Building the W3C Markup Validation Service from source

Preamble

This guide provides a step-by-step walk-through to build and install the W3C Markup Validation Service (herein referred to as Validator) from its source files. It assumes minimal prior system administration experience, uses the latest versions and includes all functionality of the validator. The target Linux distribution is Ubuntu Precise, though each step will be similar for any distribution.

Installing Ubuntu

Note: this section can be skipped if you're already familiar with installing Ubuntu and/or have a Linux system running.

We will be using a Vagrant managed VirtualBox install as a base system.

Download and install Vagrant from a pre-built package

From the command line, download an Ubuntu Precise box with:

 vagrant box add precise32 http://files.vagrantup.com/precise32.box

Create an instance of Ubuntu Precise with:

 mkdir validator
 cd validator
 vagrant init

Edit the generated Vagrantfile and tell Vagrant to use the precise32 box:
```
 config.vm.box = "precise32"
```
Launch the virtual machine:
```
 vagrant up
```

Your Ubuntu virtual machine should now be up and running. Confirm by running:

vagrant ssh

Optional: if running Validator through a web server (see Optional section below), uncomment the following line in the generated Vagrantfile to enable port forwarding:

config.vm.network "forwarded_port", guest: 80, host: 8080

... then reload the virtual machine with:

vagrant reload

Installing Validator

1. Installing Validator dependencies

Install build packages:

 sudo apt-get update
 sudo apt-get install build-essential libtool automake

Install tidyp:

 cd ~
 mkdir build
 cd build
 wget https://github.com/petdance/tidyp/archive/1.04.tar.gz
 tar -xzf tidyp-1.04.tar.gz
 cd tidyp-1.04
 ./acgo
 ./configure
 make
 sudo make install
 sudo ldconfig

Install validator dependencies:

 sudo apt-get install libosp-dev libxml2-dev zlib1g-dev libwww-perl

Install required Perl Modules:

 perl -MCPAN -e "install Bundle::W3C::Validator"

When prompted, type:

yes to pre-configure as much as possible
sudo to make modules available system-wide
yes to choose a local CPAN mirror
y to install JSON::XS

2. Install Validator

Note: optionally on this step you can choose to download a more up-to-date DTD library from the development repository at http://dvcs.w3.org/hg/markup-validator/file/tip/htdocs/sgml-lib

Download the Validator and the sgml-lib DTD library:

 cd ~/build
 wget https://github.com/w3c/markup-validator/archive/validator-1_3-release.tar.gz
 tar -xzf validator-1_3-release.tar.gz

Install the Validator:

 sudo mkdir -p /usr/local/validator
 sudo mv markup-validator-validator-1_3-release/httpd/cgi-bin /usr/local/validator
 sudo mv markup-validator-validator-1_3-release/{htdocs,share,httpd} /usr/local/validator

3. Configure and test Validator

Copy the default configuration:

 sudo mkdir -p /etc/w3c
 sudo cp /usr/local/validator/htdocs/config/* /etc/w3c/

Edit /etc/w3c/validator.conf as necessary.

Note, if you plan to validate documents on a private network, enable the following option in the validator.conf file:
```
 Allow Private IPs = yes
```

Test the Validator's command line tool:

 /usr/local/validator/cgi-bin/check uri=http://www.w3.org

If some HTML output is printed to screen, Validator is installed and ready to go.

Optional

The following sections provide guides to optional but common usages of Validator.

Validator.nu HTML5 validation

Install Validator.nu's build dependencies:
```
 sudo apt-get install git default-jdk
```

Build Validator.nu:

 cd ~/build
 git clone https://github.com/validator/validator.git validator-nu
 cd validator-nu
 export JAVA_HOME=/usr/lib/jvm/java-6-openjdk
 python build/build.py all

Note, if you encounter a Java exception, run the build script again:

 python build/build.py all

Validator.nu should now be running on port 8888:

 INFO::Started SocketConnector@0.0.0.0:8888

Confirm by killing the server (CTRL+C) and re-starting it with:

 python build/build.py run

Configure Validator to use the Validator.nu engine by adding the following line in /etc/w3c/validator.conf:
```
 <External>
 HTML5 = http://localhost:8888/
 </External>
```

Running Validator.nu at system start up

Create an unprivileged user account:

 sudo groupadd validator-nu
 sudo useradd -r -c "Validator.nu daemon" -g validator-nu \
   -d /usr/local/validator-nu -s /bin/false validator-nu

Install Validator.nu system-wide:

 sudo mv ~/build/validator-nu /usr/local/
 sudo chown -R validator-nu:validator-nu /usr/local/validator-nu

Create a launch script:

 cat << EOF | sudo tee /usr/local/bin/validator-nu
 #!/bin/sh
 cd /usr/local/validator-nu
 su -s '/bin/sh' -c 'python build/build.py run >logs/validator-nu.log 2>&1' validator-nu
 EOF
 sudo chmod +x /usr/local/bin/validator-nu

Edit /etc/rc.local and add the following line before exit 0:
```
 /usr/local/bin/validator-nu
```

After rebooting, Validator.nu should be running. Confirm by running:

sudo ps aux | grep validator-nu

Set up a web server

Here, we'll be using Apache but the same principles apply to any web server.

Install Apache and its Perl bindings:

 sudo apt-get install apache2 libapache2-mod-perl2

Copy Validator's HTTP server configuration:

 sudo cp /usr/local/validator/httpd/conf/httpd.conf /etc/w3c/
 sudo ln -s /etc/w3c/httpd.conf /etc/apache2/conf.d/w3c-validator.conf
 sudo ln -s /usr/local/validator/htdocs/ /var/www/w3c-validator

Restart Apache:
```
 sudo apachectl graceful
```

Validator should now be set up and ready to go. Browse to http://localhost/w3c-validator (or http://localhost:8080/w3c-validator if running in Vagrant) to see it running.

References

The author would like to thank the Validator Team's official install guide, which forms the basis of much of this document.

tlvince/w3c-validator-guide