Skip to content

Latest commit

 

History

History
104 lines (79 loc) · 3.17 KB

README.md

File metadata and controls

104 lines (79 loc) · 3.17 KB
Raw

README

master:CircleCI

This is the crawler component for Buzzbang, a project to enable applications to find and use Bioschemas markup, and Google-like search over it for humans. Please see https://github.com/buzzbangorg/buzzbang-doc/wiki for more information.

Usage

These instructions are for Linux. Windows is not supported.

1. Create the intermediate crawl database

./setup/bsbang-setup-sqlite.py <path-to-crawl-db>

Example:

./setup/bsbang-setup-sqlite.py data/crawl.db

2. Queue URLs for Bioschemas JSON-LD extraction by adding them directly and crawling sitemaps

./bsbang-crawl.py <path-to-crawl-db> <location>

The location can be:

  • a sitemap (e.g. http://beta.synbiomine.org/synbiomine/sitemap.xml)
  • a webpage (e.g. http://identifiers.org or file://test/examples/FAIRsharing.html)
  • a path (e.g. conf/default-targets.txt which will then crawl all the locations in that file)

Example:

./bsbang-crawl.py data/crawl.db conf/default-targets.txt

3. Extract Bioschemas JSON-LD from webpages and insert into the crawl database.

./bsbang-extract.py <path-to-crawl-db>

** To download the crawled data from the database -

./bsbang-dump.py <path-to-crawl-db> <path-to-save-jsonld>

4. Install Solr.

5. Create a Solr core named 'bsbang'

cd $SOLR/bin
./solr create -c bsbang

6. Run Solr setup

cd $BSBANG
./setup/bsbang-setup-solr.py <path-to-bsbang-config-file> --solr-core-url <URL-of-solr-endpoint>

Example:

./setup/bsbang-setup-solr.py conf/bsbang-solr-setup.xml --solr-core-url http://localhost:8983/solr/testcore/

7. Index the extracted Bioschemas JSON-LD in Solr

./bsbang-index.py <path-to-crawl-db> --solr-core-url <URL-of-solr-endpoint>

Example:

./bsbang-index.py data/crawl.db --solr-core-url http://localhost:8983/solr/testcore/

Frontend

See https://github.com/justinccdev/bsbang-frontend for a frontend project to the index.

Tests

$ python3 -m unittest discover

TODO

Future possibilities include:

  • Possibly switch to using a 3rd party crawler or components rather than this custom-built one. Please see #5
  • Make crawler periodically re-crawl.
  • Understand much more structure (e.g. DataSet elements within DataCatalog).
  • Parse other Bioschemas and schema.org types used by life sciences websites (e.g. Organization, Service, Product)
  • Instead of using Sqlite as intermediate crawl store, use something more scalable (perhaps mongodb, cassandra, etc.). But also see the item where we may want to replace parts/most of crawling infrastructure with a third party project, which will already have solved some, if not all, of the scalability issues.
  • Crawl and understand PhysicalEntity/BioChemEntity/ResearchEntity once this matures further.

Any other suggestions welcome as Github issues for discussion or as pull requests.

Hacking

Contributions welcome! Please

  • Make pull requests to the dev branch.
  • Conform to the PEP 8 style guide.

Thanks!