cldfzenodo 0.2.0

functionality to retrieve CLDF datasets deposited on Zenodo

cldfzenodo

cldfzenodo provides programmatic access to CLDF data deposited on Zenodo.

Install

pip install cldfzenodo

CLI

cldfzenodo provides a subcommand to be run from cldfbench. To make use of this command, you have to install cldfbench, which can be done via

pip install cldfzenodo[cli]

Then you can download CLDF datasets from Zenodo, using the DOI for identification. E.g.

cldfbench zenodo.download 10.5281/zenodo.4683137  --directory wals-2020.1/

will download WALS Online as CLDF dataset into wals-2020.1:

$ tree wals-2020.1/
wals-2020.1/
├── areas.csv
├── chapters.csv
├── codes.csv
├── contributors.csv
├── countries.csv
├── examples.csv
├── language_names.csv
├── languages.csv
├── parameters.csv
├── sources.bib
├── StructureDataset-metadata.json
└── values.csv

0 directories, 12 files

API

Metadata and data of (potential) CLDF datasets deposited on Zenodo is accessed via cldfzenodo.Record objects. Such objects can be obtained in various ways:

• Via DOI:

  import cldfzenodo
  rec = cldfzenodo.Record.from_doi('https://doi.org/10.5281/zenodo.4762034')
  

• From deposits grouped into a Zenodo community (and obtained through OAI-PMH):

  import cldfzenodo.oai
  for rec in cldfzenodo.oai.iter_records('dictionaria'):
    print(rec)
  

• From search results using keywords:

  import cldfzenodo
  for rec in cldfzenodo.search_wordlists():
    print(rec)
  

cldfzenodo.Record objects provide sufficient metadata to allow identification and data access:

>>> from cldfzenodo import Record
>>> print(Record.from_doi('10.5281/zenodo.4762034').bibtex)
@misc{zenodo-4762034,
  author    = {Hammarström, Harald and Forkel, Robert and Haspelmath, Martin and Bank, Sebastian},
  title     = {glottolog/glottolog: Glottolog database 4.4 as CLDF},
  keywords  = {cldf:StructureDataset, linguistics},
  publisher = {Zenodo},
  year      = {2021},
  doi       = {10.5281/zenodo.4762034},
  url       = {https://doi.org/10.5281/zenodo.4762034},
  copyright = {Creative Commons Attribution 4.0}
}

One can download the full deposit (and access - possible multiple - CLDF datasets):

from pycldf import iter_datasets
record.download('my_directory')
for cldf in iter_datasets('my_directory'):
    pass

But often, only the "pure" CLDF data is of interest - and not the additional metadata and curation context, e.g. of cldfbench-curated datasets. This can be done via

from pycldf import Dataset
mdpath = record.download_dataset('my_directory')
cldf = Dataset.from_metadata(mdpath)