Merge pull request #24 from sudoskys/dev

 feat(render): add support for task list items

README.md

@@ -110,4 +110,4 @@ print(converted)
 
 output as follows:
 
-![.github/result.png](.github/result-3.png)
+![.github/result.png](.github/result-5.png)

playground/show_send.py

@@ -28,8 +28,12 @@
 *bold _italic bold ~~italic bold strikethrough ||italic bold strikethrough spoiler||~~ __underline italic bold___ bold*
 __underline italic bold__
 [link](https://www.google.com)
+
 - [ ] Uncompleted task list item
+- [ ] __Underline__ ~~Strikethrough~~ _italic_ Item
 - [x] Completed task list item
+- [x] **Bold** ||Spoiler|| `Inline Code` Item
+
 > Quote
 
 > Multiline Quote In Markdown it's not possible to send multiline quote in telegram without using code block or html tag but telegramify_markdown can do it.

playground/use_case.py

@@ -9,14 +9,18 @@
 ~strikethrough~
 """
 quote = """>test"""
-
+task = """
+- [x] task1?
+- [x] task2?
+"""
 test_md = """
 **bold text**
 ||spoiler||
 """
-converted = telegramify_markdown.convert(quote)
+converted = telegramify_markdown.convert(task)
 print(converted)
 
 rule = re.compile(r"(?<!\\)(?:\\\\)*\|\|(.+?)\|\|", re.DOTALL)
-
+pattern = re.compile(r"^- \[([ xX])\] (.*)", re.DOTALL | re.MULTILINE)
 print(rule.findall(test_md))
+print(pattern.findall(task))

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "telegramify-markdown"
-version = "0.1.11"
+version = "0.1.12"
 description = "Convert Markdown to a format usable by Telegram."
 authors = [
     { name = "sudoskys", email = "coldlando@hotmail.com" },

src/telegramify_markdown/customize.py

@@ -11,6 +11,10 @@ class Symbol(object):
     # "🖼"
     link = "\N{LINK SYMBOL}"
     # "🔗"
+    task_completed = "\N{WHITE HEAVY CHECK MARK}"
+    # "✅"
+    task_uncompleted = "\N{BALLOT BOX WITH CHECK}"
+    # "☑️"
 
 
 markdown_symbol = Symbol()

src/telegramify_markdown/render.py

@@ -4,6 +4,7 @@
 from typing import Iterable
 
 from mistletoe import block_token, span_token
+from mistletoe.block_token import BlockToken
 from mistletoe.markdown_renderer import MarkdownRenderer, LinkReferenceDefinition, Fragment
 from mistletoe.span_token import SpanToken
 from telebot import formatting
@@ -15,6 +16,39 @@ class Spoiler(SpanToken):
     pattern = re.compile(r"(?<!\\)(?:\\\\)*\|\|(.+?)\|\|", re.DOTALL)
 
 
+class TaskListItem(BlockToken):
+    """
+    Custom block token for task list items in Markdown.
+    Matches task list items like '- [ ]' and '- [x]'.
+
+    Attributes:
+        checked (bool): whether the task is checked.
+        content (str): the description of the task list item.
+    """
+    repr_attributes = BlockToken.repr_attributes + ("checked", "content")
+    pattern = re.compile(r'^( *)(- \[([ xX])\] )(.*)')
+
+    def __init__(self, match):
+        self.indentation, self.marker, self.checked_char, self.content = match
+        self.checked = self.checked_char.lower() == 'x'
+        super().__init__(lines=self.content, tokenize_func=span_token.tokenize_inner)
+
+    @classmethod
+    def start(cls, line):
+        stripped = line.lstrip()
+        if not stripped.startswith("- [ ]") and not stripped.startswith("- [x]"):
+            return False
+        return bool(cls.pattern.match(line))
+
+    @classmethod
+    def read(cls, lines):
+        line = next(lines)
+        match = cls.pattern.match(line)
+        if match:
+            return match.groups()
+        raise ValueError("TaskListItem did not match expected format.")
+
+
 def escape_markdown(content: str, unescape_html: bool = True) -> str:
     """
     Escapes Markdown characters in a string of Markdown with optional HTML unescaping.
@@ -45,11 +79,13 @@ def __init__(self, *extras, **kwargs):
             *chain(
                 (
                     Spoiler,
+                    TaskListItem,
                 ),
                 extras
             )
         )
         self.render_map["Spoiler"] = self.render_spoiler
+        self.render_map["TaskListItem"] = self.render_task_list_item
 
     def render_quote(
             self, token: block_token.Quote, max_line_length: int
@@ -141,10 +177,26 @@ def render_strikethrough(
     def render_spoiler(self, token: Spoiler) -> Iterable[Fragment]:
         return self.embed_span(Fragment("||"), token.children)
 
+    def render_task_list_item(self,
+                              token: TaskListItem,
+                              max_line_length: int
+                              ) -> Iterable[str]:
+        symbol = markdown_symbol.task_completed if token.checked else markdown_symbol.task_uncompleted
+        if self.normalize_whitespace:
+            indentation = 0
+        else:
+            indentation = len(token.indentation)
+        lines = self.span_to_lines(
+            token.children, max_line_length=max_line_length
+        )
+        space = " " * indentation
+        return self.prefix_lines(lines or [""], f"{space}{symbol} ")
+
     def render_list_item(
             self, token: block_token.ListItem, max_line_length: int
     ) -> Iterable[str]:
-        if str(token.leader).strip().endswith("."):
+        token_origin = str(token.leader).strip()
+        if token_origin.endswith("."):
             token.leader = formatting.escape_markdown(token.leader) + " "
         else:
             token.leader = formatting.escape_markdown("⦁")
@@ -154,12 +206,13 @@ def render_link_reference_definition(
             self, token: LinkReferenceDefinition
     ) -> Iterable[Fragment]:
         yield from (
-            Fragment(markdown_symbol.link + formatting.mlink(
-                content=token.title if token.title else token.label,
-                url=token.dest,
-                escape=True
-            )
-                     ),
+            Fragment(
+                markdown_symbol.link + formatting.mlink(
+                    content=token.title if token.title else token.label,
+                    url=token.dest,
+                    escape=True
+                )
+            ),
         )
 
     def render_image(self, token: span_token.Image) -> Iterable[Fragment]: