Merge branch 'develop'

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

README.rst

@@ -139,10 +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.
@@ -167,7 +180,7 @@ 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.
 The function that handles a HTML tag named ``abc`` is called
-``convert_abc(self, el, text, convert_as_inline)`` and returns a string
+``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:
@@ -180,8 +193,8 @@ function names:
         """
         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):
@@ -195,7 +208,7 @@ function names:
         """
         Create a custom MarkdownConverter that ignores paragraphs
         """
-        def convert_p(self, el, text, convert_as_inline):
+        def convert_p(self, el, text, parent_tags):
             return ''
 
     # Create shorthand method for conversion

markdownify/__init__.py

@@ -1,16 +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'[\t \r\n]+')
-newline_whitespace_re = re.compile(r'[\t \r\n]*[\r\n][\t \r\n]*')
-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'
@@ -26,6 +51,11 @@ BACKSLASH = 'backslash'
 ASTERISK = '*'
 UNDERSCORE = '_'
 
+# Document strip styles
+LSTRIP = 'lstrip'
+RSTRIP = 'rstrip'
+STRIP = 'strip'
+
 
 def chomp(text):
     """
@@ -48,13 +78,13 @@ def abstract_inline_conversion(markup_fn):
     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):
+    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 el.find_parent(['pre', 'code', 'kbd', 'samp']):
+        if '_noformat' in parent_tags:
             return text
         prefix, suffix, text = chomp(text)
         if not text:
@@ -71,9 +101,10 @@ 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 html_heading_re.match(el.name) is not None:
+    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')
@@ -84,6 +115,41 @@ def should_remove_whitespace_outside(el):
     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
@@ -99,9 +165,11 @@ class MarkdownConverter(object):
         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
 
@@ -118,79 +186,149 @@ 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
-
-        # Remove whitespace-only textnodes just before, after or
-        # inside block-level elements.
+        # 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)
-        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 (block-level)
-            # - el is the last element in its parent (block-level)
-            # - el is adjacent to a block-level node
-            can_extract = (should_remove_inside and (not el.previous_sibling
-                                                     or not el.next_sibling)
-                           or should_remove_whitespace_outside(el.previous_sibling)
-                           or should_remove_whitespace_outside(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_strip = text.rstrip('\n')
-                newlines_left = len(text) - len(text_strip)
-                next_text = self.process_tag(el, convert_children_as_inline)
-                next_text_strip = next_text.lstrip('\n')
-                newlines_right = len(next_text) - len(next_text_strip)
-                newlines = '\n' * max(newlines_left, newlines_right)
-                text = text_strip + newlines + next_text_strip
+                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 ''
 
         # normalize whitespace if we're not inside a preformatted element
-        if not el.find_parent('pre'):
+        if 'pre' not in parent_tags:
             if self.options['wrap']:
-                text = all_whitespace_re.sub(' ', text)
+                text = re_all_whitespace.sub(' ', text)
             else:
-                text = newline_whitespace_re.sub('\n', text)
-                text = whitespace_re.sub(' ', text)
+                text = re_newline_whitespace.sub('\n', text)
+                text = re_whitespace.sub(' ', text)
 
         # escape special characters if we're not inside a preformatted or code element
-        if not el.find_parent(['pre', 'code', 'kbd', 'samp']):
-            text = self.escape(text)
+        if '_noformat' not in parent_tags:
+            text = self.escape(text, parent_tags)
 
         # remove leading whitespace at the start or just after a
         # block-level element; remove traliing whitespace at the end
@@ -198,7 +336,7 @@ class MarkdownConverter(object):
         if (should_remove_whitespace_outside(el.previous_sibling)
                 or (should_remove_whitespace_inside(el.parent)
                     and not el.previous_sibling)):
-            text = text.lstrip()
+            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)):
@@ -206,23 +344,36 @@ class MarkdownConverter(object):
 
         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:
@@ -232,38 +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.sub(r'([\\&<`[>~=+|])', r'\\\1', text)
-            # A sequence of one or more consecutive '-', preceded and
-            # followed by whitespace or start/end of fragment, might
-            # be confused with an underline of a header, or with a
-            # list marker.
-            text = re.sub(r'(\s|^)(-+(?:\s|$))', r'\1\\\2', text)
-            # A sequence of up to six consecutive '#', preceded and
-            # followed by whitespace or start/end of fragment, might
-            # be confused with an ATX heading.
-            text = re.sub(r'(\s|^)(#{1,6}(?:\s|$))', r'\1\\\2', text)
-            # '.' or ')' preceded by up to nine digits might be
-            # confused with a list item.
-            text = re.sub(r'((?:\s|^)[0-9]{1,9})([.)](?:\s|$))', r'\1\\\2',
-                          text)
+            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, columns):
-        return line_beginning_re.sub(' ' * columns, text) if text else ''
-
     def underline(self, text, pad_char):
         text = (text or '').rstrip()
         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 ''
@@ -283,15 +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.strip() + ' '
+        # 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.strip()) + '\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:
@@ -299,79 +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.strip()
         if style == UNDERLINED and n <= 2:
             line = '=' if n == 1 else '-'
             return self.underline(text, line)
-        text = all_whitespace_re.sub(' ', text)
+        text = re_all_whitespace.sub(' ', text)
         hashes = '#' * n
         if style == ATX_CLOSED:
-            return '\n%s %s %s\n\n' % (hashes, text, hashes)
-        return '\n%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
+        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") 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:
@@ -381,34 +578,44 @@ class MarkdownConverter(object):
             bullets = self.options['bullets']
             bullet = bullets[depth % len(bullets)]
         bullet = bullet + ' '
-        text = (text or '').strip()
-        text = self.indent(text, len(bullet))
-        if text:
-            text = bullet + text[len(bullet):]
+        bullet_width = len(bullet)
+        bullet_indent = ' ' * bullet_width
+
+        # 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, convert_as_inline):
-        if convert_as_inline:
-            return ' ' + text.strip() + ' '
+    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']:
             # Preserve newlines (and preceding whitespace) resulting
             # from <br> tags.  Newlines in the input have already been
             # replaced by spaces.
-            lines = text.split('\n')
-            new_lines = []
-            for line in lines:
-                line = line.lstrip()
-                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)
+            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']
@@ -416,12 +623,12 @@ 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, convert_as_inline):
+    def convert_script(self, el, text, parent_tags):
         return ''
 
-    def convert_style(self, el, text, convert_as_inline):
+    def convert_style(self, el, text, parent_tags):
         return ''
 
     convert_s = convert_del
@@ -434,38 +641,50 @@ 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_caption(self, el, text, convert_as_inline):
-        return text + '\n'
+    def convert_caption(self, el, text, parent_tags):
+        return text.strip() + '\n\n'
 
-    def convert_figcaption(self, el, text, convert_as_inline):
-        return '\n\n' + text + '\n\n'
+    def convert_figcaption(self, el, text, parent_tags):
+        return '\n\n' + text.strip() + '\n\n'
 
-    def convert_td(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, convert_as_inline):
+    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, convert_as_inline):
+    def convert_tr(self, el, text, parent_tags):
         cells = el.find_all(['td', 'th'])
+        is_first_row = el.find_previous_sibling() is None
         is_headrow = (
             all([cell.name == 'th' for cell in cells])
-            or (not el.previous_sibling and not el.parent.name == 'tbody')
-            or (not el.previous_sibling and el.parent.name == 'tbody' and len(el.parent.parent.find_all(['thead'])) < 1)
+            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
+        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():
@@ -473,13 +692,16 @@ class MarkdownConverter(object):
                 else:
                     full_colspan += 1
             underline += '| ' + ' | '.join(['---'] * full_colspan) + ' |' + '\n'
-        elif (not el.previous_sibling
-              and (el.parent.name == 'table'
-                   or (el.parent.name == 'tbody'
-                       and not el.parent.previous_sibling))):
+        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

@@ -61,6 +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

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "markdownify"
-version = "0.14.0"
+version = "1.0.0"
 authors = [{name = "Matthew Tretter", email = "m@tthewwithanm.com"}]
 description = "Convert HTML to markdown."
 readme = "README.rst"

tests/test_advanced.py

@@ -1,4 +1,4 @@
-from markdownify import markdownify as md
+from .utils import md
 
 
 def test_chomp():

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():
@@ -6,7 +6,7 @@ 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():

tests/test_conversions.py

@@ -1,4 +1,5 @@
-from markdownify import markdownify as md, ATX, ATX_CLOSED, BACKSLASH, SPACES, 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**'
 
@@ -53,11 +59,12 @@ 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'
+    assert md('<blockquote><p>Hello</p><p>Hello again</p></blockquote>') == '\n> Hello\n>\n> Hello again\n\n'
 
 
 def test_blockquote_with_paragraph():
@@ -74,11 +81,6 @@ def test_br():
     assert md('a<br />b<br />c', newline_style=BACKSLASH) == 'a\\\nb\\\nc'
 
 
-def test_caption():
-    assert md('TEXT<figure><figcaption>Caption</figcaption><span>SPAN</span></figure>') == 'TEXT\n\nCaption\n\nSPAN'
-    assert md('<figure><span>SPAN</span><figcaption>Caption</figcaption></figure>TEXT') == 'SPAN\n\nCaption\n\nTEXT'
-
-
 def test_code():
     inline_tests('code', '`')
     assert md('<code>*this_should_not_escape*</code>') == '`*this_should_not_escape*`'
@@ -99,25 +101,51 @@ def test_code():
     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### Hello\n\n'
-    assert md('<h3>Hello\n\n\nWorld</h3>') == '\n### Hello World\n\n'
-    assert md('<h4>\n\nHello</h4>') == '\n#### Hello\n\n'
-    assert md('<h5>\n\nHello</h5>') == '\n##### Hello\n\n'
-    assert md('<h5>\n\nHello\n\n</h5>') == '\n##### Hello\n\n'
-    assert md('<h5>\n\nHello   \n\n</h5>') == '\n##### Hello\n\n'
+    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():
@@ -129,22 +157,24 @@ def test_h2():
 
 
 def test_hn():
-    assert md('<h3>Hello</h3>') == '\n### Hello\n\n'
-    assert md('<h4>Hello</h4>') == '\n#### Hello\n\n'
-    assert md('<h5>Hello</h5>') == '\n##### Hello\n\n'
-    assert md('<h6>Hello</h6>') == '\n###### 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) == '\n# First\n\n## Second\n\n### Third\n\n'
-    assert md('X<h1>First</h1>', heading_style=ATX) == 'X\n# First\n\n'
-    assert md('X<h1>First</h1>', heading_style=ATX_CLOSED) == 'X\n# 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) == '\n# A P C #\n\n'
-    assert md('<h1>A <p>P</p> C </h1>', heading_style=ATX) == '\n# 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():
@@ -160,9 +190,9 @@ def test_hn_nested_simple_tag():
     ]
 
     for tag, markdown in tag_to_markdown:
-        assert md('<h3>A <' + tag + '>' + tag + '</' + tag + '> B</h3>') == '\n### 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) == '\n### 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) == '\n### A li1 li2 B\n\n'
@@ -175,18 +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>') == '\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### 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) == '\n# Hello\n\n'
-    assert md('<h2>Hello</h2>', heading_style=ATX) == '\n## 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) == '\n# Hello #\n\n'
-    assert md('<h2>Hello</h2>', heading_style=ATX_CLOSED) == '\n## 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():
@@ -214,9 +249,11 @@ def test_kbd():
 
 def test_p():
     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'
@@ -230,26 +267,31 @@ def test_p():
     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```\n*this_should_not_escape*\n```\n'
-    assert md('<pre><span>*this_should_not_escape*</span></pre>') == '\n```\n*this_should_not_escape*\n```\n'
-    assert md('<pre>\t\tthis  should\t\tnot  normalize</pre>') == '\n```\n\t\tthis  should\t\tnot  normalize\n```\n'
-    assert md('<pre><span>\t\tthis  should\t\tnot  normalize</span></pre>') == '\n```\n\t\tthis  should\t\tnot  normalize\n```\n'
-    assert md('<pre>foo<b>\nbar\n</b>baz</pre>') == '\n```\nfoo\nbar\nbaz\n```\n'
-    assert md('<pre>foo<i>\nbar\n</i>baz</pre>') == '\n```\nfoo\nbar\nbaz\n```\n'
-    assert md('<pre>foo\n<i>bar</i>\nbaz</pre>') == '\n```\nfoo\nbar\nbaz\n```\n'
-    assert md('<pre>foo<i>\n</i>baz</pre>') == '\n```\nfoo\nbaz\n```\n'
-    assert md('<pre>foo<del>\nbar\n</del>baz</pre>') == '\n```\nfoo\nbar\nbaz\n```\n'
-    assert md('<pre>foo<em>\nbar\n</em>baz</pre>') == '\n```\nfoo\nbar\nbaz\n```\n'
-    assert md('<pre>foo<code>\nbar\n</code>baz</pre>') == '\n```\nfoo\nbar\nbaz\n```\n'
-    assert md('<pre>foo<strong>\nbar\n</strong>baz</pre>') == '\n```\nfoo\nbar\nbaz\n```\n'
-    assert md('<pre>foo<s>\nbar\n</s>baz</pre>') == '\n```\nfoo\nbar\nbaz\n```\n'
-    assert md('<pre>foo<sup>\nbar\n</sup>baz</pre>', sup_symbol='^') == '\n```\nfoo\nbar\nbaz\n```\n'
-    assert md('<pre>foo<sub>\nbar\n</sub>baz</pre>', sub_symbol='^') == '\n```\nfoo\nbar\nbaz\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():
@@ -292,17 +334,17 @@ def test_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():
@@ -312,4 +354,4 @@ def test_spaces():
     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 foo \n```\nbar'
+    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

@@ -1,6 +1,6 @@
 import warnings
 from bs4 import MarkupResemblesLocatorWarning
-from markdownify import markdownify as md
+from .utils import md
 
 
 def test_asterisks():
@@ -51,7 +51,9 @@ def test_misc():
     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](link)', escape_misc=True) == r'\[text](link)'
+    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'

tests/test_lists.py

@@ -1,4 +1,4 @@
-from markdownify import markdownify as md
+from .utils import md
 
 
 nested_uls = """
@@ -42,12 +42,13 @@ nested_ols = """
 
 def test_ol():
     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'
+    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():
@@ -64,7 +65,7 @@ def test_ul():
      <li>   c
      </li>
  </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'
+    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():

tests/test_tables.py

@@ -1,4 +1,4 @@
-from markdownify import markdownify as md
+from .utils import md
 
 
 table = """<table>
@@ -141,6 +141,33 @@ table_head_body_missing_head = """<table>
     </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>
@@ -201,7 +228,10 @@ table_body = """<table>
     </tbody>
 </table>"""
 
-table_with_caption = """TEXT<table><caption>Caption</caption>
+table_with_caption = """TEXT<table>
+    <caption>
+        Caption
+    </caption>
     <tbody><tr><td>Firstname</td>
             <td>Lastname</td>
             <td>Age</td>
@@ -245,10 +275,28 @@ def test_table():
     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| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
-    assert md(table_body) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
-    assert md(table_with_caption) == 'TEXT\n\nCaption\n| Firstname | Lastname | Age |\n| --- | --- | --- |\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)