Test paper discovery (#66)

* Add tests for discover jmlr and miniconf
* Add tests for discover openalex
* Add openreview tests
* Add test on pmlr
* Add semantic scholar test
* Add v3 to CI
* Disable coverage for now

Author: satyaog

.github/workflows/python-package.yml

@@ -2,9 +2,9 @@ name: Python package
 
 on:
   push:
-    branches: [ master ]
+    branches: [ master, v3 ]
   pull_request:
-    branches: [ master ]
+    branches: [ master, v3 ]
 
 jobs:
   lint:
@@ -36,7 +36,7 @@ jobs:
         - python: '3.12'
           coverage: false
         - python: '3.13'
-          coverage: true
+          coverage: false
     steps:
     - name: Check out the code
       uses: actions/checkout@v3

src/paperoni/discovery/miniconf.py

@@ -1,3 +1,4 @@
+import math
 import re
 from datetime import datetime
 
@@ -43,9 +44,11 @@ def convert_paper(self, data, conference=None, venue_date=None):
                 name=author_data.get("fullname", ""),
                 aliases=[],
                 links=[
-                    Link(type="profile", link=author_data["url"])
-                    if author_data.get("url")
-                    else None
+                    (
+                        Link(type="profile", link=author_data["url"])
+                        if author_data.get("url")
+                        else None
+                    )
                 ],
             )
             author.links = [link for link in author.links if link is not None]
@@ -122,6 +125,9 @@ def expand_base(uri):
             links.add(Link(type="pdf", link=url))
         if url := expand_base(data.get("virtualsite_url")):
             links.add(Link(type="abstract", link=url))
+        links = sorted(
+            links, key=lambda x: ({"pdf": 0}.get(x.type, math.inf), x.type, x.link)
+        )
 
         # Add eventmedia links
         for media in data.get("eventmedia", []):

src/paperoni/discovery/openalex.py

@@ -313,9 +313,6 @@ def query(
         # Title of the paper (mutually exclusive with "exact-title")
         # [alias: -t]
         title: str = None,
-        # Exact title to query (mutually exclusive with "title")
-        # [alias: -T]
-        exact_title: str = None,
         # Page of results to display (start at 1). Need argument "per_page". By default, all results are displayed.
         page: int = None,
         # Number of results to display per page. Need argument "page". By default, all results are displayed.
@@ -354,14 +351,8 @@ def query(
                     return
             filters.append(f"institutions.id:{institution_id}")
 
-        if title and exact_title:
-            raise QueryError("Cannot query both title and exact title")
-        elif title:
+        if title:
             filters.append(f"display_name.search:{title}")
-        elif exact_title:
-            # No stemming, and quotation mark around title, to try to get exact title
-            # https://docs.openalex.org/how-to-use-the-api/get-lists-of-entities/search-entities#boolean-searches
-            filters.append(f'display_name.search.no_stem:"{exact_title}"')
 
         params = {}
         if filters:

src/paperoni/discovery/openreview.py

@@ -525,9 +525,23 @@ def query(
                 block_size=block_size,
                 limit=limit,
             )
+
             has_papers = False
-            for paper in q:
-                has_papers = True
-                yield paper
+
+            exception = None
+            try:
+                for paper in q:
+                    has_papers = True
+                    yield paper
+
+            except openreview.OpenReviewException as e:
+                # Try the next API version while holding the exception
+                exception = e
+                continue
+
             if has_papers:
                 break
+
+        else:
+            if exception is not None:
+                raise exception

src/paperoni/model/classes.py

@@ -51,12 +51,12 @@ def assimilate_date(date, infer_precision=True):
                 if year < 100:
                     year += 2000
                 return {
-                    "date": f"{year}-01-01 00:00",
+                    "date": datetime(year, 1, 1).date(),
                     "date_precision": DatePrecision.year,
                 }
             case str() as year if re.match("^[0-9]{4}$", date):
                 return {
-                    "date": f"{year}-01-01 00:00",
+                    "date": datetime(year, 1, 1).date(),
                     "date_precision": DatePrecision.year,
                 }
             case str() if m := re.match("^(....)-(..)-(..).*", date):
@@ -70,15 +70,15 @@ def assimilate_date(date, infer_precision=True):
                         else:
                             precision = DatePrecision.day
                         return {
-                            "date": f"{year}-{month}-{day} 00:00",
+                            "date": datetime(year, month, day).date(),
                             "date_precision": precision,
                         }
                     case _:  # pragma: no cover
                         assert False
             case None | "":
                 return (
                     {
-                        "date": "2000-01-01 00:00",
+                        "date": datetime(2000, 1, 1).date(),
                         "date_precision": DatePrecision.unknown,
                     }
                     if infer_precision

tests/conftest.py

@@ -0,0 +1,10 @@
+from pathlib import Path
+
+import gifnoc
+from pytest import fixture
+
+
+@fixture(scope="session", autouse=True)
+def set_config():
+    with gifnoc.use(Path(__file__).resolve().parent.parent / "config/basic.yaml"):
+        yield

tests/discovery/__init__.py

@@ -0,0 +1,21 @@
+from pytest_regressions.data_regression import DataRegressionFixture
+
+from paperoni.discovery.base import PaperInfo
+from paperoni.discovery.jmlr import JMLR
+
+from ..utils import check_papers
+
+
+def test_query(data_regression: DataRegressionFixture):
+    discoverer = JMLR()
+
+    assert "v24" in discoverer.list_volumes()
+
+    papers: list[PaperInfo] = sorted(
+        discoverer.query(volume="v24", name="Yoshua Bengio"),
+        key=lambda x: x.paper.title,
+    )
+
+    assert papers, "No papers found for Yoshua Bengio in v24"
+
+    check_papers(data_regression, papers)

tests/discovery/test_jmlr.py

@@ -0,0 +1,132 @@
+- key: jmlr:vv24:benchmarking_graph_neural_networks
+  paper:
+    abstract: null
+    authors:
+    - affiliations: []
+      author:
+        aliases: []
+        links: []
+        name: Vijay Prakash Dwivedi
+      display_name: Vijay Prakash Dwivedi
+    - affiliations: []
+      author:
+        aliases: []
+        links: []
+        name: Chaitanya K. Joshi
+      display_name: Chaitanya K. Joshi
+    - affiliations: []
+      author:
+        aliases: []
+        links: []
+        name: Anh Tuan Luu
+      display_name: Anh Tuan Luu
+    - affiliations: []
+      author:
+        aliases: []
+        links: []
+        name: Thomas Laurent
+      display_name: Thomas Laurent
+    - affiliations: []
+      author:
+        aliases: []
+        links: []
+        name: Yoshua Bengio
+      display_name: Yoshua Bengio
+    - affiliations: []
+      author:
+        aliases: []
+        links: []
+        name: Xavier Bresson
+      display_name: Xavier Bresson
+    flags: []
+    links:
+    - link: https://jmlr.org/papers/v24/22-0567.html
+      type: abstract.official
+    - link: https://jmlr.org/papers/volume24/22-0567/22-0567.pdf
+      type: pdf.official
+    - link: https://jmlr.org/papers/v24/22-0567.bib
+      type: bibtex
+    releases:
+    - pages: 1-48
+      status: published
+      venue:
+        aliases: []
+        date: '2023-01-01'
+        date_precision: 1
+        links: []
+        name: Journal of Machine Learning Research
+        open: false
+        peer_reviewed: true
+        publisher: JMLR
+        series: JMLR
+        type: journal
+        volume: null
+    title: Benchmarking Graph Neural Networks
+    topics: []
+  update_key: null
+- key: jmlr:vv24:gflownet_foundations
+  paper:
+    abstract: null
+    authors:
+    - affiliations: []
+      author:
+        aliases: []
+        links: []
+        name: Yoshua Bengio
+      display_name: Yoshua Bengio
+    - affiliations: []
+      author:
+        aliases: []
+        links: []
+        name: Salem Lahlou
+      display_name: Salem Lahlou
+    - affiliations: []
+      author:
+        aliases: []
+        links: []
+        name: Tristan Deleu
+      display_name: Tristan Deleu
+    - affiliations: []
+      author:
+        aliases: []
+        links: []
+        name: Edward J. Hu
+      display_name: Edward J. Hu
+    - affiliations: []
+      author:
+        aliases: []
+        links: []
+        name: Mo Tiwari
+      display_name: Mo Tiwari
+    - affiliations: []
+      author:
+        aliases: []
+        links: []
+        name: Emmanuel Bengio
+      display_name: Emmanuel Bengio
+    flags: []
+    links:
+    - link: https://jmlr.org/papers/v24/22-0364.html
+      type: abstract.official
+    - link: https://jmlr.org/papers/volume24/22-0364/22-0364.pdf
+      type: pdf.official
+    - link: https://jmlr.org/papers/v24/22-0364.bib
+      type: bibtex
+    releases:
+    - pages: 1-55
+      status: published
+      venue:
+        aliases: []
+        date: '2023-01-01'
+        date_precision: 1
+        links: []
+        name: Journal of Machine Learning Research
+        open: false
+        peer_reviewed: true
+        publisher: JMLR
+        series: JMLR
+        type: journal
+        volume: null
+    title: GFlowNet Foundations
+    topics: []
+  update_key: null

tests/discovery/test_jmlr/test_query.yml

@@ -0,0 +1,55 @@
+import itertools
+
+import pytest
+from pytest_regressions.data_regression import DataRegressionFixture
+
+from paperoni.discovery.base import PaperInfo
+from paperoni.discovery.miniconf import MiniConf, conference_urls
+
+from ..utils import check_papers, iter_affiliations
+
+
+@pytest.mark.parametrize(
+    ["conference", "query_params"],
+    itertools.product(
+        conference_urls, [{"affiliation": "mila"}, {"author": "Yoshua Bengio"}]
+    ),
+)
+def test_query(data_regression: DataRegressionFixture, conference, query_params):
+    discoverer = MiniConf()
+
+    papers: list[PaperInfo] = sorted(
+        discoverer.query(conference, year=2024, **query_params),
+        key=lambda x: x.paper.title,
+    )
+
+    match_found = False
+
+    for param in query_params:
+        match param:
+            case "affiliation":
+                assert all(
+                    any(
+                        query_params["affiliation"].lower() in aff.name.lower()
+                        for aff in iter_affiliations(paper.paper)
+                    )
+                    for paper in papers
+                ), (
+                    f"Some papers do not contain the affiliation {query_params['affiliation']=}"
+                )
+                match_found = True
+
+            case "author":
+                assert all(
+                    any(
+                        query_params["author"].lower() in author.author.name.lower()
+                        for author in paper.paper.authors
+                    )
+                    for paper in papers
+                ), f"Some papers do not contain the author {query_params['author']=}"
+                match_found = True
+
+    if not match_found:
+        assert False, f"Unknown query parameters: {query_params=}"
+
+    check_papers(data_regression, papers)