logarion/cli/convert.ml

open Logarion

let is_older source dest = try
	Unix.((stat dest).st_mtime < (stat source).st_mtime) with _-> true

let convert cs r (text, files) = match Text.str "Content-Type" text with
	| "" | "text/plain" ->
		let source = List.hd files in
		let dest = Filename.concat r.Conversion.dir (Text.short_id text) in
		List.fold_left
			(fun a f ->
				match f.Conversion.page with None -> false || a
				| Some page ->
					let dest = dest ^ f.Conversion.ext in
					(if is_older source dest then (File_store.file dest (page r text); true) else false)
					|| a)
			false cs
	| x -> Printf.eprintf "Can't convert Content-Type: %s file: %s" x text.Text.title; false

let converters types kv =
	let n = String.split_on_char ',' types in
	let t = [] in
	let t = if List.(mem "all" n || mem "htm" n) then (Html.converter kv)::t else t in
	let t = if List.(mem "all" n || mem "atom" n) then (Atom.converter "text/html")::t else t in
	let t = if List.(mem "all" n || mem "gmi" n) then (Gemini.converter)::t else t in
	let t = if List.(mem "all" n || mem "gmi-atom" n) then (Atom.converter "text/gemini")::t else t in
	t

let directory converters noindex dir id kv =
	let empty = Topic_set.Map.empty in
	let repo = Conversion.{ id; dir; kv; topic_roots = []; topics = empty; texts = [] } in
	let fn (ts,ls,acc) ((elt,_) as r) =
		(Topic_set.to_map ts (Text.set "topics" elt)), elt::ls,
		if convert converters repo r then acc+1 else acc in
	let topics, texts, count = File_store.(fold ~dir ~order:newest fn (empty,[],0)) in
	let topic_roots = try List.rev @@ String_set.list_of_csv (Store.KV.find "Topics" kv)
		with Not_found -> Topic_set.roots topics in
	let repo = Conversion.{ repo with topic_roots; topics; texts } in
	if not noindex then List.iter (fun c -> match c.Conversion.indices with None -> () | Some f -> f repo) converters;
	Printf.printf "Converted: %d Indexed: %d\n" count (List.length texts)

let at_path types noindex path =
	match path with "" -> prerr_endline "unspecified text file or directory"
	| dir when Sys.file_exists dir && Sys.is_directory dir ->
		let fname = Filename.concat dir "index.pck" in
		(match Header_pack.of_string @@ File_store.to_string fname with
		| Error s -> prerr_endline s
		| Ok { info; peers; _ } ->
			let kv = let f = Filename.concat dir ".convert.conf" in (* TODO: better place to store convert conf? *)
				if Sys.file_exists f then File_store.of_kv_file f else Store.KV.empty in
			let kv = if Store.KV.mem "Title" kv then kv else Store.KV.add "Title" info.Header_pack.title kv in
			let kv = Store.KV.add "Locations" (String.concat ";\n" info.Header_pack.locations) kv in
			let kv = Store.KV.add "Peers" (String.concat ";\n" Header_pack.(to_str_list peers)) kv in
			let cs = converters types kv in
			directory cs noindex dir info.Header_pack.id kv)
	| path when Sys.file_exists path ->
		let repo = Conversion.{
			id = ""; dir = ""; kv = Store.KV.empty; topic_roots = [];
			topics = Topic_set.Map.empty; texts = [] } in
		let cs = converters types repo.kv in
		(match File_store.to_text path with
		| Ok text -> ignore @@ convert cs repo (text, [path])
		| Error s -> prerr_endline s)
	| path -> Printf.eprintf "Path doesn't exist: %s" path

open Cmdliner
let term =
	let path = Arg.(value & pos 0 string "" & info [] ~docv:"path"
		~doc:"Text file or directory to convert. Ff directory is provided, it must contain an index.pck (see: txt index)") in
	let types = Arg.(value & opt string "all" & info ["t"; "type"] ~docv:"output type"
		~doc:"Convert to file type") in
	let noindex = Arg.(value & flag & info ["noindex"]
		~doc:"Don't create indices in target format") in
	Term.(const at_path $ types $ noindex $ path),
	Term.info "convert" ~doc:"convert texts"
		~man:[ `S "DESCRIPTION"; `P "Convert text or indexed texts within a directory to another format.
		If path is a directory must contain an index.pck. Run `txt index` first." ]