ForcePilot/backend/package/yuxi/channels/adapters/twitch/markdown_utils.py

56 lines
1.7 KiB
Python
Raw Normal View History

from __future__ import annotations
import re
_LINK_PATTERN = re.compile(r"\[([^\]]*)\]\([^)]*\)")
_IMG_PATTERN = re.compile(r"!\[([^\]]*)\]\([^)]*\)")
_BOLD_PATTERN = re.compile(r"\*\*(.+?)\*\*")
_BOLD_ALT_PATTERN = re.compile(r"__(.+?)__")
_ITALIC_PATTERN = re.compile(r"\*(.+?)\*")
_ITALIC_ALT_PATTERN = re.compile(r"_(.+?)_")
_STRIKETHROUGH_PATTERN = re.compile(r"~~(.+?)~~")
_FENCED_CODE_PATTERN = re.compile(r"```[^\n]*\n(.*?)```", re.DOTALL)
_INLINE_CODE_PATTERN = re.compile(r"`([^`]+)`")
_HEADING_PATTERN = re.compile(r"^#{1,6}\s+", re.MULTILINE)
_UNORDERED_LIST_PATTERN = re.compile(r"^[\-\*]\s+", re.MULTILINE)
_ORDERED_LIST_PATTERN = re.compile(r"^\d+\.\s+", re.MULTILINE)
_BLOCKQUOTE_PATTERN = re.compile(r"^>\s?", re.MULTILINE)
_HR_PATTERN = re.compile(r"^[\-\*_]{3,}\s*$", re.MULTILINE)
_MULTISPACE_PATTERN = re.compile(r"[ \t]+")
_MULTINEWLINE_PATTERN = re.compile(r"\n{3,}")
def strip_twitch_markdown(text: str) -> str:
if not text:
return text
text = _IMG_PATTERN.sub("", text)
text = _LINK_PATTERN.sub(r"\1", text)
text = _FENCED_CODE_PATTERN.sub(r"\1", text)
text = _HR_PATTERN.sub("", text)
text = _BOLD_PATTERN.sub(r"\1", text)
text = _BOLD_ALT_PATTERN.sub(r"\1", text)
text = _ITALIC_PATTERN.sub(r"\1", text)
text = _ITALIC_ALT_PATTERN.sub(r"\1", text)
text = _STRIKETHROUGH_PATTERN.sub(r"\1", text)
text = _INLINE_CODE_PATTERN.sub(r"\1", text)
text = _HEADING_PATTERN.sub("", text)
text = _ORDERED_LIST_PATTERN.sub("", text)
text = _UNORDERED_LIST_PATTERN.sub("", text)
text = _BLOCKQUOTE_PATTERN.sub("", text)
text = _MULTISPACE_PATTERN.sub(" ", text)
text = _MULTINEWLINE_PATTERN.sub("\n\n", text)
return text.strip()