aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorBryan Newbold <bnewbold@archive.org>2021-10-27 19:10:35 -0700
committerBryan Newbold <bnewbold@archive.org>2021-10-27 19:10:35 -0700
commita0e275a4bad46ef41585f0207d6dfa1e3c38bc35 (patch)
tree92dead8a85e6ff38808beefada8a42693261ceff
parent40adf5ed09d917b8a4b8f75680bbf90c147848b3 (diff)
downloadsandcrawler-a0e275a4bad46ef41585f0207d6dfa1e3c38bc35.tar.gz
sandcrawler-a0e275a4bad46ef41585f0207d6dfa1e3c38bc35.zip
remove grobid2json helper file, replace with grobid_tei_xml
-rw-r--r--python/.coveragerc1
-rwxr-xr-xpython/grobid2json.py210
-rwxr-xr-xpython/grobid_tool.py6
-rw-r--r--python/sandcrawler/grobid.py7
-rw-r--r--python/sandcrawler/persist.py2
-rwxr-xr-xpython/scripts/grobid_affiliations.py6
-rw-r--r--python/tests/test_grobid2json.py14
7 files changed, 22 insertions, 224 deletions
diff --git a/python/.coveragerc b/python/.coveragerc
index 67053a7..51038d6 100644
--- a/python/.coveragerc
+++ b/python/.coveragerc
@@ -2,4 +2,3 @@
omit = tests/*
source =
sandcrawler
- grobid2json
diff --git a/python/grobid2json.py b/python/grobid2json.py
deleted file mode 100755
index d92b351..0000000
--- a/python/grobid2json.py
+++ /dev/null
@@ -1,210 +0,0 @@
-#!/usr/bin/env python3
-"""
-NB: adapted to work as a library for PDF extraction. Will probably be
-re-written eventually to be correct, complete, and robust; this is just a
-first iteration.
-
-This script tries to extract everything from a GROBID TEI XML fulltext dump:
-
-- header metadata
-- affiliations
-- references (with context)
-- abstract
-- fulltext
-- tables, figures, equations
-
-A flag can be specified to disable copyright encumbered bits (--no-emcumbered):
-
-- abstract
-- fulltext
-- tables, figures, equations
-
-Prints JSON to stdout, errors to stderr
-"""
-
-import argparse
-import io
-import json
-import xml.etree.ElementTree as ET
-from typing import Any, AnyStr, Dict, List, Optional
-
-xml_ns = "http://www.w3.org/XML/1998/namespace"
-ns = "http://www.tei-c.org/ns/1.0"
-
-
-def all_authors(elem: Optional[ET.Element]) -> List[Dict[str, Any]]:
- if not elem:
- return []
- names = []
- for author in elem.findall(".//{%s}author" % ns):
- pn = author.find("./{%s}persName" % ns)
- if not pn:
- continue
- given_name = pn.findtext("./{%s}forename" % ns) or None
- surname = pn.findtext("./{%s}surname" % ns) or None
- full_name = " ".join(pn.itertext())
- obj: Dict[str, Any] = dict(name=full_name)
- if given_name:
- obj["given_name"] = given_name
- if surname:
- obj["surname"] = surname
- ae = author.find("./{%s}affiliation" % ns)
- if ae:
- affiliation: Dict[str, Any] = dict()
- for on in ae.findall("./{%s}orgName" % ns):
- on_type = on.get("type")
- if on_type:
- affiliation[on_type] = on.text
- addr_e = ae.find("./{%s}address" % ns)
- if addr_e:
- address = dict()
- for t in addr_e:
- address[t.tag.split("}")[-1]] = t.text
- if address:
- affiliation["address"] = address
- # affiliation['address'] = {
- # 'post_code': addr.findtext('./{%s}postCode' % ns) or None,
- # 'settlement': addr.findtext('./{%s}settlement' % ns) or None,
- # 'country': addr.findtext('./{%s}country' % ns) or None,
- # }
- obj["affiliation"] = affiliation
- names.append(obj)
- return names
-
-
-def journal_info(elem: ET.Element) -> Dict[str, Any]:
- journal = dict()
- journal["name"] = elem.findtext(".//{%s}monogr/{%s}title" % (ns, ns))
- journal["publisher"] = elem.findtext(".//{%s}publicationStmt/{%s}publisher" % (ns, ns))
- if journal["publisher"] == "":
- journal["publisher"] = None
- journal["issn"] = elem.findtext('.//{%s}idno[@type="ISSN"]' % ns)
- journal["eissn"] = elem.findtext('.//{%s}idno[@type="eISSN"]' % ns)
- journal["volume"] = elem.findtext('.//{%s}biblScope[@unit="volume"]' % ns)
- journal["issue"] = elem.findtext('.//{%s}biblScope[@unit="issue"]' % ns)
- keys = list(journal.keys())
-
- # remove empty/null keys
- for k in keys:
- if not journal[k]:
- journal.pop(k)
- return journal
-
-
-def biblio_info(elem: ET.Element) -> Dict[str, Any]:
- ref: Dict[str, Any] = dict()
- ref["id"] = elem.attrib.get("{http://www.w3.org/XML/1998/namespace}id")
- # Title stuff is messy in references...
- ref["title"] = elem.findtext(".//{%s}analytic/{%s}title" % (ns, ns))
- other_title = elem.findtext(".//{%s}monogr/{%s}title" % (ns, ns))
- if other_title:
- if ref["title"]:
- ref["journal"] = other_title
- else:
- ref["journal"] = None
- ref["title"] = other_title
- ref["authors"] = all_authors(elem)
- ref["publisher"] = elem.findtext(".//{%s}publicationStmt/{%s}publisher" % (ns, ns))
- if ref["publisher"] == "":
- ref["publisher"] = None
- date = elem.find('.//{%s}date[@type="published"]' % ns)
- ref["date"] = (date is not None) and date.attrib.get("when")
- ref["volume"] = elem.findtext('.//{%s}biblScope[@unit="volume"]' % ns)
- ref["issue"] = elem.findtext('.//{%s}biblScope[@unit="issue"]' % ns)
- el = elem.find(".//{%s}ptr[@target]" % ns)
- if el is not None:
- ref["url"] = el.attrib["target"]
- # Hand correction
- if ref["url"].endswith(".Lastaccessed"):
- ref["url"] = ref["url"].replace(".Lastaccessed", "")
- else:
- ref["url"] = None
- return ref
-
-
-def teixml2json(content: AnyStr, encumbered: bool = True) -> Dict[str, Any]:
-
- if isinstance(content, str):
- tree = ET.parse(io.StringIO(content))
- elif isinstance(content, bytes):
- tree = ET.parse(io.BytesIO(content))
-
- info: Dict[str, Any] = dict()
-
- # print(content)
- # print(content.getvalue())
- tei = tree.getroot()
-
- header = tei.find(".//{%s}teiHeader" % ns)
- if header is None:
- raise ValueError("XML does not look like TEI format")
- application_tag = header.findall(".//{%s}appInfo/{%s}application" % (ns, ns))[0]
- info["grobid_version"] = application_tag.attrib["version"].strip()
- info["grobid_timestamp"] = application_tag.attrib["when"].strip()
- info["title"] = header.findtext(".//{%s}analytic/{%s}title" % (ns, ns))
- info["authors"] = all_authors(header.find(".//{%s}sourceDesc/{%s}biblStruct" % (ns, ns)))
- info["journal"] = journal_info(header)
- date = header.find('.//{%s}date[@type="published"]' % ns)
- info["date"] = (date is not None) and date.attrib.get("when")
- info["fatcat_release"] = header.findtext('.//{%s}idno[@type="fatcat"]' % ns)
- info["doi"] = header.findtext('.//{%s}idno[@type="DOI"]' % ns)
- if info["doi"]:
- info["doi"] = info["doi"].lower()
-
- refs = []
- for (i, bs) in enumerate(tei.findall(".//{%s}listBibl/{%s}biblStruct" % (ns, ns))):
- ref = biblio_info(bs)
- ref["index"] = i
- refs.append(ref)
- info["citations"] = refs
-
- text = tei.find(".//{%s}text" % (ns))
- # print(text.attrib)
- if text and text.attrib.get("{%s}lang" % xml_ns):
- info["language_code"] = text.attrib["{%s}lang" % xml_ns] # xml:lang
-
- if encumbered:
- el = tei.find(".//{%s}profileDesc/{%s}abstract" % (ns, ns))
- info["abstract"] = (el or None) and " ".join(el.itertext()).strip()
- el = tei.find(".//{%s}text/{%s}body" % (ns, ns))
- info["body"] = (el or None) and " ".join(el.itertext()).strip()
- el = tei.find('.//{%s}back/{%s}div[@type="acknowledgement"]' % (ns, ns))
- info["acknowledgement"] = (el or None) and " ".join(el.itertext()).strip()
- el = tei.find('.//{%s}back/{%s}div[@type="annex"]' % (ns, ns))
- info["annex"] = (el or None) and " ".join(el.itertext()).strip()
-
- # remove empty/null keys
- keys = list(info.keys())
- for k in keys:
- if not info[k]:
- info.pop(k)
- return info
-
-
-def main() -> None: # pragma no cover
- parser = argparse.ArgumentParser(
- formatter_class=argparse.ArgumentDefaultsHelpFormatter,
- description="GROBID TEI XML to JSON",
- usage="%(prog)s [options] <teifile>...",
- )
- parser.add_argument(
- "--no-encumbered",
- action="store_true",
- help="don't include ambiguously copyright encumbered fields (eg, abstract, body)",
- )
- parser.add_argument("teifiles", nargs="+")
-
- args = parser.parse_args()
-
- for filename in args.teifiles:
- content = open(filename, "r").read()
- print(
- json.dumps(
- teixml2json(content, encumbered=(not args.no_encumbered)),
- sort_keys=True,
- )
- )
-
-
-if __name__ == "__main__": # pragma no cover
- main()
diff --git a/python/grobid_tool.py b/python/grobid_tool.py
index f85d243..f99a78b 100755
--- a/python/grobid_tool.py
+++ b/python/grobid_tool.py
@@ -12,7 +12,8 @@ import argparse
import json
import sys
-from grobid2json import teixml2json
+from grobid_tei_xml import parse_document_xml
+
from sandcrawler import *
@@ -75,7 +76,8 @@ def run_transform(args):
if args.metadata_only:
out = grobid_client.metadata(line)
else:
- out = teixml2json(line["tei_xml"])
+ tei_doc = parse_document_xml(line["tei_xml"])
+ out = tei_doc.to_legacy_dict()
if out:
if "source" in line:
out["source"] = line["source"]
diff --git a/python/sandcrawler/grobid.py b/python/sandcrawler/grobid.py
index 26918f6..37c4ea1 100644
--- a/python/sandcrawler/grobid.py
+++ b/python/sandcrawler/grobid.py
@@ -1,8 +1,7 @@
from typing import Any, Dict, Optional
import requests
-
-from grobid2json import teixml2json
+from grobid_tei_xml import parse_document_xml
from .ia import WaybackClient
from .misc import gen_file_metadata
@@ -71,7 +70,9 @@ class GrobidClient(object):
def metadata(self, result: Dict[str, Any]) -> Optional[Dict[str, Any]]:
if result["status"] != "success":
return None
- tei_json = teixml2json(result["tei_xml"], encumbered=False)
+ tei_doc = parse_document_xml(result["tei_xml"])
+ tei_doc.remove_encumbered()
+ tei_json = tei_doc.to_legacy_dict()
meta = dict()
biblio = dict()
for k in (
diff --git a/python/sandcrawler/persist.py b/python/sandcrawler/persist.py
index c8c0c33..f50b9d1 100644
--- a/python/sandcrawler/persist.py
+++ b/python/sandcrawler/persist.py
@@ -395,7 +395,7 @@ class PersistGrobidWorker(SandcrawlerWorker):
)
self.counts["s3-put"] += 1
- # enhance with teixml2json metadata, if available
+ # enhance with GROBID TEI-XML metadata, if available
try:
metadata = self.grobid.metadata(r)
except xml.etree.ElementTree.ParseError as xml_e:
diff --git a/python/scripts/grobid_affiliations.py b/python/scripts/grobid_affiliations.py
index b01e46a..90a0f77 100755
--- a/python/scripts/grobid_affiliations.py
+++ b/python/scripts/grobid_affiliations.py
@@ -12,7 +12,7 @@ Run in bulk like:
import json
import sys
-from grobid2json import teixml2json
+from grobid_tei_xml import parse_document_xml
def parse_hbase(line):
@@ -38,7 +38,9 @@ def run(mode="hbase"):
else:
raise NotImplementedError("parse mode: {}".format(mode))
- obj = teixml2json(tei_xml, encumbered=False)
+ tei_doc = parse_document_xml(tei_xml)
+ tei_doc.remove_encumbered()
+ obj = tei_doc.to_legacy_dict()
affiliations = []
for author in obj["authors"]:
diff --git a/python/tests/test_grobid2json.py b/python/tests/test_grobid2json.py
index 98888e8..b00a88d 100644
--- a/python/tests/test_grobid2json.py
+++ b/python/tests/test_grobid2json.py
@@ -2,23 +2,27 @@ import json
import xml
import pytest
-
-from grobid2json import *
+from grobid_tei_xml import parse_document_xml
def test_small_xml():
+ """
+ This used to be a test of grobid2json; now it is a compatability test for
+ the to_legacy_dict() feature of grobid_tei_xml.
+ """
with open("tests/files/small.xml", "r") as f:
tei_xml = f.read()
with open("tests/files/small.json", "r") as f:
json_form = json.loads(f.read())
- assert teixml2json(tei_xml) == json_form
+ tei_doc = parse_document_xml(tei_xml)
+ assert tei_doc.to_legacy_dict() == json_form
def test_invalid_xml():
with pytest.raises(xml.etree.ElementTree.ParseError):
- teixml2json("this is not XML")
+ parse_document_xml("this is not XML")
with pytest.raises(ValueError):
- teixml2json("<xml></xml>")
+ parse_document_xml("<xml></xml>")