diff options
| -rw-r--r-- | python/fatcat_tools/importers/common.py | 60 | ||||
| -rw-r--r-- | python/fatcat_tools/importers/grobid_metadata.py | 18 | ||||
| -rw-r--r-- | python/fatcat_tools/importers/matched.py | 17 |
3 files changed, 66 insertions, 29 deletions
diff --git a/python/fatcat_tools/importers/common.py b/python/fatcat_tools/importers/common.py index 89203a4f..2856b88e 100644 --- a/python/fatcat_tools/importers/common.py +++ b/python/fatcat_tools/importers/common.py @@ -47,6 +47,66 @@ def test_clean(): assert clean('<b>a&b</b>') == '<b>a&b</b>' assert clean('<b>a&b</b>', force_xml=True) == '<b>a&b</b>' +DOMAIN_REL_MAP = { + "archive.org": "repository", + "arxiv.org": "repository", + "babel.hathitrust.org": "repository", + "cds.cern.ch": "repository", + "citeseerx.ist.psu.edu": "repository", + "deepblue.lib.umich.edu": "repository", + "europepmc.org": "repository", + "hal.inria.fr": "repository", + "scielo.isciii.es": "repository", + "www.dtic.mil": "repository", + "www.jstage.jst.go.jp": "repository", + "www.jstor.org": "repository", + "www.ncbi.nlm.nih.gov": "repository", + "www.scielo.br": "repository", + "www.scielo.cl": "repository", + "www.scielo.org.mx": "repository", + "zenodo.org": "repository", + + "academic.oup.com": "publisher", + "cdn.elifesciences.org": "publisher", + "cell.com": "publisher", + "dl.acm.org": "publisher", + "downloads.hindawi.com": "publisher", + "elifesciences.org": "publisher", + "iopscience.iop.org": "publisher", + "journals.plos.org": "publisher", + "link.springer.com": "publisher", + "onlinelibrary.wiley.com": "publisher", + "works.bepress.com": "publisher", + "www.biomedcentral.com": "publisher", + "www.cell.com": "publisher", + "www.nature.com": "publisher", + "www.pnas.org": "publisher", + "www.tandfonline.com": "publisher", + + "www.researchgate.net": "social", + "academia.edu": "social", + + "wayback.archive-it.org": "webarchive", + "web.archive.org": "webarchive", + "archive.is": "webarchive", +} + +def make_rel_url(raw_url, default_link_rel="web"): + # this is where we map specific domains to rel types, and also filter out + # bad domains, invalid URLs, etc + rel = default_link_rel + for domain, domain_rel in DOMAIN_REL_MAP.items(): + if "//{}/".format(domain) in raw_url: + rel = domain_rel + break + return (rel, raw_url) + +def test_make_rel_url(): + assert make_rel_url("http://example.com/thing.pdf")[0] == "web" + assert make_rel_url("http://example.com/thing.pdf", default_link_rel="jeans")[0] == "jeans" + assert make_rel_url("https://web.archive.org/web/*/http://example.com/thing.pdf")[0] == "webarchive" + assert make_rel_url("http://cell.com/thing.pdf")[0] == "publisher" + class EntityImporter: """ Base class for fatcat entity importers. diff --git a/python/fatcat_tools/importers/grobid_metadata.py b/python/fatcat_tools/importers/grobid_metadata.py index 9d95fe0b..25f9fa89 100644 --- a/python/fatcat_tools/importers/grobid_metadata.py +++ b/python/fatcat_tools/importers/grobid_metadata.py @@ -5,7 +5,7 @@ import json import base64 import datetime import fatcat_client -from .common import EntityImporter, clean +from .common import EntityImporter, clean, make_rel_url MAX_ABSTRACT_BYTES=4096 @@ -157,18 +157,6 @@ class GrobidMetadataImporter(EntityImporter): extra=extra) return re - # TODO: make this a common function somewhere - def make_url(self, raw): - rel = self.default_link_rel - # TODO: this is where we could map specific domains to rel types, - # and also filter out bad domains, invalid URLs, etc - if "//archive.org/" in raw or "//arxiv.org/" in raw: - # TODO: special-case the arxiv.org bulk mirror? - rel = "repository" - elif "//web.archive.org/" in raw or "//archive.is/" in raw: - rel = "webarchive" - return fatcat_client.FileEntityUrls(url=raw, rel=rel) - def parse_file_metadata(self, sha1_key, cdx, mimetype, file_size): sha1 = base64.b16encode(base64.b32decode(sha1_key.replace('sha1:', ''))).decode('ascii').lower() @@ -189,9 +177,9 @@ class GrobidMetadataImporter(EntityImporter): original) fe.urls.append( fatcat_client.FileEntityUrls(url=wayback, rel="webarchive")) - original_url = self.make_url(original) + original_url = make_rel_url(original, default_link_rel=self.default_link_rel) if original_url is not None: - fe.urls.append(original_url) + fe.urls.append(fatcat_client.FileEntityUrls(rel=original_url[0], url=original_url[1])) return fe diff --git a/python/fatcat_tools/importers/matched.py b/python/fatcat_tools/importers/matched.py index 2ec6c95d..aca2cc34 100644 --- a/python/fatcat_tools/importers/matched.py +++ b/python/fatcat_tools/importers/matched.py @@ -4,7 +4,7 @@ import json import sqlite3 import itertools import fatcat_client -from .common import EntityImporter, clean +from .common import EntityImporter, clean, make_rel_url class MatchedImporter(EntityImporter): @@ -43,17 +43,6 @@ class MatchedImporter(EntityImporter): self.default_link_rel = kwargs.get("default_link_rel", "web") self.default_mime = kwargs.get("default_mime", None) - def make_url(self, raw): - rel = self.default_link_rel - # TODO: this is where we could map specific domains to rel types, - # and also filter out bad domains, invalid URLs, etc - if "//archive.org/" in raw or "//arxiv.org/" in raw: - # TODO: special-case the arxiv.org bulk mirror? - rel = "repository" - elif "//web.archive.org/" in raw or "//archive.is/" in raw: - rel = "webarchive" - return (rel, raw) - def want(self, raw_record): return True @@ -80,7 +69,7 @@ class MatchedImporter(EntityImporter): # parse URLs and CDX urls = set() for url in obj.get('url', []): - url = self.make_url(url) + url = make_rel_url(url, default_link_rel=self.default_link_rel) if url != None: urls.add(url) for cdx in obj.get('cdx', []): @@ -89,7 +78,7 @@ class MatchedImporter(EntityImporter): cdx['dt'], original) urls.add(("webarchive", wayback)) - url = self.make_url(original) + url = make_rel_url(original, default_link_rel=self.default_link_rel) if url != None: urls.add(url) urls = [fatcat_client.FileEntityUrls(rel, url) for (rel, url) in urls] |