nimbusnote: improve list conversion (indentation; checklist, numbered…

… and unnumbered lists)

src/formats/nimbus_note.py

@@ -11,23 +11,51 @@
 import markdown_lib.common
 
 
-def clean_tables(soup):
+def streamline_tables(soup: BeautifulSoup):
     for table in soup.find_all("table"):
         tags_to_remove = ["div", "span"]
         for tag in tags_to_remove:
             for element in table.find_all(tag):
                 element.unwrap()
 
 
-def assign_lists(_soup):
-    # TODO:
+def streamline_lists(soup: BeautifulSoup):
     # - all lists are unnumbered lists (ul)
     #   - type is in the class attr (list-item-number, -bullet, -checkbox)
     # - indentation is in the class attr (indent-0)
-
-    # for task_list in soup.find_all("ul", class_="checklist"):
-    #     ...
-    pass
+    for list_ in soup.find_all("ul"):
+        current_indent = 0
+        current_list = list_
+        for item in list_.find_all("li"):
+            item_type = [
+                i[len("list-item-") :]
+                for i in item["class"]
+                if i.startswith("list-item-")
+            ][0]
+            list_type = {"checkbox": "ul", "bullet": "ul", "number": "ol"}[item_type]
+            if item_type == "checkbox":
+                item.insert(0, soup.new_tag("input", type="checkbox"))
+
+            indent = [i for i in item["class"] if i.startswith("indent-")][0]
+            indent_int = int(indent[len("indent-") :])  # 1 digit number always
+            if indent_int == 0:
+                # would be sufficient to do only one time
+                current_list.name = list_type
+                if item_type == "checkbox" and "checklist" not in current_list["class"]:
+                    current_list["class"] = ["checklist"]  # drop the other classes
+            if indent_int > current_indent:
+                # new child list
+                new_list = soup.new_tag(list_type)
+                current_list.append(new_list)
+                current_list = new_list
+                current_indent = indent_int
+            elif indent_int < current_indent:
+                # find parent list at the corresponding level
+                for _ in range(current_indent - indent_int):
+                    current_list = current_list.parent
+
+            item.attrs = {}  # remove all attributes
+            current_list.append(item)
 
 
 class Converter(converter.BaseConverter):
@@ -81,8 +109,8 @@ def convert(self, file_or_folder: Path):
             )
 
             soup = BeautifulSoup(note_body_html, "html.parser")
-            clean_tables(soup)
-            assign_lists(soup)
+            streamline_tables(soup)
+            streamline_lists(soup)
 
             note_body_markdown = markdown_lib.common.markup_to_markdown(str(soup))
             resources = self.handle_markdown_links(note_body_markdown, temp_folder_note)

src/formats/zoho_notebook.py

@@ -12,7 +12,7 @@
 import markdown_lib
 
 
-def clean_tables(soup):
+def streamline_tables(soup: BeautifulSoup):
     for table in soup.find_all("table"):
         for row in table.find_all("tr"):
             for td in row.find_all("td"):
@@ -22,9 +22,7 @@ def clean_tables(soup):
                 td.append(text_only)
 
 
-def clean_task_lists(soup):
-    # TODO: Not sure why the cleaned task lists still don't work.
-    # It works online. Maybe caused by an old pandoc version.
+def streamline_checklists(soup: BeautifulSoup):
     for task_list in soup.find_all("div", class_="checklist"):
         task_list.name = "ul"
         # remove the spans
@@ -132,8 +130,8 @@ def convert_note(self, file_: Path):
 
         # convert the note body to Markdown
         if soup.body is not None:
-            clean_tables(soup)
-            clean_task_lists(soup)
+            streamline_tables(soup)
+            streamline_checklists(soup)
             body = markdown_lib.common.markup_to_markdown(str(soup))
 
             # resources and internal links