Adding empty anchor support

pdelfin/prompts/anchor.py

@@ -259,6 +259,9 @@ def _linearize_pdf_report(report: PageReport, max_length: int = 4000) -> str:
     result = ""
     result += f"Page dimensions: {report.mediabox.x1:.1f}x{report.mediabox.y1:.1f}\n"
 
+    if max_length < 20:
+        return result
+
     images = _merge_image_elements(report.image_elements)
 
     # Process image elements

tests/test_anchor.py

@@ -112,6 +112,21 @@ def testTobaccoPaperMissingParagraphs(self):
         print(len(anchor_text))
         self.assertLess(len(anchor_text), 4000)
 
+    def testAnchorOtherLengths(self):
+        local_pdf_path = os.path.join(os.path.dirname(__file__), "gnarly_pdfs", "tobacco_missed_tokens_pg1.pdf")
+
+        anchor_text = get_anchor_text(local_pdf_path, 1, pdf_engine="pdfreport", target_length=2000)
+
+        print(anchor_text)
+        print(len(anchor_text))
+        self.assertLess(len(anchor_text), 2000)
+
+        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 testFailingAnchor(self):
         local_pdf_path = os.path.join(os.path.dirname(__file__), "gnarly_pdfs", "failing_anchor_pg4.pdf")
 
@@ -121,6 +136,13 @@ def testFailingAnchor(self):
         print(len(anchor_text))
         self.assertLess(len(anchor_text), 4000)
 
+    def testEmptyAnchor(self):
+        local_pdf_path = os.path.join(os.path.dirname(__file__), "gnarly_pdfs", "tobacco_missed_tokens_pg1.pdf")
+
+        anchor_text = get_anchor_text(local_pdf_path, 1, pdf_engine="pdfreport", target_length=0)
+
+        self.assertEqual(anchor_text.strip(), "Page dimensions: 612.0x792.0")
+
 class BuildSilverTest(unittest.TestCase):
     def testSmallPage(self):
         local_pdf_path = os.path.join(os.path.dirname(__file__), "gnarly_pdfs", "small_page_size.pdf")