repos/GoogleTron/GoogleTron.cpp

// Google Tron
//
// Douglas Thrift
//
// $Id$

#include <cxx/standard.hh>

#include <api/exename.hpp>
#include <api/pcre/regex.hpp>
#include <api/process.hpp>
#include <app/simple.hpp>
#include <hop/bind.hpp>
#include <xml/textwriter.hpp>

#include "GoogleTron.hpp"
#include "Sleep.hpp"
#include "Zlib/GzipWriter.hpp"

int Main(const app::Options &options)
{
        cse::String sitemap(_B("sitemap.gz")), base;
        _L<_R<Command> > commands;

        api::Pcre::RegEx sitemap_(_B("^-sitemap=(.+)$")), base_(_B("^-base=(.+)$")), fs(_B("^-fs=(.*)$")), blog(_B("^-blog=(.*)$")), wiki(_B("^-wiki=(.*)$"));

        _foreach (const app::ArgumentList, arg, app::GetArguments())
        {
                api::Pcre::RegEx::Match match;

                if (match = sitemap_(*arg))
                        sitemap = match[1];
                else if (match = base_(*arg))
                        base = match[1];
                else if (match = fs(*arg))
                        commands.InsertLast(new _H<FileSystemCommand>(match[1]));
                else if (match = blog(*arg))
                        commands.InsertLast(new _H<BlogCommand>(match[1]));
                else if (match = wiki(*arg))
                        commands.InsertLast(new _H<WikiCommand>(match[1]));
                else
                        goto usage;
        }

        if (base.IsEmpty() || commands.IsEmpty())
        {
usage:  api::Cout << _B("Usage: ") << api::GetExecutablePath().GetName() << _B(" [-sitemap=.+] -base=.+ [-fs=.*] [-blog=.*] [-wiki=.*] [...]") << ios::NewLine;

                return 1;
        }

        GoogleTron tron(sitemap, base, commands);

        return 0;
}

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))
{
        api::Apr::CheckError(::apr_initialize());

        _foreach (const _L<_R<Command> >, command_, commands)
                if (_R<FileSystemCommand> command = dynamic_cast<FileSystemCommand *>(command_->GetValue()))
                        threads.Add(hop::BindAll(&GoogleTron::FileSystem_, this, command));
                else if (_R<BlogCommand> command = dynamic_cast<BlogCommand *>(command_->GetValue()))
                        threads.Add(hop::BindAll(&GoogleTron::Blog_, this, command));
                else if (_R<WikiCommand> command = dynamic_cast<WikiCommand *>(command_->GetValue()))
                        threads.Add(hop::BindAll(&GoogleTron::Wiki_, this, command));
}

GoogleTron::~GoogleTron()
{
        threads.Join();

        working = false;

        output.Join();

        ::apr_terminate();
}

int GoogleTron::Output()
{
        _S<Zlib::GzipWriter> gzip(sitemap);
        _S<xml::TextWriter> writer(gzip);
        xml::ScopeElement urlset(writer, _B("urlset"));

        writer.SetAttribute(_B("xmlns"), _B("http://www.google.com/schemas/sitemap/0.84"));
        writer.SetAttribute(_B("xmlns:xsi"), _B("http://www.w3.org/2001/XMLSchema-instance"));
        writer.SetAttribute(_B("xsi:schemaLocation"), _B("http://www.google.com/schemas/sitemap/0.84 http://www.google.com/schemas/sitemap/0.84/sitemap.xsd"));

        _synchronized (queueLock)
                do
                        _desynchronized (queueLock)
                        {
                                Sleep();

                                _synchronized (queueLock)
                                        if (queue.GetSize())
                                        {
                                                xml::ScopeElement url_(writer, _B("url"));
                                                const Url &url(queue.Front());

                                                {
                                                        xml::ScopeElement loc(writer, _B("loc"));

                                                        writer.OutputText(url.GetLocation());
                                                }

                                                {
                                                        xml::ScopeElement lastmod(writer, _B("lastmod"));

                                                        writer.OutputText(url.GetModified());
                                                }

                                                {
                                                        xml::ScopeElement changefreq(writer, _B("changefreq"));

                                                        writer.OutputText(url.GetFrequency());
                                                }

                                                {
                                                        xml::ScopeElement priority(writer, _B("priority"));

                                                        writer.OutputText(url.GetPriority());
                                                }

                                                api::Cout << _B("Url: ") << url.GetLocation() << ios::NewLine << _B("     ") << url.GetModified() << _B(" ") << url.GetFrequency() << _B(" ") << url.GetPriority() << ios::NewLine;

                                                queue.Pop();
                                        }
                        }
                while (working || queue.GetSize());

        return 0;
}

int GoogleTron::FileSystem_(const _R<FileSystemCommand> &command)
{
        FileSystem fs(sitemap, base, queue, queueLock, command);

        return 0;
}

int GoogleTron::Blog_(const _R<BlogCommand> &command)
{
        Blog blog(sitemap, base, queue, queueLock, command);

        return 0;
}

int GoogleTron::Wiki_(const _R<WikiCommand> &command)
{
        Wiki wiki(sitemap, base, queue, queueLock, command);

        return 0;
}
Revision:	715
Committed:	2006-03-27T02:07:13-08:00 (19 years, 2 months ago) by douglas
File size:	4096 byte(s)
Log Message:	NewLine!
#	User	Rev	Content
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	688	cse::String sitemap(_B("sitemap.gz")), base;
23	douglas	686	_L<_R<Command> > commands;
24	douglas	676
25	douglas	696	api::Pcre::RegEx sitemap_(_B("^-sitemap=(.+)$")), base_(_B("^-base=(.+)$")), fs(_B("^-fs=(.)$")), blog(_B("^-blog=(.)$")), wiki(_B("^-wiki=(.*)$"));
26	douglas	679
27			_foreach (const app::ArgumentList, arg, app::GetArguments())
28			{
29			api::Pcre::RegEx::Match match;
30
31			if (match = sitemap_(*arg))
32			sitemap = match[1];
33	douglas	688	else if (match = base_(*arg))
34			base = match[1];
35	douglas	696	else if (match = fs(*arg))
36	douglas	686	commands.InsertLast(new _H<FileSystemCommand>(match[1]));
37	douglas	696	else if (match = blog(*arg))
38	douglas	688	commands.InsertLast(new _H<BlogCommand>(match[1]));
39	douglas	696	else if (match = wiki(*arg))
40			commands.InsertLast(new _H<WikiCommand>(match[1]));
41	douglas	679	else
42	douglas	714	goto usage;
43	douglas	679	}
44
45	douglas	688	if (base.IsEmpty() \|\| commands.IsEmpty())
46	douglas	714	{
47			usage: api::Cout << _B("Usage: ") << api::GetExecutablePath().GetName() << _B(" [-sitemap=.+] -base=.+ [-fs=.] [-blog=.] [-wiki=.*] [...]") << ios::NewLine;
48	douglas	679
49	douglas	714	return 1;
50			}
51
52	douglas	688	GoogleTron tron(sitemap, base, commands);
53
54	douglas	676	return 0;
55			}
56
57	douglas	688	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))
58	douglas	676	{
59	douglas	698	api::Apr::CheckError(::apr_initialize());
60
61	douglas	686	_foreach (const _L<_R<Command> >, command_, commands)
62			if (_R<FileSystemCommand> command = dynamic_cast<FileSystemCommand *>(command_->GetValue()))
63			threads.Add(hop::BindAll(&GoogleTron::FileSystem_, this, command));
64	douglas	688	else if (_R<BlogCommand> command = dynamic_cast<BlogCommand *>(command_->GetValue()))
65			threads.Add(hop::BindAll(&GoogleTron::Blog_, this, command));
66	douglas	696	else if (_R<WikiCommand> command = dynamic_cast<WikiCommand *>(command_->GetValue()))
67			threads.Add(hop::BindAll(&GoogleTron::Wiki_, this, command));
68	douglas	676	}
69
70			GoogleTron::~GoogleTron()
71			{
72	douglas	681	threads.Join();
73
74	douglas	677	working = false;
75
76	douglas	681	output.Join();
77	douglas	698
78			::apr_terminate();
79	douglas	676	}
80
81			int GoogleTron::Output()
82			{
83	douglas	680	_S<Zlib::GzipWriter> gzip(sitemap);
84	douglas	674	_S<xml::TextWriter> writer(gzip);
85	douglas	673	xml::ScopeElement urlset(writer, _B("urlset"));
86
87			writer.SetAttribute(_B("xmlns"), _B("http://www.google.com/schemas/sitemap/0.84"));
88	douglas	679	writer.SetAttribute(_B("xmlns:xsi"), _B("http://www.w3.org/2001/XMLSchema-instance"));
89			writer.SetAttribute(_B("xsi:schemaLocation"), _B("http://www.google.com/schemas/sitemap/0.84 http://www.google.com/schemas/sitemap/0.84/sitemap.xsd"));
90	douglas	673
91	douglas	707	_synchronized (queueLock)
92			do
93			_desynchronized (queueLock)
94	douglas	677	{
95	douglas	707	Sleep();
96	douglas	677
97	douglas	707	_synchronized (queueLock)
98			if (queue.GetSize())
99			{
100			xml::ScopeElement url_(writer, _B("url"));
101			const Url &url(queue.Front());
102	douglas	677
103	douglas	707	{
104			xml::ScopeElement loc(writer, _B("loc"));
105	douglas	677
106	douglas	707	writer.OutputText(url.GetLocation());
107			}
108	douglas	677
109	douglas	707	{
110			xml::ScopeElement lastmod(writer, _B("lastmod"));
111	douglas	677
112	douglas	707	writer.OutputText(url.GetModified());
113			}
114	douglas	677
115	douglas	707	{
116			xml::ScopeElement changefreq(writer, _B("changefreq"));
117	douglas	677
118	douglas	707	writer.OutputText(url.GetFrequency());
119			}
120	douglas	677
121	douglas	707	{
122			xml::ScopeElement priority(writer, _B("priority"));
123	douglas	677
124	douglas	707	writer.OutputText(url.GetPriority());
125			}
126
127	douglas	715	api::Cout << _B("Url: ") << url.GetLocation() << ios::NewLine << _B(" ") << url.GetModified() << _B(" ") << url.GetFrequency() << _B(" ") << url.GetPriority() << ios::NewLine;
128	douglas	714
129	douglas	707	queue.Pop();
130			}
131	douglas	677	}
132	douglas	707	while (working \|\| queue.GetSize());
133	douglas	677
134	douglas	672	return 0;
135			}
136	douglas	686
137			int GoogleTron::FileSystem_(const _R<FileSystemCommand> &command)
138			{
139	douglas	688	FileSystem fs(sitemap, base, queue, queueLock, command);
140	douglas	686
141			return 0;
142			}
143	douglas	688
144			int GoogleTron::Blog_(const _R<BlogCommand> &command)
145			{
146			Blog blog(sitemap, base, queue, queueLock, command);
147
148			return 0;
149			}
150	douglas	696
151			int GoogleTron::Wiki_(const _R<WikiCommand> &command)
152			{
153			Wiki wiki(sitemap, base, queue, queueLock, command);
154
155			return 0;
156			}