Adding some gnarly 1 pager pdfs from kyle

allenai · Feb 11, 2025 · 91eef27 · 91eef27
1 parent 87cb957
commit 91eef27
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 0 deletions.
diff --git a/tests/gnarly_pdfs/dolma-page-1.pdf b/tests/gnarly_pdfs/dolma-page-1.pdf
diff --git a/tests/gnarly_pdfs/olmo-page-1.pdf b/tests/gnarly_pdfs/olmo-page-1.pdf
diff --git a/tests/test_anchor.py b/tests/test_anchor.py
@@ -163,6 +163,24 @@ def testExcessiveMapAnchor(self):
         print(len(anchor_text))
         self.assertLess(len(anchor_text), 4000)
 
+    def testKyleOnePageAnchors1(self):
+        local_pdf_path = os.path.join(os.path.dirname(__file__), "gnarly_pdfs", "dolma-page-1.pdf")
+
+        anchor_text = get_anchor_text(local_pdf_path, 1, pdf_engine="pdfreport", target_length=6000)
+
+        print(anchor_text)
+        print(len(anchor_text))
+        self.assertLess(len(anchor_text), 6000)
+
+    def testKyleOnePageAnchors2(self):
+        local_pdf_path = os.path.join(os.path.dirname(__file__), "gnarly_pdfs", "olmo-page-1.pdf")
+
+        anchor_text = get_anchor_text(local_pdf_path, 1, pdf_engine="pdfreport", target_length=6000)
+
+        print(anchor_text)
+        print(len(anchor_text))
+        self.assertLess(len(anchor_text), 6000)
+
 
 class BuildSilverTest(unittest.TestCase):
     def testSmallPage(self):