A data processing pipeline that schedules and runs content harvesters, normalizes their data, and outputs that normalized data to a variety of output streams. This is part of the SHARE project, and will be used to create a free and open dataset of research (meta)data. Data collected can be explored at https://osf.io/share/, and viewed at https:/…
Python Other
Pull request Compare This branch is 296 commits behind CenterForOpenScience:develop.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
api
img/favicons
institutions
scrapi
tests
.codeclimate.yml
.coveragerc
.gitignore
.travis.yml
CHANGELOG
LICENSE
README.md
dev-requirements.txt
manage.py
normalized.json
requirements.txt
setup.cfg
setup.py
tasks.py
tox.ini

README.md

scrapi

master build status: Build Status

develop build status: Build Status

Coverage Status Code Climate

Getting started

  • To run absolutely everyting, you will need to:
    • Install requirements
    • Install Elasticsearch
    • Install Cassandra, or Postgres, or both (optional)
    • Install RabbitMQ (optional)
  • You do not have to install RabbitMQ if you're only running the harvesters locally.
  • Both Cassandra and Postgres aren't really necessary, you can choose which one you'd like, or use both. If you install neither, you can use local storage instead. In your settings, you'll specify a CANONICAL_PROCESSOR, just make sure that one is installed.

Requirements

  • Create and enter virtual environment for scrapi, and go to the top level project directory. From there, run
$ pip install -r requirements.txt

Or, if you'd like some nicer testing and debugging utilities in addition to the core requirements, run

$ pip install -r dev-requirements.txt

This will also install the core requirements like normal.

Installing Elasticsearch

Note: Elasticsearch requires JDK 7.

Mac OSX

$ brew install homebrew/versions/elasticsearch17

Ubuntu

  1. Download and install the Public Signing Key.

    $ wget -qO - https://packages.elasticsearch.org/GPG-KEY-elasticsearch | sudo apt-key add -
  2. Add the ElasticSearch repository to yout /etc/apt/sources.list.

    $ sudo add-apt-repository "deb http://packages.elasticsearch.org/elasticsearch/1.4/debian stable main"
  3. Install the package

    $ sudo apt-get update
    $ sudo apt-get install elasticsearch

Running

$ elasticsearch

Installing Postgres

Postgres is required only if "postgres" is specified in your settings, or if RECORD_HTTP_TRANSACTIONS is set to True.

Mac OSX

$ brew install postgresql
$ ln -sfv /usr/local/homebrew/opt/postgresql/*.plist ~/Library/LaunchAgents
$ launchctl load ~/Library/LaunchAgents/homebrew.mxcl.postgresql.plist

Ubuntu

$ sudo apt-get update
$ sudo apt-get install postgresql
$ service postgresql start

Running

Inside your scrapi checkout:

$ createdb scrapi
$ invoke apidb

Installing Cassandra

Cassandra is required only if "cassandra" is specified in your settings, or if RECORD_HTTP_TRANSACTIONS is set to True.

Note: Cassandra requires JDK 7.

Mac OSX

$ brew install cassandra

Ubuntu

  1. Check which version of Java is installed by running the following command:

    $ java -version

    Use the latest version of Oracle Java 7 on all nodes.

  2. Add the DataStax Community repository to the /etc/apt/sources.list.d/cassandra.sources.list

    $ echo "deb http://debian.datastax.com/community stable main" | sudo tee -a /etc/apt/sources.list.d/cassandra.sources.list
  3. Add the DataStax repository key to your aptitude trusted keys.

    $ curl -L http://debian.datastax.com/debian/repo_key | sudo apt-key add -
  4. Install the package.

    $ sudo apt-get update
    $ sudo apt-get install cassandra

Running

$ cassandra

Or, if you'd like your cassandra session to be bound to your current session, run:

$ cassandra -f

and you should be good to go.

RabbitMQ (optional)

Note, if you're developing locally, you do not have to run RabbitMQ!

Mac OSX

$ brew install rabbitmq

Ubuntu

$ sudo apt-get install rabbitmq-server

Settings

You will need to have a local copy of the settings. Copy local-dist.py into your own version of local.py:

cp scrapi/settings/local-dist.py scrapi/settings/local.py

Copy over the api settings:

cp api/api/settings/local-dist.py api/api/settings/local.py

If you installed Cassandra, Postgres, and Elasticsearch earlier, you will want add something like the following configuration to your local.py, based on the databases you have:

RECORD_HTTP_TRANSACTIONS = True  # Only if cassandra or postgres are installed

RAW_PROCESSING = ['cassandra', 'postgres']
NORMALIZED_PROCESSING = ['cassandra', 'postgres', 'elasticsearch']
CANONICAL_PROCESSOR = 'postgres'
RESPONSE_PROCESSOR = 'postgres'

For raw and normalized processing, add the databases you have installed. Only add elasticsearch to normalized processing, as it does not have a raw processing module.

RAW_PROCESSING and NORMALIZED_PROCESSING are both lists, so you can add as many processors as you wish. CANONICAL_PROCESSOR and RESPONSE_PROCESSOR both are single processors only.

note: Cassandra processing will soon be phased out, so we recommend using Postgres for your processing needs. Either one will work for now!

If you'd like to use local storage, you will want to make sure your local.py has the following configuration:

RECORD_HTTP_TRANSACTIONS = False

NORMALIZED_PROCESSING = ['storage']
RAW_PROCESSING = ['storage']

This will save all harvested/normalized files to the directory archive/<source>/<document identifier>

note: Be careful with this, as if you harvest too many documents with the storage module enabled, you could start experiencing inode errors

If you'd like to be able to run all harvesters, you'll need to register for a PLOS API key, a Harvard Dataverse API Key, and a Springer API Key.

Add your API keys to the following line to your local.py file:

PLOS_API_KEY = 'your-api-key-here'
HARVARD_DATAVERSE_API_KEY = 'your-api-key-here'
SPRINGER_KEY = 'your-api-key-here'

Running the scheduler (optional)

  • from the top-level project directory run:
$ invoke beat

to start the scheduler, and

$ invoke worker

to start the worker.

Harvesters

Run all harvesters with

$ invoke harvesters

or, just one with

$ invoke harvester harvester-name

For testing local development, running the mit harvester is recommended.

Note: harvester-name is the same as the defined harvester "short name".

Invoke a harvester for a certain start date with the --start or -sargument. Invoke a harvester for a certain end date with the --end or -eargument.

For example, to run a harvester between the dates of March 14th and March 16th 2015, run:

$ invoke harvester harvester-name --start 2015-03-14 --end 2015-03-16

Either --start or --end can also be used on their own. Not supplying arguments will default to starting the number of days specified in settings.DAYS_BACK and ending on the current date.

If --end is given with no --start, start will default to the number of days specified in settings.DAYS_BACK before the given end date.

Automated OAI PMH Harvester Creation

Writing a harvester for inclusion with scrAPI? If the provider makes their metadata available using the OAI-PMH standard, then autooai is a utility that will do most of the work for you.

Working with the OSF

To configure scrapi to work in a local OSF dev environment:

  1. Ensure 'elasticsearch' is in the NORMALIZED_PROCESSING list in scrapi/settings/local.py
  2. Run at least one harvester
  3. Configure the share_v2 alias
  4. Generate the provider map

Aliases

Multiple SHARE indices may be used by the OSF. By default, OSF uses the share_v2 index. Activate this alias by running:

$ inv alias share share_v2

Note that aliases must be activated before the provider map is generated.

Provider Map

$ inv alias share share_v2
$ inv provider_map 

Delete the Elasticsearch index

To remove both the share and share_v2 indices from elasticsearch:

$ curl -XDELETE 'localhost:9200/share*'

Testing

  • To run the tests for the project, just type
$ invoke test

and all of the tests in the 'tests/' directory will be run.

Pitfalls

Installing with anaconda

If you're using anaconda on your system at all, using pip to install all requirements from scratch from requirements.txt and dev-requirements.txt results in an Import Error when invoking tests or harvesters.

Example:

ImportError: dlopen(/Users/username/.virtualenvs/scrapi2/lib/python2.7/site-packages/lxml/etree.so, 2): Library not loaded: libxml2.2.dylib Referenced from: /Users/username/.virtualenvs/scrapi2/lib/python2.7/site-packages/lxml/etree.so Reason: Incompatible library version: etree.so requires version 12.0.0 or later, but libxml2.2.dylib provides version 10.0.0

To fix:

  • run pip uninstall lxml
  • remove the anaconda/bin from your system path in your bash_profile
  • reinstall requirements as usual

Answer found in this stack overflow question and answer

Institutions!

Scrapi supports the addition of institutions in a separate index (institutions). Unlike data stored in the share indices, institution's metadata is updated much less frequently, meaning that simple parsers can be used to manually load data from providers instead of using scheduled harvesters.

Currently, data from GRID and IPEDS is supported:

  • GRID: Provides data on international research facilities. The currently used dataset is grid_2015_11_05.json, which can be found here or, for the full dataset, here. To use this dataset move the file to '/institutions/', or override the file path and/or name on tasks.py. This can be individually loaded using the function grid() in tasks.py.
  • IPEDS: Provides data on secondary education institutions in the US. The currently used dataset is hd2014.csv, which can be found here, by clicking on Survey Data -> Complete data files -> 2014 -> Institutional Characteristics -> Directory information, or can be downloaded directly here. This will give you a file named HD2014.zip, which can be unzipped into hd2014.csv by running unzip HD2014.zip. To use this dataset move the file to '/institutions/', or override the file path and/or name on tasks.py. This can be individually loaded using the function ipeds() in tasks.py.

Running invoke institutions will properly load up institution data into elastic search provided the datasets are provided.