Database and search service for copyright records and renewals
 
 
Go to file
Mike Benowitz 5e02c87657 Select most complete date
in `copyrightEntry` XML entries there are frequently repeating date elements that have the same values. These can be different and more or less complete. The most common example of this is an entry having two `pub_date` values of `1932` and `1932-06-15` for example.

Previously this process simply took the first date. This updates the date parser to sort all dates of the same type by length, meaning the most complete dates are parsed. If they cannot be parsed the other, less complete, dates are used.

This was prompted by the presence of false January 1st dates and also the discovery that publication dates are often used as replacement registration dates in renewals.
2019-08-21 11:29:22 -04:00
.ebextensions Add swap space configuration to EBS instance 2019-07-09 10:47:39 -04:00
api SFR-539 Add new fields to source block 2019-08-12 15:46:45 -04:00
helpers Add error handling for 404 and 500 results from PostgreSQL 2019-08-05 15:20:27 -04:00
model Fix bug in publisher and author updates 2019-08-07 09:46:24 -04:00
tests Select most complete date 2019-08-21 11:29:22 -04:00
.coveragerc Further test additions and linting fixes 2019-07-10 17:27:42 -04:00
.gitignore Updates to maintenance script for EBS compatibility 2019-07-08 12:18:37 -04:00
README.md SFR-479 Update README.md 2019-06-04 17:04:37 -04:00
builder.py Select most complete date 2019-08-21 11:29:22 -04:00
config.yaml-dist SFR-453 Initial Commit 2019-05-29 12:47:03 -04:00
dev-requirements.txt Initial commit of testing configuration using `pytest` 2019-07-09 17:32:22 -04:00
esIndexer.py Fix missing index issue for indexing updates 2019-08-05 14:40:32 -04:00
main.py Initial commit of testing configuration using `pytest` 2019-07-09 17:32:22 -04:00
renBuilder.py SFR-453 Initial Commit 2019-05-29 12:47:03 -04:00
requirements.txt Add flasgger to requirements.txt 2019-07-05 14:52:17 -04:00
sessionManager.py SFR-453 Initial Commit 2019-05-29 12:47:03 -04:00

README.md

Copyright Registrations and Renewals DB Maintenance Script

Introduction

This script creates and updates a database of copyright records, both original registrations and renewals, drawn from two projects supported by the NYPL to digitize and gather this source data. The goal of this project is to create a single source for all copyright information for works registered after 1922 and renewed before 1978 (see Duration of Copyright for more information). This should provide the connections between registrations and renewals and allow users to see the connections between related registrations and renewals.

The two projects that this data is drawn from are hosted on GitHub, more information about the source of data for these projects can be found in each repository:

This repository also includes an API for querying the created database using ElasticSearch as a search layer and Flask as the application framework. See below for more detail on running and using the search API.

Database

This script generates and updates a Postgresql database and ElasticSearch search layer. It scans the above repository for updated source files and inserts/updates the appropriate records, ensuring that the search layer is kept up to date following these changes.

Structure

The database is relatively simple but contains some complexities to describe the range of conditions that exist in the copyright data. The tables currently are:

  • cce The core table that contains one copyright entry per row, identified by a UUIDv4. This includes a title and other descriptive information
  • renewal The core table that contains one copyright renewal per row, identified by a UUIDv4 and a renewal number.
  • registration This table contains copyright registration numbers (e.g. A999999) and registration dates. There can be multiple registration numbers per entry and renewal.
  • volume Contains information describing the CCE volume that specific records were drawn from, including a URL where a digitized version of the volume can be found. This can be used to locate the original text of a copyright entry.
  • author The authors associated with a copyright entry. For each entry one author is designated as the primary author, generally drawn from the entry heading and used for lookup and sorting purposes
  • publisher The publishers associated with a copyright entry. Each entry is tagged with a boolean claimant field that designates if the publisher was the copyright claimant. These entries can overlap with the author table due to discrepancies in how data was entered.
  • lccn Normalized LCCN numbers associated with a copyright entry
  • renewal_claimant Claimants who have created a renewal record
  • xml The source XML string for copyright entries. This is stored in an XML field and can be queried with xpath. This table also stores successive versions of the source XML for each entry, allowing changes to be found./visualized.
  • error_cce Contains copyright entries that could not be properly parsed. These are stored to see if the issue lies in the parser or the source data.

Relationships

The primary relationships are between the registration table, entries and renewals. Through registration numbers and dates, connections can be found between entries and renewals. It should be noted that an individual entry can have multiple renewals and vice-versa, related through multiple registration numbers

ElasticSearch

The ElasticSearch instance is designed to provide a simple, lightweight search layer. The documents in the ElasticSearch do not contain a full set of metadata, simply enough to return results that can be fully realized with an additional query to the database.

The ES instance is comprised of two indexes. cce for entries and ccr for renewals. These indexes can be queried separately or together

Configuration

This script can be configured to run locally using the NYPL data sources or your own forks of these sources. A config.yaml-dist file is included here that contains the structure for the necessary environment variables. The following will need to be supplied:

  • Connection details for a PostgreSQL database
  • Connection details for an ElasticSearch instance
  • A GitHub Personal Access Token for interacting with the GitHub API
  • Repository names for the entry and renewal repositories

Setup

This script requires Python 3.6+

To create a local instance it is recommended to create a virtualenv and install dependencies with pip install -r requirements.txt

Running

The utility script can be invoked simply with python main.py to initialize the database and pull down the full set of source data. Please note that this will take 2-4 hours to fully download, parse and index the source data.

Several command-line arguments control the options for the execution of this script

  • --REINITIALIZE Will drop the existing database and rebuild it from scratch. WARNING THIS WILL DELETE ALL CURRENT DATA
  • -t or --time The time ago in seconds to check for updated records in the GitHub repositories. This allows for only updating changed records
  • -y or --year A specific year to load from either the entries or renewals
  • -x or --exclude Set to exclude either the entries (with cce) or the renewals (with ccr) from the current execution. Useful when used in conjunction with the year parameter to control what records are updated.

API

This is a basic API that allows for a limited set of queries to be executed against the database. It allows for lookups by fulltext search, registration/renewal numbers and internal UUID numbers (to retrieve specific records). The returned objects show relationships between registrations and renewals and can optionally return the source data from which each record was created.

Running the API

To start the api, ensure that that you've updated your virtualenv with the most recent version of the requirements file and run the following commands:

  1. export FLASK_ENV=development
  2. export FLASK_APP=api/app.py
  3. python -m flask run

This will start the flask application at localhost:5000. Accessing that page will redirect you to a SwaggerDocs page that describes the available endpoints, their parameters, response object and allow users a chance to experiment with the endpoints.

Using the API

The API provides 5 endpoints for retrieving registration and renewal records. These are split between Search and Lookup endpoints

The search endpoints return many or no object depending on what search terms are used. All three search endpoints share 3 query parameters:

  • page: The page of results to return. Defaults to 0
  • per_page: The number of results to return per page. Defaults to 10
  • source: A flag to set the return of the source XML/CSV data. Defaults to false

The individual endpoints are:

  • /search/fulltext?query=<query string>: A full text query
  • /search/registration/<regnum>: A search for a specific registration number
  • /search/renewal/<rennum>: A search for a specific renewal number

Lookup

The lookup endpoints return data for a specific Registration or Renewal record. These do not accept additional parameters, but return the source data for any record. These endpoints use the internally generated UUID numbers to ensure a globally unique lookup value

  • /registration/<uuid>: Returns a single Registration record
  • /renewal/<uuid>: Returns a single Renewal record