(*
- * Copyright (C) 2000, HELM Team.
+ * Copyright (C) 2003:
+ * Stefano Zacchiroli <zack@cs.unibo.it>
+ * for the HELM Team http://helm.cs.unibo.it/
*
* This file is part of HELM, an Hypertextual, Electronic
* Library of Mathematics, developed at the Computer Science
* MA 02111-1307, USA.
*
* For details, see the HELM World-Wide-Web page,
- * http://cs.unibo.it/helm/.
+ * http://helm.cs.unibo.it/
*)
+open Http_getter_debugger;;
+open Printf;;
+
let fold_file f init fname =
let inchan = open_in fname in
let rec fold_lines' value =
in
List.fold_left (fun acc k -> f k (Hashtbl.find tbl k) acc) init sorted_keys
+let cp src dst =
+ Shell.call
+ ~stdout:Shell.to_dev_null ~stderr:Shell.to_dev_null
+ [Shell.cmd "cp" [src; dst]]
+
+let file_scheme_RE = Pcre.regexp "^file://"
+let wget ?output url =
+ let use_wget () =
+ let flags =
+ (match output with Some file -> ["-O"; file] | None -> []) @ [url]
+ in
+ debug_print ("wget " ^ String.concat " " flags);
+ Shell.call
+ ~stdout:Shell.to_dev_null ~stderr:Shell.to_dev_null
+ [Shell.cmd "wget" flags]
+ in
+ if Pcre.pmatch ~rex:file_scheme_RE url then begin (* file:// URL *)
+ let src_fname = Pcre.replace ~rex:file_scheme_RE url in
+ match output with
+ | Some dst_fname -> cp src_fname dst_fname
+ | None ->
+ let dst_fname = Filename.basename src_fname in
+ if src_fname <> dst_fname then
+ cp src_fname dst_fname
+ else (* src and dst are the same: do nothing *)
+ ()
+ end else (* other URL, pass it to wget *)
+ use_wget ()
+
+let bufsiz = 16384 (* for g{,un}zip *)
+let trailing_dot_gz_RE = Pcre.regexp "\\.gz$" (* for g{,un}zip *)
+
+let gzip ?(keep = false) fname =
+ debug_print (sprintf "gzipping %s (keep: %b)" fname keep);
+ let (ic, oc) = (open_in fname, Gzip.open_out (fname ^ ".gz")) in
+ let buf = String.create bufsiz in
+ (try
+ while true do
+ let bytes = input ic buf 0 bufsiz in
+ if bytes = 0 then raise End_of_file else Gzip.output oc buf 0 bytes
+ done
+ with End_of_file -> ());
+ close_in ic;
+ Gzip.close_out oc;
+ if not keep then Sys.remove fname
+
+let gunzip ?(keep = false) fname =
+ debug_print (sprintf "gunzipping %s (keep: %b)" fname keep);
+ let basename = Pcre.replace ~rex:trailing_dot_gz_RE fname in
+ assert (basename <> fname);
+ let (ic, oc) = (Gzip.open_in fname, open_out basename) in
+ let buf = String.create bufsiz in
+ (try
+ while true do
+ let bytes = Gzip.input ic buf 0 bufsiz in
+ if bytes = 0 then raise End_of_file else output oc buf 0 bytes
+ done
+ with End_of_file -> ());
+ Gzip.close_in ic;
+ close_out oc;
+ if not keep then Sys.remove fname
+
+let tempfile () =
+ let buf = Buffer.create 28 in (* strlen("/tmp/fileSzb3Mw_http_getter") *)
+ Shell.call
+ ~stdout:(Shell.to_buffer buf)
+ [Shell.cmd "tempfile" ["--suffix=_http_getter"]];
+ Pcre.replace ~pat:"\n" (Buffer.contents buf)
+
+let mkdir ?(parents = false) dirname =
+ if not (Sys.file_exists dirname) then begin
+ let flags = if parents then ["-p"; dirname] else [dirname] in
+ debug_print ("mkdir " ^ String.concat " " flags);
+ Shell.call [Shell.cmd "mkdir" flags]
+ end
+
+let string_of_proc_status = function
+ | Unix.WEXITED code -> sprintf "[Exited: %d]" code
+ | Unix.WSIGNALED sg -> sprintf "[Killed: %d]" sg
+ | Unix.WSTOPPED sg -> sprintf "[Stopped: %d]" sg
+
+let http_get url =
+ if Pcre.pmatch ~rex:file_scheme_RE url then begin
+ (* file:// URL. Read data from file system *)
+ let fname = Pcre.replace ~rex:file_scheme_RE url in
+ try
+ let size = (Unix.stat fname).Unix.st_size in
+ let buf = String.create size in
+ let ic = open_in fname in
+ really_input ic buf 0 size;
+ close_in ic;
+ Some buf
+ with Unix.Unix_error (Unix.ENOENT, "stat", _) -> None
+ end else (* other URL, pass it to netclient *)
+ try
+ Some (Http_client.Convenience.http_get url)
+ with Http_client.Http_error (code, _) -> None
+