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

        return 1;
}

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=(.*)$")), blog_(_B("^-blog=(.*)$")), exclude(_B("^-exclude=(.+)$"));

        _foreach (const app::ArgumentList, arg, app::GetArguments())
        {
                api::Pcre::RegEx::Match match;
                _R<FileSystemCommand> fs;
                _R<BlogCommand> blog;

                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
                        return Usage();
        }

        if (base.IsEmpty() || commands.IsEmpty())
                return Usage();

        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))
{
        _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));
}

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

        working = false;

        output.Join();
}

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"));

        do
        {
                Sleep();

                _synchronized (queueLock)
                        if (queue.size())
                        {
                                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());
                                }

                                queue.pop();
                        }
        }
        while (working);

        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;
}
Revision:	688
Committed:	2006-03-07T02:05:22-08:00 (19 years, 3 months ago) by douglas
File size:	3417 byte(s)
Log Message:	Blog, some logic, removed some comment weirdness.
#	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	688	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	douglas	672	int Main(const app::Options &options)
28			{
29	douglas	688	cse::String sitemap(_B("sitemap.gz")), base;
30	douglas	686	_L<_R<Command> > commands;
31	douglas	676
32	douglas	688	api::Pcre::RegEx sitemap_(_B("^-sitemap=(.+)$")), base_(_B("^-base=(.+)$")), fs_(_B("^-fs=(.)$")), blog_(_B("^-blog=(.)$")), exclude(_B("^-exclude=(.+)$"));
33	douglas	679
34			_foreach (const app::ArgumentList, arg, app::GetArguments())
35			{
36			api::Pcre::RegEx::Match match;
37	douglas	687	_R<FileSystemCommand> fs;
38	douglas	688	_R<BlogCommand> blog;
39	douglas	679
40			if (match = sitemap_(*arg))
41			sitemap = match[1];
42	douglas	688	else if (match = base_(*arg))
43			base = match[1];
44	douglas	687	else if (match = fs_(*arg))
45	douglas	686	commands.InsertLast(new _H<FileSystemCommand>(match[1]));
46	douglas	688	else if (match = blog_(*arg))
47			commands.InsertLast(new _H<BlogCommand>(match[1]));
48	douglas	679	else
49	douglas	688	return Usage();
50	douglas	679	}
51
52	douglas	688	if (base.IsEmpty() \|\| commands.IsEmpty())
53			return Usage();
54	douglas	679
55	douglas	688	GoogleTron tron(sitemap, base, commands);
56
57	douglas	676	return 0;
58			}
59
60	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))
61	douglas	676	{
62	douglas	686	_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	douglas	688	else if (_R<BlogCommand> command = dynamic_cast<BlogCommand *>(command_->GetValue()))
66			threads.Add(hop::BindAll(&GoogleTron::Blog_, this, command));
67	douglas	676	}
68
69			GoogleTron::~GoogleTron()
70			{
71	douglas	681	threads.Join();
72
73	douglas	677	working = false;
74
75	douglas	681	output.Join();
76	douglas	676	}
77
78			int GoogleTron::Output()
79			{
80	douglas	680	_S<Zlib::GzipWriter> gzip(sitemap);
81	douglas	674	_S<xml::TextWriter> writer(gzip);
82	douglas	673	xml::ScopeElement urlset(writer, _B("urlset"));
83
84			writer.SetAttribute(_B("xmlns"), _B("http://www.google.com/schemas/sitemap/0.84"));
85	douglas	679	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	douglas	673
88	douglas	677	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	douglas	672	return 0;
128			}
129	douglas	686
130			int GoogleTron::FileSystem_(const _R<FileSystemCommand> &command)
131			{
132	douglas	688	FileSystem fs(sitemap, base, queue, queueLock, command);
133	douglas	686
134			return 0;
135			}
136	douglas	688
137			int GoogleTron::Blog_(const _R<BlogCommand> &command)
138			{
139			Blog blog(sitemap, base, queue, queueLock, command);
140
141			return 0;
142			}