- added debugging log messages in g{,un}zip

[helm.git] / helm / http_getter / http_getter_misc.ml

diff --git a/helm/http_getter/http_getter_misc.ml b/helm/http_getter/http_getter_misc.ml
index ab517e20ed66fdf57a188591b69e984d2ced1b10..0bc701685242d7c009c685587dfcef9ca2168687 100644 (file)
--- a/helm/http_getter/http_getter_misc.ml
+++ b/helm/http_getter/http_getter_misc.ml
@@ -1,5 +1,7 @@
 (*
- *  Copyright (C) 2003, 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
@@ -21,7 +23,7 @@
  *  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;;
@@ -46,41 +48,67 @@ let hashtbl_sorted_fold f tbl init =
   in
   List.fold_left (fun acc k -> f k (Hashtbl.find tbl k) acc) init sorted_keys
 
-let wget ?output url =
-  let flags =
-    (match output with Some file -> ["-O"; file] | None -> []) @ [url]
-  in
-  debug_print ("wget " ^ String.concat " " flags);
+let cp src dst =
   Shell.call
     ~stdout:Shell.to_dev_null ~stderr:Shell.to_dev_null
-    [Shell.cmd "wget" flags]
+    [Shell.cmd "cp" [src; dst]]
 
-  (* stderr shown as usual *)
-let gzip ?(keep = false) fname =
-  if keep then begin  (* keep original file *)
-    debug_print ("gzip -f -c " ^ fname);
+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_file (fname ^ ".gz"))
-      [Shell.cmd "gzip" ["-f"; "-c"; fname]]
-  end else begin  (* don't keep original file *)
-    debug_print ("gzip -f " ^ fname);
-    Shell.call [Shell.cmd "gzip" ["-f"; fname]]
-  end
+      ~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
 
-  (* stderr shown as usual *)
 let gunzip ?(keep = false) fname =
-  if not (Pcre.pmatch ~pat:"\\.gz$" fname) then
-    failwith "gunzip: source file doesn't end with '.gz'";
-  let basename = Pcre.replace ~pat:"\\.gz$" fname in
-  if keep then begin  (* keep original file *)
-    debug_print ("gunzip -f -c " ^ fname);
-    Shell.call
-      ~stdout:(Shell.to_file basename)
-      [Shell.cmd "gunzip" ["-f"; "-c"; fname]]
-  end else begin  (* don't keep original file *)
-    debug_print ("gunzip -f " ^ fname);
-    Shell.call [Shell.cmd "gunzip" ["-f"; fname]]
-  end
+  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") *)
@@ -102,7 +130,19 @@ let string_of_proc_status = function
   | Unix.WSTOPPED sg -> sprintf "[Stopped: %d]" sg
 
 let http_get url =
-  try
-    Some (Http_client.Convenience.http_get url)
-  with Http_client.Http_error (code, _) -> None
+  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