Merge pull request #18 from jlevy/feature/fix-multiline-tag-closing

Fix multi-line opening tags placing closing tags on own line

Author: jlevy

src/flowmark/linewrapping/tag_handling.py

@@ -253,15 +253,18 @@ def add_tag_newline_handling(base_wrapper: LineWrapper) -> LineWrapper:
     """
 
     def enhanced_wrapper(text: str, initial_indent: str, subsequent_indent: str) -> str:
-        # If no newlines, nothing to preserve
+        # If no newlines in input, just wrap and apply post-processing fixes.
+        # The base_wrapper may produce multi-line output that needs fixing.
         if "\n" not in text:
-            return base_wrapper(text, initial_indent, subsequent_indent)
+            result = base_wrapper(text, initial_indent, subsequent_indent)
+            return _fix_multiline_opening_tag_with_closing(result)
 
         lines = text.split("\n")
 
-        # If only one line after split, nothing to preserve
+        # If only one line after split, same as above
         if len(lines) <= 1:
-            return base_wrapper(text, initial_indent, subsequent_indent)
+            result = base_wrapper(text, initial_indent, subsequent_indent)
+            return _fix_multiline_opening_tag_with_closing(result)
 
         # Check if there are any tags in the text - only apply block content
         # heuristics when tags are present to avoid changing normal markdown behavior
@@ -299,7 +302,8 @@ def enhanced_wrapper(text: str, initial_indent: str, subsequent_indent: str) ->
 
         # If we only have one segment, no tag boundaries were found
         if len(segments) == 1:
-            return base_wrapper(text, initial_indent, subsequent_indent)
+            result = base_wrapper(text, initial_indent, subsequent_indent)
+            return _fix_multiline_opening_tag_with_closing(result)
 
         # Wrap each segment separately
         wrapped_segments: list[str] = []
@@ -345,6 +349,10 @@ def enhanced_wrapper(text: str, initial_indent: str, subsequent_indent: str) ->
         # The Markdown parser may indent closing tags due to lazy continuation.
         result = _fix_closing_tag_spacing(result)
 
+        # Fix multi-line opening tags that have closing tags on the same line.
+        # This works around a Markdoc parser bug (see GitHub issue #17).
+        result = _fix_multiline_opening_tag_with_closing(result)
+
         return result
 
     return enhanced_wrapper
@@ -394,3 +402,79 @@ def _fix_closing_tag_spacing(text: str) -> str:
             fixed_lines.append(line)
 
     return "\n".join(fixed_lines)
+
+
+# Pattern to detect closing delimiter of opening tag followed by a closing tag.
+# This handles cases like:  %}{% /tag %}  or  --><!-- /tag -->
+# where a multi-line opening tag ends and a closing tag follows on the same line.
+# Uses named group "closing_tag" to capture the start of the closing tag.
+_multiline_closing_pattern: re.Pattern[str] = re.compile(
+    rf"{JINJA_TAG_CLOSE_RE}\s*(?P<closing_tag>{JINJA_TAG_OPEN_RE}\s*/)|"  # %}{% /
+    rf"{JINJA_COMMENT_CLOSE_RE}\s*(?P<closing_comment>{JINJA_COMMENT_OPEN_RE}\s*/)|"  # #}{# /
+    rf"{JINJA_VAR_CLOSE_RE}\s*(?P<closing_var>{JINJA_VAR_OPEN_RE}\s*/)|"  # }}{{ /
+    rf"{HTML_COMMENT_CLOSE_RE}\s*(?P<closing_html>{HTML_COMMENT_OPEN_RE}\s*/)"  # --><!-- /
+)
+
+
+def _fix_multiline_opening_tag_with_closing(text: str) -> str:
+    """
+    Ensure closing tags are on their own line when the opening tag spans multiple lines.
+
+    This works around a Markdoc parser bug where multi-line opening tags with
+    closing tags on the same line cause incorrect AST parsing.
+
+    Problem pattern (triggers Markdoc bug):
+        {% tag attr1=value1
+        attr2=value2 %}{% /tag %}
+
+    Fixed pattern:
+        {% tag attr1=value1
+        attr2=value2 %}
+        {% /tag %}
+
+    Single-line paired tags like `{% field %}{% /field %}` are NOT affected.
+    Tags in the middle of prose like `Before {% field %}{% /field %} after` are
+    also NOT affected because the line contains content before the tag.
+    """
+    # Only apply fix if there are multiple lines - single line input means
+    # no multi-line tags to fix
+    if "\n" not in text:
+        return text
+
+    lines = text.split("\n")
+    result_lines: list[str] = []
+
+    for i, line in enumerate(lines):
+        # Skip the first line - it can't be a continuation of a multi-line tag
+        if i == 0:
+            result_lines.append(line)
+            continue
+
+        stripped = line.lstrip()
+
+        # Only process lines that are continuations (don't start with a tag opener).
+        # If a line starts with a tag opener, the tag began on that line, not a continuation.
+        is_tag_start = (
+            stripped.startswith(JINJA_TAG_OPEN)
+            or stripped.startswith(JINJA_COMMENT_OPEN)
+            or stripped.startswith(JINJA_VAR_OPEN)
+            or stripped.startswith(HTML_COMMENT_OPEN)
+        )
+
+        if not is_tag_start:
+            match = _multiline_closing_pattern.search(line)
+            if match:
+                # Find which named group matched and split at the closing tag
+                for group_name in ["closing_tag", "closing_comment", "closing_var", "closing_html"]:
+                    if match.group(group_name) is not None:
+                        split_pos = match.start(group_name)
+                        before = line[:split_pos].rstrip()
+                        closing = line[split_pos:].lstrip()
+                        result_lines.append(before)
+                        result_lines.append(closing)
+                        break
+                continue
+
+        result_lines.append(line)
+
+    return "\n".join(result_lines)

tests/test_tag_formatting.py

@@ -419,3 +419,81 @@ def test_smart_quotes_multiline_tag_with_prose():
     # Tag quotes should NOT be converted
     assert 'kind="string"' in result
     assert 'label="Full Name"' in result
+
+
+def test_multiline_opening_tag_closing_on_own_line():
+    """
+    Test that closing tags are placed on their own line after multiline opening tags.
+
+    This is a regression test for GitHub issue #17: When an opening tag spans
+    multiple lines, having the closing tag on the same line as the opening tag's
+    closing delimiter breaks Markdoc's parser.
+    """
+    from flowmark.linewrapping.tag_handling import (
+        _fix_multiline_opening_tag_with_closing,  # pyright: ignore[reportPrivateUsage]
+    )
+
+    # Pattern that triggers Markdoc bug: multi-line opening tag with closing on same line
+    problematic = "{% field kind='string'\nrequired=true %}{% /field %}"
+    result = _fix_multiline_opening_tag_with_closing(problematic)
+
+    # Closing tag should be on its own line
+    assert "%}\n{% /field %}" in result, f"Closing tag not on own line: {result}"
+
+
+def test_single_line_paired_tags_not_split():
+    """
+    Test that single-line paired tags like {% field %}{% /field %} are NOT split.
+
+    This is a regression test to ensure the fix for issue #17 doesn't affect
+    single-line tags.
+    """
+    from flowmark.linewrapping.tag_handling import (
+        _fix_multiline_opening_tag_with_closing,  # pyright: ignore[reportPrivateUsage]
+    )
+
+    # Single-line paired tag - should NOT be split
+    single_line = "{% field kind='string' %}{% /field %}"
+    result = _fix_multiline_opening_tag_with_closing(single_line)
+
+    # Should remain on single line
+    assert result == single_line, f"Single-line tag was incorrectly split: {result}"
+
+
+def test_multiline_tag_through_pipeline():
+    """Test multiline tags with closing on same line through the full pipeline."""
+    # Use a tag that's long enough to actually trigger wrapping at width 88
+    # This should produce the problematic pattern that triggers the Markdoc bug
+    long_tag = (
+        '{% field kind="string" id="name" label="Full Name" role="user" '
+        'required=true minLength=2 maxLength=100 placeholder="Enter your full name" %}'
+        "{% /field %}"
+    )
+
+    result = fill_markdown(long_tag, semantic=True, width=88)
+    lines = result.strip().split("\n")
+
+    # If the tag wrapped (longer than line width), closing tag should be on its own line
+    if len(lines) >= 2:
+        # Last line should be the closing tag on its own
+        assert lines[-1].strip() == "{% /field %}", (
+            f"Last line should be closing tag, got: {lines[-1]}"
+        )
+        # The line before closing tag should end with %}
+        assert lines[-2].strip().endswith("%}"), (
+            f"Line before closing should end with %}}, got: {lines[-2]}"
+        )
+
+
+def test_html_comment_multiline_closing():
+    """Test HTML comment tags with multi-line opening and closing on same line."""
+    from flowmark.linewrapping.tag_handling import (
+        _fix_multiline_opening_tag_with_closing,  # pyright: ignore[reportPrivateUsage]
+    )
+
+    # HTML comment pattern
+    text = "<!-- f:field kind='string'\nlabel='Name' --><!-- /f:field -->"
+    result = _fix_multiline_opening_tag_with_closing(text)
+
+    # Closing comment should be on its own line
+    assert "-->\n<!-- /f:field -->" in result, f"HTML closing tag not split: {result}"

tests/testdocs/testdoc.expected.auto.md

@@ -1693,6 +1693,29 @@ Self-closing tags with tables:
 
 <!-- end-section -->
 
+### Issue 7: Multi-Line Opening Tags (GitHub Issue #17)
+
+When an opening tag is long enough to wrap across multiple lines, the closing tag should
+be placed on its own line to avoid triggering a Markdoc parser bug.
+
+This tag is long enough to wrap and should have its closing tag on a separate line:
+
+{% field kind="number" id="age" label="Your Age" role="user" required=true min=0 max=150
+integer=true placeholder="Enter your age" %}
+{% /field %}
+
+HTML comment version:
+
+<!-- f:field kind="number" id="score" label="Score" role="user" required=true min=0
+max=100 integer=true placeholder="Enter score" -->
+<!-- /f:field -->
+
+Short tags that fit on one line should remain together:
+
+{% field kind="string" id="name" %}{% /field %}
+
+<!-- f:field id="email" --><!-- /f:field -->
+
 ### Mixed Content Test
 
 A form with various content types:

tests/testdocs/testdoc.expected.cleaned.md

@@ -1693,6 +1693,29 @@ Self-closing tags with tables:
 
 <!-- end-section -->
 
+### Issue 7: Multi-Line Opening Tags (GitHub Issue #17)
+
+When an opening tag is long enough to wrap across multiple lines, the closing tag should
+be placed on its own line to avoid triggering a Markdoc parser bug.
+
+This tag is long enough to wrap and should have its closing tag on a separate line:
+
+{% field kind="number" id="age" label="Your Age" role="user" required=true min=0 max=150
+integer=true placeholder="Enter your age" %}
+{% /field %}
+
+HTML comment version:
+
+<!-- f:field kind="number" id="score" label="Score" role="user" required=true min=0
+max=100 integer=true placeholder="Enter score" -->
+<!-- /f:field -->
+
+Short tags that fit on one line should remain together:
+
+{% field kind="string" id="name" %}{% /field %}
+
+<!-- f:field id="email" --><!-- /f:field -->
+
 ### Mixed Content Test
 
 A form with various content types:

tests/testdocs/testdoc.expected.plain.md

@@ -1644,6 +1644,29 @@ Self-closing tags with tables:
 
 <!-- end-section -->
 
+### Issue 7: Multi-Line Opening Tags (GitHub Issue #17)
+
+When an opening tag is long enough to wrap across multiple lines, the closing tag should
+be placed on its own line to avoid triggering a Markdoc parser bug.
+
+This tag is long enough to wrap and should have its closing tag on a separate line:
+
+{% field kind="number" id="age" label="Your Age" role="user" required=true min=0 max=150
+integer=true placeholder="Enter your age" %}
+{% /field %}
+
+HTML comment version:
+
+<!-- f:field kind="number" id="score" label="Score" role="user" required=true min=0
+max=100 integer=true placeholder="Enter score" -->
+<!-- /f:field -->
+
+Short tags that fit on one line should remain together:
+
+{% field kind="string" id="name" %}{% /field %}
+
+<!-- f:field id="email" --><!-- /f:field -->
+
 ### Mixed Content Test
 
 A form with various content types:

tests/testdocs/testdoc.expected.semantic.md

@@ -1693,6 +1693,29 @@ Self-closing tags with tables:
 
 <!-- end-section -->
 
+### Issue 7: Multi-Line Opening Tags (GitHub Issue #17)
+
+When an opening tag is long enough to wrap across multiple lines, the closing tag should
+be placed on its own line to avoid triggering a Markdoc parser bug.
+
+This tag is long enough to wrap and should have its closing tag on a separate line:
+
+{% field kind="number" id="age" label="Your Age" role="user" required=true min=0 max=150
+integer=true placeholder="Enter your age" %}
+{% /field %}
+
+HTML comment version:
+
+<!-- f:field kind="number" id="score" label="Score" role="user" required=true min=0
+max=100 integer=true placeholder="Enter score" -->
+<!-- /f:field -->
+
+Short tags that fit on one line should remain together:
+
+{% field kind="string" id="name" %}{% /field %}
+
+<!-- f:field id="email" --><!-- /f:field -->
+
 ### Mixed Content Test
 
 A form with various content types:

tests/testdocs/testdoc.orig.md

@@ -1299,6 +1299,24 @@ Self-closing tags with tables:
 | X | 1 |
 <!-- end-section -->
 
+### Issue 7: Multi-Line Opening Tags (GitHub Issue #17)
+
+When an opening tag is long enough to wrap across multiple lines, the closing tag should be placed on its own line to avoid triggering a Markdoc parser bug.
+
+This tag is long enough to wrap and should have its closing tag on a separate line:
+
+{% field kind="number" id="age" label="Your Age" role="user" required=true min=0 max=150 integer=true placeholder="Enter your age" %}{% /field %}
+
+HTML comment version:
+
+<!-- f:field kind="number" id="score" label="Score" role="user" required=true min=0 max=100 integer=true placeholder="Enter score" --><!-- /f:field -->
+
+Short tags that fit on one line should remain together:
+
+{% field kind="string" id="name" %}{% /field %}
+
+<!-- f:field id="email" --><!-- /f:field -->
+
 ### Mixed Content Test
 
 A form with various content types: