added escape_backlashes function with test (#858)

Author: JoaquinPolonuer

paperqa/core.py

@@ -20,10 +20,14 @@ def llm_parse_json(text: str) -> dict:
     def escape_newlines(match: re.Match) -> str:
         return match.group(0).replace("\n", "\\n")
 
+    def escape_double_backslashes(match: re.Match) -> str:
+        return match.group(0).replace("\\", "\\\\")
+
     # Match anything between double quotes
     # including escaped quotes and other escaped characters.
     # https://regex101.com/r/VFcDmB/1
     pattern = r'"(?:[^"\\]|\\.)*"'
+    ptext = re.sub(pattern, escape_double_backslashes, ptext)
     ptext = re.sub(pattern, escape_newlines, ptext)
     try:
         return json.loads(ptext)

tests/test_paperqa.py

@@ -376,35 +376,44 @@ def test_extract_score() -> None:
 
 
 @pytest.mark.parametrize(
-    "example",
+    ("example", "expected"),
     [
-        """Sure here is the json you asked for!
+        (
+            """Sure here is the json you asked for!
 
     {
     "example": "json"
     }
 
     Did you like it?""",
-        '{"example": "json"}',
-        """
+            {"example": "json"},
+        ),
+        (
+            """
 ```json
 {
     "example": "json"
 }
 ```
 
 I have written the json you asked for.""",
-        """
+            {"example": "json"},
+        ),
+        (
+            """
 
 {
     "example": "json"
 }
 
 """,
+            {"example": "json"},
+        ),
+        ('{"example": "\\json"}', {"example": "\\json"}),
     ],
 )
-def test_llm_parse_json(example: str) -> None:
-    assert llm_parse_json(example) == {"example": "json"}
+def test_llm_parse_json(example: str, expected: dict) -> None:
+    assert llm_parse_json(example) == expected
 
 
 def test_llm_parse_json_newlines() -> None: