comprehensive completion generation: native, manpage, --help

Three-strategy pipeline with priority: native completion generators
(e.g. CMD completions nushell) > manpage parsing > --help fallback.
Single `generate` command produces one module-wrapped .nu file per
command. Parallel execution scaled to cores, 200ms timeouts, ELF
string scanning to skip binaries without -h support, native gzip
decompression via camlzip, SYNOPSIS-based subcommand detection,
nix3 manpage strategy, deduplication, nushell builtin exclusion.
This commit is contained in:
atagen 2026-03-21 02:07:46 +11:00
parent 01ccf64efc
commit 7f0ec8ab4d
9 changed files with 937 additions and 265 deletions

View file

@ -1,6 +1,90 @@
open Parser
(* Map a param name/type hint to a nushell type *)
(* Nushell built-in commands and keywords *)
let nushell_builtins = [
"alias"; "all"; "ansi"; "any"; "append"; "ast"; "attr";
"bits"; "break"; "bytes";
"cal"; "cd"; "char"; "chunk-by"; "chunks"; "clear"; "collect";
"columns"; "commandline"; "compact"; "complete"; "config"; "const";
"continue"; "cp";
"date"; "debug"; "decode"; "def"; "default"; "describe"; "detect";
"do"; "drop"; "du";
"each"; "echo"; "encode"; "enumerate"; "error"; "every"; "exec";
"exit"; "explain"; "explore"; "export"; "export-env"; "extern";
"fill"; "filter"; "find"; "first"; "flatten"; "for"; "format"; "from";
"generate"; "get"; "glob"; "grid"; "group-by";
"hash"; "headers"; "help"; "hide"; "hide-env"; "histogram";
"history"; "http";
"if"; "ignore"; "input"; "insert"; "inspect"; "interleave"; "into";
"is-admin"; "is-empty"; "is-not-empty"; "is-terminal"; "items";
"job"; "join";
"keybindings"; "kill";
"last"; "length"; "let"; "let-env"; "lines"; "load-env"; "loop"; "ls";
"match"; "math"; "merge"; "metadata"; "mkdir"; "mktemp"; "module";
"move"; "mut"; "mv";
"nu-check"; "nu-highlight";
"open"; "overlay";
"panic"; "par-each"; "parse"; "path"; "plugin"; "port"; "prepend"; "print"; "ps";
"query";
"random"; "reduce"; "reject"; "rename"; "return"; "reverse"; "rm";
"roll"; "rotate"; "run-external";
"save"; "schema"; "scope"; "select"; "seq"; "shuffle"; "skip"; "sleep";
"slice"; "sort"; "sort-by"; "source"; "source-env"; "split"; "start";
"stor"; "str"; "sys";
"table"; "take"; "tee"; "term"; "timeit"; "to"; "touch"; "transpose";
"try"; "tutor";
"ulimit"; "umask"; "uname"; "uniq"; "uniq-by"; "unlet"; "update";
"upsert"; "url"; "use";
"values"; "version"; "view";
"watch"; "where"; "which"; "while"; "whoami"; "window"; "with-env"; "wrap";
"zip";
]
let builtin_set = lazy (
let tbl = Hashtbl.create (List.length nushell_builtins) in
List.iter (fun s -> Hashtbl.replace tbl s true) nushell_builtins;
tbl)
let is_nushell_builtin cmd =
Hashtbl.mem (Lazy.force builtin_set) cmd
let dedup_entries entries =
let key_of entry =
match entry.switch with
| Short c -> Printf.sprintf "-%c" c
| Long l | Both (_, l) -> Printf.sprintf "--%s" l
in
let score entry =
let sw = match entry.switch with Both _ -> 10 | _ -> 0 in
let p = match entry.param with Some _ -> 5 | None -> 0 in
let d = min 5 (String.length entry.desc / 10) in
sw + p + d
in
let best = Hashtbl.create 64 in
List.iter (fun e ->
let k = key_of e in
match Hashtbl.find_opt best k with
| Some prev when score prev >= score e -> ()
| _ -> Hashtbl.replace best k e
) entries;
let covered_shorts = Hashtbl.create 16 in
Hashtbl.iter (fun _ e ->
match e.switch with
| Both (c, _) -> Hashtbl.replace covered_shorts c true
| _ -> ()
) best;
let seen = Hashtbl.create 64 in
List.filter_map (fun e ->
let k = key_of e in
if Hashtbl.mem seen k then None
else
match e.switch with
| Short c when Hashtbl.mem covered_shorts c -> None
| _ ->
Hashtbl.add seen k true;
Hashtbl.find_opt best k
) entries
let nushell_type_of_param = function
| "FILE" | "file" | "PATH" | "path" | "DIR" | "dir" | "DIRECTORY"
| "FILENAME" | "PATTERNFILE" -> "path"
@ -8,75 +92,63 @@ let nushell_type_of_param = function
| "LINES" | "DEPTH" | "depth" -> "int"
| _ -> "string"
(* Escape a nushell string: wrap in double quotes, escape inner quotes *)
let escape_nu s =
let buf = Buffer.create (String.length s + 2) in
String.iter (fun c ->
match c with
| '"' -> Buffer.add_string buf "\\\""
| '\\' -> Buffer.add_string buf "\\\\"
| _ -> Buffer.add_char buf c
) s;
Buffer.contents buf
if not (String.contains s '"') && not (String.contains s '\\') then s
else begin
let buf = Buffer.create (String.length s + 4) in
String.iter (fun c -> match c with
| '"' -> Buffer.add_string buf "\\\""
| '\\' -> Buffer.add_string buf "\\\\"
| _ -> Buffer.add_char buf c
) s;
Buffer.contents buf
end
(* Format a single flag for nushell extern *)
let format_flag entry =
let buf = Buffer.create 64 in
Buffer.add_string buf " ";
(* Flag name *)
(match entry.switch with
| Both (s, l) ->
Buffer.add_string buf (Printf.sprintf "--%s(-%c)" l s)
| Long l ->
Buffer.add_string buf (Printf.sprintf "--%s" l)
| Short s ->
Buffer.add_string buf (Printf.sprintf "-%c" s));
(* Type annotation *)
(match entry.param with
| Some (Mandatory name) ->
Buffer.add_string buf ": ";
Buffer.add_string buf (nushell_type_of_param name)
| Some (Optional name) ->
Buffer.add_string buf ": ";
Buffer.add_string buf (nushell_type_of_param name)
| None -> ());
(* Description as comment *)
if String.length entry.desc > 0 then begin
(* Pad to align comments *)
let current_len = Buffer.length buf in
let target = max (current_len + 1) 40 in
for _ = current_len to target - 1 do
Buffer.add_char buf ' '
done;
Buffer.add_string buf "# ";
Buffer.add_string buf entry.desc
end;
Buffer.contents buf
let name = match entry.switch with
| Both (s, l) -> Printf.sprintf "--%s(-%c)" l s
| Long l -> Printf.sprintf "--%s" l
| Short s -> Printf.sprintf "-%c" s
in
let typed = match entry.param with
| Some (Mandatory p) | Some (Optional p) -> ": " ^ nushell_type_of_param p
| None -> ""
in
let flag = " " ^ name ^ typed in
if String.length entry.desc = 0 then flag
else
let pad_len = max 1 (40 - String.length flag) in
flag ^ String.make pad_len ' ' ^ "# " ^ entry.desc
let write_extern buf cmd_name result =
let entries = dedup_entries result.entries in
Printf.bprintf buf "export extern \"%s\" [\n" (escape_nu cmd_name);
List.iter (fun e ->
Buffer.add_string buf (format_flag e); Buffer.add_char buf '\n'
) entries;
Buffer.add_string buf "]\n";
List.iter (fun (sc : subcommand) ->
Printf.bprintf buf "\nexport extern \"%s %s\" [ # %s\n]\n"
(escape_nu cmd_name) (escape_nu sc.name) (escape_nu sc.desc)
) result.subcommands
(* Generate nushell extern definition for a command *)
let generate_extern cmd_name result =
let buf = Buffer.create 1024 in
(* Main extern with flags *)
Buffer.add_string buf (Printf.sprintf "export extern \"%s\" [\n" (escape_nu cmd_name));
List.iter (fun entry ->
Buffer.add_string buf (format_flag entry);
Buffer.add_char buf '\n'
) result.entries;
Buffer.add_string buf "]\n";
(* Subcommand externs *)
List.iter (fun (sc : subcommand) ->
Buffer.add_string buf
(Printf.sprintf "\nexport extern \"%s %s\" [ # %s\n]\n"
(escape_nu cmd_name) (escape_nu sc.name) (escape_nu sc.desc))
) result.subcommands;
write_extern buf cmd_name result;
Buffer.contents buf
(* Generate a complete nushell module *)
let generate_module cmd_name result =
Printf.sprintf "module %s-completions {\n%s}\n"
cmd_name (generate_extern cmd_name result)
let module_name_of cmd_name =
let s = String.map (function
| ('a'..'z' | 'A'..'Z' | '0'..'9' | '-' | '_') as c -> c | _ -> '-') cmd_name in
s ^ "-completions"
let generate_module cmd_name result =
let m = module_name_of cmd_name in
let buf = Buffer.create 1024 in
Printf.bprintf buf "module %s {\n" m;
write_extern buf cmd_name result;
Printf.bprintf buf "}\n\nuse %s *\n" m;
Buffer.contents buf
(* Generate from manpage entries (no subcommands) *)
let generate_extern_from_entries cmd_name entries =
let result = { entries; subcommands = [] } in
generate_extern cmd_name result
generate_extern cmd_name { entries; subcommands = [] }