inshellah/test/test_inshellah.ml

open Inshellah.Parser
open Inshellah.Manpage
open Inshellah.Nushell

let failures = ref 0
let passes = ref 0

let check name condition =
  if condition then begin
    incr passes;
    Printf.printf "  PASS: %s\n" name
  end else begin
    incr failures;
    Printf.printf "  FAIL: %s\n" name
  end

let parse txt =
  match parse_help txt with
  | Ok r -> r
  | Error msg -> failwith (Printf.sprintf "parse_help failed: %s" msg)

(* --- Help parser tests --- *)

let test_gnu_basic () =
  Printf.printf "\n== GNU basic flags ==\n";
  let r = parse "  -a, --all                  do not ignore entries starting with .\n" in
  check "one entry" (List.length r.entries = 1);
  let e = List.hd r.entries in
  check "both switch" (e.switch = Both ('a', "all"));
  check "no param" (e.param = None);
  check "desc" (String.length e.desc > 0)

let test_gnu_eq_param () =
  Printf.printf "\n== GNU = param ==\n";
  let r = parse "      --block-size=SIZE      scale sizes by SIZE\n" in
  check "one entry" (List.length r.entries = 1);
  let e = List.hd r.entries in
  check "long switch" (e.switch = Long "block-size");
  check "mandatory param" (e.param = Some (Mandatory "SIZE"))

let test_gnu_opt_param () =
  Printf.printf "\n== GNU optional param ==\n";
  let r = parse "      --color[=WHEN]         color the output WHEN\n" in
  check "one entry" (List.length r.entries = 1);
  let e = List.hd r.entries in
  check "long switch" (e.switch = Long "color");
  check "optional param" (e.param = Some (Optional "WHEN"))

let test_underscore_param () =
  Printf.printf "\n== Underscore in param (TIME_STYLE) ==\n";
  let r = parse "      --time-style=TIME_STYLE  time/date format\n" in
  check "one entry" (List.length r.entries = 1);
  let e = List.hd r.entries in
  check "param with underscore" (e.param = Some (Mandatory "TIME_STYLE"))

let test_short_only () =
  Printf.printf "\n== Short-only flag ==\n";
  let r = parse "  -v                       verbose output\n" in
  check "one entry" (List.length r.entries = 1);
  check "short switch" ((List.hd r.entries).switch = Short 'v')

let test_long_only () =
  Printf.printf "\n== Long-only flag ==\n";
  let r = parse "      --help                 display help\n" in
  check "one entry" (List.length r.entries = 1);
  check "long switch" ((List.hd r.entries).switch = Long "help")

let test_multiline_desc () =
  Printf.printf "\n== Multi-line description ==\n";
  let r = parse {|      --block-size=SIZE      with -l, scale sizes by SIZE when printing them;
                               e.g., '--block-size=M'; see SIZE format below
|} in
  check "one entry" (List.length r.entries = 1);
  let e = List.hd r.entries in
  check "desc includes continuation" (String.length e.desc > 50)

let test_multiple_entries () =
  Printf.printf "\n== Multiple entries ==\n";
  let r = parse {|  -a, --all                  do not ignore entries starting with .
  -A, --almost-all           do not list implied . and ..
      --author               with -l, print the author of each file
|} in
  check "three entries" (List.length r.entries = 3)

let test_clap_short_sections () =
  Printf.printf "\n== Clap short with section headers ==\n";
  let r = parse {|INPUT OPTIONS:
  -e, --regexp=PATTERN       A pattern to search for.
  -f, --file=PATTERNFILE     Search for patterns from the given file.
SEARCH OPTIONS:
  -s, --case-sensitive       Search case sensitively.
|} in
  check "three entries" (List.length r.entries = 3);
  let e = List.hd r.entries in
  check "first is regexp" (e.switch = Both ('e', "regexp"));
  check "first has param" (e.param = Some (Mandatory "PATTERN"))

let test_clap_long_style () =
  Printf.printf "\n== Clap long style (desc below flag) ==\n";
  let r = parse {|  -H, --hidden
          Include hidden directories and files.

      --no-ignore
          Do not respect ignore files.
|} in
  check "two entries" (List.length r.entries = 2);
  let e = List.hd r.entries in
  check "hidden switch" (e.switch = Both ('H', "hidden"));
  check "desc below" (String.length e.desc > 0)

let test_clap_long_angle_param () =
  Printf.printf "\n== Clap long angle bracket param ==\n";
  let r = parse {|      --nonprintable-notation <notation>
          Set notation for non-printable characters.
|} in
  check "one entry" (List.length r.entries = 1);
  let e = List.hd r.entries in
  check "long switch" (e.switch = Long "nonprintable-notation");
  check "angle param" (e.param = Some (Mandatory "notation"))

let test_space_upper_param () =
  Printf.printf "\n== Space-separated ALL_CAPS param ==\n";
  let r = parse "  -f, --foo FOO  foo help\n" in
  check "one entry" (List.length r.entries = 1);
  let e = List.hd r.entries in
  check "switch" (e.switch = Both ('f', "foo"));
  check "space param" (e.param = Some (Mandatory "FOO"))

let test_go_cobra_flags () =
  Printf.printf "\n== Go/Cobra flags ==\n";
  let r = parse {|Flags:
  -D, --debug              Enable debug mode
  -H, --host string        Daemon socket to connect to
  -v, --version            Print version information
|} in
  check "three flag entries" (List.length r.entries = 3);
  (* Check the host flag has a type param *)
  let host = List.nth r.entries 1 in
  check "host switch" (host.switch = Both ('H', "host"));
  check "host type param" (host.param = Some (Mandatory "string"))

let test_go_cobra_subcommands () =
  Printf.printf "\n== Go/Cobra subcommands ==\n";
  let r = parse {|Common Commands:
  run         Create and run a new container from an image
  exec        Execute a command in a running container
  build       Build an image from a Dockerfile
|} in
  check "has subcommands" (List.length r.subcommands > 0)

let test_busybox_tab () =
  Printf.printf "\n== Busybox tab-indented ==\n";
  let r = parse "\t-1\tOne column output\n\t-a\tInclude names starting with .\n" in
  check "two entries" (List.length r.entries = 2);
  check "first is -1" ((List.hd r.entries).switch = Short '1')

let test_no_debug_prints () =
  Printf.printf "\n== No debug side effects ==\n";
  (* The old parser had print_endline at module load time.
     If we got here without "opt param is running" on stdout, we're good. *)
  check "no debug prints" true

(* --- Manpage parser tests --- *)

let test_manpage_tp_style () =
  Printf.printf "\n== Manpage .TP style ==\n";
  let groff = {|.SH OPTIONS
.TP
\fB\-a\fR, \fB\-\-all\fR
do not ignore entries starting with .
.TP
\fB\-A\fR, \fB\-\-almost\-all\fR
do not list implied . and ..
.TP
\fB\-\-block\-size\fR=\fISIZE\fR
with \fB\-l\fR, scale sizes by SIZE
.SH AUTHOR
Written by someone.
|} in
  let result = parse_manpage_string groff in
  check "three entries" (List.length result.entries = 3);
  if List.length result.entries >= 1 then begin
    let e = List.hd result.entries in
    check "first is -a/--all" (e.switch = Both ('a', "all"));
    check "first desc" (String.length e.desc > 0)
  end;
  if List.length result.entries >= 3 then begin
    let e = List.nth result.entries 2 in
    check "block-size switch" (e.switch = Long "block-size");
    check "block-size param" (e.param = Some (Mandatory "SIZE"))
  end

let test_manpage_ip_style () =
  Printf.printf "\n== Manpage .IP style ==\n";
  let groff = {|.SH OPTIONS
.IP "\fB\-k\fR, \fB\-\-insecure\fR"
Allow insecure connections.
.IP "\fB\-o\fR, \fB\-\-output\fR \fIfile\fR"
Write output to file.
.SH SEE ALSO
|} in
  let result = parse_manpage_string groff in
  check "two entries" (List.length result.entries = 2);
  if List.length result.entries >= 1 then begin
    let e = List.hd result.entries in
    check "first is -k/--insecure" (e.switch = Both ('k', "insecure"))
  end

let test_manpage_groff_stripping () =
  Printf.printf "\n== Groff escape stripping ==\n";
  let s = strip_groff_escapes {|\fB\-\-color\fR[=\fIWHEN\fR]|} in
  check "font escapes removed" (not (String.contains s 'f' && String.contains s 'B'));
  check "dashes converted" (String.contains s '-');
  let s2 = strip_groff_escapes {|\(aqhello\(aq|} in
  check "aq -> quote" (String.contains s2 '\'')

let test_manpage_empty_options () =
  Printf.printf "\n== Manpage with no OPTIONS section ==\n";
  let groff = {|.SH NAME
foo \- does stuff
.SH DESCRIPTION
Does stuff.
|} in
  let result = parse_manpage_string groff in
  check "no entries" (List.length result.entries = 0)

let test_slash_switch_separator () =
  Printf.printf "\n== Slash switch separator (--long / -s) ==\n";
  let r = parse "  --verbose / -v             Increase verbosity\n" in
  check "one entry" (List.length r.entries = 1);
  let e = List.hd r.entries in
  check "both switch" (e.switch = Both ('v', "verbose"));
  check "no param" (e.param = None);
  check "desc" (e.desc = "Increase verbosity")

let test_manpage_nix3_style () =
  Printf.printf "\n== Manpage nix3 style ==\n";
  let groff = {|.SH Options
.SS Logging-related options
.IP "\(bu" 3
.UR #opt-verbose
\f(CR--verbose\fR
.UE
/ \f(CR-v\fR
.IP
Increase the logging verbosity level.
.IP "\(bu" 3
.UR #opt-quiet
\f(CR--quiet\fR
.UE
.IP
Decrease the logging verbosity level.
.SH SEE ALSO
|} in
  let result = parse_manpage_string groff in
  check "two entries" (List.length result.entries = 2);
  if List.length result.entries >= 1 then begin
    let e = List.hd result.entries in
    check "verbose is Both" (e.switch = Both ('v', "verbose"));
    check "verbose desc" (String.length e.desc > 0)
  end;
  if List.length result.entries >= 2 then begin
    let e = List.nth result.entries 1 in
    check "quiet is Long" (e.switch = Long "quiet");
    check "quiet desc" (String.length e.desc > 0)
  end

let test_manpage_nix3_with_params () =
  Printf.printf "\n== Manpage nix3 with params ==\n";
  let groff = {|.SH Options
.IP "\(bu" 3
.UR #opt-arg
\f(CR--arg\fR
.UE
\fIname\fR \fIexpr\fR
.IP
Pass the value as the argument name to Nix functions.
.IP "\(bu" 3
.UR #opt-include
\f(CR--include\fR
.UE
/ \f(CR-I\fR \fIpath\fR
.IP
Add path to search path entries.
.IP
This option may be given multiple times.
.SH SEE ALSO
|} in
  let result = parse_manpage_string groff in
  check "two entries" (List.length result.entries = 2);
  if List.length result.entries >= 1 then begin
    let e = List.hd result.entries in
    check "arg is Long" (e.switch = Long "arg");
    check "arg has param" (e.param <> None)
  end;
  if List.length result.entries >= 2 then begin
    let e = List.nth result.entries 1 in
    check "include is Both" (e.switch = Both ('I', "include"));
    check "include has path param" (e.param = Some (Mandatory "path"))
  end

let test_synopsis_subcommand () =
  Printf.printf "\n== SYNOPSIS subcommand detection ==\n";
  let groff = {|.SH "SYNOPSIS"
.sp
.nf
\fBgit\fR \fBcommit\fR [\fB\-a\fR | \fB\-\-interactive\fR]
.fi
.SH "DESCRIPTION"
|} in
  let cmd = extract_synopsis_command groff in
  check "detected git commit" (cmd = Some "git commit")

let test_synopsis_standalone () =
  Printf.printf "\n== SYNOPSIS standalone command ==\n";
  let groff = {|.SH Synopsis
.LP
\f(CRnix-build\fR [\fIpaths\fR]
.SH Description
|} in
  let cmd = extract_synopsis_command groff in
  check "detected nix-build" (cmd = Some "nix-build")

let test_synopsis_nix3 () =
  Printf.printf "\n== SYNOPSIS nix3 subcommand ==\n";
  let groff = {|.SH Synopsis
.LP
\f(CRnix run\fR [\fIoption\fR] \fIinstallable\fR
.SH Description
|} in
  let cmd = extract_synopsis_command groff in
  check "detected nix run" (cmd = Some "nix run")

(* --- Nushell generation tests --- *)

let contains s sub =
  try
    let _ = Str.search_forward (Str.regexp_string sub) s 0 in true
  with Not_found -> false

let test_nushell_basic () =
  Printf.printf "\n== Nushell basic extern ==\n";
  let r = parse "  -a, --all                  do not ignore entries starting with .\n" in
  let nu = generate_extern "ls" r in
  check "has extern" (contains nu "export extern \"ls\"");
  check "has --all(-a)" (contains nu "--all(-a)");
  check "has comment" (contains nu "# do not ignore")

let test_nushell_param_types () =
  Printf.printf "\n== Nushell param type mapping ==\n";
  let r = parse {|  -w, --width=COLS           set output width
      --block-size=SIZE      scale sizes
  -o, --output FILE          output file
|} in
  let nu = generate_extern "ls" r in
  check "COLS -> int" (contains nu "--width(-w): int");
  check "SIZE -> string" (contains nu "--block-size: string");
  check "FILE -> path" (contains nu "--output(-o): path")

let test_nushell_subcommands () =
  Printf.printf "\n== Nushell subcommands ==\n";
  let r = parse {|Common Commands:
  run         Create and run a new container
  exec        Execute a command

Flags:
  -D, --debug              Enable debug mode
|} in
  let nu = generate_extern "docker" r in
  check "has main extern" (contains nu "export extern \"docker\"");
  check "has --debug" (contains nu "--debug(-D)");
  check "has run subcommand" (contains nu "export extern \"docker run\"");
  check "has exec subcommand" (contains nu "export extern \"docker exec\"")

let test_nushell_from_manpage () =
  Printf.printf "\n== Nushell from manpage ==\n";
  let groff = {|.SH OPTIONS
.TP
\fB\-a\fR, \fB\-\-all\fR
do not ignore entries starting with .
.TP
\fB\-\-block\-size\fR=\fISIZE\fR
scale sizes by SIZE
.SH AUTHOR
|} in
  let result = parse_manpage_string groff in
  let nu = generate_extern "ls" result in
  check "has extern" (contains nu "export extern \"ls\"");
  check "has --all(-a)" (contains nu "--all(-a)");
  check "has --block-size" (contains nu "--block-size: string")

let test_nushell_module () =
  Printf.printf "\n== Nushell module wrapper ==\n";
  let r = parse "  -v, --verbose              verbose output\n" in
  let nu = generate_module "myapp" r in
  check "has module" (contains nu "module myapp-completions");
  check "has extern inside" (contains nu "export extern \"myapp\"");
  check "has flag" (contains nu "--verbose(-v)")

let test_dedup_entries () =
  Printf.printf "\n== Deduplication ==\n";
  let r = parse {|  -v, --verbose              verbose output
  --verbose                  verbose mode
  -v                         be verbose
|} in
  let nu = generate_extern "test" r in
  (* Count occurrences of --verbose *)
  let count =
    let re = Str.regexp_string "--verbose" in
    let n = ref 0 in
    let i = ref 0 in
    (try while true do
       let _ = Str.search_forward re nu !i in
       incr n; i := Str.match_end ()
     done with Not_found -> ());
    !n
  in
  check "verbose appears once" (count = 1);
  check "best version kept (Both)" (contains nu "--verbose(-v)")

let test_dedup_manpage () =
  Printf.printf "\n== Dedup from manpage ==\n";
  let groff = {|.SH OPTIONS
.TP
\fB\-v\fR, \fB\-\-verbose\fR
Be verbose.
.SH DESCRIPTION
Use \fB\-v\fR for verbose output.
Use \fB\-\-verbose\fR to see more.
|} in
  let result = parse_manpage_string groff in
  let nu = generate_extern "test" result in
  check "has --verbose(-v)" (contains nu "--verbose(-v)");
  (* Should not have standalone -v or duplicate --verbose *)
  let lines = String.split_on_char '\n' nu in
  let verbose_lines = List.filter (fun l -> contains l "verbose") lines in
  check "only one verbose line" (List.length verbose_lines = 1)

let test_commands_section_subcommands () =
  Printf.printf "\n== COMMANDS section subcommand extraction ==\n";
  (* manpages like systemctl have a COMMANDS section with bold command names
   * inside .PP + .RS/.RE blocks. these should be extracted as subcommands
   * and treated as leaf nodes (no entries of their own). *)
  let groff = {|.SH OPTIONS
.TP
\fB\-\-user\fR
Talk to the service manager of the calling user.
.TP
\fB\-\-system\fR
Talk to the service manager of the system.
.SH COMMANDS
.PP
\fBstart\fR \fIUNIT\fR\&...
.RS 4
Start (activate) one or more units.
.RE
.PP
\fBstop\fR \fIUNIT\fR\&...
.RS 4
Stop (deactivate) one or more units.
.RE
.PP
\fBreload\fR \fIUNIT\fR\&...
.RS 4
Asks all units to reload their configuration.
.RE
.SH SEE ALSO
|} in
  let result = parse_manpage_string groff in
  check "has options entries" (List.length result.entries = 2);
  check "has subcommands" (List.length result.subcommands = 3);
  let sc_names = List.map (fun (sc : subcommand) -> sc.name) result.subcommands in
  check "has start" (List.mem "start" sc_names);
  check "has stop" (List.mem "stop" sc_names);
  check "has reload" (List.mem "reload" sc_names);
  (* verify subcommand descriptions are extracted *)
  let start_sc = List.find (fun (sc : subcommand) -> sc.name = "start") result.subcommands in
  check "start has desc" (String.length start_sc.desc > 0)

let test_self_listing_detection () =
  Printf.printf "\n== Self-listing subcommand detection ==\n";
  (* when a subcommand's --help shows the parent's help text,
   * the subcommand name appears in its own subcommand list.
   * the parser should detect this — tested via parse_help. *)
  let help_text = {|systemctl [OPTIONS...] COMMAND ...

Unit Commands:
  start UNIT...                       Start (activate) one or more units
  stop UNIT...                        Stop (deactivate) one or more units
  status [PATTERN...]                 Show runtime status

Options:
  --user                              Talk to the user service manager
  --system                            Talk to the system service manager
|} in
  let r = parse help_text in
  let has_start = List.exists (fun (sc : subcommand) -> sc.name = "start") r.subcommands in
  check "detected start as subcommand" has_start;
  (* the self-listing logic (in main.ml) would check: is "start" in r.subcommands?
   * here we just verify the parser extracts it correctly. *)
  check "has entries too" (List.length r.entries >= 2)

let test_nu_file_parsing () =
  Printf.printf "\n== .nu file parsing ==\n";
  let nu_source = {|module completions {

  # Unofficial CLI tool
  export extern mytool [
    --help(-h)                # Print help
    --version(-V)             # Print version
  ]

  # List all items
  export extern "mytool list" [
    --raw                     # Output as JSON
    --format(-f): string      # Output format
    --help(-h)                # Print help
    name?: string             # Filter by name
  ]

}

use completions *
|} in
  let r = Inshellah.Store.parse_nu_completions "mytool" nu_source in
  check "has entries" (List.length r.entries = 2);
  check "has subcommands" (List.length r.subcommands >= 1);
  let list_sc = List.find_opt (fun (sc : subcommand) -> sc.name = "list") r.subcommands in
  check "has list subcommand" (list_sc <> None);
  check "description" (r.description = "Unofficial CLI tool");
  (* test subcommand lookup *)
  let r2 = Inshellah.Store.parse_nu_completions "mytool list" nu_source in
  check "list has entries" (List.length r2.entries = 3);
  let has_format = List.exists (fun (e : entry) ->
    e.switch = Both ('f', "format")) r2.entries in
  check "list has --format(-f)" has_format;
  check "list has positional" (List.length r2.positionals >= 1)

let test_italic_synopsis () =
  Printf.printf "\n== Italic in SYNOPSIS ==\n";
  let groff = {|.SH Synopsis
.LP
\f(CRnix-env\fR \fIoperation\fR [\fIoptions\fR] [\fIarguments…\fR]
.SH Description
|} in
  let cmd = extract_synopsis_command groff in
  check "no phantom operation" (cmd = Some "nix-env")

let test_font_boundary_spacing () =
  Printf.printf "\n== Font boundary spacing ==\n";
  (* \fB--max-results\fR\fIcount\fR should become "--max-results count" *)
  let s = strip_groff_escapes {|\fB\-\-max\-results\fR\fIcount\fR|} in
  check "has space before param" (contains s "--max-results count");
  (* \fB--color\fR[=\fIWHEN\fR] should NOT insert space before = *)
  let s2 = strip_groff_escapes {|\fB\-\-color\fR[=\fIWHEN\fR]|} in
  check "no space before =" (contains s2 "--color[=WHEN]")

let () =
  Printf.printf "Running help parser tests...\n";
  test_gnu_basic ();
  test_gnu_eq_param ();
  test_gnu_opt_param ();
  test_underscore_param ();
  test_short_only ();
  test_long_only ();
  test_multiline_desc ();
  test_multiple_entries ();
  test_clap_short_sections ();
  test_clap_long_style ();
  test_clap_long_angle_param ();
  test_space_upper_param ();
  test_go_cobra_flags ();
  test_go_cobra_subcommands ();
  test_busybox_tab ();
  test_no_debug_prints ();

  Printf.printf "\nRunning manpage parser tests...\n";
  test_manpage_tp_style ();
  test_manpage_ip_style ();
  test_manpage_groff_stripping ();
  test_manpage_empty_options ();
  test_slash_switch_separator ();
  test_manpage_nix3_style ();
  test_manpage_nix3_with_params ();
  test_synopsis_subcommand ();
  test_synopsis_standalone ();
  test_synopsis_nix3 ();

  Printf.printf "\nRunning nushell generation tests...\n";
  test_nushell_basic ();
  test_nushell_param_types ();
  test_nushell_subcommands ();
  test_nushell_from_manpage ();
  test_nushell_module ();

  Printf.printf "\nRunning dedup and font tests...\n";
  test_dedup_entries ();
  test_dedup_manpage ();
  test_font_boundary_spacing ();

  Printf.printf "\nRunning COMMANDS section tests...\n";
  test_commands_section_subcommands ();
  test_self_listing_detection ();

  Printf.printf "\nRunning .nu and synopsis tests...\n";
  test_nu_file_parsing ();
  test_italic_synopsis ();

  Printf.printf "\n=== Results: %d passed, %d failed ===\n" !passes !failures;
  if !failures > 0 then exit 1