T M4.5: default Python LSP (basedpyright) + LSP-independent filetypes

First real language on the now-correct async + UTF-16 substrate.

- pmacs.lsp.config.python → `basedpyright-langserver --stdio`.
  basedpyright (MIT fork of pyright) re-enables inlay hints /
  semantic tokens in the OSS server that upstream pyright withholds
  for Pylance — matches the deferred-feature roadmap. No init_options:
  strictness is project config (pyrightconfig.json / [tool.pyright]);
  pmacs does not yet advertise workspace/configuration, so an
  editor-side typeCheckingMode would not be honoured regardless
  (documented in-line, with the upstream-pyright one-field override).

- LSP language detection separated from tree-sitter. pmacs.parse's
  extension registry is grammar-gated (rejects "python" — no bundled
  grammar). New user-extensible pmacs.lsp.filetypes map (py/pyi →
  python); active_buffer_language() tries grammar-backed parse first
  (rust/.rs etc. unchanged) then falls back to the map, so a language
  with a server but no grammar still auto-attaches.

- PATH-gated acceptance test mirroring m4_5_rust_analyzer_initializes;
  unique assertion: a real basedpyright must negotiate a
  positionEncoding pmacs can encode — validates Option B against a
  real strict server, not just the fake. Skips when absent.

Gate: fmt clean; clippy --all-targets -D warnings clean; lib
1229/0; m4_acceptance 61/0; m9_1 18/0; m8_1/m8_9/m8_10 green.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

builtin/runtime/lsp.lua

@@ -28,6 +28,33 @@ pmacs.lsp.config.rust = pmacs.lsp.config.rust or {
   },
 }
 
+-- Default Python config: basedpyright (an MIT fork of pyright that
+-- re-enables inlay hints / semantic tokens in the open-source server,
+-- which upstream pyright withholds for Pylance). `--stdio` is the
+-- LSP transport. No `init_options`: basedpyright/pyright take their
+-- strictness from project config (`pyrightconfig.json` /
+-- `[tool.pyright]` in `pyproject.toml`), and pmacs does not yet
+-- advertise `workspace/configuration` (a deferred capability), so an
+-- editor-side `typeCheckingMode` would not be honoured anyway. Until
+-- the project pins it, basedpyright's stricter defaults can make the
+-- diagnostics gutter noisier than upstream pyright — documented, not
+-- a bug. Users override any field from init.lua before a .py opens;
+-- swapping to upstream pyright is just `command = "pyright-langserver"`.
+pmacs.lsp.config.python = pmacs.lsp.config.python or {
+  command = "basedpyright-langserver",
+  args = { "--stdio" },
+}
+
+-- LSP-side extension → language map, deliberately independent of the
+-- tree-sitter detection in `pmacs.parse` (which is grammar-gated:
+-- Python has an LSP server but no bundled grammar). Consulted only
+-- when `pmacs.parse.language_for_path` finds nothing, so grammar-
+-- backed languages keep their existing detection. Extensible from
+-- init.lua: `pmacs.lsp.filetypes.foo = "bar"`.
+pmacs.lsp.filetypes = pmacs.lsp.filetypes or {}
+pmacs.lsp.filetypes.py = pmacs.lsp.filetypes.py or "python"
+pmacs.lsp.filetypes.pyi = pmacs.lsp.filetypes.pyi or "python"
+
 -- Per-buffer attachment record: { language, server, uri, version }.
 -- Keyed by `tostring(BufferIdLua)` because BufferIdLua hands out fresh
 -- userdata each call (so two handles to the same buffer wouldn't hash
@@ -69,7 +96,13 @@ end
 local function active_buffer_language()
   local path = active_buffer_path()
   if not path then return nil end
-  return pmacs.parse.language_for_path(path)
+  -- Grammar-backed detection first (keeps rust/.rs etc. exactly as
+  -- before); fall back to the LSP-only filetype map so languages
+  -- with a server but no tree-sitter grammar (Python) still attach.
+  local lang = pmacs.parse.language_for_path(path)
+  if lang then return lang end
+  local ext = path:match("%.([%w_]+)$")
+  return ext and pmacs.lsp.filetypes[ext] or nil
 end
 
 local function ensure_server(language)

tests/m4_acceptance.rs

@@ -1144,6 +1144,52 @@ fn m4_5_rust_analyzer_initializes() {
     let _ = mgr.borrow_mut().stop(sid);
 }
 
+/// PATH-gated, mirrors `m4_5_rust_analyzer_initializes` for the
+/// default Python server (basedpyright). Validates the whole stack
+/// against a real, strict-by-default server: the registry command +
+/// `--stdio` launches, the LSP handshake completes, and the server
+/// negotiates a `positionEncoding` against the
+/// `general.positionEncodings: ["utf-8","utf-16"]` we advertise
+/// (Option B) — proving negotiation round-trips with a real server,
+/// not only the fake. Skips cleanly when basedpyright is absent.
+#[test]
+fn m4_5_basedpyright_initializes_and_negotiates_encoding() {
+    let Ok(_) = which_binary("basedpyright-langserver") else {
+        eprintln!("basedpyright-langserver not on PATH; skipping");
+        return;
+    };
+    let (sup, mgr) = make_lsp_test_manager();
+    let mut spec = LspServerSpec::new("basedpyright", "python", "basedpyright-langserver");
+    spec.args = vec!["--stdio".into()];
+    spec.restart = LspRestartPolicy::Never;
+    let sid = mgr.borrow_mut().spawn(spec).expect("spawn basedpyright");
+    let evs = drain_lsp_until(&sup, &mgr, sid, Duration::from_secs(20), |evs| {
+        evs.iter()
+            .any(|e| matches!(e.kind, LspEventKind::Initialized { .. }))
+    });
+    let caps = evs
+        .iter()
+        .find_map(|e| match &e.kind {
+            LspEventKind::Initialized { capabilities } => Some(capabilities.clone()),
+            _ => None,
+        })
+        .expect("must observe Initialized event");
+    assert!(caps.is_object(), "capabilities should be a JSON object");
+    // If basedpyright implements LSP 3.17 position-encoding it echoes
+    // its choice, which must be one we can actually encode. A server
+    // predating 3.17 omits the field — correct too, since pmacs then
+    // defaults to UTF-16 (the spec default). What must never happen:
+    // a third encoding we don't handle.
+    let enc = caps.get("positionEncoding").and_then(|v| v.as_str());
+    assert!(
+        matches!(enc, None | Some("utf-8" | "utf-16")),
+        "basedpyright negotiated an encoding pmacs cannot handle: {enc:?}"
+    );
+    let state = mgr.borrow().state(sid).cloned();
+    assert!(matches!(state, Some(LspClientState::Initialized { .. })));
+    let _ = mgr.borrow_mut().stop(sid);
+}
+
 /// Helper: scan PATH for a binary by name. Returns the absolute
 /// path if found.
 fn which_binary(name: &str) -> std::io::Result<std::path::PathBuf> {