diff options
Diffstat (limited to 'python/fatcat_tools/importers')
| -rw-r--r-- | python/fatcat_tools/importers/common.py | 32 | ||||
| -rw-r--r-- | python/fatcat_tools/importers/crossref.py | 16 |
2 files changed, 37 insertions, 11 deletions
diff --git a/python/fatcat_tools/importers/common.py b/python/fatcat_tools/importers/common.py index 40c7abc0..5c33ebc9 100644 --- a/python/fatcat_tools/importers/common.py +++ b/python/fatcat_tools/importers/common.py @@ -4,6 +4,7 @@ import sys import csv import json import itertools +import subprocess from collections import Counter import pykafka @@ -37,19 +38,32 @@ class FatcatImporter: Base class for fatcat importers """ - def __init__(self, host_url, issn_map_file=None): - conf = fatcat_client.Configuration() - conf.host = host_url - self.api = fatcat_client.DefaultApi(fatcat_client.ApiClient(conf)) + def __init__(self, api, **kwargs): + + eg_extra = kwargs.get('editgroup_extra', dict()) + eg_extra['git_rev'] = eg_extra.get('git_rev', + subprocess.check_output(["git", "describe", "--always"]).strip()).decode('utf-8') + + self.api = api + self._editgroup_description = kwargs.get('editgroup_description') + self._editgroup_extra = kwargs.get('editgroup_extra') + issn_map_file = kwargs.get('issn_map_file') + self._issnl_id_map = dict() self._orcid_id_map = dict() self._doi_id_map = dict() - self._issn_issnl_map = None - self._orcid_regex = re.compile("^\\d{4}-\\d{4}-\\d{4}-\\d{3}[\\dX]$") if issn_map_file: self.read_issn_map_file(issn_map_file) + self._orcid_regex = re.compile("^\\d{4}-\\d{4}-\\d{4}-\\d{3}[\\dX]$") self.counts = Counter({'insert': 0, 'update': 0, 'processed_lines': 0}) + def _editgroup(self): + eg = fatcat_client.Editgroup( + description=self._editgroup_description, + extra=self._editgroup_extra, + ) + return self.api.create_editgroup(eg) + def describe_run(self): print("Processed {} lines, inserted {}, updated {}.".format( self.counts['processed_lines'], self.counts['insert'], self.counts['update'])) @@ -64,13 +78,13 @@ class FatcatImporter: def process_source(self, source, group_size=100): """Creates and auto-accepts editgroup every group_size rows""" - eg = self.api.create_editgroup(fatcat_client.Editgroup()) + eg = self._editgroup() i = 0 for i, row in enumerate(source): self.create_row(row, editgroup_id=eg.editgroup_id) if i > 0 and (i % group_size) == 0: self.api.accept_editgroup(eg.editgroup_id) - eg = self.api.create_editgroup(fatcat_client.Editgroup()) + eg = self._editgroup() self.counts['processed_lines'] += 1 if i == 0 or (i % group_size) != 0: self.api.accept_editgroup(eg.editgroup_id) @@ -81,7 +95,7 @@ class FatcatImporter: if decode_kafka: rows = [msg.value.decode('utf-8') for msg in rows] self.counts['processed_lines'] += len(rows) - eg = self.api.create_editgroup(fatcat_client.Editgroup()) + eg = self._editgroup() self.create_batch(rows, editgroup_id=eg.editgroup_id) def process_csv_source(self, source, group_size=100, delimiter=','): diff --git a/python/fatcat_tools/importers/crossref.py b/python/fatcat_tools/importers/crossref.py index 05543590..4f7faf59 100644 --- a/python/fatcat_tools/importers/crossref.py +++ b/python/fatcat_tools/importers/crossref.py @@ -4,6 +4,7 @@ import json import sqlite3 import datetime import itertools +import subprocess import fatcat_client from .common import FatcatImporter @@ -40,8 +41,19 @@ class CrossrefImporter(FatcatImporter): See https://github.com/CrossRef/rest-api-doc for JSON schema notes """ - def __init__(self, host_url, issn_map_file, extid_map_file=None, create_containers=True, check_existing=True): - super().__init__(host_url, issn_map_file) + def __init__(self, api, issn_map_file, **kwargs): + + eg_desc = kwargs.get('editgroup_description', + "Automated import of Crossref DOI metadata, harvested from REST API") + eg_extra = kwargs.get('editgroup_extra', dict()) + eg_extra['agent'] = eg_extra.get('agent', 'CrossrefImporter') + super().__init__(api, + issn_map_file=issn_map_file, + editgroup_description=eg_desc, + editgroup_extra=eg_extra) + extid_map_file = kwargs.get('extid_map_file') + create_containers = kwargs.get('create_containers') + check_existing = kwargs.get('check_existing') self.extid_map_db = None if extid_map_file: db_uri = "file:{}?mode=ro".format(extid_map_file) |