Replace XML codeblock spaces with tabs

doc/tools/make_rst.py

@@ -1840,41 +1840,10 @@ def format_text_block(
     context: DefinitionBase,
     state: State,
 ) -> str:
-    # Linebreak + tabs in the XML should become two line breaks unless in a "codeblock"
-    pos = 0
-    while True:
-        pos = text.find("\n", pos)
-        if pos == -1:
-            break
-
-        pre_text = text[:pos]
-        indent_level = 0
-        while pos + 1 < len(text) and text[pos + 1] == "\t":
-            pos += 1
-            indent_level += 1
-        post_text = text[pos + 1 :]
-
-        # Handle codeblocks
-        if (
-            post_text.startswith("[codeblock]")
-            or post_text.startswith("[codeblock ")
-            or post_text.startswith("[gdscript]")
-            or post_text.startswith("[gdscript ")
-            or post_text.startswith("[csharp]")
-            or post_text.startswith("[csharp ")
-        ):
-            tag_text = post_text[1:].split("]", 1)[0]
-            tag_state = get_tag_and_args(tag_text)
-            result = format_codeblock(tag_state, post_text, indent_level, state)
-            if result is None:
-                return ""
-            text = f"{pre_text}{result[0]}"
-            pos += result[1] - indent_level
-
-        # Handle normal text
-        else:
-            text = f"{pre_text}\n\n{post_text}"
-            pos += 2 - indent_level
+    result = preformat_text_block(text, state)
+    if result is None:
+        return ""
+    text = result
 
     next_brac_pos = text.find("[")
     text = escape_rst(text, next_brac_pos)
@@ -2426,6 +2395,56 @@ def format_text_block(
     return text
 
 
+def preformat_text_block(text: str, state: State) -> Optional[str]:
+    result = ""
+    codeblock_tag = ""
+    indent_level = 0
+
+    for line in text.splitlines():
+        stripped_line = line.lstrip("\t")
+        tab_count = len(line) - len(stripped_line)
+
+        if codeblock_tag:
+            if line == "":
+                result += "\n"
+                continue
+
+            if tab_count < indent_level:
+                print_error(f"{state.current_class}.xml: Invalid indentation.", state)
+                return None
+
+            if stripped_line.startswith("[/" + codeblock_tag):
+                result += stripped_line
+                codeblock_tag = ""
+            else:
+                # Remove extraneous tabs and replace remaining tabs with spaces.
+                result += "\n" + "    " * (tab_count - indent_level + 1) + stripped_line
+        else:
+            if (
+                stripped_line.startswith("[codeblock]")
+                or stripped_line.startswith("[codeblock ")
+                or stripped_line.startswith("[gdscript]")
+                or stripped_line.startswith("[gdscript ")
+                or stripped_line.startswith("[csharp]")
+                or stripped_line.startswith("[csharp ")
+            ):
+                if result:
+                    result += "\n"
+                result += stripped_line
+
+                tag_text = stripped_line[1:].split("]", 1)[0]
+                tag_state = get_tag_and_args(tag_text)
+                codeblock_tag = tag_state.name
+                indent_level = tab_count
+            else:
+                # A line break in XML should become two line breaks (unless in a code block).
+                if result:
+                    result += "\n\n"
+                result += stripped_line
+
+    return result
+
+
 def format_context_name(context: Union[DefinitionBase, None]) -> str:
     context_name: str = "unknown context"
     if context is not None:
@@ -2468,50 +2487,6 @@ def escape_rst(text: str, until_pos: int = -1) -> str:
     return text
 
 
-def format_codeblock(
-    tag_state: TagState, post_text: str, indent_level: int, state: State
-) -> Union[Tuple[str, int], None]:
-    end_pos = post_text.find("[/" + tag_state.name + "]")
-    if end_pos == -1:
-        print_error(
-            f"{state.current_class}.xml: Tag depth mismatch for [{tag_state.name}]: no closing [/{tag_state.name}].",
-            state,
-        )
-        return None
-
-    opening_formatted = tag_state.name
-    if len(tag_state.arguments) > 0:
-        opening_formatted += " " + tag_state.arguments
-
-    code_text = post_text[len(f"[{opening_formatted}]") : end_pos]
-    post_text = post_text[end_pos:]
-
-    # Remove extraneous tabs
-    code_pos = 0
-    while True:
-        code_pos = code_text.find("\n", code_pos)
-        if code_pos == -1:
-            break
-
-        to_skip = 0
-        while code_pos + to_skip + 1 < len(code_text) and code_text[code_pos + to_skip + 1] == "\t":
-            to_skip += 1
-
-        if to_skip > indent_level:
-            print_error(
-                f"{state.current_class}.xml: Four spaces should be used for indentation within [{tag_state.name}].",
-                state,
-            )
-
-        if len(code_text[code_pos + to_skip + 1 :]) == 0:
-            code_text = f"{code_text[:code_pos]}\n"
-            code_pos += 1
-        else:
-            code_text = f"{code_text[:code_pos]}\n    {code_text[code_pos + to_skip + 1 :]}"
-            code_pos += 5 - to_skip
-    return (f"\n[{opening_formatted}]{code_text}{post_text}", len(f"\n[{opening_formatted}]{code_text}"))
-
-
 def format_table(f: TextIO, data: List[Tuple[Optional[str], ...]], remove_empty_columns: bool = False) -> None:
     if len(data) == 0:
         return