Fix GeoSearch elink using retmode xml (#200)

* fix GeoSearch elink using retmode xml

* Format

* Update sphinx

---------

Co-authored-by: Saket Choudhary <[email protected]>

Author: jggatter

.github/workflows/pull_request.yml

@@ -39,5 +39,5 @@ jobs:
         codecov
     - uses: ammaraskar/sphinx-action@master
       with:
-        pre-build-command: "pip install -r requirements.txt && pip install . && pip install sphinx myst-parser && pip install sphinxcontrib-gtagjs ipython numpydoc sphinx-tabs sphinx_rtd_theme nbsphinx ipython pydata-sphinx-theme nbsphinx-link sphinx-panels"
+        pre-build-command: "pip install -r requirements.txt && pip install . && pip install -U sphinx myst-parser && pip install sphinxcontrib-gtagjs ipython numpydoc sphinx-tabs sphinx_rtd_theme nbsphinx ipython pydata-sphinx-theme nbsphinx-link sphinx-panels"
         docs-folder: "docs/"

.gitignore

@@ -108,3 +108,5 @@ venv.bak/
 .mypy_cache/
 *.sqlite
 *.sqlite.gz
+
+geoweb_downloads/

pysradb/search.py

@@ -1660,6 +1660,7 @@ def _format_request(self):
         return payload
 
     def search(self):
+        """Sends the user query via requests to SRA, GEO DataSets, or both"""
         if not self.search_geo:
             super().search()
         else:
@@ -1679,7 +1680,7 @@ def search(self):
                 elink_payload = {
                     "dbfrom": "gds",
                     "db": "sra",
-                    "retmode": "json",
+                    "retmode": "xml",
                     "query_key": query_key,
                     "WebEnv": web_env,
                 }
@@ -1690,9 +1691,12 @@ def search(self):
                 )
                 r.raise_for_status()
                 try:
-                    data = r.json()
-                    uids_from_geo = data["linksets"][0]["linksetdbs"][0]["links"]
-                except (JSONDecodeError, KeyError, IndexError):
+                    root = Et.fromstring(r.text)
+                    uids_from_geo = [
+                        elem.text
+                        for elem in root.findall(".//LinkSet/LinkSetDb/Link/Id")
+                    ]
+                except (Et.ParseError, TypeError, ValueError):
                     uids_from_geo = []
                 # Step 2: Retrieve list of uids from SRA and
                 # Find the intersection of both lists of uids

tests/test_search.py

@@ -1179,8 +1179,10 @@ def test_geo_search_1():
     instance = GeoSearch(3, 1000, geo_query="human")
     instance.search()
     df = instance.get_df()
-    df = instance.get_df()["experiment_accession"].to_list()
-    assert len(df) > 10
+    assert not df.empty
+
+    experiment_accessions = instance.get_df()["experiment_accession"].to_list()
+    assert len(experiment_accessions) > 10
     # with open("./tests/data/test_search/geo_search_test1.txt", "r") as f:
     #    expected_accessions = f.read().splitlines()