def to_xml(text, sourcepos=False, smart=False):
"""Convert markup to XML.
Parameters
----------
text: str
Text marked up with `CommonMark <http://commonmark.org>`_.
sourcepos: bool
If ``True``, add ``sourcepos`` attribute to all block elements
(that is, use ``CMARK_OPT_SOURCEPOS``).
smart: bool
Use :py:data:`~paka.cmark.lowlevel.OPT_SMART`.
Returns
-------
str
XML
"""
opts = _add_smart_to_opts(
smart, _add_sourcepos_to_opts(sourcepos, _lowlevel.OPT_DEFAULT))
text_bytes = _lowlevel.text_to_c(text)
parsed = _lowlevel.parse_document(text_bytes, len(text_bytes), opts)
root = _ffi.gc(parsed, _lowlevel.node_free)
return _lowlevel.text_from_c(_lowlevel.render_xml(root, opts))
def to_html(text, breaks=False, safe=False, sourcepos=False, smart=False):
r"""Convert markup to HTML.
Parameters
----------
text: str
Text marked up with `CommonMark <http://commonmark.org>`_.
breaks: bool or LineBreaks
How line breaks in text will be rendered. If ``True``,
``"soft"``, or :py:attr:`LineBreaks.soft` -- as newlines
(``\n``). If ``False`` -- as spaces. If ``"hard"`` or
:py:attr:`LineBreaks.hard` -- as ``<br />``\ s.
safe: bool
If ``True``, replace raw HTML (that was present in ``text``)
with HTML comment.
sourcepos: bool
If ``True``, add ``data-sourcepos`` attribute to block elements
(that is, use ``CMARK_OPT_SOURCEPOS``).
smart: bool
Use :py:data:`~paka.cmark.lowlevel.OPT_SMART`.
Returns
-------
str
HTML
"""
opts = _add_sourcepos_to_opts(
sourcepos, _add_breaks_to_opts(breaks, _lowlevel.OPT_DEFAULT))
opts = _add_smart_to_opts(smart, opts)
if safe:
opts |= _lowlevel.OPT_SAFE
text_bytes = _lowlevel.text_to_c(text)
return _lowlevel.text_from_c(
_lowlevel.markdown_to_html(text_bytes, len(text_bytes), opts))
def to_latex(text, breaks=False, width=0, smart=False):
r"""Convert markup to LaTeX.
Parameters
----------
text: str
Text marked up with `CommonMark <http://commonmark.org>`_.
breaks: bool or LineBreaks
How line breaks will be rendered. If ``True``,
``"soft"``, or :py:attr:`LineBreaks.soft` -- as newlines.
If ``False`` -- “soft break nodes” (single newlines) are
rendered as spaces. If ``"hard"`` or :py:attr:`LineBreaks.hard`
-- “soft break nodes” are rendered as ``\\\n``.
width: int
Wrap width of output by inserting line breaks (default is
``0``—no wrapping). Has no effect if ``breaks`` are ``False``.
smart: bool
Use :py:data:`~paka.cmark.lowlevel.OPT_SMART`.
Returns
-------
str
LaTeX document.
"""
opts = _add_smart_to_opts(
smart, _add_breaks_to_opts(breaks, _lowlevel.OPT_DEFAULT))
text_bytes = _lowlevel.text_to_c(text)
parsed = _lowlevel.parse_document(text_bytes, len(text_bytes), opts)
root = _ffi.gc(parsed, _lowlevel.node_free)
return _lowlevel.text_from_c(_lowlevel.render_latex(root, opts, width))
def _invoke_html_block_callback(node):
"""Take HTML block, highlight code (if possible), set new literal."""
def _get_html(body):
assert body.tag == "body"
for old_el in body.iter("pre"):
children = list(old_el)
first_child = children[0]
if not isinstance(first_child, lxml.html.HtmlComment):
continue
first_child_match = _match_highlighting_comment(
first_child.text.strip())
if not first_child_match:
continue
fence_info = first_child_match.group("fence_info")
code = children[1].text_content()
highlighted = callback(fence_info, code)
new_el = lxml.html.fragment_fromstring(highlighted)
old_el.getparent().replace(old_el, new_el)
for el in body:
yield lxml.html.tostring(el, encoding="unicode")
root = lxml.html.fragment_fromstring(_lowlevel.text_from_c(
_lowlevel.node_get_literal(node)),
create_parent="body")
buf = io.StringIO()
buf.writelines(_get_html(root))
new_literal = _lowlevel.text_to_c(buf.getvalue())
assert _lowlevel.node_set_literal(node, new_literal) == 1
def render_commonmark(text):
"""Render CommonMark as HTML.
Highlights code in code blocks.
"""
text_bytes = _lowlevel.text_to_c(text)
root = _lowlevel.parse_document(text_bytes, len(text_bytes), _OPTS)
try:
_substitute_code_blocks(root, _highlight)
result = _lowlevel.text_from_c(_lowlevel.render_html(root, _OPTS))
finally:
_lowlevel.node_free(root)
return result
def _invoke_code_block_callback(old_node):
old_contents = _lowlevel.text_from_c(
_lowlevel.node_get_literal(old_node))
fence_info = _lowlevel.text_from_c(
_lowlevel.node_get_fence_info(old_node))
new_contents = callback(fence_info, old_contents)
if new_contents is None: # can't create new contents, skip
return
new_node = _lowlevel.node_new(_lowlevel.NODE_HTML_BLOCK)
try:
assert _lowlevel.node_set_literal(
new_node, _lowlevel.text_to_c(new_contents)) == 1
assert _lowlevel.node_replace(old_node, new_node) == 1
except Exception:
_lowlevel.node_free(new_node)
raise
_lowlevel.node_free(old_node)