From e045e1533c48629566affe03ad436117818f55ae Mon Sep 17 00:00:00 2001
From: christoph_xd <c.brandau91@gmail.com>
Date: Fri, 18 Jul 2025 22:49:49 +0200
Subject: [PATCH] add tcl parser

---
 server/src/lsp_server.py    | 27 ++++++++++--------
 server/src/parser/parser.py | 57 +++++++++++++++++++++----------------
 server/src/parser/tcl.lark  | 38 +++++++++++--------------
 3 files changed, 63 insertions(+), 59 deletions(-)

diff --git a/server/src/lsp_server.py b/server/src/lsp_server.py
index 07cec2a..288306d 100644
--- a/server/src/lsp_server.py
+++ b/server/src/lsp_server.py
@@ -13,6 +13,7 @@ import sys
 import sysconfig
 import traceback
 from typing import Any, Optional, Sequence
+from parser.parser import parse_tcl, format_tree
 
 
 # **********************************************************
@@ -179,18 +180,20 @@ def _get_severity(*_codes: list[str]) -> lsp.DiagnosticSeverity:
 @LSP_SERVER.feature(lsp.TEXT_DOCUMENT_FORMATTING)
 def formatting(params: lsp.DocumentFormattingParams) -> list[lsp.TextEdit] | None:
     """LSP handler for textDocument/formatting request."""
-    # If your tool is a formatter you can use this handler to provide
-    # formatting support on save. You have to return an array of lsp.TextEdit
-    # objects, to provide your formatted results.
-
-    document = LSP_SERVER.workspace.get_document(params.text_document.uri)
-    edits = _formatting_helper(document)
-    if edits:
-        return edits
-
-    # NOTE: If you provide [] array, VS Code will clear the file of all contents.
-    # To indicate no changes to file return None.
-    return None
+    text = LSP_SERVER.workspace.get_text_document(params.text_document.uri)
+    tree, _ = parse_tcl(text)
+    if not tree:
+        return []
+    new_text = format_tree(tree)
+    return [
+        {
+            "range": {
+                "start": {"line": 0, "character": 0},
+                "end": {"line": len(text.splitlines()), "character": 0},
+            },
+            "newText": new_text,
+        }
+    ]
 
 
 def _formatting_helper(document: workspace.Document) -> list[lsp.TextEdit] | None:
diff --git a/server/src/parser/parser.py b/server/src/parser/parser.py
index 366230d..0481917 100644
--- a/server/src/parser/parser.py
+++ b/server/src/parser/parser.py
@@ -1,32 +1,39 @@
-from lark import Lark, Transformer, Token, v_args
 from pathlib import Path
+from lark import Lark, Tree, Token, UnexpectedInput
 
-GRAMMAR_FILE = Path(__file__).parent.joinpath("tcl.lark")
-TCL_GRAMMAR = GRAMMAR_FILE.read_text(encoding="utf-8")
+tcl_parser = Lark.open(
+    Path(__file__).parent.joinpath("tcl.lark"), parser="lalr", propagate_positions=True
+)
 
 
-@v_args(inline=True)
-class TCLTransformer(Transformer):
-    def start(self, *stmts):
-        return list(stmts)
+def parse_tcl(source: str):
+    try:
+        tree = tcl_parser.parse(source)
+        return tree, []
+    except UnexpectedInput as e:
+        # e.line, e.column enthalten Position
+        diagnostic = {
+            "range": {
+                "start": {"line": e.line - 1, "character": e.column - 1},
+                "end": {"line": e.line - 1, "character": e.column},
+            },
+            "message": f"Syntaxfehler: {e}",
+            "severity": 1,  # Error
+        }
+        return None, [diagnostic]
 
-    def command(self, *elements):
-        return list(elements)
 
-    def braced(self, *content):
-        return "{" + "".join(c for c in content) + "}"
+def format_tree(tree):
+    def fmt(node, indent=0):
+        if isinstance(node, Token):
+            return node.value
+        elif isinstance(node, Tree):
+            if node.data == "cmd":
+                parts = []
+                for child in node.children:
+                    parts.append(fmt(child, indent))
+                return " ".join(parts) + ";\n"
+            # weitere Node-Typen …
+        return ""
 
-    def quoted(self, *content):
-        return '"' + "".join(c for c in content) + '"'
-
-    def variable(self, token):
-        return token.value
-
-    def cmdsubst(self, *inner):
-        return "[" + " ".join(inner) + "]"
-
-    def bare(self, token):
-        return token.value
-
-    def COMMENT(self, token):
-        return token.value
+    return "".join(fmt(tree))
diff --git a/server/src/parser/tcl.lark b/server/src/parser/tcl.lark
index b8a7353..3674b1b 100644
--- a/server/src/parser/tcl.lark
+++ b/server/src/parser/tcl.lark
@@ -1,27 +1,21 @@
-start: statement*
+// tcl.lark
+// sehr vereinfachtes Beispiel – für volle Abdeckung musst du die offizielle TCL-Spec implementieren
 
-?statement: command | COMMENT
+?start: script
 
-COMMENT: /#[^\n]*/
+script: stmt*
+stmt: command ";"?                      -> cmd
+     | newline                        -> empty
 
-command: element+
+command: WORD arg*
+arg: WORD                             -> bareword
+   | STRING                           -> string
+   | BRACED                            -> braced
 
-?element: braced
-| quoted
-| variable
-| cmdsubst
-| bare
+WORD: /[^\s\{\}";]+/
+STRING: "\"" /([^"\\]|\\.)*/ "\""     // "…"
+BRACED:  "{" /([^{}\\]|\\.)*/ "}"     // {…}
 
-braced: "{" inner_braced* "}"
-inner_braced: /[^{}]+/ | braced
-
-quoted: '"' inner_quoted* '"'
-inner_quoted: /[^"{}]+/ | braced
-
-variable: "$" /[A-Za-z_][A-Za-z0-9_]*/
-cmdsubst: "[" start "]"
-
-bare: /[^\s{}$"]+/
-
-%import common.WS_INLINE
-%ignore WS_INLINE
\ No newline at end of file
+%import common.NEWLINE   -> newline
+%import common.WS
+%ignore WS