Ensure unique DOI from PubMed API (#71)

* fix: ensure unique DOIs in PubMed

* chore: patch release prep

* fix: skip None

* chore: bump pymed

Author: jannisborn

paperscraper/__init__.py

@@ -1,7 +1,7 @@
 """Initialize the module."""
 
 __name__ = "paperscraper"
-__version__ = "0.2.15"
+__version__ = "0.2.16"
 
 import logging
 import os

paperscraper/pubmed/pubmed.py

@@ -26,6 +26,7 @@
     "date": lambda date: (
         date.strftime("%Y-%m-%d") if isinstance(date, datetime.date) else date
     ),
+    "doi": lambda doi: doi.split("\n")[0] if isinstance(doi, str) else doi,
 }
 
 

paperscraper/pubmed/tests/test_pubmed.py

@@ -8,7 +8,6 @@
 
 
 class TestPubMed:
-
     def test_get_and_dump_pubmed(self):
         with tempfile.TemporaryDirectory() as temp_dir:
             output_filepath = os.path.join(temp_dir, "tmp.jsonl")
@@ -30,3 +29,8 @@ def test_email(self):
         df = get_pubmed_papers(query, fields=["emails", "title", "authors"])
         assert "emails" in df.columns
 
+    def test_doi(self):
+        query = "CRISPR-based gene editing in plants: Focus on reagents and their delivery tools"
+        df = get_pubmed_papers(query, fields=["doi", "title", "authors"])
+        for i, r in df.iterrows():
+            assert "\n" not in r.doi

requirements.txt

@@ -1,5 +1,5 @@
 arxiv>=1.4.7
-pymed-paperscraper>=1.0.2
+pymed-paperscraper>=1.0.3
 pandas>=1.0.4
 requests==2.32.0
 tqdm>=4.51.0

setup.py

@@ -28,7 +28,7 @@
     license="MIT",
     install_requires=[
         "arxiv>=1.4.2",
-        "pymed-paperscraper>=1.0.2",
+        "pymed-paperscraper>=1.0.3",
         "pandas",
         "requests",
         "tqdm",