Semantic Indexer

Introduction

This repository contains the semantic indexer and a tool for sparql endpoints discovering registered in datahub.io.

Datahub Usage

If you want to update the list of sparql endpoints. You can run datahub/find_sparql_endpoints.py, be sure to add your api key to datahub/datahub-credentials.json before doing so.

Indexer Deployment Guide

Clone the project to your server and follow the following instructions. This guide describes how to deploy the project on debian. In other unix or linux distribution package names may differ.

If you are familiar with rabbitmq and psql configuration feel free to skip the relevant sections.

Install Requirements.

Install postgresql, the development package is required by the python connector.
```
 apt-get install postgresql-9.4
 apt-get install postgresql-server-dev-9.4
```
Install rabbitmq.
```
 apt-get install rabbitmq-server
```

Install python.

 apt-get install python
 apt-get install python-pip

Install python requirements. You will probably want to install these in a virtualenv.
```
 pip install -r indexer/requirements.txt
```

Postgres configuration.

It is important to note that the application assumes that peer authentication is used and will attempt to connect without using a password.

The default debian postgres installation is configured to use peer authentication for all users. If you are not familiar with postres's authentication configuration see https://www.postgresql.org/docs/9.4/static/auth-methods.html. All you need to know is that under the default configuration in order to connect to a database the system username must match the database user username.

For the simplest configuration do the following.

Create a new db user with the current username.

Run
```
 createuser
```
Create a database. user name should match the system usersname Run
```
 sudo -u postgres createdb {databasename} -O {username}
```
Test that the user and database are correctly created.

Run
```
 psql -d {databasename}
```
You should now see a prompt connected to the database. If you are having problems go over the postresql documentation

Rabbitmq configuration

On a fresh installation the user "guest" is created with password "guest" and read and write permissions to the "/" vhost.

If you are not familiar with the rabbitmq server see its documentation

If you wish to create another user and vhost use the following commands.

Run

rabbitmqctl add_user {username} {password}

rabbitmqctl add_vhost {vhost}

rabbitmqctl set_permissions [-p vhost] {username} ".*" ".*" ".*"

The last command sets full configure, write and read. permissions to the user for the specified host. By default new users have no permissions.

Indexer configuration

Important note:

The application assumes that postgresql is setup to use peer authentication and will NOT attempt to use a password.

Ensure settings.json is correct.
1. database name and database user name. Set the database name and the database owners name the user name should match the system users username.
  
  i.e
```
 "database": {
 	"name": "backlinks",
 	"user": "pavlov"
 }
```
2. Rabbitmq credentials Set the rabbitmq server credentials to use.
  
  If you have not created a new user and vhost you can use the following.
```
 "rabbitmq": {
 	"user": "guest",
 	"pass": "guest",
 	"vhost": "/"
 }
```
3. Data-paths This section contains only one item the path to the file containing the sparql endpoints to query.
4. Worker counts. db-workers the count of processes writing to the database. web-workers the count of processes querying sparql endpoints and processing results. Keep in mind processing the results is cpu intensive
5. Queue names db-queue, web-worker-endpoints-queue, log-event-queue. These names set the names of the rabbitmq queues used for communication between the different workers.
6. Logging These are python logging settings if you are not familiar with the python logging system see the [relevant python documentation] (https://docs.python.org/2/library/logging.html)
Ensure the database tables are created.
1. run create-tables.py to create the database tables.

Running indexer/index.py

index.py will output a unique session id, that has been passed as an argument to all child processes. In order to kill all the processes use the following command.

ps ax | grep {session_id} | sed 's/^ //' | cut -d ' ' -f 1 | xargs kill

i.e

ps ax | grep 10169789888 | sed 's/^ //' | cut -d ' ' -f 1 | xargs kill

Pavlov123/semantic_indexer