Better guesses for ambiguous quotation marks + quote continuer edge case fix

Author: Ben King

machine/corpora/fallback_quotation_mark_resolver.py

@@ -42,6 +42,19 @@ def _resolve_quotation_mark(
             else:
                 self._issues.add(QuotationMarkResolutionIssue.UNEXPECTED_QUOTATION_MARK)
         else:
+            # Make a reasonable guess about the direction of the quotation mark
+            if (
+                self._last_quotation_mark is None
+                or self._last_quotation_mark.direction is QuotationMarkDirection.CLOSING
+            ):
+                quotation_mark: Optional[QuotationMarkMetadata] = self._resolve_opening_mark(quotation_mark_match)
+                if quotation_mark is not None:
+                    yield quotation_mark
+            else:
+                quotation_mark: Optional[QuotationMarkMetadata] = self._resolve_closing_mark(quotation_mark_match)
+                if quotation_mark is not None:
+                    yield quotation_mark
+
             self._issues.add(QuotationMarkResolutionIssue.AMBIGUOUS_QUOTATION_MARK)
 
     def _is_opening_quotation_mark(

machine/corpora/punctuation_analysis/depth_based_quotation_mark_resolver.py

@@ -176,6 +176,9 @@ def _meets_quote_continuer_prerequisites(
         self,
         quotation_mark_match: QuotationMarkStringMatch,
     ) -> bool:
+        if self._quote_continuer_state.current_depth >= self._quotation_mark_resolver_state.current_depth:
+            return False
+
         if (
             self._settings.should_rely_on_paragraph_markers
             and not quotation_mark_match._text_segment.marker_is_in_preceding_context(UsfmMarkerType.PARAGRAPH)

tests/corpora/test_fallback_quotation_mark_resolver.py

@@ -29,7 +29,7 @@ def test_reset():
     assert len(basic_quotation_mark_resolver._issues) == 0
 
 
-def test_simple_quotation_mark_resolution():
+def test_simple_quotation_mark_resolution_with_no_previous_mark():
     english_quote_convention = STANDARD_QUOTE_CONVENTIONS.get_quote_convention_by_name("standard_english")
     assert english_quote_convention is not None
 
@@ -40,17 +40,75 @@ def test_simple_quotation_mark_resolution():
     actual_resolved_quotation_marks = list(
         basic_quotation_mark_resolver.resolve_quotation_marks(
             [
-                QuotationMarkStringMatch(TextSegment.Builder().set_text('"test text"').build(), 0, 1),
-                QuotationMarkStringMatch(TextSegment.Builder().set_text('"test text"').build(), 10, 11),
+                QuotationMarkStringMatch(TextSegment.Builder().set_text('test " text').build(), 5, 6),
             ]
         )
     )
     expected_resolved_quotation_marks = [
         QuotationMarkMetadata(
-            '"', 1, QuotationMarkDirection.OPENING, TextSegment.Builder().set_text('"test text"').build(), 0, 1
+            '"', 1, QuotationMarkDirection.OPENING, TextSegment.Builder().set_text('test " text').build(), 5, 6
+        ),
+    ]
+
+    assert_resolved_quotation_marks_equal(
+        actual_resolved_quotation_marks,
+        expected_resolved_quotation_marks,
+    )
+
+
+def test_simple_quotation_mark_resolution_with_previous_opening_mark():
+    english_quote_convention = STANDARD_QUOTE_CONVENTIONS.get_quote_convention_by_name("standard_english")
+    assert english_quote_convention is not None
+
+    basic_quotation_mark_resolver = FallbackQuotationMarkResolver(
+        QuotationMarkUpdateResolutionSettings(english_quote_convention.normalize())
+    )
+
+    actual_resolved_quotation_marks = list(
+        basic_quotation_mark_resolver.resolve_quotation_marks(
+            [
+                QuotationMarkStringMatch(TextSegment.Builder().set_text('"test " text').build(), 0, 1),
+                QuotationMarkStringMatch(TextSegment.Builder().set_text('"test " text').build(), 6, 7),
+            ]
+        )
+    )
+    expected_resolved_quotation_marks = [
+        QuotationMarkMetadata(
+            '"', 1, QuotationMarkDirection.OPENING, TextSegment.Builder().set_text('"test " text').build(), 0, 1
+        ),
+        QuotationMarkMetadata(
+            '"', 1, QuotationMarkDirection.CLOSING, TextSegment.Builder().set_text('"test " text').build(), 6, 7
+        ),
+    ]
+
+    assert_resolved_quotation_marks_equal(
+        actual_resolved_quotation_marks,
+        expected_resolved_quotation_marks,
+    )
+
+
+def test_simple_quotation_mark_resolution_with_previous_closing_mark():
+    english_quote_convention = STANDARD_QUOTE_CONVENTIONS.get_quote_convention_by_name("standard_english")
+    assert english_quote_convention is not None
+
+    basic_quotation_mark_resolver = FallbackQuotationMarkResolver(
+        QuotationMarkUpdateResolutionSettings(english_quote_convention.normalize())
+    )
+
+    actual_resolved_quotation_marks = list(
+        basic_quotation_mark_resolver.resolve_quotation_marks(
+            [
+                QuotationMarkStringMatch(TextSegment.Builder().set_text('test" " text').build(), 4, 5),
+                QuotationMarkStringMatch(TextSegment.Builder().set_text('test" " text').build(), 6, 7),
+            ]
+        )
+    )
+    expected_resolved_quotation_marks = [
+        QuotationMarkMetadata(
+            '"', 1, QuotationMarkDirection.CLOSING, TextSegment.Builder().set_text('test" " text').build(), 4, 5
         ),
         QuotationMarkMetadata(
-            '"', 1, QuotationMarkDirection.CLOSING, TextSegment.Builder().set_text('"test text"').build(), 10, 11
+            '"', 1, QuotationMarkDirection.OPENING, TextSegment.Builder().set_text('test" " text').build(), 6, 7
         ),
     ]