Group source comments in logic views

src/docforge/treesitter_logic.py

@@ -20,7 +20,7 @@ from tree_sitter import Language, Node, Parser
 from .errors import DocForgeError
 from .models import LogicEdge, LogicNode, LogicProjection
 
-_TRIVIA_NODE_TYPES = frozenset({"comment"})
+_COMMENT_NODE_TYPES = frozenset({"comment"})
 
 
 @dataclass(frozen=True)
@@ -421,10 +421,29 @@ class _TreeSitterFunctionBuilder:
         control: _Control | None,
     ) -> tuple[_Tail, ...]:
         tails = incoming
-        for statement in statements:
+        items = tuple(statements)
+        index = 0
+        while index < len(items):
             if not tails:
                 break
+            statement = items[index]
+            if statement.type in _COMMENT_NODE_TYPES:
+                comments = [statement]
+                index += 1
+                while index < len(items):
+                    candidate = items[index]
+                    previous = comments[-1]
+                    if (
+                        candidate.type not in _COMMENT_NODE_TYPES
+                        or candidate.start_point.row > previous.end_point.row + 1
+                    ):
+                        break
+                    comments.append(candidate)
+                    index += 1
+                tails = self._comment_block(tuple(comments), tails)
+                continue
             tails = self._statement(statement, tails, control=control)
+            index += 1
         return tails
 
     def _statement(
@@ -434,8 +453,8 @@ class _TreeSitterFunctionBuilder:
         *,
         control: _Control | None,
     ) -> tuple[_Tail, ...]:
-        if statement.type in _TRIVIA_NODE_TYPES:
-            return incoming
+        if statement.type in _COMMENT_NODE_TYPES:
+            return self._comment_block((statement,), incoming)
         if statement.type in self.profile.block_types:
             return self._statements(statement.named_children, incoming, control=control)
         if statement.type == "if_statement":
@@ -493,6 +512,19 @@ class _TreeSitterFunctionBuilder:
         self._connect(incoming, node_id)
         return (_Tail(node_id),)
 
+    def _comment_block(
+        self,
+        comments: tuple[Node, ...],
+        incoming: tuple[_Tail, ...],
+    ) -> tuple[_Tail, ...]:
+        node_id = self._node(
+            "comment",
+            _compact(" ".join(_comment_text(comment, self.raw) for comment in comments), 480),
+            comments[0],
+        )
+        self._connect(incoming, node_id)
+        return (_Tail(node_id),)
+
     def _if(
         self,
         statement: Node,
@@ -830,6 +862,15 @@ def _text(node: Node, raw: bytes) -> str:
     return raw[node.start_byte : node.end_byte].decode("utf-8", errors="replace")
 
 
+def _comment_text(node: Node, raw: bytes) -> str:
+    value = _text(node, raw).strip()
+    if value.startswith("//"):
+        return value[2:].strip()
+    if value.startswith("/*") and value.endswith("*/"):
+        value = value[2:-2]
+    return " ".join(line.strip().removeprefix("*").strip() for line in value.splitlines()).strip()
+
+
 def _compact(value: str, limit: int = 240) -> str:
     compact = " ".join(value.strip().split())
     return compact if len(compact) <= limit else f"{compact[: limit - 1]}…"