scielo-clea

SciELO Publishing Schema XML document front matter metadata reader/sanitizer


Keywords
labs
License
BSD-2-Clause
Install
pip install scielo-clea==0.4.4

Documentation

Clea

This project is an XML front matter metadata reader for documents that almost follows the SciELO Publishing Schema, extracting and sanitizing the values regarding the affiliations.

Installation

One can install Clea with either:

pip install scielo-clea          # Minimal
pip install scielo-clea[cli]     # Clea with CLI (recommended)
pip install scielo-clea[server]  # Clea with the testing/example server
pip install scielo-clea[all]     # Clea with both CLI and the server

Actually all these commands installs everything, only the dependencies aren't the same. The first is an installation with minimal requirements, intended for use within Python, as an imported package.

Running the command line interface

The CLI is a way to use Clea as an article XML to JSONL converter (one JSON output line for each XML input):

clea -o output.jsonl article1.xml article2.xml article3.xml

The same can be done with python -m clea instead of clea. The output is the standard output stream. See clea --help for more information.

Running the testing server

You can run the development server using the flask CLI. For example, for listening at 8080 from every host:

FLASK_APP=clea.server flask run -h 0.0.0.0 -p 8080

In a production server with 4 worker processes for handling requests, you can, for example:

  • Install gunicorn (it's not a dependency)
  • Run gunicorn -b 0.0.0.0:8080 -w 4 clea.server:app

Clea as a library

A simple example to see all the extracted data is:

from clea import Article
from pprint import pprint

art = Article("some_file.xml")
pprint(art.data_full)

That's a dictionary of lists with all the "raw" extracted data. The keys of that dictionary can be directly accessed, so one can avoid extracting everything from the XML by getting just the specific items/attributes (e.g. art["journal_meta"][0].data_full or art.journal_meta[0].data_full instead of art.data_full["journal_meta"][0]). These items/attributes are always lists, for example:

  • art["aff"]: List of clea.core.Branch instances
  • art["sub_article"]: List of clea.core.SubArticle instances
  • art["contrib"][0]["contrib_name"]: List of strings

Where the art["contrib"][0] is a Branch instance, and all such instances behave in the same way (there's no nested branches). That can be seen as another way to navigate in the former dictionary, the last example should return the same list one would get with art.data_full["contrib"][0]["contrib_name"], but without extracting everything else that appears in the art.data_full dictionary.

More simple stuff that can be done:

len(art.aff)              # Number of <aff> entries
len(art.sub_article)      # Number of <sub-article>
art.contrib[0].data_full  # Data from the first contributor as a dict

# Something like {"type": ["translation"], "lang": ["en"]},
# the content from <sub-article> attributes
art["sub_article"][0]["article"][0].data_full

# A string with the article title, accessing just the desired content
art["article_meta"][0]["article_title"][0]

All SubArticle, Article and Branch instances have the data_full property and the get method, the latter being internally used for item/attribute getting. Their behavior is:

  • Branch.get always returns a list of strings
  • Article.get("sub_article") returns a list of SubArticle
  • Article.get(...) returns a list of Branch
  • SubArticle behaves like Article

The extracted information is not exhaustive! Its result should not be seen as a replacement of the raw XML.

One of the goals of this library was to help on creating a tabular data from a given XML with as many rows as required to have a pair of a matching <aff> and <contrib> in each row. These are the Article methods/properties that does that matching:

  • art.aff_contrib_inner_gen()
  • art.aff_contrib_full_gen()
  • art.aff_contrib_inner
  • art.aff_contrib_full
  • art.aff_contrib_inner_indices
  • art.aff_contrib_full_indices

The most useful ones are probably the last ones, which return a list of pairs (tuples) of indices (ints), so one can use a (ai, ci) result to access the (art.aff[ai], art.contrib[ci]) pair, unless the index is -1 (not found). The ones with the _gen suffix are generator functions that yields a tuple with two Branch entries (or None), the ones without a suffix return a list of merged dictionaries in an almost tabular format (dictionary of lists of strings). Each list regarding these elements for these specific elements should usually have at most one string, but that's not always the case even for these specific elements, then one should be careful when using the data property.

The inner and full in the names regards to INNER JOIN and FULL OUTER JOIN from SQL, meaning the unmatched elements (all <aff> and <contrib> unreferred nodes) are discarded in the former strategy, whereas they're forcefully matched with None in the latter.

To print all the extracted data from a XML including the indices of matching <aff> and <contrib> pairs performed in the FULL OUTER JOIN sense, similar to the test server response:

pprint({
    **article.data_full,
    "aff_contrib_pairs": article.aff_contrib_full_indices,
})