Merge branch 'develop'

Signed-off-by: chrispy <chrispy@synopsys.com>

.github/workflows/python-app.yml

@@ -23,11 +23,10 @@ jobs:
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
-        pip install flake8==3.8.4 pytest
-        if [ -f requirements.txt ]; then pip install -r requirements.txt; fi
-    - name: Lint with flake8
+        pip install --upgrade setuptools setuptools_scm wheel build tox
+    - name: Lint and test
       run: |
-        python setup.py lint
-    - name: Test with pytest
+        tox
+    - name: Build
       run: |
-        python setup.py test
+        python -m build -nwsx .

.github/workflows/python-publish.yml

@@ -21,11 +21,11 @@ jobs:
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
-        pip install setuptools wheel twine
+        pip install --upgrade setuptools setuptools_scm wheel build twine
     - name: Build and publish
       env:
         TWINE_USERNAME: ${{ secrets.PYPI_USERNAME }}
         TWINE_PASSWORD: ${{ secrets.PYPI_PASSWORD }}
       run: |
-        python setup.py sdist bdist_wheel
+        python -m build -nwsx .
         twine upload dist/*

.gitignore

@@ -9,3 +9,4 @@
 /venv
 build/
 .vscode/settings.json
+.tox/

MANIFEST.in

@@ -1 +1,2 @@
 include README.rst
+prune tests

README.rst

@@ -1,8 +1,8 @@
 |build| |version| |license| |downloads|
 
-.. |build| image:: https://img.shields.io/github/workflow/status/matthewwithanm/python-markdownify/Python%20application/develop
+.. |build| image:: https://img.shields.io/github/actions/workflow/status/matthewwithanm/python-markdownify/python-app.yml?branch=develop
     :alt: GitHub Workflow Status
-    :target: https://github.com/matthewwithanm/python-markdownify/actions?query=workflow%3A%22Python+application%22
+    :target: https://github.com/matthewwithanm/python-markdownify/actions/workflows/python-app.yml?query=workflow%3A%22Python+application%22
 
 .. |version| image:: https://img.shields.io/pypi/v/markdownify
     :alt: Pypi version
@@ -87,7 +87,11 @@ strong_em_symbol
 sub_symbol, sup_symbol
   Define the chars that surround ``<sub>`` and ``<sup>`` text. Defaults to an
   empty string, because this is non-standard behavior. Could be something like
-  ``~`` and ``^`` to result in ``~sub~`` and ``^sup^``.
+  ``~`` and ``^`` to result in ``~sub~`` and ``^sup^``.  If the value starts
+  with ``<`` and ends with ``>``, it is treated as an HTML tag and a ``/`` is
+  inserted after the ``<`` in the string used after the text; this allows
+  specifying ``<sub>`` to use raw HTML in the output for subscripts, for
+  example.
 
 newline_style
   Defines the style of marking linebreaks (``<br>``) in markdown. The default
@@ -123,6 +127,11 @@ escape_underscores
   If set to ``False``, do not escape ``_`` to ``\_`` in text.
   Defaults to ``True``.
 
+escape_misc
+  If set to ``True``, escape miscellaneous punctuation characters
+  that sometimes have Markdown significance in text.
+  Defaults to ``False``.
+
 keep_inline_images_in
   Images are converted to their alt-text when the images are located inside
   headlines or table cells. If some inline images should be converted to
@@ -130,10 +139,23 @@ keep_inline_images_in
   that should be allowed to contain inline images, for example ``['td']``.
   Defaults to an empty list.
 
+table_infer_header
+  Controls handling of tables with no header row (as indicated by ``<thead>``
+  or ``<th>``). When set to ``True``, the first body row is used as the header row.
+  Defaults to ``False``, which leaves the header row empty.
+
 wrap, wrap_width
   If ``wrap`` is set to ``True``, all text paragraphs are wrapped at
   ``wrap_width`` characters. Defaults to ``False`` and ``80``.
   Use with ``newline_style=BACKSLASH`` to keep line breaks in paragraphs.
+  A `wrap_width` value of `None` reflows lines to unlimited line length.
+
+strip_document
+  Controls whether leading and/or trailing separation newlines are removed from
+  the final converted document. Supported values are ``LSTRIP`` (leading),
+  ``RSTRIP`` (trailing), ``STRIP`` (both), and ``None`` (neither). Newlines
+  within the document are unaffected.
+  Defaults to ``STRIP``.
 
 Options may be specified as kwargs to the ``markdownify`` function, or as a
 nested ``Options`` class in ``MarkdownConverter`` subclasses.
@@ -156,7 +178,12 @@ Creating Custom Converters
 
 If you have a special usecase that calls for a special conversion, you can
 always inherit from ``MarkdownConverter`` and override the method you want to
-change:
+change.
+The function that handles a HTML tag named ``abc`` is called
+``convert_abc(self, el, text, parent_tags)`` and returns a string
+containing the converted HTML tag.
+The ``MarkdownConverter`` object will handle the conversion based on the
+function names:
 
 .. code:: python
 
@@ -166,16 +193,31 @@ change:
         """
         Create a custom MarkdownConverter that adds two newlines after an image
         """
-        def convert_img(self, el, text, convert_as_inline):
-            return super().convert_img(el, text, convert_as_inline) + '\n\n'
+        def convert_img(self, el, text, parent_tags):
+            return super().convert_img(el, text, parent_tags) + '\n\n'
 
     # Create shorthand method for conversion
     def md(html, **options):
         return ImageBlockConverter(**options).convert(html)
 
+.. code:: python
+
+    from markdownify import MarkdownConverter
+
+    class IgnoreParagraphsConverter(MarkdownConverter):
+        """
+        Create a custom MarkdownConverter that ignores paragraphs
+        """
+        def convert_p(self, el, text, parent_tags):
+            return ''
+
+    # Create shorthand method for conversion
+    def md(html, **options):
+        return IgnoreParagraphsConverter(**options).convert(html)
+
 
 Command Line Interface
-=====================
+======================
 
 Use ``markdownify example.html > example.md`` or pipe input from stdin
 (``cat example.html | markdownify > example.md``).
@@ -186,10 +228,4 @@ They are the same as listed above and take the same arguments.
 Development
 ===========
 
-To run tests:
-
-``python setup.py test``
-
-To lint:
-
-``python setup.py lint``
+To run tests and the linter run ``pip install tox`` once, then ``tox``.

markdownify/__init__.py

@@ -1,15 +1,41 @@
-from bs4 import BeautifulSoup, NavigableString, Comment, Doctype
+from bs4 import BeautifulSoup, Comment, Doctype, NavigableString, Tag
 from textwrap import fill
 import re
 import six
 
 
-convert_heading_re = re.compile(r'convert_h(\d+)')
-line_beginning_re = re.compile(r'^', re.MULTILINE)
-whitespace_re = re.compile(r'[\t ]+')
-all_whitespace_re = re.compile(r'[\s]+')
-html_heading_re = re.compile(r'h[1-6]')
+# General-purpose regex patterns
+re_convert_heading = re.compile(r'convert_h(\d+)')
+re_line_with_content = re.compile(r'^(.*)', flags=re.MULTILINE)
+re_whitespace = re.compile(r'[\t ]+')
+re_all_whitespace = re.compile(r'[\t \r\n]+')
+re_newline_whitespace = re.compile(r'[\t \r\n]*[\r\n][\t \r\n]*')
+re_html_heading = re.compile(r'h(\d+)')
 
+# Pattern for creating convert_<tag> function names from tag names
+re_make_convert_fn_name = re.compile(r'[\[\]:-]')
+
+# Extract (leading_nl, content, trailing_nl) from a string
+# (functionally equivalent to r'^(\n*)(.*?)(\n*)$', but greedy is faster than reluctant here)
+re_extract_newlines = re.compile(r'^(\n*)((?:.*[^\n])?)(\n*)$', flags=re.DOTALL)
+
+# Escape miscellaneous special Markdown characters
+re_escape_misc_chars = re.compile(r'([]\\&<`[>~=+|])')
+
+# Escape sequence of one or more consecutive '-', preceded
+# and followed by whitespace or start/end of fragment, as it
+# might be confused with an underline of a header, or with a
+# list marker
+re_escape_misc_dash_sequences = re.compile(r'(\s|^)(-+(?:\s|$))')
+
+# Escape sequence of up to six consecutive '#', preceded
+# and followed by whitespace or start/end of fragment, as
+# it might be confused with an ATX heading
+re_escape_misc_hashes = re.compile(r'(\s|^)(#{1,6}(?:\s|$))')
+
+# Escape '.' or ')' preceded by up to nine digits, as it might be
+# confused with a list item
+re_escape_misc_list_items = re.compile(r'((?:\s|^)[0-9]{1,9})([.)](?:\s|$))')
 
 # Heading styles
 ATX = 'atx'
@@ -25,6 +51,11 @@ BACKSLASH = 'backslash'
 ASTERISK = '*'
 UNDERSCORE = '_'
 
+# Document strip styles
+LSTRIP = 'lstrip'
+RSTRIP = 'rstrip'
+STRIP = 'strip'
+
 
 def chomp(text):
     """
@@ -43,15 +74,22 @@ def abstract_inline_conversion(markup_fn):
     """
     This abstracts all simple inline tags like b, em, del, ...
     Returns a function that wraps the chomped text in a pair of the string
-    that is returned by markup_fn. markup_fn is necessary to allow for
+    that is returned by markup_fn, with '/' inserted in the string used after
+    the text if it looks like an HTML tag. markup_fn is necessary to allow for
     references to self.strong_em_symbol etc.
     """
-    def implementation(self, el, text, convert_as_inline):
-        markup = markup_fn(self)
+    def implementation(self, el, text, parent_tags):
+        markup_prefix = markup_fn(self)
+        if markup_prefix.startswith('<') and markup_prefix.endswith('>'):
+            markup_suffix = '</' + markup_prefix[1:]
+        else:
+            markup_suffix = markup_prefix
+        if '_noformat' in parent_tags:
+            return text
         prefix, suffix, text = chomp(text)
         if not text:
             return ''
-        return '%s%s%s%s%s' % (prefix, markup, text, markup, suffix)
+        return '%s%s%s%s%s' % (prefix, markup_prefix, text, markup_suffix, suffix)
     return implementation
 
 
@@ -59,6 +97,59 @@ def _todict(obj):
     return dict((k, getattr(obj, k)) for k in dir(obj) if not k.startswith('_'))
 
 
+def should_remove_whitespace_inside(el):
+    """Return to remove whitespace immediately inside a block-level element."""
+    if not el or not el.name:
+        return False
+    if re_html_heading.match(el.name) is not None:
+        return True
+    return el.name in ('p', 'blockquote',
+                       'article', 'div', 'section',
+                       'ol', 'ul', 'li',
+                       'table', 'thead', 'tbody', 'tfoot',
+                       'tr', 'td', 'th')
+
+
+def should_remove_whitespace_outside(el):
+    """Return to remove whitespace immediately outside a block-level element."""
+    return should_remove_whitespace_inside(el) or (el and el.name == 'pre')
+
+
+def _is_block_content_element(el):
+    """
+    In a block context, returns:
+
+    - True for content elements (tags and non-whitespace text)
+    - False for non-content elements (whitespace text, comments, doctypes)
+    """
+    if isinstance(el, Tag):
+        return True
+    elif isinstance(el, (Comment, Doctype)):
+        return False  # (subclasses of NavigableString, must test first)
+    elif isinstance(el, NavigableString):
+        return el.strip() != ''
+    else:
+        return False
+
+
+def _prev_block_content_sibling(el):
+    """Returns the first previous sibling that is a content element, else None."""
+    while el is not None:
+        el = el.previous_sibling
+        if _is_block_content_element(el):
+            return el
+    return None
+
+
+def _next_block_content_sibling(el):
+    """Returns the first next sibling that is a content element, else None."""
+    while el is not None:
+        el = el.next_sibling
+        if _is_block_content_element(el):
+            return el
+    return None
+
+
 class MarkdownConverter(object):
     class DefaultOptions:
         autolinks = True
@@ -69,13 +160,16 @@ class MarkdownConverter(object):
         default_title = False
         escape_asterisks = True
         escape_underscores = True
+        escape_misc = False
         heading_style = UNDERLINED
         keep_inline_images_in = []
         newline_style = SPACES
         strip = None
+        strip_document = STRIP
         strong_em_symbol = ASTERISK
         sub_symbol = ''
         sup_symbol = ''
+        table_infer_header = False
         wrap = False
         wrap_width = 80
 
@@ -92,102 +186,194 @@ class MarkdownConverter(object):
             raise ValueError('You may specify either tags to strip or tags to'
                              ' convert, but not both.')
 
+        # Initialize the conversion function cache
+        self.convert_fn_cache = {}
+
     def convert(self, html):
         soup = BeautifulSoup(html, 'html.parser')
         return self.convert_soup(soup)
 
     def convert_soup(self, soup):
-        return self.process_tag(soup, convert_as_inline=False, children_only=True)
+        return self.process_tag(soup, parent_tags=set())
 
-    def process_tag(self, node, convert_as_inline, children_only=False):
-        text = ''
+    def process_element(self, node, parent_tags=None):
+        if isinstance(node, NavigableString):
+            return self.process_text(node, parent_tags=parent_tags)
+        else:
+            return self.process_tag(node, parent_tags=parent_tags)
 
-        # markdown headings or cells can't include
-        # block elements (elements w/newlines)
-        isHeading = html_heading_re.match(node.name) is not None
-        isCell = node.name in ['td', 'th']
-        convert_children_as_inline = convert_as_inline
+    def process_tag(self, node, parent_tags=None):
+        # For the top-level element, initialize the parent context with an empty set.
+        if parent_tags is None:
+            parent_tags = set()
 
-        if not children_only and (isHeading or isCell):
-            convert_children_as_inline = True
+        # Collect child elements to process, ignoring whitespace-only text elements
+        # adjacent to the inner/outer boundaries of block elements.
+        should_remove_inside = should_remove_whitespace_inside(node)
 
-        # Remove whitespace-only textnodes in purely nested nodes
-        def is_nested_node(el):
-            return el and el.name in ['ol', 'ul', 'li',
-                                      'table', 'thead', 'tbody', 'tfoot',
-                                      'tr', 'td', 'th']
-
-        if is_nested_node(node):
-            for el in node.children:
-                # Only extract (remove) whitespace-only text node if any of the
-                # conditions is true:
-                # - el is the first element in its parent
-                # - el is the last element in its parent
-                # - el is adjacent to an nested node
-                can_extract = (not el.previous_sibling
-                               or not el.next_sibling
-                               or is_nested_node(el.previous_sibling)
-                               or is_nested_node(el.next_sibling))
-                if (isinstance(el, NavigableString)
-                        and six.text_type(el).strip() == ''
-                        and can_extract):
-                    el.extract()
-
-        # Convert the children first
-        for el in node.children:
-            if isinstance(el, Comment) or isinstance(el, Doctype):
-                continue
+        def _can_ignore(el):
+            if isinstance(el, Tag):
+                # Tags are always processed.
+                return False
+            elif isinstance(el, (Comment, Doctype)):
+                # Comment and Doctype elements are always ignored.
+                # (subclasses of NavigableString, must test first)
+                return True
             elif isinstance(el, NavigableString):
-                text += self.process_text(el)
+                if six.text_type(el).strip() != '':
+                    # Non-whitespace text nodes are always processed.
+                    return False
+                elif should_remove_inside and (not el.previous_sibling or not el.next_sibling):
+                    # Inside block elements (excluding <pre>), ignore adjacent whitespace elements.
+                    return True
+                elif should_remove_whitespace_outside(el.previous_sibling) or should_remove_whitespace_outside(el.next_sibling):
+                    # Outside block elements (including <pre>), ignore adjacent whitespace elements.
+                    return True
+                else:
+                    return False
+            elif el is None:
+                return True
             else:
-                text += self.process_tag(el, convert_children_as_inline)
+                raise ValueError('Unexpected element type: %s' % type(el))
 
-        if not children_only:
-            convert_fn = getattr(self, 'convert_%s' % node.name, None)
-            if convert_fn and self.should_convert_tag(node.name):
-                text = convert_fn(node, text, convert_as_inline)
+        children_to_convert = [el for el in node.children if not _can_ignore(el)]
+
+        # Create a copy of this tag's parent context, then update it to include this tag
+        # to propagate down into the children.
+        parent_tags_for_children = set(parent_tags)
+        parent_tags_for_children.add(node.name)
+
+        # if this tag is a heading or table cell, add an '_inline' parent pseudo-tag
+        if (
+            re_html_heading.match(node.name) is not None  # headings
+            or node.name in {'td', 'th'}  # table cells
+        ):
+            parent_tags_for_children.add('_inline')
+
+        # if this tag is a preformatted element, add a '_noformat' parent pseudo-tag
+        if node.name in {'pre', 'code', 'kbd', 'samp'}:
+            parent_tags_for_children.add('_noformat')
+
+        # Convert the children elements into a list of result strings.
+        child_strings = [
+            self.process_element(el, parent_tags=parent_tags_for_children)
+            for el in children_to_convert
+        ]
+
+        # Remove empty string values.
+        child_strings = [s for s in child_strings if s]
+
+        # Collapse newlines at child element boundaries, if needed.
+        if node.name == 'pre' or node.find_parent('pre'):
+            # Inside <pre> blocks, do not collapse newlines.
+            pass
+        else:
+            # Collapse newlines at child element boundaries.
+            updated_child_strings = ['']  # so the first lookback works
+            for child_string in child_strings:
+                # Separate the leading/trailing newlines from the content.
+                leading_nl, content, trailing_nl = re_extract_newlines.match(child_string).groups()
+
+                # If the last child had trailing newlines and this child has leading newlines,
+                # use the larger newline count, limited to 2.
+                if updated_child_strings[-1] and leading_nl:
+                    prev_trailing_nl = updated_child_strings.pop()  # will be replaced by the collapsed value
+                    num_newlines = min(2, max(len(prev_trailing_nl), len(leading_nl)))
+                    leading_nl = '\n' * num_newlines
+
+                # Add the results to the updated child string list.
+                updated_child_strings.extend([leading_nl, content, trailing_nl])
+
+            child_strings = updated_child_strings
+
+        # Join all child text strings into a single string.
+        text = ''.join(child_strings)
+
+        # apply this tag's final conversion function
+        convert_fn = self.get_conv_fn_cached(node.name)
+        if convert_fn is not None:
+            text = convert_fn(node, text, parent_tags=parent_tags)
 
         return text
 
-    def process_text(self, el):
+    def convert__document_(self, el, text, parent_tags):
+        """Final document-level formatting for BeautifulSoup object (node.name == "[document]")"""
+        if self.options['strip_document'] == LSTRIP:
+            text = text.lstrip('\n')  # remove leading separation newlines
+        elif self.options['strip_document'] == RSTRIP:
+            text = text.rstrip('\n')  # remove trailing separation newlines
+        elif self.options['strip_document'] == STRIP:
+            text = text.strip('\n')  # remove leading and trailing separation newlines
+        elif self.options['strip_document'] is None:
+            pass  # leave leading and trailing separation newlines as-is
+        else:
+            raise ValueError('Invalid value for strip_document: %s' % self.options['strip_document'])
+
+        return text
+
+    def process_text(self, el, parent_tags=None):
+        # For the top-level element, initialize the parent context with an empty set.
+        if parent_tags is None:
+            parent_tags = set()
+
         text = six.text_type(el) or ''
 
-        # dont remove any whitespace when handling pre or code in pre
-        if not (el.parent.name == 'pre'
-                or (el.parent.name == 'code'
-                    and el.parent.parent.name == 'pre')):
-            text = whitespace_re.sub(' ', text)
+        # normalize whitespace if we're not inside a preformatted element
+        if 'pre' not in parent_tags:
+            if self.options['wrap']:
+                text = re_all_whitespace.sub(' ', text)
+            else:
+                text = re_newline_whitespace.sub('\n', text)
+                text = re_whitespace.sub(' ', text)
 
-        if el.parent.name != 'code' and el.parent.name != 'pre':
-            text = self.escape(text)
+        # escape special characters if we're not inside a preformatted or code element
+        if '_noformat' not in parent_tags:
+            text = self.escape(text, parent_tags)
 
-        # remove trailing whitespaces if any of the following condition is true:
-        # - current text node is the last node in li
-        # - current text node is followed by an embedded list
-        if (el.parent.name == 'li'
-                and (not el.next_sibling
-                     or el.next_sibling.name in ['ul', 'ol'])):
+        # remove leading whitespace at the start or just after a
+        # block-level element; remove traliing whitespace at the end
+        # or just before a block-level element.
+        if (should_remove_whitespace_outside(el.previous_sibling)
+                or (should_remove_whitespace_inside(el.parent)
+                    and not el.previous_sibling)):
+            text = text.lstrip(' \t\r\n')
+        if (should_remove_whitespace_outside(el.next_sibling)
+                or (should_remove_whitespace_inside(el.parent)
+                    and not el.next_sibling)):
             text = text.rstrip()
 
         return text
 
-    def __getattr__(self, attr):
-        # Handle headings
-        m = convert_heading_re.match(attr)
-        if m:
-            n = int(m.group(1))
+    def get_conv_fn_cached(self, tag_name):
+        """Given a tag name, return the conversion function using the cache."""
+        # If conversion function is not in cache, add it
+        if tag_name not in self.convert_fn_cache:
+            self.convert_fn_cache[tag_name] = self.get_conv_fn(tag_name)
 
-            def convert_tag(el, text, convert_as_inline):
-                return self.convert_hn(n, el, text, convert_as_inline)
+        # Return the cached entry
+        return self.convert_fn_cache[tag_name]
 
-            convert_tag.__name__ = 'convert_h%s' % n
-            setattr(self, convert_tag.__name__, convert_tag)
-            return convert_tag
+    def get_conv_fn(self, tag_name):
+        """Given a tag name, find and return the conversion function."""
+        tag_name = tag_name.lower()
 
-        raise AttributeError(attr)
+        # Handle strip/convert exclusion options
+        if not self.should_convert_tag(tag_name):
+            return None
+
+        # Handle headings with _convert_hn() function
+        match = re_html_heading.match(tag_name)
+        if match:
+            n = int(match.group(1))
+            return lambda el, text, parent_tags: self._convert_hn(n, el, text, parent_tags)
+
+        # For other tags, look up their conversion function by tag name
+        convert_fn_name = "convert_%s" % re_make_convert_fn_name.sub('_', tag_name)
+        convert_fn = getattr(self, convert_fn_name, None)
+        return convert_fn
 
     def should_convert_tag(self, tag):
-        tag = tag.lower()
+        """Given a tag name, return whether to convert based on strip/convert options."""
         strip = self.options['strip']
         convert = self.options['convert']
         if strip is not None:
@@ -197,23 +383,28 @@ class MarkdownConverter(object):
         else:
             return True
 
-    def escape(self, text):
+    def escape(self, text, parent_tags):
         if not text:
             return ''
+        if self.options['escape_misc']:
+            text = re_escape_misc_chars.sub(r'\\\1', text)
+            text = re_escape_misc_dash_sequences.sub(r'\1\\\2', text)
+            text = re_escape_misc_hashes.sub(r'\1\\\2', text)
+            text = re_escape_misc_list_items.sub(r'\1\\\2', text)
+
         if self.options['escape_asterisks']:
             text = text.replace('*', r'\*')
         if self.options['escape_underscores']:
             text = text.replace('_', r'\_')
         return text
 
-    def indent(self, text, level):
-        return line_beginning_re.sub('\t' * level, text) if text else ''
-
     def underline(self, text, pad_char):
         text = (text or '').rstrip()
-        return '%s\n%s\n\n' % (text, pad_char * len(text)) if text else ''
+        return '\n\n%s\n%s\n\n' % (text, pad_char * len(text)) if text else ''
 
-    def convert_a(self, el, text, convert_as_inline):
+    def convert_a(self, el, text, parent_tags):
+        if '_noformat' in parent_tags:
+            return text
         prefix, suffix, text = chomp(text)
         if not text:
             return ''
@@ -233,15 +424,24 @@ class MarkdownConverter(object):
 
     convert_b = abstract_inline_conversion(lambda self: 2 * self.options['strong_em_symbol'])
 
-    def convert_blockquote(self, el, text, convert_as_inline):
+    def convert_blockquote(self, el, text, parent_tags):
+        # handle some early-exit scenarios
+        text = (text or '').strip(' \t\r\n')
+        if '_inline' in parent_tags:
+            return ' ' + text + ' '
+        if not text:
+            return "\n"
 
-        if convert_as_inline:
-            return text
+        # indent lines with blockquote marker
+        def _indent_for_blockquote(match):
+            line_content = match.group(1)
+            return '> ' + line_content if line_content else '>'
+        text = re_line_with_content.sub(_indent_for_blockquote, text)
 
-        return '\n' + (line_beginning_re.sub('> ', text) + '\n\n') if text else ''
+        return '\n' + text + '\n\n'
 
-    def convert_br(self, el, text, convert_as_inline):
-        if convert_as_inline:
+    def convert_br(self, el, text, parent_tags):
+        if '_inline' in parent_tags:
             return ""
 
         if self.options['newline_style'].lower() == BACKSLASH:
@@ -249,78 +449,126 @@ class MarkdownConverter(object):
         else:
             return '  \n'
 
-    def convert_code(self, el, text, convert_as_inline):
-        if el.parent.name == 'pre':
+    def convert_code(self, el, text, parent_tags):
+        if 'pre' in parent_tags:
             return text
         converter = abstract_inline_conversion(lambda self: '`')
-        return converter(self, el, text, convert_as_inline)
+        return converter(self, el, text, parent_tags)
 
     convert_del = abstract_inline_conversion(lambda self: '~~')
 
+    def convert_div(self, el, text, parent_tags):
+        if '_inline' in parent_tags:
+            return ' ' + text.strip() + ' '
+        text = text.strip()
+        return '\n\n%s\n\n' % text if text else ''
+
+    convert_article = convert_div
+
+    convert_section = convert_div
+
     convert_em = abstract_inline_conversion(lambda self: self.options['strong_em_symbol'])
 
     convert_kbd = convert_code
 
-    def convert_hn(self, n, el, text, convert_as_inline):
-        if convert_as_inline:
+    def convert_dd(self, el, text, parent_tags):
+        text = (text or '').strip()
+        if '_inline' in parent_tags:
+            return ' ' + text + ' '
+        if not text:
+            return '\n'
+
+        # indent definition content lines by four spaces
+        def _indent_for_dd(match):
+            line_content = match.group(1)
+            return '    ' + line_content if line_content else ''
+        text = re_line_with_content.sub(_indent_for_dd, text)
+
+        # insert definition marker into first-line indent whitespace
+        text = ':' + text[1:]
+
+        return '%s\n' % text
+
+    def convert_dt(self, el, text, parent_tags):
+        # remove newlines from term text
+        text = (text or '').strip()
+        text = re_all_whitespace.sub(' ', text)
+        if '_inline' in parent_tags:
+            return ' ' + text + ' '
+        if not text:
+            return '\n'
+
+        # TODO - format consecutive <dt> elements as directly adjacent lines):
+        #   https://michelf.ca/projects/php-markdown/extra/#def-list
+
+        return '\n%s\n' % text
+
+    def _convert_hn(self, n, el, text, parent_tags):
+        """ Method name prefixed with _ to prevent <hn> to call this """
+        if '_inline' in parent_tags:
             return text
 
+        # prevent MemoryErrors in case of very large n
+        n = max(1, min(6, n))
+
         style = self.options['heading_style'].lower()
-        text = text.rstrip()
+        text = text.strip()
         if style == UNDERLINED and n <= 2:
             line = '=' if n == 1 else '-'
             return self.underline(text, line)
+        text = re_all_whitespace.sub(' ', text)
         hashes = '#' * n
         if style == ATX_CLOSED:
-            return '%s %s %s\n\n' % (hashes, text, hashes)
-        return '%s %s\n\n' % (hashes, text)
+            return '\n\n%s %s %s\n\n' % (hashes, text, hashes)
+        return '\n\n%s %s\n\n' % (hashes, text)
 
-    def convert_hr(self, el, text, convert_as_inline):
+    def convert_hr(self, el, text, parent_tags):
         return '\n\n---\n\n'
 
     convert_i = convert_em
 
-    def convert_img(self, el, text, convert_as_inline):
+    def convert_img(self, el, text, parent_tags):
         alt = el.attrs.get('alt', None) or ''
         src = el.attrs.get('src', None) or ''
         title = el.attrs.get('title', None) or ''
         title_part = ' "%s"' % title.replace('"', r'\"') if title else ''
-        if (convert_as_inline
+        if ('_inline' in parent_tags
                 and el.parent.name not in self.options['keep_inline_images_in']):
             return alt
 
         return '![%s](%s%s)' % (alt, src, title_part)
 
-    def convert_list(self, el, text, convert_as_inline):
+    def convert_list(self, el, text, parent_tags):
 
         # Converting a list to inline is undefined.
-        # Ignoring convert_to_inline for list.
+        # Ignoring inline conversion parents for list.
 
-        nested = False
         before_paragraph = False
-        if el.next_sibling and el.next_sibling.name not in ['ul', 'ol']:
+        next_sibling = _next_block_content_sibling(el)
+        if next_sibling and next_sibling.name not in ['ul', 'ol']:
             before_paragraph = True
-        while el:
-            if el.name == 'li':
-                nested = True
-                break
-            el = el.parent
-        if nested:
-            # remove trailing newline if nested
-            return '\n' + self.indent(text, 1).rstrip()
-        return text + ('\n' if before_paragraph else '')
+        if 'li' in parent_tags:
+            # remove trailing newline if we're in a nested list
+            return '\n' + text.rstrip()
+        return '\n\n' + text + ('\n' if before_paragraph else '')
 
     convert_ul = convert_list
     convert_ol = convert_list
 
-    def convert_li(self, el, text, convert_as_inline):
+    def convert_li(self, el, text, parent_tags):
+        # handle some early-exit scenarios
+        text = (text or '').strip()
+        if not text:
+            return "\n"
+
+        # determine list item bullet character to use
         parent = el.parent
         if parent is not None and parent.name == 'ol':
-            if parent.get("start"):
+            if parent.get("start") and str(parent.get("start")).isnumeric():
                 start = int(parent.get("start"))
             else:
                 start = 1
-            bullet = '%s.' % (start + parent.index(el))
+            bullet = '%s.' % (start + len(el.find_previous_siblings('li')))
         else:
             depth = -1
             while el:
@@ -329,19 +577,45 @@ class MarkdownConverter(object):
                 el = el.parent
             bullets = self.options['bullets']
             bullet = bullets[depth % len(bullets)]
-        return '%s %s\n' % (bullet, (text or '').strip())
+        bullet = bullet + ' '
+        bullet_width = len(bullet)
+        bullet_indent = ' ' * bullet_width
 
-    def convert_p(self, el, text, convert_as_inline):
-        if convert_as_inline:
-            return text
+        # indent content lines by bullet width
+        def _indent_for_li(match):
+            line_content = match.group(1)
+            return bullet_indent + line_content if line_content else ''
+        text = re_line_with_content.sub(_indent_for_li, text)
+
+        # insert bullet into first-line indent whitespace
+        text = bullet + text[bullet_width:]
+
+        return '%s\n' % text
+
+    def convert_p(self, el, text, parent_tags):
+        if '_inline' in parent_tags:
+            return ' ' + text.strip(' \t\r\n') + ' '
+        text = text.strip(' \t\r\n')
         if self.options['wrap']:
-            text = fill(text,
-                        width=self.options['wrap_width'],
-                        break_long_words=False,
-                        break_on_hyphens=False)
-        return '%s\n\n' % text if text else ''
+            # Preserve newlines (and preceding whitespace) resulting
+            # from <br> tags.  Newlines in the input have already been
+            # replaced by spaces.
+            if self.options['wrap_width'] is not None:
+                lines = text.split('\n')
+                new_lines = []
+                for line in lines:
+                    line = line.lstrip(' \t\r\n')
+                    line_no_trailing = line.rstrip()
+                    trailing = line[len(line_no_trailing):]
+                    line = fill(line,
+                                width=self.options['wrap_width'],
+                                break_long_words=False,
+                                break_on_hyphens=False)
+                    new_lines.append(line + trailing)
+                text = '\n'.join(new_lines)
+        return '\n\n%s\n\n' % text if text else ''
 
-    def convert_pre(self, el, text, convert_as_inline):
+    def convert_pre(self, el, text, parent_tags):
         if not text:
             return ''
         code_language = self.options['code_language']
@@ -349,7 +623,13 @@ class MarkdownConverter(object):
         if self.options['code_language_callback']:
             code_language = self.options['code_language_callback'](el) or code_language
 
-        return '\n```%s\n%s\n```\n' % (code_language, text)
+        return '\n\n```%s\n%s\n```\n\n' % (code_language, text)
+
+    def convert_script(self, el, text, parent_tags):
+        return ''
+
+    def convert_style(self, el, text, parent_tags):
+        return ''
 
     convert_s = convert_del
 
@@ -361,30 +641,67 @@ class MarkdownConverter(object):
 
     convert_sup = abstract_inline_conversion(lambda self: self.options['sup_symbol'])
 
-    def convert_table(self, el, text, convert_as_inline):
-        return '\n\n' + text + '\n'
+    def convert_table(self, el, text, parent_tags):
+        return '\n\n' + text.strip() + '\n\n'
 
-    def convert_td(self, el, text, convert_as_inline):
-        return ' ' + text + ' |'
+    def convert_caption(self, el, text, parent_tags):
+        return text.strip() + '\n\n'
 
-    def convert_th(self, el, text, convert_as_inline):
-        return ' ' + text + ' |'
+    def convert_figcaption(self, el, text, parent_tags):
+        return '\n\n' + text.strip() + '\n\n'
 
-    def convert_tr(self, el, text, convert_as_inline):
+    def convert_td(self, el, text, parent_tags):
+        colspan = 1
+        if 'colspan' in el.attrs and el['colspan'].isdigit():
+            colspan = int(el['colspan'])
+        return ' ' + text.strip().replace("\n", " ") + ' |' * colspan
+
+    def convert_th(self, el, text, parent_tags):
+        colspan = 1
+        if 'colspan' in el.attrs and el['colspan'].isdigit():
+            colspan = int(el['colspan'])
+        return ' ' + text.strip().replace("\n", " ") + ' |' * colspan
+
+    def convert_tr(self, el, text, parent_tags):
         cells = el.find_all(['td', 'th'])
-        is_headrow = all([cell.name == 'th' for cell in cells])
+        is_first_row = el.find_previous_sibling() is None
+        is_headrow = (
+            all([cell.name == 'th' for cell in cells])
+            or (el.parent.name == 'thead'
+                # avoid multiple tr in thead
+                and len(el.parent.find_all('tr')) == 1)
+        )
+        is_head_row_missing = (
+            (is_first_row and not el.parent.name == 'tbody')
+            or (is_first_row and el.parent.name == 'tbody' and len(el.parent.parent.find_all(['thead'])) < 1)
+        )
         overline = ''
         underline = ''
-        if is_headrow and not el.previous_sibling:
-            # first row and is headline: print headline underline
-            underline += '| ' + ' | '.join(['---'] * len(cells)) + ' |' + '\n'
-        elif (not el.previous_sibling
-              and (el.parent.name == 'table'
-                   or (el.parent.name == 'tbody'
-                       and not el.parent.previous_sibling))):
+        if ((is_headrow
+             or (is_head_row_missing
+                 and self.options['table_infer_header']))
+                and is_first_row):
+            # first row and:
+            # - is headline or
+            # - headline is missing and header inference is enabled
+            # print headline underline
+            full_colspan = 0
+            for cell in cells:
+                if 'colspan' in cell.attrs and cell['colspan'].isdigit():
+                    full_colspan += int(cell["colspan"])
+                else:
+                    full_colspan += 1
+            underline += '| ' + ' | '.join(['---'] * full_colspan) + ' |' + '\n'
+        elif ((is_head_row_missing
+               and not self.options['table_infer_header'])
+              or (is_first_row
+                  and (el.parent.name == 'table'
+                       or (el.parent.name == 'tbody'
+                           and not el.parent.find_previous_sibling())))):
+            # headline is missing and header inference is disabled or:
             # first row, not headline, and:
-            # - the parent is table or
-            # - the parent is tbody at the beginning of a table.
+            #  - the parent is table or
+            #  - the parent is tbody at the beginning of a table.
             # print empty headline above this row
             overline += '| ' + ' | '.join([''] * len(cells)) + ' |' + '\n'
             overline += '| ' + ' | '.join(['---'] * len(cells)) + ' |' + '\n'

markdownify/main.py

@@ -3,7 +3,8 @@
 import argparse
 import sys
 
-from markdownify import markdownify
+from markdownify import markdownify, ATX, ATX_CLOSED, UNDERLINED, \
+    SPACES, BACKSLASH, ASTERISK, UNDERSCORE
 
 
 def main(argv=sys.argv[1:]):
@@ -28,16 +29,23 @@ def main(argv=sys.argv[1:]):
     parser.add_argument('--default-title', action='store_false',
                         help="A boolean to enable setting the title of a link to its "
                         "href, if no title is given.")
-    parser.add_argument('--heading-style',
-                        choices=('ATX', 'ATX_CLOSED', 'SETEXT', 'UNDERLINED'),
+    parser.add_argument('--heading-style', default=UNDERLINED,
+                        choices=(ATX, ATX_CLOSED, UNDERLINED),
                         help="Defines how headings should be converted.")
     parser.add_argument('-b', '--bullets', default='*+-',
                         help="A string of bullet styles to use; the bullet will "
                         "alternate based on nesting level.")
+    parser.add_argument('--strong-em-symbol', default=ASTERISK,
+                        choices=(ASTERISK, UNDERSCORE),
+                        help="Use * or _ to convert strong and italics text"),
     parser.add_argument('--sub-symbol', default='',
                         help="Define the chars that surround '<sub>'.")
     parser.add_argument('--sup-symbol', default='',
                         help="Define the chars that surround '<sup>'.")
+    parser.add_argument('--newline-style', default=SPACES,
+                        choices=(SPACES, BACKSLASH),
+                        help="Defines the style of <br> conversions: two spaces "
+                        "or backslash at the and of the line thet should break.")
     parser.add_argument('--code-language', default='',
                         help="Defines the language that should be assumed for all "
                         "'<pre>' sections.")
@@ -53,6 +61,10 @@ def main(argv=sys.argv[1:]):
                         "should be converted to markdown images instead, this option can "
                         "be set to a list of parent tags that should be allowed to "
                         "contain inline images.")
+    parser.add_argument('--table-infer-header', dest='table_infer_header',
+                        action='store_true',
+                        help="When a table has no header row (as indicated by '<thead>' "
+                        "or '<th>'), use the first body row as the header row.")
     parser.add_argument('-w', '--wrap', action='store_true',
                         help="Wrap all text paragraphs at --wrap-width characters.")
     parser.add_argument('--wrap-width', type=int, default=80)

pyproject.toml

@@ -0,0 +1,45 @@
+[build-system]
+requires = ["setuptools>=61.2", "setuptools_scm[toml]>=3.4.3"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "markdownify"
+version = "1.0.0"
+authors = [{name = "Matthew Tretter", email = "m@tthewwithanm.com"}]
+description = "Convert HTML to markdown."
+readme = "README.rst"
+classifiers = [
+    "Environment :: Web Environment",
+    "Framework :: Django",
+    "Intended Audience :: Developers",
+    "License :: OSI Approved :: MIT License",
+    "Operating System :: OS Independent",
+    "Programming Language :: Python :: 2.5",
+    "Programming Language :: Python :: 2.6",
+    "Programming Language :: Python :: 2.7",
+    "Programming Language :: Python :: 3.6",
+    "Programming Language :: Python :: 3.7",
+    "Programming Language :: Python :: 3.8",
+    "Topic :: Utilities",
+]
+dependencies = [
+    "beautifulsoup4>=4.9,<5",
+    "six>=1.15,<2"
+]
+
+[project.urls]
+Homepage = "http://github.com/matthewwithanm/python-markdownify"
+Download = "http://github.com/matthewwithanm/python-markdownify/tarball/master"
+
+[project.scripts]
+markdownify = "markdownify.main:main"
+
+[tool.setuptools]
+zip-safe = false
+include-package-data = true
+
+[tool.setuptools.packages.find]
+include = ["markdownify", "markdownify.*"]
+namespaces = false
+
+[tool.setuptools_scm]

setup.cfg

@@ -1,2 +0,0 @@
-[flake8]
-ignore = E501 W503

setup.py

@@ -1,104 +0,0 @@
-#/usr/bin/env python
-import codecs
-import os
-from setuptools import setup, find_packages
-from setuptools.command.test import test as TestCommand, Command
-
-
-read = lambda filepath: codecs.open(filepath, 'r', 'utf-8').read()
-
-pkgmeta = {
-    '__title__': 'markdownify',
-    '__author__': 'Matthew Tretter',
-    '__version__': '0.11.2',
-}
-
-
-class PyTest(TestCommand):
-    def finalize_options(self):
-        TestCommand.finalize_options(self)
-        self.test_args = ['tests', '-s']
-        self.test_suite = True
-
-    def run_tests(self):
-        import pytest
-        errno = pytest.main(self.test_args)
-        raise SystemExit(errno)
-
-
-class LintCommand(Command):
-    """
-    A copy of flake8's Flake8Command
-
-    """
-    description = "Run flake8 on modules registered in setuptools"
-    user_options = []
-
-    def initialize_options(self):
-        pass
-
-    def finalize_options(self):
-        pass
-
-    def distribution_files(self):
-        if self.distribution.packages:
-            for package in self.distribution.packages:
-                yield package.replace(".", os.path.sep)
-
-        if self.distribution.py_modules:
-            for filename in self.distribution.py_modules:
-                yield "%s.py" % filename
-
-    def run(self):
-        from flake8.api.legacy import get_style_guide
-        flake8_style = get_style_guide(config_file='setup.cfg')
-        paths = self.distribution_files()
-        report = flake8_style.check_files(paths)
-        raise SystemExit(report.total_errors > 0)
-
-
-setup(
-    name='markdownify',
-    description='Convert HTML to markdown.',
-    long_description=read(os.path.join(os.path.dirname(__file__), 'README.rst')),
-    version=pkgmeta['__version__'],
-    author=pkgmeta['__author__'],
-    author_email='m@tthewwithanm.com',
-    url='http://github.com/matthewwithanm/python-markdownify',
-    download_url='http://github.com/matthewwithanm/python-markdownify/tarball/master',
-    packages=find_packages(),
-    zip_safe=False,
-    include_package_data=True,
-    setup_requires=[
-        'flake8>=3.8,<5',
-    ],
-    tests_require=[
-        'pytest>=6.2,<7',
-    ],
-    install_requires=[
-        'beautifulsoup4>=4.9,<5', 'six>=1.15,<2'
-    ],
-    classifiers=[
-        'Environment :: Web Environment',
-        'Framework :: Django',
-        'Intended Audience :: Developers',
-        'License :: OSI Approved :: MIT License',
-        'Operating System :: OS Independent',
-        'Programming Language :: Python :: 2.5',
-        'Programming Language :: Python :: 2.6',
-        'Programming Language :: Python :: 2.7',
-        'Programming Language :: Python :: 3.6',
-        'Programming Language :: Python :: 3.7',
-        'Programming Language :: Python :: 3.8',
-        'Topic :: Utilities'
-    ],
-    cmdclass={
-        'test': PyTest,
-        'lint': LintCommand,
-    },
-    entry_points={
-        'console_scripts': [
-            'markdownify = markdownify.main:main'
-        ]
-    }
-)

shell.nix

@@ -0,0 +1,10 @@
+{ pkgs ? import <nixpkgs> {} }:
+pkgs.mkShell {
+  name = "python-shell";
+  buildInputs = with pkgs; [
+    python38
+    python38Packages.tox
+    python38Packages.setuptools
+    python38Packages.virtualenv
+  ];
+}

tests/test_advanced.py

@@ -1,4 +1,4 @@
-from markdownify import markdownify as md
+from .utils import md
 
 
 def test_chomp():
@@ -14,7 +14,7 @@ def test_chomp():
 
 def test_nested():
     text = md('<p>This is an <a href="http://example.com/">example link</a>.</p>')
-    assert text == 'This is an [example link](http://example.com/).\n\n'
+    assert text == '\n\nThis is an [example link](http://example.com/).\n\n'
 
 
 def test_ignore_comments():

tests/test_args.py

@@ -2,7 +2,8 @@
 Test whitelisting/blacklisting of specific tags.
 
 """
-from markdownify import markdownify as md
+from markdownify import markdownify, LSTRIP, RSTRIP, STRIP
+from .utils import md
 
 
 def test_strip():
@@ -23,3 +24,11 @@ def test_convert():
 def test_do_not_convert():
     text = md('<a href="https://github.com/matthewwithanm">Some Text</a>', convert=[])
     assert text == 'Some Text'
+
+
+def test_strip_document():
+    assert markdownify("<p>Hello</p>") == "Hello"  # test default of STRIP
+    assert markdownify("<p>Hello</p>", strip_document=LSTRIP) == "Hello\n\n"
+    assert markdownify("<p>Hello</p>", strip_document=RSTRIP) == "\n\nHello"
+    assert markdownify("<p>Hello</p>", strip_document=STRIP) == "Hello"
+    assert markdownify("<p>Hello</p>", strip_document=None) == "\n\nHello\n\n"

tests/test_basic.py

@@ -1,4 +1,4 @@
-from markdownify import markdownify as md
+from .utils import md
 
 
 def test_single_tag():
@@ -6,8 +6,9 @@ def test_single_tag():
 
 
 def test_soup():
-    assert md('<div><span>Hello</div></span>') == 'Hello'
+    assert md('<div><span>Hello</div></span>') == '\n\nHello\n\n'
 
 
 def test_whitespace():
     assert md(' a  b \t\t c ') == ' a b c '
+    assert md(' a  b \n\n c ') == ' a b\nc '

tests/test_conversions.py

@@ -1,4 +1,5 @@
-from markdownify import markdownify as md, ATX, ATX_CLOSED, BACKSLASH, UNDERSCORE
+from markdownify import ATX, ATX_CLOSED, BACKSLASH, SPACES, UNDERSCORE
+from .utils import md
 
 
 def inline_tests(tag, markup):
@@ -39,6 +40,11 @@ def test_a_no_autolinks():
     assert md('<a href="https://google.com">https://google.com</a>', autolinks=False) == '[https://google.com](https://google.com)'
 
 
+def test_a_in_code():
+    assert md('<code><a href="https://google.com">Google</a></code>') == '`Google`'
+    assert md('<pre><a href="https://google.com">Google</a></pre>') == '\n\n```\nGoogle\n```\n\n'
+
+
 def test_b():
     assert md('<b>Hello</b>') == '**Hello**'
 
@@ -52,6 +58,13 @@ def test_b_spaces():
 
 def test_blockquote():
     assert md('<blockquote>Hello</blockquote>') == '\n> Hello\n\n'
+    assert md('<blockquote>\nHello\n</blockquote>') == '\n> Hello\n\n'
+    assert md('<blockquote>&nbsp;Hello</blockquote>') == '\n> \u00a0Hello\n\n'
+
+
+def test_blockquote_with_nested_paragraph():
+    assert md('<blockquote><p>Hello</p></blockquote>') == '\n> Hello\n\n'
+    assert md('<blockquote><p>Hello</p><p>Hello again</p></blockquote>') == '\n> Hello\n>\n> Hello again\n\n'
 
 
 def test_blockquote_with_paragraph():
@@ -60,7 +73,7 @@ def test_blockquote_with_paragraph():
 
 def test_blockquote_nested():
     text = md('<blockquote>And she was like <blockquote>Hello</blockquote></blockquote>')
-    assert text == '\n> And she was like \n> > Hello\n> \n> \n\n'
+    assert text == '\n> And she was like\n> > Hello\n\n'
 
 
 def test_br():
@@ -70,44 +83,98 @@ def test_br():
 
 def test_code():
     inline_tests('code', '`')
-    assert md('<code>this_should_not_escape</code>') == '`this_should_not_escape`'
+    assert md('<code>*this_should_not_escape*</code>') == '`*this_should_not_escape*`'
+    assert md('<kbd>*this_should_not_escape*</kbd>') == '`*this_should_not_escape*`'
+    assert md('<samp>*this_should_not_escape*</samp>') == '`*this_should_not_escape*`'
+    assert md('<code><span>*this_should_not_escape*</span></code>') == '`*this_should_not_escape*`'
+    assert md('<code>this  should\t\tnormalize</code>') == '`this should normalize`'
+    assert md('<code><span>this  should\t\tnormalize</span></code>') == '`this should normalize`'
+    assert md('<code>foo<b>bar</b>baz</code>') == '`foobarbaz`'
+    assert md('<kbd>foo<i>bar</i>baz</kbd>') == '`foobarbaz`'
+    assert md('<samp>foo<del> bar </del>baz</samp>') == '`foo bar baz`'
+    assert md('<samp>foo <del>bar</del> baz</samp>') == '`foo bar baz`'
+    assert md('<code>foo<em> bar </em>baz</code>') == '`foo bar baz`'
+    assert md('<code>foo<code> bar </code>baz</code>') == '`foo bar baz`'
+    assert md('<code>foo<strong> bar </strong>baz</code>') == '`foo bar baz`'
+    assert md('<code>foo<s> bar </s>baz</code>') == '`foo bar baz`'
+    assert md('<code>foo<sup>bar</sup>baz</code>', sup_symbol='^') == '`foobarbaz`'
+    assert md('<code>foo<sub>bar</sub>baz</code>', sub_symbol='^') == '`foobarbaz`'
+
+
+def test_dl():
+    assert md('<dl><dt>term</dt><dd>definition</dd></dl>') == '\nterm\n:   definition\n'
+    assert md('<dl><dt><p>te</p><p>rm</p></dt><dd>definition</dd></dl>') == '\nte rm\n:   definition\n'
+    assert md('<dl><dt>term</dt><dd><p>definition-p1</p><p>definition-p2</p></dd></dl>') == '\nterm\n:   definition-p1\n\n    definition-p2\n'
+    assert md('<dl><dt>term</dt><dd><p>definition 1</p></dd><dd><p>definition 2</p></dd></dl>') == '\nterm\n:   definition 1\n:   definition 2\n'
+    assert md('<dl><dt>term 1</dt><dd>definition 1</dd><dt>term 2</dt><dd>definition 2</dd></dl>') == '\nterm 1\n:   definition 1\nterm 2\n:   definition 2\n'
+    assert md('<dl><dt>term</dt><dd><blockquote><p>line 1</p><p>line 2</p></blockquote></dd></dl>') == '\nterm\n:   > line 1\n    >\n    > line 2\n'
+    assert md('<dl><dt>term</dt><dd><ol><li><p>1</p><ul><li>2a</li><li>2b</li></ul></li><li><p>3</p></li></ol></dd></dl>') == '\nterm\n:   1. 1\n\n       * 2a\n       * 2b\n    2. 3\n'
 
 
 def test_del():
     inline_tests('del', '~~')
 
 
-def test_div():
-    assert md('Hello</div> World') == 'Hello World'
+def test_div_section_article():
+    for tag in ['div', 'section', 'article']:
+        assert md(f'<{tag}>456</{tag}>') == '\n\n456\n\n'
+        assert md(f'123<{tag}>456</{tag}>789') == '123\n\n456\n\n789'
+        assert md(f'123<{tag}>\n 456 \n</{tag}>789') == '123\n\n456\n\n789'
+        assert md(f'123<{tag}><p>456</p></{tag}>789') == '123\n\n456\n\n789'
+        assert md(f'123<{tag}>\n<p>456</p>\n</{tag}>789') == '123\n\n456\n\n789'
+        assert md(f'123<{tag}><pre>4 5 6</pre></{tag}>789') == '123\n\n```\n4 5 6\n```\n\n789'
+        assert md(f'123<{tag}>\n<pre>4 5 6</pre>\n</{tag}>789') == '123\n\n```\n4 5 6\n```\n\n789'
+        assert md(f'123<{tag}>4\n5\n6</{tag}>789') == '123\n\n4\n5\n6\n\n789'
+        assert md(f'123<{tag}>\n4\n5\n6\n</{tag}>789') == '123\n\n4\n5\n6\n\n789'
+        assert md(f'123<{tag}>\n<p>\n4\n5\n6\n</p>\n</{tag}>789') == '123\n\n4\n5\n6\n\n789'
+        assert md(f'<{tag}><h1>title</h1>body</{{tag}}>', heading_style=ATX) == '\n\n# title\n\nbody\n\n'
 
 
 def test_em():
     inline_tests('em', '*')
 
 
+def test_figcaption():
+    assert (md("TEXT<figure><figcaption>\nCaption\n</figcaption><span>SPAN</span></figure>") == "TEXT\n\nCaption\n\nSPAN")
+    assert (md("<figure><span>SPAN</span><figcaption>\nCaption\n</figcaption></figure>TEXT") == "SPAN\n\nCaption\n\nTEXT")
+
+
+def test_header_with_space():
+    assert md('<h3>\n\nHello</h3>') == '\n\n### Hello\n\n'
+    assert md('<h3>Hello\n\n\nWorld</h3>') == '\n\n### Hello World\n\n'
+    assert md('<h4>\n\nHello</h4>') == '\n\n#### Hello\n\n'
+    assert md('<h5>\n\nHello</h5>') == '\n\n##### Hello\n\n'
+    assert md('<h5>\n\nHello\n\n</h5>') == '\n\n##### Hello\n\n'
+    assert md('<h5>\n\nHello   \n\n</h5>') == '\n\n##### Hello\n\n'
+
+
 def test_h1():
-    assert md('<h1>Hello</h1>') == 'Hello\n=====\n\n'
+    assert md('<h1>Hello</h1>') == '\n\nHello\n=====\n\n'
 
 
 def test_h2():
-    assert md('<h2>Hello</h2>') == 'Hello\n-----\n\n'
+    assert md('<h2>Hello</h2>') == '\n\nHello\n-----\n\n'
 
 
 def test_hn():
-    assert md('<h3>Hello</h3>') == '### Hello\n\n'
-    assert md('<h4>Hello</h4>') == '#### Hello\n\n'
-    assert md('<h5>Hello</h5>') == '##### Hello\n\n'
-    assert md('<h6>Hello</h6>') == '###### Hello\n\n'
+    assert md('<h3>Hello</h3>') == '\n\n### Hello\n\n'
+    assert md('<h4>Hello</h4>') == '\n\n#### Hello\n\n'
+    assert md('<h5>Hello</h5>') == '\n\n##### Hello\n\n'
+    assert md('<h6>Hello</h6>') == '\n\n###### Hello\n\n'
+    assert md('<h10>Hello</h10>') == md('<h6>Hello</h6>')
+    assert md('<hn>Hello</hn>') == md('Hello')
 
 
 def test_hn_chained():
-    assert md('<h1>First</h1>\n<h2>Second</h2>\n<h3>Third</h3>', heading_style=ATX) == '# First\n\n\n## Second\n\n\n### Third\n\n'
-    assert md('X<h1>First</h1>', heading_style=ATX) == 'X# First\n\n'
+    assert md('<h1>First</h1>\n<h2>Second</h2>\n<h3>Third</h3>', heading_style=ATX) == '\n\n# First\n\n## Second\n\n### Third\n\n'
+    assert md('X<h1>First</h1>', heading_style=ATX) == 'X\n\n# First\n\n'
+    assert md('X<h1>First</h1>', heading_style=ATX_CLOSED) == 'X\n\n# First #\n\n'
+    assert md('X<h1>First</h1>') == 'X\n\nFirst\n=====\n\n'
 
 
 def test_hn_nested_tag_heading_style():
-    assert md('<h1>A <p>P</p> C </h1>', heading_style=ATX_CLOSED) == '# A P C #\n\n'
-    assert md('<h1>A <p>P</p> C </h1>', heading_style=ATX) == '# A P C\n\n'
+    assert md('<h1>A <p>P</p> C </h1>', heading_style=ATX_CLOSED) == '\n\n# A P C #\n\n'
+    assert md('<h1>A <p>P</p> C </h1>', heading_style=ATX) == '\n\n# A P C\n\n'
 
 
 def test_hn_nested_simple_tag():
@@ -123,12 +190,12 @@ def test_hn_nested_simple_tag():
     ]
 
     for tag, markdown in tag_to_markdown:
-        assert md('<h3>A <' + tag + '>' + tag + '</' + tag + '> B</h3>') == '### A ' + markdown + ' B\n\n'
+        assert md('<h3>A <' + tag + '>' + tag + '</' + tag + '> B</h3>') == '\n\n### A ' + markdown + ' B\n\n'
 
-    assert md('<h3>A <br>B</h3>', heading_style=ATX) == '### A B\n\n'
+    assert md('<h3>A <br>B</h3>', heading_style=ATX) == '\n\n### A B\n\n'
 
     # Nested lists not supported
-    # assert md('<h3>A <ul><li>li1</i><li>l2</li></ul></h3>', heading_style=ATX) == '### A li1 li2 B\n\n'
+    # assert md('<h3>A <ul><li>li1</i><li>l2</li></ul></h3>', heading_style=ATX) == '\n### A li1 li2 B\n\n'
 
 
 def test_hn_nested_img():
@@ -138,18 +205,23 @@ def test_hn_nested_img():
         ("alt='Alt Text' title='Optional title'", "Alt Text", " \"Optional title\""),
     ]
     for image_attributes, markdown, title in image_attributes_to_markdown:
-        assert md('<h3>A <img src="/path/to/img.jpg" ' + image_attributes + '/> B</h3>') == '### A ' + markdown + ' B\n\n'
-        assert md('<h3>A <img src="/path/to/img.jpg" ' + image_attributes + '/> B</h3>', keep_inline_images_in=['h3']) == '### A ![' + markdown + '](/path/to/img.jpg' + title + ') B\n\n'
+        assert md('<h3>A <img src="/path/to/img.jpg" ' + image_attributes + '/> B</h3>') == '\n\n### A' + (' ' + markdown + ' ' if markdown else ' ') + 'B\n\n'
+        assert md('<h3>A <img src="/path/to/img.jpg" ' + image_attributes + '/> B</h3>', keep_inline_images_in=['h3']) == '\n\n### A ![' + markdown + '](/path/to/img.jpg' + title + ') B\n\n'
 
 
 def test_hn_atx_headings():
-    assert md('<h1>Hello</h1>', heading_style=ATX) == '# Hello\n\n'
-    assert md('<h2>Hello</h2>', heading_style=ATX) == '## Hello\n\n'
+    assert md('<h1>Hello</h1>', heading_style=ATX) == '\n\n# Hello\n\n'
+    assert md('<h2>Hello</h2>', heading_style=ATX) == '\n\n## Hello\n\n'
 
 
 def test_hn_atx_closed_headings():
-    assert md('<h1>Hello</h1>', heading_style=ATX_CLOSED) == '# Hello #\n\n'
-    assert md('<h2>Hello</h2>', heading_style=ATX_CLOSED) == '## Hello ##\n\n'
+    assert md('<h1>Hello</h1>', heading_style=ATX_CLOSED) == '\n\n# Hello #\n\n'
+    assert md('<h2>Hello</h2>', heading_style=ATX_CLOSED) == '\n\n## Hello ##\n\n'
+
+
+def test_hn_newlines():
+    assert md("<h1>H1-1</h1>TEXT<h2>H2-2</h2>TEXT<h1>H1-2</h1>TEXT", heading_style=ATX) == '\n\n# H1-1\n\nTEXT\n\n## H2-2\n\nTEXT\n\n# H1-2\n\nTEXT'
+    assert md('<h1>H1-1</h1>\n<p>TEXT</p>\n<h2>H2-2</h2>\n<p>TEXT</p>\n<h1>H1-2</h1>\n<p>TEXT</p>', heading_style=ATX) == '\n\n# H1-1\n\nTEXT\n\n## H2-2\n\nTEXT\n\n# H1-2\n\nTEXT\n\n'
 
 
 def test_head():
@@ -159,7 +231,7 @@ def test_head():
 def test_hr():
     assert md('Hello<hr>World') == 'Hello\n\n---\n\nWorld'
     assert md('Hello<hr />World') == 'Hello\n\n---\n\nWorld'
-    assert md('<p>Hello</p>\n<hr>\n<p>World</p>') == 'Hello\n\n\n\n\n---\n\n\nWorld\n\n'
+    assert md('<p>Hello</p>\n<hr>\n<p>World</p>') == '\n\nHello\n\n---\n\nWorld\n\n'
 
 
 def test_i():
@@ -176,18 +248,58 @@ def test_kbd():
 
 
 def test_p():
-    assert md('<p>hello</p>') == 'hello\n\n'
-    assert md('<p>123456789 123456789</p>') == '123456789 123456789\n\n'
-    assert md('<p>123456789 123456789</p>', wrap=True, wrap_width=10) == '123456789\n123456789\n\n'
-    assert md('<p><a href="https://example.com">Some long link</a></p>', wrap=True, wrap_width=10) == '[Some long\nlink](https://example.com)\n\n'
-    assert md('<p>12345<br />67890</p>', wrap=True, wrap_width=10, newline_style=BACKSLASH) == '12345\\\n67890\n\n'
-    assert md('<p>12345678901<br />12345</p>', wrap=True, wrap_width=10, newline_style=BACKSLASH) == '12345678901\\\n12345\n\n'
+    assert md('<p>hello</p>') == '\n\nhello\n\n'
+    assert md("<p><p>hello</p></p>") == "\n\nhello\n\n"
+    assert md('<p>123456789 123456789</p>') == '\n\n123456789 123456789\n\n'
+    assert md('<p>123456789\n\n\n123456789</p>') == '\n\n123456789\n123456789\n\n'
+    assert md('<p>123456789\n\n\n123456789</p>', wrap=True, wrap_width=80) == '\n\n123456789 123456789\n\n'
+    assert md('<p>123456789\n\n\n123456789</p>', wrap=True, wrap_width=None) == '\n\n123456789 123456789\n\n'
+    assert md('<p>123456789 123456789</p>', wrap=True, wrap_width=10) == '\n\n123456789\n123456789\n\n'
+    assert md('<p><a href="https://example.com">Some long link</a></p>', wrap=True, wrap_width=10) == '\n\n[Some long\nlink](https://example.com)\n\n'
+    assert md('<p>12345<br />67890</p>', wrap=True, wrap_width=10, newline_style=BACKSLASH) == '\n\n12345\\\n67890\n\n'
+    assert md('<p>12345<br />67890</p>', wrap=True, wrap_width=50, newline_style=BACKSLASH) == '\n\n12345\\\n67890\n\n'
+    assert md('<p>12345<br />67890</p>', wrap=True, wrap_width=10, newline_style=SPACES) == '\n\n12345  \n67890\n\n'
+    assert md('<p>12345<br />67890</p>', wrap=True, wrap_width=50, newline_style=SPACES) == '\n\n12345  \n67890\n\n'
+    assert md('<p>12345678901<br />12345</p>', wrap=True, wrap_width=10, newline_style=BACKSLASH) == '\n\n12345678901\\\n12345\n\n'
+    assert md('<p>12345678901<br />12345</p>', wrap=True, wrap_width=50, newline_style=BACKSLASH) == '\n\n12345678901\\\n12345\n\n'
+    assert md('<p>12345678901<br />12345</p>', wrap=True, wrap_width=10, newline_style=SPACES) == '\n\n12345678901  \n12345\n\n'
+    assert md('<p>12345678901<br />12345</p>', wrap=True, wrap_width=50, newline_style=SPACES) == '\n\n12345678901  \n12345\n\n'
+    assert md('<p>1234 5678 9012<br />67890</p>', wrap=True, wrap_width=10, newline_style=BACKSLASH) == '\n\n1234 5678\n9012\\\n67890\n\n'
+    assert md('<p>1234 5678 9012<br />67890</p>', wrap=True, wrap_width=10, newline_style=SPACES) == '\n\n1234 5678\n9012  \n67890\n\n'
+    assert md('First<p>Second</p><p>Third</p>Fourth') == 'First\n\nSecond\n\nThird\n\nFourth'
+    assert md('<p>&nbsp;x y</p>', wrap=True, wrap_width=80) == '\n\n\u00a0x y\n\n'
 
 
 def test_pre():
-    assert md('<pre>test\n    foo\nbar</pre>') == '\n```\ntest\n    foo\nbar\n```\n'
-    assert md('<pre><code>test\n    foo\nbar</code></pre>') == '\n```\ntest\n    foo\nbar\n```\n'
-    assert md('<pre>this_should_not_escape</pre>') == '\n```\nthis_should_not_escape\n```\n'
+    assert md('<pre>test\n    foo\nbar</pre>') == '\n\n```\ntest\n    foo\nbar\n```\n\n'
+    assert md('<pre><code>test\n    foo\nbar</code></pre>') == '\n\n```\ntest\n    foo\nbar\n```\n\n'
+    assert md('<pre>*this_should_not_escape*</pre>') == '\n\n```\n*this_should_not_escape*\n```\n\n'
+    assert md('<pre><span>*this_should_not_escape*</span></pre>') == '\n\n```\n*this_should_not_escape*\n```\n\n'
+    assert md('<pre>\t\tthis  should\t\tnot  normalize</pre>') == '\n\n```\n\t\tthis  should\t\tnot  normalize\n```\n\n'
+    assert md('<pre><span>\t\tthis  should\t\tnot  normalize</span></pre>') == '\n\n```\n\t\tthis  should\t\tnot  normalize\n```\n\n'
+    assert md('<pre>foo<b>\nbar\n</b>baz</pre>') == '\n\n```\nfoo\nbar\nbaz\n```\n\n'
+    assert md('<pre>foo<i>\nbar\n</i>baz</pre>') == '\n\n```\nfoo\nbar\nbaz\n```\n\n'
+    assert md('<pre>foo\n<i>bar</i>\nbaz</pre>') == '\n\n```\nfoo\nbar\nbaz\n```\n\n'
+    assert md('<pre>foo<i>\n</i>baz</pre>') == '\n\n```\nfoo\nbaz\n```\n\n'
+    assert md('<pre>foo<del>\nbar\n</del>baz</pre>') == '\n\n```\nfoo\nbar\nbaz\n```\n\n'
+    assert md('<pre>foo<em>\nbar\n</em>baz</pre>') == '\n\n```\nfoo\nbar\nbaz\n```\n\n'
+    assert md('<pre>foo<code>\nbar\n</code>baz</pre>') == '\n\n```\nfoo\nbar\nbaz\n```\n\n'
+    assert md('<pre>foo<strong>\nbar\n</strong>baz</pre>') == '\n\n```\nfoo\nbar\nbaz\n```\n\n'
+    assert md('<pre>foo<s>\nbar\n</s>baz</pre>') == '\n\n```\nfoo\nbar\nbaz\n```\n\n'
+    assert md('<pre>foo<sup>\nbar\n</sup>baz</pre>', sup_symbol='^') == '\n\n```\nfoo\nbar\nbaz\n```\n\n'
+    assert md('<pre>foo<sub>\nbar\n</sub>baz</pre>', sub_symbol='^') == '\n\n```\nfoo\nbar\nbaz\n```\n\n'
+    assert md('<pre>foo<sub>\nbar\n</sub>baz</pre>', sub_symbol='^') == '\n\n```\nfoo\nbar\nbaz\n```\n\n'
+
+    assert md('foo<pre>bar</pre>baz', sub_symbol='^') == 'foo\n\n```\nbar\n```\n\nbaz'
+    assert md("<p>foo</p>\n<pre>bar</pre>\n</p>baz</p>", sub_symbol="^") == "\n\nfoo\n\n```\nbar\n```\n\nbaz"
+
+
+def test_script():
+    assert md('foo <script>var foo=42;</script> bar') == 'foo  bar'
+
+
+def test_style():
+    assert md('foo <style>h1 { font-size: larger }</style> bar') == 'foo  bar'
 
 
 def test_s():
@@ -212,22 +324,34 @@ def test_strong_em_symbol():
 def test_sub():
     assert md('<sub>foo</sub>') == 'foo'
     assert md('<sub>foo</sub>', sub_symbol='~') == '~foo~'
+    assert md('<sub>foo</sub>', sub_symbol='<sub>') == '<sub>foo</sub>'
 
 
 def test_sup():
     assert md('<sup>foo</sup>') == 'foo'
     assert md('<sup>foo</sup>', sup_symbol='^') == '^foo^'
+    assert md('<sup>foo</sup>', sup_symbol='<sup>') == '<sup>foo</sup>'
 
 
 def test_lang():
-    assert md('<pre>test\n    foo\nbar</pre>', code_language='python') == '\n```python\ntest\n    foo\nbar\n```\n'
-    assert md('<pre><code>test\n    foo\nbar</code></pre>', code_language='javascript') == '\n```javascript\ntest\n    foo\nbar\n```\n'
+    assert md('<pre>test\n    foo\nbar</pre>', code_language='python') == '\n\n```python\ntest\n    foo\nbar\n```\n\n'
+    assert md('<pre><code>test\n    foo\nbar</code></pre>', code_language='javascript') == '\n\n```javascript\ntest\n    foo\nbar\n```\n\n'
 
 
 def test_lang_callback():
     def callback(el):
         return el['class'][0] if el.has_attr('class') else None
 
-    assert md('<pre class="python">test\n    foo\nbar</pre>', code_language_callback=callback) == '\n```python\ntest\n    foo\nbar\n```\n'
-    assert md('<pre class="javascript"><code>test\n    foo\nbar</code></pre>', code_language_callback=callback) == '\n```javascript\ntest\n    foo\nbar\n```\n'
-    assert md('<pre class="javascript"><code class="javascript">test\n    foo\nbar</code></pre>', code_language_callback=callback) == '\n```javascript\ntest\n    foo\nbar\n```\n'
+    assert md('<pre class="python">test\n    foo\nbar</pre>', code_language_callback=callback) == '\n\n```python\ntest\n    foo\nbar\n```\n\n'
+    assert md('<pre class="javascript"><code>test\n    foo\nbar</code></pre>', code_language_callback=callback) == '\n\n```javascript\ntest\n    foo\nbar\n```\n\n'
+    assert md('<pre class="javascript"><code class="javascript">test\n    foo\nbar</code></pre>', code_language_callback=callback) == '\n\n```javascript\ntest\n    foo\nbar\n```\n\n'
+
+
+def test_spaces():
+    assert md('<p> a b </p> <p> c d </p>') == '\n\na b\n\nc d\n\n'
+    assert md('<p> <i>a</i> </p>') == '\n\n*a*\n\n'
+    assert md('test <p> again </p>') == 'test\n\nagain\n\n'
+    assert md('test <blockquote> text </blockquote> after') == 'test\n> text\n\nafter'
+    assert md(' <ol> <li> x </li> <li> y </li> </ol> ') == '\n\n1. x\n2. y\n'
+    assert md(' <ul> <li> x </li> <li> y </li> </ol> ') == '\n\n* x\n* y\n'
+    assert md('test <pre> foo </pre> bar') == 'test\n\n```\n foo \n```\n\nbar'

tests/test_custom_converter.py

@@ -2,21 +2,28 @@ from markdownify import MarkdownConverter
 from bs4 import BeautifulSoup
 
 
-class ImageBlockConverter(MarkdownConverter):
+class UnitTestConverter(MarkdownConverter):
     """
-    Create a custom MarkdownConverter that adds two newlines after an image
+    Create a custom MarkdownConverter for unit tests
     """
-    def convert_img(self, el, text, convert_as_inline):
-        return super().convert_img(el, text, convert_as_inline) + '\n\n'
+    def convert_img(self, el, text, parent_tags):
+        """Add two newlines after an image"""
+        return super().convert_img(el, text, parent_tags) + '\n\n'
+
+    def convert_custom_tag(self, el, text, parent_tags):
+        """Ensure conversion function is found for tags with special characters in name"""
+        return "FUNCTION USED: %s" % text
 
 
-def test_img():
+def test_custom_conversion_functions():
     # Create shorthand method for conversion
     def md(html, **options):
-        return ImageBlockConverter(**options).convert(html)
+        return UnitTestConverter(**options).convert(html)
 
-    assert md('<img src="/path/to/img.jpg" alt="Alt text" title="Optional title" />') == '![Alt text](/path/to/img.jpg "Optional title")\n\n'
-    assert md('<img src="/path/to/img.jpg" alt="Alt text" />') == '![Alt text](/path/to/img.jpg)\n\n'
+    assert md('<img src="/path/to/img.jpg" alt="Alt text" title="Optional title" />text') == '![Alt text](/path/to/img.jpg "Optional title")\n\ntext'
+    assert md('<img src="/path/to/img.jpg" alt="Alt text" />text') == '![Alt text](/path/to/img.jpg)\n\ntext'
+
+    assert md("<custom-tag>text</custom-tag>") == "FUNCTION USED: text"
 
 
 def test_soup():

tests/test_escaping.py

@@ -1,4 +1,6 @@
-from markdownify import markdownify as md
+import warnings
+from bs4 import MarkupResemblesLocatorWarning
+from .utils import md
 
 
 def test_asterisks():
@@ -12,7 +14,7 @@ def test_underscore():
 
 
 def test_xml_entities():
-    assert md('&') == '&'
+    assert md('&', escape_misc=True) == r'\&'
 
 
 def test_named_entities():
@@ -25,4 +27,51 @@ def test_hexadecimal_entities():
 
 
 def test_single_escaping_entities():
-    assert md('&') == '&'
+    assert md('&', escape_misc=True) == r'\&'
+
+
+def test_misc():
+    # ignore the bs4 warning that "1.2" or "*" looks like a filename
+    warnings.filterwarnings("ignore", category=MarkupResemblesLocatorWarning)
+
+    assert md('\\*', escape_misc=True) == r'\\\*'
+    assert md('&lt;foo>', escape_misc=True) == r'\<foo\>'
+    assert md('# foo', escape_misc=True) == r'\# foo'
+    assert md('#5', escape_misc=True) == r'#5'
+    assert md('5#', escape_misc=True) == '5#'
+    assert md('####### foo', escape_misc=True) == r'####### foo'
+    assert md('> foo', escape_misc=True) == r'\> foo'
+    assert md('~~foo~~', escape_misc=True) == r'\~\~foo\~\~'
+    assert md('foo\n===\n', escape_misc=True) == 'foo\n\\=\\=\\=\n'
+    assert md('---\n', escape_misc=True) == '\\---\n'
+    assert md('- test', escape_misc=True) == r'\- test'
+    assert md('x - y', escape_misc=True) == r'x \- y'
+    assert md('test-case', escape_misc=True) == 'test-case'
+    assert md('x-', escape_misc=True) == 'x-'
+    assert md('-y', escape_misc=True) == '-y'
+    assert md('+ x\n+ y\n', escape_misc=True) == '\\+ x\n\\+ y\n'
+    assert md('`x`', escape_misc=True) == r'\`x\`'
+    assert md('[text](notalink)', escape_misc=True) == r'\[text\](notalink)'
+    assert md('<a href="link">text]</a>', escape_misc=True) == r'[text\]](link)'
+    assert md('<a href="link">[text]</a>', escape_misc=True) == r'[\[text\]](link)'
+    assert md('1. x', escape_misc=True) == r'1\. x'
+    # assert md('1<span>.</span> x', escape_misc=True) == r'1\. x'
+    assert md('<span>1.</span> x', escape_misc=True) == r'1\. x'
+    assert md(' 1. x', escape_misc=True) == r' 1\. x'
+    assert md('123456789. x', escape_misc=True) == r'123456789\. x'
+    assert md('1234567890. x', escape_misc=True) == r'1234567890. x'
+    assert md('A1. x', escape_misc=True) == r'A1. x'
+    assert md('1.2', escape_misc=True) == r'1.2'
+    assert md('not a number. x', escape_misc=True) == r'not a number. x'
+    assert md('1) x', escape_misc=True) == r'1\) x'
+    # assert md('1<span>)</span> x', escape_misc=True) == r'1\) x'
+    assert md('<span>1)</span> x', escape_misc=True) == r'1\) x'
+    assert md(' 1) x', escape_misc=True) == r' 1\) x'
+    assert md('123456789) x', escape_misc=True) == r'123456789\) x'
+    assert md('1234567890) x', escape_misc=True) == r'1234567890) x'
+    assert md('(1) x', escape_misc=True) == r'(1) x'
+    assert md('A1) x', escape_misc=True) == r'A1) x'
+    assert md('1)x', escape_misc=True) == r'1)x'
+    assert md('not a number) x', escape_misc=True) == r'not a number) x'
+    assert md('|not table|', escape_misc=True) == r'\|not table\|'
+    assert md(r'\ &lt;foo> &amp;amp; | ` `', escape_misc=False) == r'\ <foo> &amp; | ` `'

tests/test_lists.py

@@ -1,4 +1,4 @@
-from markdownify import markdownify as md
+from .utils import md
 
 
 nested_uls = """
@@ -41,16 +41,22 @@ nested_ols = """
 
 
 def test_ol():
-    assert md('<ol><li>a</li><li>b</li></ol>') == '1. a\n2. b\n'
-    assert md('<ol start="3"><li>a</li><li>b</li></ol>') == '3. a\n4. b\n'
+    assert md('<ol><li>a</li><li>b</li></ol>') == '\n\n1. a\n2. b\n'
+    assert md('<ol><!--comment--><li>a</li><span/><li>b</li></ol>') == '\n\n1. a\n2. b\n'
+    assert md('<ol start="3"><li>a</li><li>b</li></ol>') == '\n\n3. a\n4. b\n'
+    assert md('foo<ol start="3"><li>a</li><li>b</li></ol>bar') == 'foo\n\n3. a\n4. b\n\nbar'
+    assert md('<ol start="-1"><li>a</li><li>b</li></ol>') == '\n\n1. a\n2. b\n'
+    assert md('<ol start="foo"><li>a</li><li>b</li></ol>') == '\n\n1. a\n2. b\n'
+    assert md('<ol start="1.5"><li>a</li><li>b</li></ol>') == '\n\n1. a\n2. b\n'
+    assert md('<ol start="1234"><li><p>first para</p><p>second para</p></li><li><p>third para</p><p>fourth para</p></li></ol>') == '\n\n1234. first para\n\n      second para\n1235. third para\n\n      fourth para\n'
 
 
 def test_nested_ols():
-    assert md(nested_ols) == '\n1. 1\n\t1. a\n\t\t1. I\n\t\t2. II\n\t\t3. III\n\t2. b\n\t3. c\n2. 2\n3. 3\n'
+    assert md(nested_ols) == '\n\n1. 1\n   1. a\n      1. I\n      2. II\n      3. III\n   2. b\n   3. c\n2. 2\n3. 3\n'
 
 
 def test_ul():
-    assert md('<ul><li>a</li><li>b</li></ul>') == '* a\n* b\n'
+    assert md('<ul><li>a</li><li>b</li></ul>') == '\n\n* a\n* b\n'
     assert md("""<ul>
      <li>
              a
@@ -58,11 +64,13 @@ def test_ul():
      <li> b </li>
      <li>   c
      </li>
- </ul>""") == '* a\n* b\n* c\n'
+ </ul>""") == '\n\n* a\n* b\n* c\n'
+    assert md('<ul><li><p>first para</p><p>second para</p></li><li><p>third para</p><p>fourth para</p></li></ul>') == '\n\n* first para\n\n  second para\n* third para\n\n  fourth para\n'
 
 
 def test_inline_ul():
-    assert md('<p>foo</p><ul><li>a</li><li>b</li></ul><p>bar</p>') == 'foo\n\n* a\n* b\n\nbar\n\n'
+    assert md('<p>foo</p><ul><li>a</li><li>b</li></ul><p>bar</p>') == '\n\nfoo\n\n* a\n* b\n\nbar\n\n'
+    assert md('foo<ul><li>bar</li></ul>baz') == 'foo\n\n* bar\n\nbaz'
 
 
 def test_nested_uls():
@@ -70,12 +78,12 @@ def test_nested_uls():
     Nested ULs should alternate bullet characters.
 
     """
-    assert md(nested_uls) == '\n* 1\n\t+ a\n\t\t- I\n\t\t- II\n\t\t- III\n\t+ b\n\t+ c\n* 2\n* 3\n'
+    assert md(nested_uls) == '\n\n* 1\n  + a\n    - I\n    - II\n    - III\n  + b\n  + c\n* 2\n* 3\n'
 
 
 def test_bullets():
-    assert md(nested_uls, bullets='-') == '\n- 1\n\t- a\n\t\t- I\n\t\t- II\n\t\t- III\n\t- b\n\t- c\n- 2\n- 3\n'
+    assert md(nested_uls, bullets='-') == '\n\n- 1\n  - a\n    - I\n    - II\n    - III\n  - b\n  - c\n- 2\n- 3\n'
 
 
 def test_li_text():
-    assert md('<ul><li>foo <a href="#">bar</a></li><li>foo bar  </li><li>foo <b>bar</b>   <i>space</i>.</ul>') == '* foo [bar](#)\n* foo bar\n* foo **bar** *space*.\n'
+    assert md('<ul><li>foo <a href="#">bar</a></li><li>foo bar  </li><li>foo <b>bar</b>   <i>space</i>.</ul>') == '\n\n* foo [bar](#)\n* foo bar\n* foo **bar** *space*.\n'

tests/test_tables.py

@@ -1,4 +1,4 @@
-from markdownify import markdownify as md
+from .utils import md
 
 
 table = """<table>
@@ -57,6 +57,26 @@ table_with_paragraphs = """<table>
     </tr>
 </table>"""
 
+table_with_linebreaks = """<table>
+    <tr>
+        <th>Firstname</th>
+        <th>Lastname</th>
+        <th>Age</th>
+    </tr>
+    <tr>
+        <td>Jill</td>
+        <td>Smith
+        Jackson</td>
+        <td>50</td>
+    </tr>
+    <tr>
+        <td>Eve</td>
+        <td>Jackson
+        Smith</td>
+        <td>94</td>
+    </tr>
+</table>"""
+
 
 table_with_header_column = """<table>
     <tr>
@@ -99,6 +119,55 @@ table_head_body = """<table>
     </tbody>
 </table>"""
 
+table_head_body_missing_head = """<table>
+    <thead>
+        <tr>
+            <td>Firstname</td>
+            <td>Lastname</td>
+            <td>Age</td>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Jill</td>
+            <td>Smith</td>
+            <td>50</td>
+        </tr>
+        <tr>
+            <td>Eve</td>
+            <td>Jackson</td>
+            <td>94</td>
+        </tr>
+    </tbody>
+</table>"""
+
+table_head_body_multiple_head = """<table>
+    <thead>
+        <tr>
+            <td>Creator</td>
+            <td>Editor</td>
+            <td>Server</td>
+        </tr>
+        <tr>
+            <td>Operator</td>
+            <td>Manager</td>
+            <td>Engineer</td>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Bob</td>
+            <td>Oliver</td>
+            <td>Tom</td>
+        </tr>
+        <tr>
+            <td>Thomas</td>
+            <td>Lucas</td>
+            <td>Ethan</td>
+        </tr>
+    </tbody>
+</table>"""
+
 table_missing_text = """<table>
     <thead>
         <tr>
@@ -159,13 +228,75 @@ table_body = """<table>
     </tbody>
 </table>"""
 
+table_with_caption = """TEXT<table>
+    <caption>
+        Caption
+    </caption>
+    <tbody><tr><td>Firstname</td>
+            <td>Lastname</td>
+            <td>Age</td>
+        </tr>
+    </tbody>
+</table>"""
+
+table_with_colspan = """<table>
+    <tr>
+        <th colspan="2">Name</th>
+        <th>Age</th>
+    </tr>
+    <tr>
+        <td colspan="1">Jill</td>
+        <td>Smith</td>
+        <td>50</td>
+    </tr>
+    <tr>
+        <td>Eve</td>
+        <td>Jackson</td>
+        <td>94</td>
+    </tr>
+</table>"""
+
+table_with_undefined_colspan = """<table>
+    <tr>
+        <th colspan="undefined">Name</th>
+        <th>Age</th>
+    </tr>
+    <tr>
+        <td colspan="-1">Jill</td>
+        <td>Smith</td>
+    </tr>
+</table>"""
+
 
 def test_table():
     assert md(table) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
     assert md(table_with_html_content) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| **Jill** | *Smith* | [50](#) |\n| Eve | Jackson | 94 |\n\n'
     assert md(table_with_paragraphs) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
+    assert md(table_with_linebreaks) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith Jackson | 50 |\n| Eve | Jackson Smith | 94 |\n\n'
     assert md(table_with_header_column) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
     assert md(table_head_body) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
+    assert md(table_head_body_multiple_head) == '\n\n|  |  |  |\n| --- | --- | --- |\n| Creator | Editor | Server |\n| Operator | Manager | Engineer |\n| Bob | Oliver | Tom |\n| Thomas | Lucas | Ethan |\n\n'
+    assert md(table_head_body_missing_head) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
     assert md(table_missing_text) == '\n\n|  | Lastname | Age |\n| --- | --- | --- |\n| Jill |  | 50 |\n| Eve | Jackson | 94 |\n\n'
     assert md(table_missing_head) == '\n\n|  |  |  |\n| --- | --- | --- |\n| Firstname | Lastname | Age |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
     assert md(table_body) == '\n\n|  |  |  |\n| --- | --- | --- |\n| Firstname | Lastname | Age |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
+    assert md(table_with_caption) == 'TEXT\n\nCaption\n\n|  |  |  |\n| --- | --- | --- |\n| Firstname | Lastname | Age |\n\n'
+    assert md(table_with_colspan) == '\n\n| Name | | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
+    assert md(table_with_undefined_colspan) == '\n\n| Name | Age |\n| --- | --- |\n| Jill | Smith |\n\n'
+
+
+def test_table_infer_header():
+    assert md(table, table_infer_header=True) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
+    assert md(table_with_html_content, table_infer_header=True) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| **Jill** | *Smith* | [50](#) |\n| Eve | Jackson | 94 |\n\n'
+    assert md(table_with_paragraphs, table_infer_header=True) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
+    assert md(table_with_linebreaks, table_infer_header=True) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith Jackson | 50 |\n| Eve | Jackson Smith | 94 |\n\n'
+    assert md(table_with_header_column, table_infer_header=True) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
+    assert md(table_head_body, table_infer_header=True) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
+    assert md(table_head_body_multiple_head, table_infer_header=True) == '\n\n| Creator | Editor | Server |\n| --- | --- | --- |\n| Operator | Manager | Engineer |\n| Bob | Oliver | Tom |\n| Thomas | Lucas | Ethan |\n\n'
+    assert md(table_head_body_missing_head, table_infer_header=True) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
+    assert md(table_missing_text, table_infer_header=True) == '\n\n|  | Lastname | Age |\n| --- | --- | --- |\n| Jill |  | 50 |\n| Eve | Jackson | 94 |\n\n'
+    assert md(table_missing_head, table_infer_header=True) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
+    assert md(table_body, table_infer_header=True) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
+    assert md(table_with_caption, table_infer_header=True) == 'TEXT\n\nCaption\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n\n'
+    assert md(table_with_colspan, table_infer_header=True) == '\n\n| Name | | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
+    assert md(table_with_undefined_colspan, table_infer_header=True) == '\n\n| Name | Age |\n| --- | --- |\n| Jill | Smith |\n\n'

tests/utils.py

@@ -0,0 +1,9 @@
+from markdownify import MarkdownConverter
+
+
+# for unit testing, disable document-level stripping by default so that
+# separation newlines are included in testing
+def md(html, **options):
+    options = {"strip_document": None, **options}
+
+    return MarkdownConverter(**options).convert(html)

tox.ini

@@ -0,0 +1,15 @@
+[tox]
+envlist = py38
+
+[testenv]
+passenv = PYTHONPATH
+deps =
+	pytest==8
+	flake8
+	restructuredtext_lint
+	Pygments
+commands =
+	pytest
+	flake8 --ignore=E501,W503 markdownify tests
+	restructuredtext-lint README.rst
+