Implement extraction command

Requires !9 (merged)

Closes #9 (closed)

dan/datasets/extract/arkindex_utils.py

-# -*- coding: utf-8 -*-
-
-"""
-    The arkindex_utils module
-    ======================
-"""
-
-import errno
-import logging
-import sys
-
-from apistar.exceptions import ErrorResponse
-
-
-def retrieve_corpus(client, corpus_name: str) -> str:
-    """
-    Retrieve the corpus id from the corpus name.
-    :param client: The arkindex client.
-    :param corpus_name: The name of the corpus to retrieve.
-    :return target_corpus: The id of the retrieved corpus.
-    """
-    for corpus in client.request("ListCorpus"):
-        if corpus["name"] == corpus_name:
-            target_corpus = corpus["id"]
-    try:
-        logging.info(f"Corpus id retrieved: {target_corpus}")
-    except NameError:
-        logging.error(f"Corpus {corpus_name} not found")
-        sys.exit(errno.EINVAL)
-
-    return target_corpus
-
-
-def retrieve_subsets(
-    client, corpus: str, parents_types: list, parents_names: list
-) -> list:
-    """
-    Retrieve the requested subsets.
-    :param client: The arkindex client.
-    :param corpus: The id of the retrieved corpus.
-    :param parents_types: The types of parents of the elements to retrieve.
-    :param parents_names: The names of parents of the elements to retrieve.
-    :return subsets: The retrieved subsets.
-    """
-    subsets = []
-    for parent_type in parents_types:
-        try:
-            subsets.extend(
-                client.request("ListElements", corpus=corpus, type=parent_type)[
-                    "results"
-                ]
-            )
-        except ErrorResponse as e:
-            logging.error(f"{e.content}: {parent_type}")
-            sys.exit(errno.EINVAL)
-    # Retrieve subsets with name in parents-names. If no parents-names given, keep all subsets.
-    if parents_names is not None:
-        logging.info(f"Retrieving {parents_names} subset(s)")
-        subsets = [subset for subset in subsets if subset["name"] in parents_names]
-    else:
-        logging.info("Retrieving all subsets")
-
-    if len(subsets) == 0:
-        logging.info("No subset found")
-
-    return subsets