fix(highlighter): escape regex keyword patterns, improve comment parsing and plugin cleanup

Author: lukisch

CHANGELOG.md

@@ -13,6 +13,15 @@ Format basiert auf [Keep a Changelog](https://keepachangelog.com/de/1.1.0/).
 
 ## [0.1.2] - 2026-08-16
 
+### Bugfixes & Plugin-Resilienz (2026-08-20)
+
+- `core/highlighter.py`: Regex-Mustererstellung in `UniversalHighlighter` (`_keyword_pattern`) gehärtet, sodass Keywords und Builtins mit Satzzeichen/Metazeichen (z.B. Rubys `defined?` oder C++-Symbole) mit `re.escape()` maskiert und mit sicheren Wortgrenzen gematcht werden. Verhindert Fehl-Highlighting von Variablennamen (`define`) und unvollständiges Keyword-Highlighting.
+- `languages/declarative.py`: Parsing von `comment_style` in `DeclarativeLanguageProvider.from_dict` erweitert, sodass Einzelstring- (`#`), 1-Element-Listen (`["--"]`), 3-Element-Flachlisten (`["--", "--[[", "--]]"]`) und Dictionary-Formate deterministisch ausgewertet werden.
+- `features/plugin_manager.py`: `discover_and_load_all()` bereinigt gelöschte Plugin-Dateien beim Re-Scan; `_load_python_plugin()` registriert Provider aus `PluginInfo`-Rückgaben von `setup()` ab.
+- `languages/__init__.py`: Null- und Whitespace-Sicherheit für `get_provider_for_extension`, `get_provider_by_name` und `is_provider_registered`.
+- `ui/plugins_dialog.py`: Abbruch-Verhalten bei Vorlagenerstellung korrigiert; Detailanzeige gegen leere Provider-Felder abgesichert.
+- `tests/test_plugin_system.py`: 4 neue Testsuiten für Keyword-Escaping, Kommentarstil-Variationen, Provider-Lookups und Plugin-Dateibereinigung (103 passed, 1 skipped).
+
 ### Technische Hygiene, Metadaten & Discoverability (2026-08-16)
 
 - `tests/test_metadata.py`: Automatisierte Metadaten-, Manifest- und Plugin-Integritätstestsuite ergänzt (Version-Parität `pyproject.toml`, `version.py`, `CHANGELOG.md`, Required-Fields, Core-Docs, Plugin-JSON-Validierung).

README.md

@@ -8,7 +8,7 @@
 [![Ecosystem: dev-bricks](https://img.shields.io/badge/ecosystem-dev--bricks-blue.svg)](https://github.com/dev-bricks)
 [![Part of: open-bricks](https://img.shields.io/badge/part%20of-open--bricks-blue.svg)](https://github.com/open-bricks)
 [![LSP Ready](https://img.shields.io/badge/LSP-ready-purple.svg)]()
-[![Tests](https://img.shields.io/badge/tests-99%20passed-brightgreen.svg)]()
+[![Tests](https://img.shields.io/badge/tests-103%20passed-brightgreen.svg)]()
 [![llms.txt](https://img.shields.io/badge/llms.txt-available-green.svg)](llms.txt)
 
 [Deutsch](README_de.md) | English

README_de.md

@@ -8,7 +8,7 @@
 [![Ecosystem: dev-bricks](https://img.shields.io/badge/ecosystem-dev--bricks-blue.svg)](https://github.com/dev-bricks)
 [![Part of: open-bricks](https://img.shields.io/badge/part%20of-open--bricks-blue.svg)](https://github.com/open-bricks)
 [![LSP Ready](https://img.shields.io/badge/LSP-ready-purple.svg)]()
-[![Tests](https://img.shields.io/badge/tests-99%20passed-brightgreen.svg)]()
+[![Tests](https://img.shields.io/badge/tests-103%20passed-brightgreen.svg)]()
 [![llms.txt](https://img.shields.io/badge/llms.txt-available-green.svg)](llms.txt)
 
 [English](README.md) | Deutsch

core/highlighter.py

@@ -27,31 +27,43 @@ def set_provider(self, provider):
             self._build_rules()
         self.rehighlight()
 
+    @staticmethod
+    def _keyword_pattern(word: str) -> str:
+        """Erzeugt ein sicheres Regex-Muster mit passenden Wortgrenzen für Keywords."""
+        prefix = r'\b' if re.match(r'^\w', word) else r'(?<!\S)'
+        suffix = r'\b' if re.search(r'\w$', word) else r'(?!\w)'
+        return f"{prefix}{re.escape(word)}{suffix}"
+
     def _build_rules(self):
         """Erstellt Highlighting-Regeln aus dem Provider"""
         self.highlighting_rules = []
+        if not self.provider:
+            return
 
         # Keywords (Blau, Fett)
         kw_fmt = QTextCharFormat()
         kw_fmt.setForeground(QColor(86, 156, 214))
         kw_fmt.setFontWeight(QFont.Weight.Bold)
         for word in self.provider.get_keywords():
-            self.highlighting_rules.append(
-                (QRegularExpression(r'\b' + word + r'\b'), kw_fmt)
-            )
+            if word:
+                self.highlighting_rules.append(
+                    (QRegularExpression(self._keyword_pattern(word)), kw_fmt)
+                )
 
         # Builtins (Gelb)
         bi_fmt = QTextCharFormat()
         bi_fmt.setForeground(QColor(220, 220, 170))
         for word in self.provider.get_builtins():
-            self.highlighting_rules.append(
-                (QRegularExpression(r'\b' + word + r'\b'), bi_fmt)
-            )
+            if word:
+                self.highlighting_rules.append(
+                    (QRegularExpression(self._keyword_pattern(word)), bi_fmt)
+                )
 
         # Decorators / Preprocessor (Lila)
         dec_fmt = QTextCharFormat()
         dec_fmt.setForeground(QColor(189, 147, 249))
-        comment_char = self.provider.get_comment_style()[0]
+        comment_style = self.provider.get_comment_style()
+        comment_char = comment_style[0] if (comment_style and len(comment_style) > 0) else ""
         if comment_char == '#':
             self.highlighting_rules.append(
                 (QRegularExpression(r'@[^\n]+'), dec_fmt)

features/plugin_manager.py

@@ -99,7 +99,15 @@ def get_primary_plugin_dir(self) -> Path:
     def discover_and_load_all(self) -> List[PluginInfo]:
         """
         Durchsucht alle konfigurierten Verzeichnisse nach .json- und .py-Plugins.
+        Bereinigt zuvor geladene Plugins, deren Dateien nicht mehr existieren.
         """
+        # Vorhandene Plugin-Dateien auf Existenz prüfen und gelöschte bereinigen
+        for name, info in list(self._plugins.items()):
+            if info.file_path and not Path(info.file_path).exists():
+                self.unload_plugin(name)
+                self._plugins.pop(name, None)
+                self._failed_plugins.pop(info.file_path, None)
+
         loaded = []
         for directory in self._plugin_dirs:
             if directory.exists() and directory.is_dir():
@@ -186,7 +194,12 @@ def _load_python_plugin(self, file_path: Path) -> PluginInfo:
                 if isinstance(setup_res, LanguageProvider):
                     provider = setup_res
                 elif isinstance(setup_res, PluginInfo):
+                    if setup_res.provider:
+                        register_provider(setup_res.provider, override=True)
+                    if not setup_res.file_path:
+                        setup_res.file_path = str(file_path)
                     self._plugins[setup_res.name] = setup_res
+                    self._failed_plugins.pop(str(file_path), None)
                     return setup_res
                 else:
                     provider = getattr(module, "PROVIDER", None)

languages/__init__.py

@@ -139,9 +139,11 @@ def reset_providers() -> None:
     _notify_listeners()
 
 
-def is_provider_registered(name: str) -> bool:
+def is_provider_registered(name: Optional[str]) -> bool:
     """Prüft, ob eine Sprache unter dem Namen registriert ist."""
-    return name in PROVIDERS_BY_NAME
+    if not name or not isinstance(name, str):
+        return False
+    return name.strip() in PROVIDERS_BY_NAME
 
 
 def add_provider_listener(callback: Callable[[], None]) -> None:
@@ -156,14 +158,21 @@ def remove_provider_listener(callback: Callable[[], None]) -> None:
         _LISTENERS.remove(callback)
 
 
-def get_provider_for_extension(ext: str) -> Optional[LanguageProvider]:
+def get_provider_for_extension(ext: Optional[str]) -> Optional[LanguageProvider]:
     """Returns the LanguageProvider for a file extension (without dot)."""
-    return PROVIDERS.get(ext.lower().lstrip("."))
+    if not ext or not isinstance(ext, str):
+        return None
+    normalized = ext.strip().lower().lstrip(".")
+    if not normalized:
+        return None
+    return PROVIDERS.get(normalized)
 
 
-def get_provider_by_name(name: str) -> Optional[LanguageProvider]:
+def get_provider_by_name(name: Optional[str]) -> Optional[LanguageProvider]:
     """Returns the LanguageProvider by language name."""
-    return PROVIDERS_BY_NAME.get(name)
+    if not name or not isinstance(name, str):
+        return None
+    return PROVIDERS_BY_NAME.get(name.strip())
 
 
 def get_all_providers() -> List[LanguageProvider]:

languages/declarative.py

@@ -153,18 +153,34 @@ def from_dict(cls, data: dict) -> DeclarativeLanguageProvider:
         comment_multi_start = "/*"
         comment_multi_end = "*/"
         comment_style = data.get("comment_style")
-        if isinstance(comment_style, (list, tuple)) and len(comment_style) >= 1:
-            comment_single = str(comment_style[0])
-            if len(comment_style) >= 2 and isinstance(comment_style[1], (list, tuple)) and len(comment_style[1]) >= 2:
-                comment_multi_start = str(comment_style[1][0])
-                comment_multi_end = str(comment_style[1][1])
-            elif len(comment_style) >= 2 and comment_style[1] is None:
+        if isinstance(comment_style, str):
+            comment_single = comment_style
+            comment_multi_start = None
+            comment_multi_end = None
+        elif isinstance(comment_style, (list, tuple)):
+            if len(comment_style) >= 1:
+                comment_single = str(comment_style[0])
+            if len(comment_style) >= 2:
+                second = comment_style[1]
+                if isinstance(second, (list, tuple)) and len(second) >= 2:
+                    comment_multi_start = str(second[0]) if second[0] else None
+                    comment_multi_end = str(second[1]) if second[1] else None
+                elif second is None:
+                    comment_multi_start = None
+                    comment_multi_end = None
+                elif len(comment_style) >= 3 and isinstance(second, str) and isinstance(comment_style[2], str):
+                    comment_multi_start = str(second)
+                    comment_multi_end = str(comment_style[2])
+                else:
+                    comment_multi_start = None
+                    comment_multi_end = None
+            else:
                 comment_multi_start = None
                 comment_multi_end = None
         elif isinstance(comment_style, dict):
             comment_single = str(comment_style.get("single", "//"))
-            comment_multi_start = comment_style.get("multi_start", "/*")
-            comment_multi_end = comment_style.get("multi_end", "*/")
+            comment_multi_start = comment_style.get("multi_start")
+            comment_multi_end = comment_style.get("multi_end")
 
         return cls(
             name=name,

llms.txt

@@ -1,6 +1,6 @@
 # CodeBox - local PySide6 desktop code editor
 
-## Last-checked: 2026-08-16
+## Last-checked: 2026-08-20
 
 > CodeBox is a local-first desktop IDE for Windows developers who want a lightweight PySide6 code editor with tabs, a project tree, an integrated terminal, Git helpers, syntax highlighting, Language Server Protocol diagnostics, and an extensible JSON/Python language plugin system.
 
@@ -31,7 +31,7 @@ Part of the dev-bricks family. Python, MIT.
 - `python main.py`: Launches the CodeBox desktop interface.
 - `python main.py --open <file>`: Launches and directly opens a target file path.
 - `build_exe.bat`: Uses PyInstaller to bundle the application into a standalone executable.
-- `python -m pytest`: Runs the automated test suite (99 passed, 1 skipped).
+- `python -m pytest`: Runs the automated test suite (103 passed, 1 skipped).
 
 ## Related Projects (dev-bricks & ellmos family)
 
@@ -69,4 +69,4 @@ CodeBox local desktop IDE
 - Web search showed GitHub organization/topic visibility, but no robust independent external listing for the direct repo yet.
 - The strongest disambiguators are `dev-bricks`, `PySide6`, `local-first`, `Windows desktop IDE`, and `Language Server Protocol`.
 
-## Last-checked: 2026-08-14
+## Last-checked: 2026-08-20

tests/test_plugin_system.py

@@ -261,3 +261,90 @@ def test_main_window_updates_on_provider_change(qapp):
     assert window.lang_combo.findText("CustomLang") == -1
 
     window.close()
+
+
+def test_highlighter_keyword_escaping_and_boundaries(qapp):
+    from PySide6.QtGui import QTextDocument
+    from core.highlighter import UniversalHighlighter
+    from pathlib import Path
+
+    ruby_json = Path(__file__).resolve().parent.parent / "plugins" / "ruby_plugin.json"
+    assert ruby_json.exists()
+    provider = DeclarativeLanguageProvider.from_json_file(ruby_json)
+
+    doc = QTextDocument()
+    hl = UniversalHighlighter(doc, provider)
+    doc.setPlainText("if defined? x\n  define = 1\nend")
+
+    # Prüfe Zeile 2: '  define = 1'
+    # 'define' darf NICHT als Keyword 'defined?' gematcht werden
+    block2 = doc.findBlockByLineNumber(1)
+    for r, fmt in hl.highlighting_rules:
+        it = r.globalMatch(block2.text())
+        while it.hasNext():
+            m = it.next()
+            assert m.captured() != "define", "Variable 'define' must not match 'defined?'"
+
+    # Prüfe Zeile 1: 'if defined? x' -> 'defined?' muss vollständig matchen
+    block1 = doc.findBlockByLineNumber(0)
+    matched_keywords = []
+    for r, fmt in hl.highlighting_rules:
+        it = r.globalMatch(block1.text())
+        while it.hasNext():
+            m = it.next()
+            matched_keywords.append(m.captured())
+    assert "defined?" in matched_keywords
+    assert "if" in matched_keywords
+
+
+def test_declarative_comment_style_variations():
+    # String format
+    p1 = DeclarativeLanguageProvider.from_dict({"name": "Lang1", "extensions": ["l1"], "comment_style": "#"})
+    assert p1.get_comment_style() == ("#", None)
+
+    # 1-Element list
+    p2 = DeclarativeLanguageProvider.from_dict({"name": "Lang2", "extensions": ["l2"], "comment_style": ["--"]})
+    assert p2.get_comment_style() == ("--", None)
+
+    # Flat 3-element list
+    p3 = DeclarativeLanguageProvider.from_dict({
+        "name": "Lang3", "extensions": ["l3"], "comment_style": ["--", "--[[", "--]]"]
+    })
+    assert p3.get_comment_style() == ("--", ("--[[", "--]]"))
+
+    # Dict format without multi
+    p4 = DeclarativeLanguageProvider.from_dict({
+        "name": "Lang4", "extensions": ["l4"], "comment_style": {"single": ";"}
+    })
+    assert p4.get_comment_style() == (";", None)
+
+
+def test_safe_provider_lookups():
+    assert get_provider_for_extension(None) is None
+    assert get_provider_for_extension("") is None
+    assert get_provider_for_extension("   ") is None
+    assert get_provider_by_name(None) is None
+    assert get_provider_by_name("") is None
+    assert is_provider_registered(None) is False
+    assert is_provider_registered("") is False
+
+
+def test_plugin_manager_cleanup_on_vanished_file(tmp_path):
+    mgr = PluginManager(plugin_dirs=[tmp_path])
+    json_plugin = tmp_path / "temp_plugin.json"
+    json_plugin.write_text(json.dumps({
+        "name": "TempLang",
+        "extensions": ["tmpl"],
+        "keywords": ["test"],
+    }), encoding="utf-8")
+
+    loaded = mgr.discover_and_load_all()
+    assert len(loaded) == 1
+    assert is_provider_registered("TempLang")
+
+    # Datei löschen und erneut scannen
+    json_plugin.unlink()
+    loaded_after = mgr.discover_and_load_all()
+    assert len(loaded_after) == 0
+    assert not is_provider_registered("TempLang")
+

ui/plugins_dialog.py

@@ -177,15 +177,22 @@ def _on_selection_changed(self):
             kw_count = len(provider.get_keywords())
             bi_count = len(provider.get_builtins())
             snip_count = len(provider.get_snippets())
-            run_cmd = " ".join(provider.get_run_command("beispiel.ext"))
+            run_cmd_list = provider.get_run_command("beispiel.ext")
+            run_cmd = " ".join(str(part) for part in run_cmd_list) if run_cmd_list else "-"
             comment = provider.get_comment_style()
+            comment_single = comment[0] if (comment and len(comment) > 0 and comment[0]) else "-"
+            comment_multi = (
+                f"{comment[1][0]} ... {comment[1][1]}"
+                if (comment and len(comment) > 1 and comment[1] and len(comment[1]) >= 2)
+                else "Keine"
+            )
 
             lines = [
                 f"<b>Sprache:</b> {name}",
                 f"<b>Dateiendungen:</b> {exts}",
                 f"<b>Keywords:</b> {kw_count} | <b>Built-ins:</b> {bi_count} | <b>Snippets:</b> {snip_count}",
                 f"<b>Ausführen-Befehl:</b> <code>{run_cmd}</code>",
-                f"<b>Kommentar-Zeichen:</b> <code>{comment[0]}</code>",
+                f"<b>Kommentar-Zeichen:</b> <code>{comment_single}</code> (Mehrzeilig: <code>{comment_multi}</code>)",
             ]
             if plugin_info:
                 if plugin_info.description:
@@ -239,9 +246,14 @@ def _create_plugin_template(self):
         ext, ok = QInputDialog.getText(
             self, "Dateiendungen", f"Dateiendungen für {name} kommagetrennt (z.B. {name.lower()}, {name.lower()}s):"
         )
-        if not ok or not ext.strip():
-            ext = name.lower()
-        extensions = [e.strip().lstrip(".") for e in ext.split(",") if e.strip()]
+        if not ok:
+            return
+        if not ext.strip():
+            extensions = [name.lower()]
+        else:
+            extensions = [e.strip().lstrip(".") for e in ext.split(",") if e.strip()]
+        if not extensions:
+            extensions = [name.lower()]
 
         try:
             primary_dir = self.plugin_manager.get_primary_plugin_dir()