Fix hard wrapping of long headings

AUTHORS.rst

@@ -40,6 +40,7 @@ The AUTHORS/Contributors are (and/or have been):
 * Edward Ross <gh: EdwardJRoss>
 * Gregory Anders <gh: gpanders>
 * Alex Vandiver <gh: alexmv>
+* Ryan Chou <gh: ryanchou1994>
 
 Maintainer:
 

ChangeLog.rst

@@ -1,3 +1,8 @@
+Unreleased
+==========
+
+* Fix #424: Keep long headings on one line, including headings in blockquotes.
+
 2025.4.15
 =========
 ----

html2text/config.py

@@ -62,6 +62,7 @@
 
 RE_ORDERED_LIST_MATCHER = re.compile(r"\d+\.\s")
 RE_UNORDERED_LIST_MATCHER = re.compile(r"[-\*\+]\s")
+RE_MD_HEADING_MATCHER = re.compile(r"(?:>[ \t]*)*#{1,6}[ \t]")
 RE_MD_CHARS_MATCHER = re.compile(r"([\\\[\]\(\)])")
 RE_MD_CHARS_MATCHER_ALL = re.compile(r"([`\*_{}\[\]\(\)#!])")
 
@@ -104,8 +105,7 @@
     r"""
     (\\)          # match one slash
     (?=[%s])      # followed by a char that requires escaping
-    """
-    % re.escape(RE_SLASH_CHARS),
+    """ % re.escape(RE_SLASH_CHARS),
     flags=re.VERBOSE,
 )
 

html2text/utils.py

@@ -202,6 +202,10 @@ def skipwrap(
     if para[0:4] == "    " or para[0] == "\t":
         return True
 
+    # ATX headings, including quoted headings, must stay on a single line.
+    if config.RE_MD_HEADING_MATCHER.match(para.lstrip(" \t")):
+        return True
+
     # If the text begins with only two "--", possibly preceded by
     # whitespace, that's an emdash; so wrap.
     stripped = para.lstrip()

test/long_headings.html

@@ -0,0 +1,7 @@
+<h1>A long heading with enough words to exceed the default line width while preserving all of its text</h1>
+<h2><em>A long heading with enough words to exceed the default line width while preserving all of its text</em></h2>
+<h3>A long heading with a <a href="https://example.com/">link containing several words that must stay together</a> and more text</h3>
+<h4>A long heading with enough words to exceed the default line width while preserving all of its text</h4>
+<h5>A long heading with enough words to exceed the default line width while preserving all of its text</h5>
+<h6>A long heading with enough words to exceed the default line width while preserving all of its text</h6>
+<blockquote><h3>A long heading with enough words to exceed the default line width while preserving all of its text</h3></blockquote>

test/long_headings.md

@@ -0,0 +1,13 @@
+# A long heading with enough words to exceed the default line width while preserving all of its text
+
+## _A long heading with enough words to exceed the default line width while preserving all of its text_
+
+### A long heading with a [link containing several words that must stay together](https://example.com/) and more text
+
+#### A long heading with enough words to exceed the default line width while preserving all of its text
+
+##### A long heading with enough words to exceed the default line width while preserving all of its text
+
+###### A long heading with enough words to exceed the default line width while preserving all of its text
+
+> ### A long heading with enough words to exceed the default line width while preserving all of its text

test/test_html2text.py

@@ -254,3 +254,37 @@ def test_strong_emptied() -> None:
     h.strong_mark = ""
     string = "A <b>B</b> <i>C</i>."
     assert h.handle(string) == "A B _C_.\n\n"
+
+
+@pytest.mark.parametrize("level", range(1, 7))
+@pytest.mark.parametrize("depth", [0, 1, 2])
+@pytest.mark.parametrize("body_width", [12, 0])
+def test_heading_is_not_wrapped(level, depth, body_width):
+    heading = "A long heading with several words that must stay together"
+    source = (
+        "<blockquote>" * depth
+        + f"<h{level}>{heading}</h{level}>"
+        + "</blockquote>" * depth
+    )
+    expected = "> " * depth + "#" * level + " " + heading
+    assert html2text.HTML2Text(bodywidth=body_width).handle(source).strip() == expected
+
+
+@pytest.mark.parametrize("prefix", ["", "#hashtag ", "####### "])
+def test_non_heading_paragraph_still_wraps(prefix):
+    text = prefix + "Some ordinary paragraph words that should still wrap."
+    result = html2text.html2text(f"<p>{text}</p>", bodywidth=20).strip()
+    assert "\n" in result
+    assert " ".join(result.splitlines()) == text
+    assert all(len(line) <= 20 for line in result.splitlines())
+
+
+@pytest.mark.parametrize("prefix", ["#&nbsp;", "&gt;&nbsp;# ", "&nbsp;# "])
+def test_non_ascii_heading_whitespace_still_wraps(prefix):
+    h = html2text.HTML2Text(bodywidth=20)
+    h.unicode_snob = True
+    result = h.handle(
+        "<p>" + prefix + "Several ordinary words that should still wrap.</p>"
+    ).strip()
+    assert "\n" in result
+    assert all(len(line) <= 20 for line in result.splitlines())