From 6518255e3551938ca6c224a65790f5b9a3d31ee0 Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Sun, 3 Aug 2025 22:17:33 +0000
Subject: [PATCH 1/3] Initial plan
From c3af79be74da19befcbc682e9e40f4aecdedb16d Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Sun, 3 Aug 2025 22:32:37 +0000
Subject: [PATCH 2/3] Implement keep_order feature - preserve element order
instead of grouping by type
Co-authored-by: MrDebugger <25988388+MrDebugger@users.noreply.github.com>
---
bs2json/__init__.py | 4 +++-
bs2json/bs2json.py | 56 +++++++++++++++++++++++++++++++++++----------
2 files changed, 47 insertions(+), 13 deletions(-)
diff --git a/bs2json/__init__.py b/bs2json/__init__.py
index 77a6523..0f41633 100644
--- a/bs2json/__init__.py
+++ b/bs2json/__init__.py
@@ -13,6 +13,7 @@ def to_json(
tag: element.Tag,
include_comments: bool=True,
strip: bool=True,
+ keep_order: bool=False,
attr_name: str="attrs",
text_name: str="text",
comment_name: str="comment",
@@ -27,7 +28,8 @@ def to_json(
kwargs = dict(
**name_kwargs,
include_comments=include_comments,
- strip=strip
+ strip=strip,
+ keep_order=keep_order
)
diff --git a/bs2json/bs2json.py b/bs2json/bs2json.py
index e467d4d..1681e38 100644
--- a/bs2json/bs2json.py
+++ b/bs2json/bs2json.py
@@ -14,6 +14,7 @@ class BS2Json:
"""
include_comments = True
strip = True
+ keep_order = False
__labels: Dict = {}
soup: BeautifulSoup = None
last_obj: Dict = {}
@@ -24,6 +25,7 @@ def __init__(self,
*,
include_comments: Union[bool, str]=True,
strip: bool=True,
+ keep_order: bool=False,
**kwargs
) -> NoReturn:
"""Initialize the instance of bs2json class.
@@ -34,6 +36,8 @@ def __init__(self,
include_comments (bool, optional): Whether to include comments in the JSON
representation. Defaults to True.
strip (bool): Whether to remove whitespaces from the start and end of text.
+ keep_order (bool): Whether to preserve the original order of elements instead of
+ grouping them by type. Defaults to False.
**kwargs: Keyword arguments for initializing BeautifulSoup.
"""
@@ -56,6 +60,7 @@ def __init__(self,
self.soup = soup
self.include_comments = include_comments
self.strip = strip
+ self.keep_order = keep_order
self.labels(attrs=attr_name, text=text_name, comment=comment_name)
@@ -262,7 +267,13 @@ def __tag(self, element):
if isinstance(value, dict):
json[element.name].update(value)
elif isinstance(value, list):
- if element.attrs:
+ # When keep_order=True, simplify single-text elements
+ if (self.keep_order and len(value) == 1 and
+ isinstance(value[0], dict) and len(value[0]) == 1 and
+ text_name in value[0] and not element.attrs):
+ # Single text content without attributes: return just the text
+ return value[0][text_name]
+ elif element.attrs:
value.append(json[element.name])
json[element.name] = value
else:
@@ -298,7 +309,9 @@ def to_json(self,
comment_name = self.__labels['comment']
if isinstance(element,Element.Tag):
json[element.name] = self.__tag(element)
- if json[element.name].get(text_name) and len(json[element.name]) == 1:
+ if (isinstance(json[element.name], dict) and
+ json[element.name].get(text_name) and
+ len(json[element.name]) == 1):
return json[element.name][text_name]
json = json[element.name]
elif isinstance(element, Element.Comment) and self.include_comments:
@@ -314,14 +327,33 @@ def to_json(self,
json['doctype'] = str(element)
json.update(self.to_json(element.next_element))
elif isinstance(element, (Iterator, Iterable)):
- for elem in element:
- name = self.__get_name(elem)
- value = self.to_json(elem) or None
- if not value and name == text_name:
- continue
- if name in json:
- json[name].append(value)
- else:
- json[name] = [value]
- self.__fix(json)
+ if self.keep_order:
+ # Return a list preserving order instead of grouping by type
+ ordered_list = []
+ for elem in element:
+ name = self.__get_name(elem)
+ value = self.to_json(elem) or None
+ if not value and name == text_name:
+ continue
+
+ # Simplify single-text elements like {"h3": [{"text": "chapter 1"}]} to {"h3": "chapter 1"}
+ if (isinstance(value, list) and len(value) == 1 and
+ isinstance(value[0], dict) and len(value[0]) == 1 and
+ text_name in value[0]):
+ value = value[0][text_name]
+
+ ordered_list.append({name: value})
+ return ordered_list
+ else:
+ # Original behavior: group by type
+ for elem in element:
+ name = self.__get_name(elem)
+ value = self.to_json(elem) or None
+ if not value and name == text_name:
+ continue
+ if name in json:
+ json[name].append(value)
+ else:
+ json[name] = [value]
+ self.__fix(json)
return json
From de4e9845f22b2fa0fb51991e11a55a9a6f2e4ab3 Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Sun, 3 Aug 2025 22:35:09 +0000
Subject: [PATCH 3/3] Add comprehensive tests and complete keep_order feature
implementation
Co-authored-by: MrDebugger <25988388+MrDebugger@users.noreply.github.com>
---
tests/tests.py | 37 +++++++++++++++++++++++++++++++++++++
1 file changed, 37 insertions(+)
diff --git a/tests/tests.py b/tests/tests.py
index 41645ec..3e96dba 100644
--- a/tests/tests.py
+++ b/tests/tests.py
@@ -72,5 +72,42 @@ def test_prettify(self):
result = out['stdout']
self.assertEqual(result, expected_4)
+ def test_keep_order_feature(self):
+ """Test the keep_order feature preserves element order."""
+ html_ordered = '''
+first heading
+paragraph
+second heading
+
+'''
+
+ # Test default behavior (grouping)
+ bs2json_default = BS2Json(html_ordered)
+ result_default = bs2json_default.convert()
+ self.assertEqual(result_default['html']['body']['h3'], ['first heading', 'second heading'])
+
+ # Test keep_order behavior (preserving order)
+ bs2json_ordered = BS2Json(html_ordered, keep_order=True)
+ result_ordered = bs2json_ordered.convert()
+
+ # Extract body content
+ body_content = None
+ for item in result_ordered['html']:
+ if 'body' in item:
+ body_content = item['body']
+ break
+
+ self.assertIsNotNone(body_content)
+ self.assertIsInstance(body_content, list)
+
+ # Verify order is preserved: h3, p, h3, hr
+ expected_order = ['h3', 'p', 'h3', 'hr']
+ actual_order = [list(element.keys())[0] for element in body_content]
+ self.assertEqual(actual_order, expected_order)
+
+ # Verify h3 elements are separate and simplified
+ self.assertEqual(body_content[0]['h3'], 'first heading')
+ self.assertEqual(body_content[2]['h3'], 'second heading')
+
if __name__ == "__main__":
unittest.main()
\ No newline at end of file