InsertLast(new _H<FileSystemCommand>(match[1])); |
| 56 |
else if (match = blog(*arg)) |
| 57 |
commands.InsertLast(new _H<BlogCommand>(match[1])); |
| 58 |
else if (match = wiki(*arg)) |
| 59 |
commands.InsertLast(new _H<WikiCommand>(match[1])); |
| 60 |
else |
| 61 |
goto usage; |
| 62 |
} |
| 63 |
|
| 64 |
if (base.IsEmpty() || commands.IsEmpty()) |
| 65 |
{ |
| 66 |
usage: api::Cout << _B("Usage: ") << api::GetExecutablePath().GetName() << _B(" [-sitemap=.+] -base=.+ [-fs=.*] [-blog=.*] [-wiki=.*] [...]") << ios::NewLine; |
| 67 |
|
| 68 |
return 1; |
| 69 |
} |
| 70 |
|
| 71 |
GoogleTron tron(sitemap, base, commands); |
| 72 |
|
| 73 |
return 0; |
| 74 |
} |
| 75 |
|
| 76 |
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)) |
| 77 |
{ |
| 78 |
_foreach (const _L<_R<Command> >, command_, commands) |
| 79 |
if (_R<FileSystemCommand> command = dynamic_cast<FileSystemCommand *>(command_->GetValue())) |
| 80 |
threads.Add(hop::BindAll(&GoogleTron::FileSystem_, this, command)); |
| 81 |
else if (_R<BlogCommand> command = dynamic_cast<BlogCommand *>(command_->GetValue())) |
| 82 |
threads.Add(hop::BindAll(&GoogleTron::Blog_, this, command)); |
| 83 |
else if (_R<WikiCommand> command = dynamic_cast<WikiCommand *>(command_->GetValue())) |
| 84 |
threads.Add(hop::BindAll(&GoogleTron::Wiki_, this, command)); |
| 85 |
} |
| 86 |
|
| 87 |
GoogleTron::~GoogleTron() |
| 88 |
{ |
| 89 |
threads.Join(); |
| 90 |
|
| 91 |
working = false; |
| 92 |
|
| 93 |
output.Join(); |
| 94 |
} |
| 95 |
|
| 96 |
int GoogleTron::Output() |
| 97 |
{ |
| 98 |
_S<Zlib::GzipWriter> gzip(sitemap); |
| 99 |
_S<xml::TextWriter> writer(gzip); |
| 100 |
xml::ScopeElement urlset(writer, _B("urlset")); |
| 101 |
|
| 102 |
writer.SetAttribute(_B("xmlns"), _B("http://www.google.com/schemas/sitemap/0.84")); |
| 103 |
writer.SetAttribute(_B("xmlns:xsi"), _B("http://www.w3.org/2001/XMLSchema-instance")); |
| 104 |
writer.SetAttribute(_B("xsi:schemaLocation"), _B("http://www.google.com/schemas/sitemap/0.84 http://www.google.com/schemas/sitemap/0.84/sitemap.xsd")); |
| 105 |
|
| 106 |
_synchronized (queueLock) |
| 107 |
do |
| 108 |
_desynchronized (queueLock) |
| 109 |
{ |
| 110 |
Sleep(); |
| 111 |
|
| 112 |
_synchronized (queueLock) |
| 113 |
if (queue.GetSize()) |
| 114 |
{ |
| 115 |
xml::ScopeElement url_(writer, _B("url")); |
| 116 |
const Url &url(queue.Front()); |
| 117 |
|
| 118 |
{ |
| 119 |
xml::ScopeElement loc(writer, _B("loc")); |
| 120 |
|
| 121 |
writer.OutputText(url.GetLocation()); |
| 122 |
} |
| 123 |
|
| 124 |
{ |
| 125 |
xml::ScopeElement lastmod(writer, _B("lastmod")); |
| 126 |
|
| 127 |
writer.OutputText(url.GetModified()); |
| 128 |
} |
| 129 |
|
| 130 |
{ |
| 131 |
xml::ScopeElement changefreq(writer, _B("changefreq")); |
| 132 |
|
| 133 |
writer.OutputText(url.GetFrequency()); |
| 134 |
} |
| 135 |
|
| 136 |
{ |
| 137 |
xml::ScopeElement priority(writer, _B("priority")); |
| 138 |
|
| 139 |
writer.OutputText(url.GetPriority()); |
| 140 |
} |
| 141 |
|
| 142 |
api::Cout << _B("Url: ") << url.GetLocation() << ios::NewLine << _B(" ") << url.GetModified() << _B(" ") << url.GetFrequency() << _B(" ") << url.GetPriority() << ios::NewLine; |
| 143 |
|
| 144 |
queue.Pop(); |
| 145 |
} |
| 146 |
} |
| 147 |
while (working || queue.GetSize()); |
| 148 |
|
| 149 |
return 0; |
| 150 |
} |
| 151 |
|
| 152 |
int GoogleTron::FileSystem_(const _R<FileSystemCommand> &command) |
| 153 |
{ |
| 154 |
FileSystem fs(sitemap, base, queue, queueLock, command); |
| 155 |
|
| 156 |
return 0; |
| 157 |
} |
| 158 |
|
| 159 |
int GoogleTron::Blog_(const _R<BlogCommand> &command) |
| 160 |
{ |
| 161 |
Blog blog(sitemap, base, queue, queueLock, command); |
| 162 |
|
| 163 |
return 0; |
| 164 |
} |
| 165 |
|
| 166 |
int GoogleTron::Wiki_(const _R<WikiCommand> &command) |
| 167 |
{ |
| 168 |
Wiki wiki(sitemap, base, queue, queueLock, command); |
| 169 |
|
| 170 |
return 0; |
| 171 |
} |
