clap subcommands

lib/manpage.ml

@@ -717,6 +717,65 @@ let parse_manpage_string contents =
     | Some d -> d | None -> "" in
   { result with description }
 
+(* --- Clap-style SUBCOMMAND section extraction --- *)
+(* Manpages generated by clap (Rust) put each subcommand under its own
+   .SH SUBCOMMAND header with a Usage: line giving the name. *)
+
+let extract_subcommand_sections contents =
+  let lines = String.split_on_char '\n' contents in
+  let classified = List.map classify_line lines in
+  (* Split into sections at .SH boundaries *)
+  let rec collect_sections acc current_name current_lines = function
+    | [] ->
+      let acc = match current_name with
+        | Some n -> (n, List.rev current_lines) :: acc
+        | None -> acc in
+      List.rev acc
+    | Macro ("SH", args) :: rest ->
+      let acc = match current_name with
+        | Some n -> (n, List.rev current_lines) :: acc
+        | None -> acc in
+      let name = String.uppercase_ascii (String.trim args) in
+      if name = "SUBCOMMAND" || name = "SUBCOMMANDS" then
+        collect_sections acc (Some name) [] rest
+      else
+        collect_sections acc None [] rest
+    | line :: rest ->
+      collect_sections acc current_name (line :: current_lines) rest
+  in
+  let sections = collect_sections [] None [] classified in
+  (* For each SUBCOMMAND section, extract name from Usage: line and parse entries *)
+  let usage_re = Str.regexp {|Usage: \([a-zA-Z0-9_-]+\)|} in
+  List.filter_map (fun (_header, section_lines) ->
+    (* Find subcommand name from Usage: line *)
+    let name = ref None in
+    let desc_lines = ref [] in
+    List.iter (fun line ->
+      if !name = None then
+        match line with
+        | Text s ->
+          if try ignore (Str.search_forward usage_re s 0); true
+             with Not_found -> false
+          then name := Some (Str.matched_group 1 s)
+          else desc_lines := s :: !desc_lines
+        | Macro (("TP" | "B" | "BI" | "BR"), args) ->
+          let s = strip_inline_macro_args args |> strip_groff_escapes |> String.trim in
+          if try ignore (Str.search_forward usage_re s 0); true
+             with Not_found -> false
+          then name := Some (Str.matched_group 1 s)
+        | _ -> ()
+    ) section_lines;
+    match !name with
+    | None -> None
+    | Some subcmd_name ->
+      let entries = extract_entries section_lines in
+      let desc = String.concat " " (List.rev !desc_lines)
+                 |> strip_groff_escapes |> String.trim in
+      (* Remove backtick quoting common in clap output *)
+      let desc = Str.global_replace (Str.regexp "`\\([^`]*\\)`") "\\1" desc in
+      Some (subcmd_name, desc, { entries; subcommands = []; positionals = []; description = desc })
+  ) sections
+
 let read_manpage_file path =
   if Filename.check_suffix path ".gz" then begin
     let ic = Gzip.open_in path in