def markdown_to_notion(markdown):
if not isinstance(markdown, str):
markdown = str(markdown)
# commonmark doesn't support strikethrough, so we need to handle it ourselves
while markdown.count("~~") >= 2:
markdown = markdown.replace("~~", "<s>", 1)
markdown = markdown.replace("~~", "</s>", 1)
parser = commonmark.Parser()
ast = prepare(parser.parse(markdown))
format = set()
notion = []
for section in ast:
_, ended_format = _extract_text_and_format_from_ast(section)
if ended_format and ended_format in format:
format.remove(ended_format)
if section["type"] == "paragraph":
notion.append(["\n\n"])
for item in section.get("children", []):
literal, new_format = _extract_text_and_format_from_ast(item)
if new_format:
format.add(new_format)
if item["type"] == "html_inline" and literal == "</s>":
format.remove(("s",))
literal = ""
if item["type"] == "softbreak":
literal = "\n"
if literal:
notion.append([literal, [list(f) for f in sorted(format)]] if format else [literal])
# in the ast format, code blocks are meant to be immediately self-closing
if ("c",) in format:
format.remove(("c",))
# remove any trailing newlines from automatic closing paragraph markers
if notion:
notion[-1][0] = notion[-1][0].rstrip("\n")
# consolidate any adjacent text blocks with identical styles
consolidated = []
for item in notion:
if consolidated and _get_format(consolidated[-1], as_set=True) == _get_format(item, as_set=True):
consolidated[-1][0] += item[0]
elif item[0]:
consolidated.append(item)
return consolidated
def markdown_to_notion(markdown):
if not isinstance(markdown, str):
markdown = str(markdown)
# commonmark doesn't support strikethrough, so we need to handle it ourselves
while markdown.count("~~") >= 2:
markdown = markdown.replace("~~", "<s>", 1)
markdown = markdown.replace("~~", "</s>", 1)
# commonmark doesn't support latex blocks, so we need to handle it ourselves
def handle_latex(match):
return f'<latex equation="{html.escape(match.group(0)[2:-2])}">\u204d</latex>'
markdown = re.sub(
r'(?<!\\\\|\$\$)(?:\\\\)*((\$\$)+)(?!(\$\$))(.+?)(?<!(\$\$))\1(?!(\$\$))',
handle_latex, markdown)
# we don't want to touch dashes, so temporarily replace them here
markdown = markdown.replace("-", "⸻")
parser = commonmark.Parser()
ast = prepare(parser.parse(markdown))
format = set()
notion = []
for section in ast:
_, ended_format = _extract_text_and_format_from_ast(section)
if ended_format and ended_format in format:
format.remove(ended_format)
if section["type"] == "paragraph":
notion.append(["\n\n"])
for item in section.get("children", []):
literal, new_format = _extract_text_and_format_from_ast(item)
if new_format:
format.add(new_format)
if item["type"] == "html_inline" and literal == "</s>":
format.remove(("s", ))
literal = ""
if item["type"] == "html_inline" and literal == "</latex>":
for f in filter(lambda f: f[0] == 'e', format):
format.remove(f)
break
literal = ""
if item["type"] == "softbreak":
literal = "\n"
if literal:
notion.append([literal, [list(f) for f in sorted(format)]]
if format else [literal])
# in the ast format, code blocks are meant to be immediately self-closing
if ("c", ) in format:
format.remove(("c", ))
# remove any trailing newlines from automatic closing paragraph markers
if notion:
notion[-1][0] = notion[-1][0].rstrip("\n")
# consolidate any adjacent text blocks with identical styles
consolidated = []
for item in notion:
if consolidated and _get_format(
consolidated[-1], as_set=True) == _get_format(item,
as_set=True):
consolidated[-1][0] += item[0]
elif item[0]:
consolidated.append(item)
return cleanup_dashes(consolidated)
def markdown_to_notion(markdown: str) -> list:
"""
Convert Markdown formatted string to Notion.
Arguments
---------
markdown : str
Text to convert.
Returns
-------
list of Block
Blocks converted from input.
"""
# commonmark doesn't support strikethrough,
# so we need to handle it ourselves
while markdown.count("~~") >= 2:
markdown = markdown.replace("~~", "<s>", 1)
markdown = markdown.replace("~~", "</s>", 1)
# we don't want to touch dashes, so temporarily replace them here
markdown = markdown.replace("-", "⸻")
parser = Parser()
ast = prepare(parser.parse(markdown))
format = set()
notion = []
for section in ast:
_, ended_format = _extract_text_and_format_from_ast(section)
if ended_format and ended_format in format:
format.remove(ended_format)
if section["type"] == "paragraph":
notion.append(["\n\n"])
for item in section.get("children", []):
literal, new_format = _extract_text_and_format_from_ast(item)
if new_format:
format.add(new_format)
if item["type"] == "html_inline" and literal == "</s>":
format.remove(("s", ))
literal = ""
if item["type"] == "softbreak":
literal = "\n"
if literal:
notion.append([literal, [list(f) for f in sorted(format)]]
if format else [literal])
# in the ast format, code blocks are meant
# to be immediately self-closing
if ("c", ) in format:
format.remove(("c", ))
# remove any trailing newlines from automatic closing paragraph markers
if notion:
notion[-1][0] = notion[-1][0].rstrip("\n")
# consolidate any adjacent text blocks with identical styles
consolidated = []
for item in notion:
if consolidated and _get_format(
consolidated[-1], as_set=True) == _get_format(item,
as_set=True):
consolidated[-1][0] += item[0]
elif item[0]:
consolidated.append(item)
return _cleanup_dashes(consolidated)
