From 9ad2f5ced43c98600e222959f7a950a33a095a0f Mon Sep 17 00:00:00 2001
From: s-heppner <mail@s-heppner.com>
Date: Sun, 4 May 2025 19:50:59 +0200
Subject: [PATCH] parser.py: Allow equal signs in field values

Previously, the parser did not account for the possibility of equal
signs (`=`) in the values of fields and would result in crashing the
script with a non descriptive error message, due to expecting a simple
`string.split("=")` to work.

However, it is completely possible to have equal signs in the value,
for example in an URL field: `URL = "https://example.org/query?x=1"`.

This adapts the parsing logic, making it stable against equal signs in
field values. Furthermore, it cleans up the `BibTeXEntry.from_string`
method a little bit and adds additional test cases to the unittests.

Fixes #3
---
 bibtex_linter/parser.py | 20 +++++++++++++-----
 test/test_parser.py     | 47 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 62 insertions(+), 5 deletions(-)

diff --git a/bibtex_linter/parser.py b/bibtex_linter/parser.py
index f52b50d..1930ce8 100644
--- a/bibtex_linter/parser.py
+++ b/bibtex_linter/parser.py
@@ -1,4 +1,4 @@
-from typing import List, Dict
+from typing import List, Dict, Tuple
 import dataclasses
 import enum
 import re
@@ -85,10 +85,7 @@ def from_string(cls, entry_string: str) -> "BibTeXEntry":
         raw_fields = cls._split_fields(entry_string)
         fields: Dict[str, str] = {}
         for raw_field in raw_fields:
-            raw_key, raw_value = raw_field.split("=")
-            # Clean up key and value
-            key = raw_key.strip(" ").lower()
-            value = cls._parse_field_value(raw_value)
+            key, value = cls._split_field_into_key_and_value(raw_field)
             fields[key] = value
 
         return BibTeXEntry(
@@ -157,6 +154,19 @@ def _parse_field_value(raw_value: str) -> str:
 
         return raw_value
 
+    @staticmethod
+    def _split_field_into_key_and_value(raw_field: str) -> Tuple[str, str]:
+        """
+        Splits a field, such as `author = {{John Doe}},` into the field's key and value and cleans up both.
+
+        :param raw_field:
+        :return:
+        """
+        parts = raw_field.split("=", 1)
+        key = parts[0].strip().lower()
+        value = parts[1].strip() if len(parts) > 1 else ""
+        return key, BibTeXEntry._parse_field_value(value)
+
 
 def split_entries(raw_content: str) -> List[str]:
     """
diff --git a/test/test_parser.py b/test/test_parser.py
index fd40738..497b787 100644
--- a/test/test_parser.py
+++ b/test/test_parser.py
@@ -110,6 +110,53 @@ def test_split_fields_missing_open_brace(self) -> None:
         with self.assertRaises(KeyError):
             BibTeXEntry._split_fields(entry)
 
+    def test_field_with_equals_in_value(self) -> None:
+        bibtex_string = """@misc{test_entry,
+          note = {This URL has equals: https://example.com/?id=123&lang=en},
+        }"""
+        entry = BibTeXEntry.from_string(bibtex_string)
+        actual = entry.fields.get("note")
+        expected = "This URL has equals: https://example.com/?id=123&lang=en"
+        self.assertEqual(expected, actual)
+
+    def test_double_braces(self) -> None:
+        bibtex_string = """@misc{test_entry,
+          title = {{Title with {{extra}} braces}},
+        }"""
+        entry = BibTeXEntry.from_string(bibtex_string)
+        actual = entry.fields.get("title")
+        expected = "Title with {{extra}} braces"
+        self.assertEqual(expected, actual)
+
+    def test_quoted_field(self) -> None:
+        bibtex_string = """@misc{test_entry,
+          author = "Jane Doe",
+        }"""
+        entry = BibTeXEntry.from_string(bibtex_string)
+        actual = entry.fields.get("author")
+        expected = "Jane Doe"
+        self.assertEqual(expected, actual)
+
+    def test_multiline_field(self) -> None:
+        bibtex_string = """@misc{test_entry,
+          note = {This is a
+                  multi-line
+                  note.},
+        }"""
+        entry = BibTeXEntry.from_string(bibtex_string)
+        actual = entry.fields.get("note")
+        expected = "This is a\n                  multi-line\n                  note."
+        self.assertEqual(expected, actual)
+
+    def test_field_with_url_and_brackets(self) -> None:
+        bibtex_string = """@misc{test_entry,
+          howpublished = {\\url{https://example.org/query?x=1&y=2}},
+        }"""
+        entry = BibTeXEntry.from_string(bibtex_string)
+        actual = entry.fields.get("howpublished")
+        expected = "\\url{https://example.org/query?x=1&y=2}"
+        self.assertEqual(expected, actual)
+
 
 class TestSplitEntries(unittest.TestCase):
     def test_single_entry(self) -> None: