diff options
Diffstat (limited to 'python/fatcat_tools')
| -rw-r--r-- | python/fatcat_tools/importers/__init__.py | 1 | ||||
| -rw-r--r-- | python/fatcat_tools/importers/common.py | 158 | ||||
| -rw-r--r-- | python/fatcat_tools/importers/crossref.py | 2 | ||||
| -rw-r--r-- | python/fatcat_tools/importers/datacite.py | 2 | ||||
| -rw-r--r-- | python/fatcat_tools/importers/doaj_article.py | 358 | ||||
| -rw-r--r-- | python/fatcat_tools/normal.py | 332 | ||||
| -rw-r--r-- | python/fatcat_tools/transforms/elasticsearch.py | 3 |
7 files changed, 697 insertions, 159 deletions
diff --git a/python/fatcat_tools/importers/__init__.py b/python/fatcat_tools/importers/__init__.py index c08e04c2..d2928d09 100644 --- a/python/fatcat_tools/importers/__init__.py +++ b/python/fatcat_tools/importers/__init__.py @@ -30,3 +30,4 @@ from .cdl_dash_dat import auto_cdl_dash_dat from .ingest import IngestFileResultImporter, SavePaperNowFileImporter, IngestWebResultImporter from .shadow import ShadowLibraryImporter from .file_meta import FileMetaImporter +from .doaj_article import DoajArticleImporter diff --git a/python/fatcat_tools/importers/common.py b/python/fatcat_tools/importers/common.py index 14415683..3c810391 100644 --- a/python/fatcat_tools/importers/common.py +++ b/python/fatcat_tools/importers/common.py @@ -3,12 +3,9 @@ import re import sys import csv import json -import ftfy -import base64 import sqlite3 import datetime import subprocess -import unicodedata from collections import Counter from confluent_kafka import Consumer, KafkaException import xml.etree.ElementTree as ET @@ -18,162 +15,13 @@ from bs4 import BeautifulSoup import fatcat_openapi_client from fatcat_openapi_client.rest import ApiException +# TODO: refactor so remove need for this (re-imports for backwards compatibility) +from fatcat_tools.normal import (clean_str as clean, is_cjk, b32_hex, LANG_MAP_MARC) # noqa: F401 DATE_FMT = "%Y-%m-%d" SANE_MAX_RELEASES = 200 SANE_MAX_URLS = 100 -# These are very close, but maybe not exactly 1-to-1 with 639-2? Some mix of -# 2/T and 2/B? -# PubMed/MEDLINE and JSTOR use these MARC codes -# https://www.loc.gov/marc/languages/language_name.html -LANG_MAP_MARC = { - 'afr': 'af', - 'alb': 'sq', - 'amh': 'am', - 'ara': 'ar', - 'arm': 'hy', - 'aze': 'az', - 'ben': 'bn', - 'bos': 'bs', - 'bul': 'bg', - 'cat': 'ca', - 'chi': 'zh', - 'cze': 'cs', - 'dan': 'da', - 'dut': 'nl', - 'eng': 'en', - 'epo': 'eo', - 'est': 'et', - 'fin': 'fi', - 'fre': 'fr', - 'geo': 'ka', - 'ger': 'de', - 'gla': 'gd', - 'gre': 'el', - 'heb': 'he', - 'hin': 'hi', - 'hrv': 'hr', - 'hun': 'hu', - 'ice': 'is', - 'ind': 'id', - 'ita': 'it', - 'jpn': 'ja', - 'kin': 'rw', - 'kor': 'ko', - 'lat': 'la', - 'lav': 'lv', - 'lit': 'lt', - 'mac': 'mk', - 'mal': 'ml', - 'mao': 'mi', - 'may': 'ms', - 'nor': 'no', - 'per': 'fa', - 'per': 'fa', - 'pol': 'pl', - 'por': 'pt', - 'pus': 'ps', - 'rum': 'ro', - 'rus': 'ru', - 'san': 'sa', - 'slo': 'sk', - 'slv': 'sl', - 'spa': 'es', - 'srp': 'sr', - 'swe': 'sv', - 'tha': 'th', - 'tur': 'tr', - 'ukr': 'uk', - 'urd': 'ur', - 'vie': 'vi', - 'wel': 'cy', - -# additions - 'gle': 'ga', # "Irish" (Gaelic) - 'jav': 'jv', # Javanese - 'welsh': 'cy', # Welsh - 'oci': 'oc', # Occitan - -# Don't have ISO 639-1 codes - 'grc': 'el', # Ancient Greek; map to modern greek - 'map': None, # Austronesian (collection) - 'syr': None, # Syriac, Modern - 'gem': None, # Old Saxon - 'non': None, # Old Norse - 'emg': None, # Eastern Meohang - 'neg': None, # Negidal - 'mul': None, # Multiple languages - 'und': None, # Undetermined -} - - -def clean(thing, force_xml=False): - """ - This function is appropriate to be called on any random, non-markup string, - such as author names, titles, etc. - - It will try to clean up common unicode mangles, HTML characters, etc. - - This will detect XML/HTML and "do the right thing" (aka, not remove - entities like '&' if there are tags in the string), unless you pass the - 'force_xml' parameter, which might be appropriate for, eg, names and - titles, which generally should be projected down to plain text. - - Also strips extra whitespace. - """ - if not thing: - return None - fix_entities = 'auto' - if force_xml: - fix_entities = True - fixed = ftfy.fix_text(thing, fix_entities=fix_entities).strip() - if not fixed or len(fixed) <= 1: - # wasn't zero-length before, but is now; return None - return None - return fixed - -def test_clean(): - - assert clean(None) == None - assert clean('') == None - assert clean('1') == None - assert clean('123') == '123' - assert clean('a&b') == 'a&b' - assert clean('<b>a&b</b>') == '<b>a&b</b>' - assert clean('<b>a&b</b>', force_xml=True) == '<b>a&b</b>' - -def b32_hex(s): - s = s.strip().split()[0].lower() - if s.startswith("sha1:"): - s = s[5:] - if len(s) != 32: - return s - return base64.b16encode(base64.b32decode(s.upper())).lower().decode('utf-8') - -def is_cjk(s): - if not s: - return False - for c in s: - if c.isalpha(): - lang_prefix = unicodedata.name(c).split()[0] - return lang_prefix in ('CJK', 'HIRAGANA', 'KATAKANA', 'HANGUL') - return False - -def test_is_cjk(): - assert is_cjk(None) is False - assert is_cjk('') is False - assert is_cjk('blah') is False - assert is_cjk('岡, 鹿, 梨, 阜, 埼') is True - assert is_cjk('[岡, 鹿, 梨, 阜, 埼]') is True - assert is_cjk('菊') is True - assert is_cjk('岡, 鹿, 梨, 阜, 埼 with eng after') is True - assert is_cjk('水道') is True - assert is_cjk('オウ, イク') is True # kanji - assert is_cjk('ひヒ') is True - assert is_cjk('き゚ゅ') is True - assert is_cjk('ㄴ, ㄹ, ㅁ, ㅂ, ㅅ') is True - DOMAIN_REL_MAP = { "archive.org": "archive", # LOCKSS, Portico, DuraSpace, etc would also be "archive" @@ -444,6 +292,7 @@ class EntityImporter: raise NotImplementedError def is_orcid(self, orcid): + # TODO: replace with clean_orcid() from fatcat_tools.normal return self._orcid_regex.match(orcid) is not None def lookup_orcid(self, orcid): @@ -464,6 +313,7 @@ class EntityImporter: return creator_id def is_doi(self, doi): + # TODO: replace with clean_doi() from fatcat_tools.normal return doi.startswith("10.") and doi.count("/") >= 1 def lookup_doi(self, doi): diff --git a/python/fatcat_tools/importers/crossref.py b/python/fatcat_tools/importers/crossref.py index 71f08952..e77fa65e 100644 --- a/python/fatcat_tools/importers/crossref.py +++ b/python/fatcat_tools/importers/crossref.py @@ -487,8 +487,6 @@ class CrossrefImporter(EntityImporter): except fatcat_openapi_client.rest.ApiException as err: if err.status != 404: raise err - # doesn't exist, need to update - return True # eventually we'll want to support "updates", but for now just skip if # entity already exists diff --git a/python/fatcat_tools/importers/datacite.py b/python/fatcat_tools/importers/datacite.py index 5cdc5577..70f8db86 100644 --- a/python/fatcat_tools/importers/datacite.py +++ b/python/fatcat_tools/importers/datacite.py @@ -781,8 +781,6 @@ class DataciteImporter(EntityImporter): except fatcat_openapi_client.rest.ApiException as err: if err.status != 404: raise err - # doesn't exist, need to update - return True # eventually we'll want to support "updates", but for now just skip if # entity already exists diff --git a/python/fatcat_tools/importers/doaj_article.py b/python/fatcat_tools/importers/doaj_article.py new file mode 100644 index 00000000..03752484 --- /dev/null +++ b/python/fatcat_tools/importers/doaj_article.py @@ -0,0 +1,358 @@ +""" +Importer for DOAJ article-level metadata, schema v1. + +DOAJ API schema and docs: https://doaj.org/api/v1/docs +""" + +import warnings +import datetime +from typing import List, Optional + +import fatcat_openapi_client +from fatcat_tools.normal import (clean_doi, clean_str, parse_month, + clean_orcid, detect_text_lang, parse_lang_name, parse_country_name, + clean_pmid, clean_pmcid) +from fatcat_tools.importers.common import EntityImporter + +# Cutoff length for abstracts. +MAX_ABSTRACT_LENGTH = 2048 + + +class DoajArticleImporter(EntityImporter): + + def __init__(self, + api, + issn_map_file, + **kwargs): + + eg_desc = kwargs.get( + 'editgroup_description', + "Automated import of DOAJ article metadata, harvested from REST API or bulk dumps" + ) + eg_extra = kwargs.get('editgroup_extra', dict()) + eg_extra['agent'] = eg_extra.get('agent', + 'fatcat_tools.DoajArticleImporter') + # ensure default is to not do updates with this worker (override super() default) + kwargs['do_updates'] = kwargs.get("do_updates", False) + super().__init__(api, + issn_map_file=issn_map_file, + editgroup_description=eg_desc, + editgroup_extra=eg_extra, + **kwargs) + + self.this_year = datetime.datetime.now().year + self.read_issn_map_file(issn_map_file) + + def want(self, obj): + return True + + def parse_record(self, obj): + """ + bibjson { + abstract (string, optional), + author (Array[bibjson.author], optional), + identifier (Array[bibjson.identifier]), + journal (bibjson.journal, optional), + keywords (Array[string], optional), + link (Array[bibjson.link], optional), + month (string, optional), + subject (Array[bibjson.subject], optional), + title (string), + year (string, optional) + } + bibjson.journal { + country (string, optional), + end_page (string, optional), + language (Array[string], optional), + license (Array[bibjson.journal.license], optional), + number (string, optional), + publisher (string, optional), + start_page (string, optional), + title (string, optional), + volume (string, optional) + } + """ + + if not obj or not isinstance(obj, dict) or not 'bibjson' in obj: + self.counts['skip-empty'] += 1 + return None + + bibjson = obj['bibjson'] + + title = clean_str(bibjson.get('title'), force_xml=True) + if not title: + self.counts['skip-title'] += 1 + return False + + container_name = clean_str(bibjson['journal']['title']) + container_id = None + # NOTE: 'issns' not documented in API schema + for issn in bibjson['journal']['issns']: + issnl = self.issn2issnl(issn) + if issnl: + container_id = self.lookup_issnl(self.issn2issnl(issn)) + if container_id: + # don't store container_name when we have an exact match + container_name = None + break + + volume = clean_str(bibjson['journal'].get('volume')) + # NOTE: this schema seems to use "number" as "issue number" + issue = clean_str(bibjson['journal'].get('number')) + publisher = clean_str(bibjson['journal'].get('publisher')) + + try: + release_year = int(bibjson.get('year')) + except (TypeError, ValueError): + release_year = None + release_month = parse_month(clean_str(bibjson.get('month'))) + + # block bogus far-future years/dates + if release_year is not None and (release_year > (self.this_year + 5) or release_year < 1000): + release_month = None + release_year = None + + license_slug = self.doaj_license_slug(bibjson['journal'].get('license')) + country = parse_country_name(bibjson['journal'].get('country')) + language = None + for raw in bibjson['journal'].get('language') or []: + language = parse_lang_name(raw) + if language: + break + + # pages + # NOTE: error in API docs? seems like start_page not under 'journal' object + start_page = clean_str(bibjson['journal'].get('start_page')) or clean_str(bibjson.get('start_page')) + end_page = clean_str(bibjson['journal'].get('end_page')) or clean_str(bibjson.get('end_page')) + pages: Optional[str] = None + if start_page and end_page: + pages = f"{start_page}-{end_page}" + elif start_page: + pages = start_page + + doaj_article_id = obj['id'].lower() + ext_ids = self.doaj_ext_ids(bibjson['identifier'], doaj_article_id) + abstracts = self.doaj_abstracts(bibjson) + contribs = self.doaj_contribs(bibjson.get('author') or []) + + # DOAJ-specific extra + doaj_extra = dict() + if bibjson.get('subject'): + doaj_extra['subject'] = bibjson.get('subject') + if bibjson.get('keywords'): + doaj_extra['keywords'] = [k for k in [clean_str(s) for s in bibjson.get('keywords')] if k] + + # generic extra + extra = dict() + if country: + extra['country'] = country + if not container_id and container_name: + extra['container_name'] = container_name + if release_year and release_month: + # TODO: schema migration + extra['release_month'] = release_month + + if doaj_extra: + extra['doaj'] = doaj_extra + if not extra: + extra = None + + re = fatcat_openapi_client.ReleaseEntity( + work_id=None, + container_id=container_id, + release_type='article-journal', + release_stage='published', + title=title, + release_year=release_year, + #release_date, + publisher=publisher, + ext_ids=ext_ids, + contribs=contribs, + volume=volume, + issue=issue, + pages=pages, + language=language, + abstracts=abstracts, + extra=extra, + license_slug=license_slug, + ) + re = self.biblio_hacks(re) + return re + + @staticmethod + def biblio_hacks(re): + """ + This function handles known special cases. For example, + publisher-specific or platform-specific workarounds. + """ + return re + + def try_update(self, re): + + # lookup existing release by DOAJ article id + existing = None + try: + existing = self.api.lookup_release(doaj=re.ext_ids.doaj) + except fatcat_openapi_client.rest.ApiException as err: + if err.status != 404: + raise err + + # then try other ext_id lookups + if not existing: + for extid_type in ('doi', 'pmid', 'pmcid'): + extid_val = getattr(re.ext_ids, extid_type) + if not extid_val: + continue + #print(f" lookup release type: {extid_type} val: {extid_val}") + try: + existing = self.api.lookup_release(**{extid_type: extid_val}) + except fatcat_openapi_client.rest.ApiException as err: + if err.status != 404: + raise err + if existing: + if existing.ext_ids.doaj: + warn_str = f"unexpected DOAJ ext_id match after lookup failed doaj={re.ext_ids.doaj} ident={existing.ident}" + warnings.warn(warn_str) + self.counts["skip-doaj-id-mismatch"] += 1 + return False + break + + # TODO: in the future could do fuzzy match here, eg using elasticsearch + + # create entity + if not existing: + return True + + # other logic could go here about skipping updates + if not self.do_updates or existing.ext_ids.doaj: + self.counts['exists'] += 1 + return False + + # fields to copy over for update + existing.ext_ids.doaj = existing.ext_ids.doaj or re.ext_ids.doaj + existing.release_type = existing.release_type or re.release_type + existing.release_stage = existing.release_stage or re.release_stage + existing.container_id = existing.container_id or re.container_id + existing.abstracts = existing.abstracts or re.abstracts + existing.extra['doaj'] = re.extra['doaj'] + existing.volume = existing.volume or re.volume + existing.issue = existing.issue or re.issue + existing.pages = existing.pages or re.pages + existing.language = existing.language or re.language + + try: + self.api.update_release(self.get_editgroup_id(), existing.ident, existing) + self.counts['update'] += 1 + except fatcat_openapi_client.rest.ApiException as err: + # there is a code path where we try to update the same release + # twice in a row; if that happens, just skip + # NOTE: API behavior might change in the future? + if "release_edit_editgroup_id_ident_id_key" in err.body: + self.counts['skip-update-conflict'] += 1 + return False + else: + raise err + + return False + + def insert_batch(self, batch): + self.api.create_release_auto_batch(fatcat_openapi_client.ReleaseAutoBatch( + editgroup=fatcat_openapi_client.Editgroup( + description=self.editgroup_description, + extra=self.editgroup_extra), + entity_list=batch)) + + def doaj_abstracts(self, bibjson: dict) -> List[fatcat_openapi_client.ReleaseAbstract]: + text = clean_str(bibjson.get('abstract')) + if not text or len(text) < 10: + return [] + if len(text) > MAX_ABSTRACT_LENGTH: + text = text[:MAX_ABSTRACT_LENGTH] + " [...]" + + lang = detect_text_lang(text) + + abstract = fatcat_openapi_client.ReleaseAbstract( + mimetype="text/plain", + content=text, + lang=lang, + ) + + return [abstract,] + + def doaj_contribs(self, authors: List[dict]) -> List[fatcat_openapi_client.ReleaseContrib]: + """ + bibjson.author { + affiliation (string, optional), + name (string), + orcid_id (string, optional) + } + """ + contribs = [] + index = 0 + for author in authors: + if not author.get('name'): + continue + creator_id = None + orcid = clean_orcid(author.get('orcid_id')) + if orcid: + creator_id = self.lookup_orcid(orcid) + contribs.append(fatcat_openapi_client.ReleaseContrib( + raw_name=author.get('name'), + role='author', + index=index, + creator_id=creator_id, + raw_affiliation=clean_str(author.get('affiliation')), + )) + index += 1 + return contribs + + def doaj_ext_ids(self, identifiers: List[dict], doaj_article_id: str) -> fatcat_openapi_client.ReleaseExtIds: + """ + bibjson.identifier { + id (string), + type (string) + } + """ + + assert doaj_article_id.isalnum() and len(doaj_article_id) == 32 + + doi: Optional[str] = None + pmid: Optional[str] = None + pmcid: Optional[str] = None + for id_obj in identifiers: + if not id_obj.get('id'): + continue + if id_obj['type'].lower() == 'doi': + doi = clean_doi(id_obj['id']) + elif id_obj['type'].lower() == 'pmid': + pmid = clean_pmid(id_obj['id']) + elif id_obj['type'].lower() == 'pmcid': + pmcid = clean_pmcid(id_obj['id']) + + return fatcat_openapi_client.ReleaseExtIds( + doaj=doaj_article_id, + doi=doi, + pmid=pmid, + pmcid=pmcid, + ) + + def doaj_license_slug(self, license_list: List[dict]) -> Optional[str]: + """ + bibjson.journal.license { + open_access (boolean, optional), + title (string, optional), + type (string, optional), + url (string, optional), + version (string, optional) + } + """ + if not license_list: + return None + for license in license_list: + if not license.get('open_access'): + continue + slug = license.get('type') + if slug.startswith('CC '): + slug = slug.replace('CC ', 'cc-').lower() + return slug + return None diff --git a/python/fatcat_tools/normal.py b/python/fatcat_tools/normal.py index 10a90dba..a3d6bccc 100644 --- a/python/fatcat_tools/normal.py +++ b/python/fatcat_tools/normal.py @@ -5,6 +5,13 @@ free-form input, titles, etc. """ import re +import base64 +from typing import Optional +import unicodedata + +import ftfy +import langdetect +import pycountry DOI_REGEX = re.compile(r"^10.\d{3,6}/\S+$") @@ -55,7 +62,8 @@ def clean_doi(raw): # will likely want to expand DOI_REGEX to exclude non-ASCII characters, but # for now block specific characters so we can get PubMed importer running # again. - if 'ä' in raw or '\u200e' in raw: + # known characters to skip: ä á \u200e \u2043 \u2012 + if not raw.isascii(): return None return raw @@ -72,6 +80,10 @@ def test_clean_doi(): assert clean_doi("10.4149/gpb¬_2017042") == None # "logical negation" character assert clean_doi("10.6002/ect.2020.häyry") == None # this example via pubmed (pmid:32519616) assert clean_doi("10.30466/vrf.2019.98547.2350\u200e") == None + assert clean_doi("10.12016/j.issn.2096⁃1456.2017.06.014") == None + assert clean_doi("10.4025/diálogos.v17i2.36030") == None + assert clean_doi("10.19027/jai.10.106‒115") == None + assert clean_doi("10.15673/атбп2312-3125.17/2014.26332") == None ARXIV_ID_REGEX = re.compile(r"^(\d{4}.\d{4,5}|[a-z\-]+(\.[A-Z]{2})?/\d{7})(v\d+)?$") @@ -233,3 +245,321 @@ def test_clean_orcid(): assert clean_orcid("0123-4567-3456-6789 ") == "0123-4567-3456-6789" assert clean_orcid("01234567-3456-6780") == None assert clean_orcid("0x23-4567-3456-6780") == None + + +def clean_str(thing: Optional[str], force_xml: bool = False) -> Optional[str]: + """ + This function is appropriate to be called on any random, non-markup string, + such as author names, titles, etc. + + It will try to clean up common unicode mangles, HTML characters, etc. + + This will detect XML/HTML and "do the right thing" (aka, not remove + entities like '&' if there are tags in the string), unless you pass the + 'force_xml' parameter, which might be appropriate for, eg, names and + titles, which generally should be projected down to plain text. + + Also strips extra whitespace. + """ + if not thing: + return None + fix_entities = 'auto' + if force_xml: + fix_entities = True + fixed = ftfy.fix_text(thing, fix_entities=fix_entities).strip() + if not fixed or len(fixed) <= 1: + # wasn't zero-length before, but is now; return None + return None + return fixed + +def test_clean_str(): + + assert clean_str(None) == None + assert clean_str('') == None + assert clean_str('1') == None + assert clean_str('123') == '123' + assert clean_str('a&b') == 'a&b' + assert clean_str('<b>a&b</b>') == '<b>a&b</b>' + assert clean_str('<b>a&b</b>', force_xml=True) == '<b>a&b</b>' + +def b32_hex(s): + s = s.strip().split()[0].lower() + if s.startswith("sha1:"): + s = s[5:] + if len(s) != 32: + return s + return base64.b16encode(base64.b32decode(s.upper())).lower().decode('utf-8') + +def is_cjk(s): + if not s: + return False + for c in s: + if c.isalpha(): + lang_prefix = unicodedata.name(c).split()[0] + return lang_prefix in ('CJK', 'HIRAGANA', 'KATAKANA', 'HANGUL') + return False + +def test_is_cjk(): + assert is_cjk(None) is False + assert is_cjk('') is False + assert is_cjk('blah') is False + assert is_cjk('岡, 鹿, 梨, 阜, 埼') is True + assert is_cjk('[岡, 鹿, 梨, 阜, 埼]') is True + assert is_cjk('菊') is True + assert is_cjk('岡, 鹿, 梨, 阜, 埼 with eng after') is True + assert is_cjk('水道') is True + assert is_cjk('オウ, イク') is True # kanji + assert is_cjk('ひヒ') is True + assert is_cjk('き゚ゅ') is True + assert is_cjk('ㄴ, ㄹ, ㅁ, ㅂ, ㅅ') is True + +MONTH_MAP = { + "jan": 1, "january": 1, + "feb": 2, "febuary": 2, + "mar": 3, "march": 3, + "apr": 4, "april": 4, + "may": 5, "may": 5, + "jun": 6, "june": 6, + "jul": 7, "july": 7, + "aug": 8, "august": 8, + "sep": 9, "september": 9, + "oct": 10, "october": 10, + "nov": 11, "nov": 11, + "dec": 12, "december": 12, +} + +def parse_month(raw: Optional[str]) -> Optional[int]: + """ + Parses a string into a month number (1 to 12) + """ + if not raw: + return None + raw = raw.strip().lower() + if raw.isdigit(): + raw_int = int(raw) + if raw_int >= 1 and raw_int <= 12: + return raw_int + else: + return None + if raw in MONTH_MAP: + return MONTH_MAP[raw] + return None + +def test_parse_month() -> None: + + assert parse_month(None) == None + assert parse_month("") == None + assert parse_month("0") == None + assert parse_month("10") == 10 + assert parse_month("jan") == 1 + assert parse_month("September") == 9 + +def detect_text_lang(raw: str) -> Optional[str]: + """ + Tries to determine language of, eg, an abstract. + + Returns an ISO 631 2-char language code, or None. + """ + if not raw: + return None + try: + lang = langdetect.detect(raw) + lang = lang.split('-')[0] + assert len(lang) == 2 + return lang + except (langdetect.lang_detect_exception.LangDetectException, TypeError): + return None + return None + +def test_detect_text_lang() -> None: + assert detect_text_lang("") == None + EN_SAMPLE = "this is a string of English text for testing" + assert detect_text_lang(EN_SAMPLE) == "en" + JA_SAMPLE = "モーラの種類は、以下に示すように111程度存在する。ただし、研究者により数え方が少しずつ異なる。" + assert detect_text_lang(JA_SAMPLE) == "ja" + ZH_SAMPLE = "随着分布式清洁能源的普及,通信技术在协调各个分布式电源的控制中显得尤为重要。在电力信息传输的过程中,不同的网络状态下表现出不同的通信特性,严重的甚至会发生信息错乱丢包等行为,这对电网的实时控制产生严重影响。为研究信息系统对电力物理系统的实时影响,搭建了电力信息物理融合仿真平台,运用RT-LAB与OPNET两款实时仿真器,通过TCP/IP进行数据交互,对微电网电压、频率的集中式恢复与分布式恢复问题展开研究。仿真结果表明,该平台能有效地反映通信网络对电网控制的影响,提供了一种可靠的未来电力信息物理融合系统研究技术。随着分布式清洁能源的普及,通信技术在协调各个分布式电源的控制中显得尤为重要。在电力信息传输的过程中,不同的网络状态下表现出不同的通信特性,严重的甚至会发生信息错乱丢包等行为,这对电网的实时控制产生严重影响。为研究信息系统对电力物理系统的实时影响,搭建了电力信息物理融合仿真平台,运用RT-LAB与OPNET两款实时仿真器,通过TCP/IP进行数据交互,对微电网电压、频率的集中式恢复与分布式恢复问题展开研究。仿真结果表明,该平台能有效地反映通信网络对电网控制的影响,提供了一种可靠的未来电力信息物理融合系统研究技术。" + assert detect_text_lang(ZH_SAMPLE) == "zh" + +def parse_lang_name(raw: Optional[str]) -> Optional[str]: + """ + Parses a language name and returns a 2-char ISO 631 language code. + """ + if not raw: + return None + try: + lang = pycountry.languages.lookup(raw) + if lang.alpha_3 in ("mul", "mis"): + return None + return lang.alpha_2.lower() + except LookupError: + #print(f" unknown language: '{raw}', file=sys.stderr) + return None + except AttributeError: + #print(f" partial language metadata: '{lang}', file=sys.stderr) + return None + return None + +def test_parse_lang_name() -> None: + + assert parse_lang_name(None) == None + assert parse_lang_name("") == None + assert parse_lang_name("asdf ") == None + assert parse_lang_name("english") == "en" + assert parse_lang_name("ENGLISH") == "en" + assert parse_lang_name("asdf blah") is None + assert parse_lang_name("en") == "en" + assert parse_lang_name("EN") == "en" + assert parse_lang_name("ENG") == "en" + assert parse_lang_name("English") == "en" + assert parse_lang_name("Portuguese") == "pt" + + +def parse_country_name(s: Optional[str]) -> Optional[str]: + """ + Parses a country name into a ISO country code (2-char). + + This version copied from the chocula repository. + """ + if not s or s in ("Unknown"): + return None + + s = s.strip() + if s.lower() in ("usa", "new york (state)", "washington (state)"): + return "us" + if s.lower() in ("russia (federation)", "russia"): + return "ru" + if s == "Québec (Province)": + s = "Canada" + if s == "China (Republic : 1949- )": + return "tw" + if s == "Brunei": + return "bn" + if s.startswith("Congo "): + s = "Congo" + if s.lower() == "iran": + return "ir" + if s.lower() == "bermuda islands": + return "bm" + if s.lower() == "burma": + s = "myanmar" + if s.lower() in ("korea (south)", "south korea"): + return "kr" + if s.lower() in ("england", "scotland", "wales"): + return "uk" + s = s.replace(" (Republic)", "").replace(" (Federation)", "") + + try: + country = pycountry.countries.lookup(s) + except LookupError: + country = None + + if country: + return country.alpha_2.lower() + try: + sub = pycountry.subdivisions.lookup(s) + except LookupError: + sub = None + + s = s.replace(" (State)", "").replace(" (Province)", "") + if sub: + return sub.country_code.lower() + + else: + # print(f"unknown country: {s}", file=sys.stderr) + return None + + +def test_parse_country_name(): + assert parse_country_name("") is None + assert parse_country_name("asdf blah") is None + assert parse_country_name("us") == "us" + assert parse_country_name("USA") == "us" + assert parse_country_name("United States of America") == "us" + assert parse_country_name("united States") == "us" + assert parse_country_name("Massachusetts") == "us" + assert parse_country_name("Russia") == "ru" + assert parse_country_name("Japan") == "jp" + +# These are very close, but maybe not exactly 1-to-1 with 639-2? Some mix of +# 2/T and 2/B? +# PubMed/MEDLINE and JSTOR use these MARC codes +# https://www.loc.gov/marc/languages/language_name.html +LANG_MAP_MARC = { + 'afr': 'af', + 'alb': 'sq', + 'amh': 'am', + 'ara': 'ar', + 'arm': 'hy', + 'aze': 'az', + 'ben': 'bn', + 'bos': 'bs', + 'bul': 'bg', + 'cat': 'ca', + 'chi': 'zh', + 'cze': 'cs', + 'dan': 'da', + 'dut': 'nl', + 'eng': 'en', + 'epo': 'eo', + 'est': 'et', + 'fin': 'fi', + 'fre': 'fr', + 'geo': 'ka', + 'ger': 'de', + 'gla': 'gd', + 'gre': 'el', + 'heb': 'he', + 'hin': 'hi', + 'hrv': 'hr', + 'hun': 'hu', + 'ice': 'is', + 'ind': 'id', + 'ita': 'it', + 'jpn': 'ja', + 'kin': 'rw', + 'kor': 'ko', + 'lat': 'la', + 'lav': 'lv', + 'lit': 'lt', + 'mac': 'mk', + 'mal': 'ml', + 'mao': 'mi', + 'may': 'ms', + 'nor': 'no', + 'per': 'fa', + 'per': 'fa', + 'pol': 'pl', + 'por': 'pt', + 'pus': 'ps', + 'rum': 'ro', + 'rus': 'ru', + 'san': 'sa', + 'slo': 'sk', + 'slv': 'sl', + 'spa': 'es', + 'srp': 'sr', + 'swe': 'sv', + 'tha': 'th', + 'tur': 'tr', + 'ukr': 'uk', + 'urd': 'ur', + 'vie': 'vi', + 'wel': 'cy', + +# additions + 'gle': 'ga', # "Irish" (Gaelic) + 'jav': 'jv', # Javanese + 'welsh': 'cy', # Welsh + 'oci': 'oc', # Occitan + +# Don't have ISO 639-1 codes + 'grc': 'el', # Ancient Greek; map to modern greek + 'map': None, # Austronesian (collection) + 'syr': None, # Syriac, Modern + 'gem': None, # Old Saxon + 'non': None, # Old Norse + 'emg': None, # Eastern Meohang + 'neg': None, # Negidal + 'mul': None, # Multiple languages + 'und': None, # Undetermined +} diff --git a/python/fatcat_tools/transforms/elasticsearch.py b/python/fatcat_tools/transforms/elasticsearch.py index dfb5f799..96a5b96b 100644 --- a/python/fatcat_tools/transforms/elasticsearch.py +++ b/python/fatcat_tools/transforms/elasticsearch.py @@ -213,6 +213,9 @@ def release_to_elasticsearch(entity, force_bool=True): t['best_pdf_url'] = best_pdf_url or good_pdf_url or any_pdf_url t['ia_pdf_url'] = ia_pdf_url + if release.ext_ids.doaj: + is_oa = True + if release.license_slug: # TODO: more/better checks here, particularly strict *not* OA licenses if release.license_slug.startswith("CC-"): |