open Printf
-open Http_getter_debugger
+let file_scheme_prefix = "file://"
let trailing_dot_gz_RE = Pcre.regexp "\\.gz$" (* for g{,un}zip *)
let url_RE = Pcre.regexp "^([\\w.-]+)(:(\\d+))?(/.*)?$"
let http_scheme_RE = Pcre.regexp ~flags:[`CASELESS] "^http://"
-let file_scheme_RE = Pcre.regexp ~flags:[`CASELESS] "^file://"
+let file_scheme_RE = Pcre.regexp ~flags:[`CASELESS] ("^" ^ file_scheme_prefix)
let dir_sep_RE = Pcre.regexp "/"
let heading_slash_RE = Pcre.regexp "^/"
+let local_url =
+ let rex = Pcre.regexp ("^(" ^ file_scheme_prefix ^ ")(.*)(.gz)$") in
+ fun s ->
+ try
+ Some ((Pcre.extract ~rex s).(2))
+ with Not_found -> None
+
let bufsiz = 16384 (* for file system I/O *)
let tcp_bufsiz = 4096 (* for TCP I/O *)
let iter_file f = fold_file (fun line _ -> f line) ()
+let iter_buf_size = 10240
+
+let iter_file_data f fname =
+ let ic = open_in fname in
+ let buf = String.create iter_buf_size in
+ try
+ while true do
+ let bytes = input ic buf 0 iter_buf_size in
+ if bytes = 0 then raise End_of_file;
+ f (String.sub buf 0 bytes)
+ done
+ with End_of_file -> close_in ic
+
let hashtbl_sorted_fold f tbl init =
let sorted_keys =
List.sort compare (Hashtbl.fold (fun key _ keys -> key::keys) tbl [])
let sorted_keys =
List.sort compare (Hashtbl.fold (fun key _ keys -> key::keys) tbl [])
in
- List.iter (fun k -> f k (Hashtbl.find tbl k)) sorted_keys
+ List.iter (fun k -> f k (Hashtbl.find tbl k)) sorted_keys
let cp src dst =
- let (ic, oc) = (open_in src, open_out dst) 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 output oc buf 0 bytes
- done
- with End_of_file -> ());
- close_in ic; close_out oc
-
-let parse_url url =
- try
- let subs =
- Pcre.extract ~rex:url_RE (Pcre.replace ~rex:http_scheme_RE url)
- in
- (subs.(1),
- (if subs.(2) = "" then 80 else int_of_string subs.(3)),
- (if subs.(4) = "" then "/" else subs.(4)))
- with exc ->
- failwith
- (sprintf "Can't parse url: %s (exception: %s)"
- url (Printexc.to_string exc))
-let init_socket addr port =
- let inet_addr = (Unix.gethostbyname addr).Unix.h_addr_list.(0) in
- let sockaddr = Unix.ADDR_INET (inet_addr, port) in
- let suck = Unix.socket Unix.PF_INET Unix.SOCK_STREAM 0 in
- Unix.connect suck sockaddr;
- let outchan = Unix.out_channel_of_descr suck in
- let inchan = Unix.in_channel_of_descr suck in
- (inchan, outchan)
-let http_get_iter_buf ~callback url =
- let (address, port, path) = parse_url url in
- let buf = String.create tcp_bufsiz in
- let (inchan, outchan) = init_socket address port in
- output_string outchan (sprintf "GET %s\r\n" path);
- flush outchan;
- (try
- while true do
- match input inchan buf 0 tcp_bufsiz with
- | 0 -> raise End_of_file
- | bytes when bytes = tcp_bufsiz -> (* buffer full, no need to slice it *)
- callback buf
- | bytes when bytes < tcp_bufsiz -> (* buffer not full, slice it *)
- callback (String.sub buf 0 bytes)
- | _ -> (* ( bytes < 0 ) || ( bytes > tcp_bufsiz ) *)
- assert false
- done
- with End_of_file -> ());
- close_in inchan (* close also outchan, same fd *)
+ try
+ let ic = open_in src in
+ try
+ let oc = open_out dst 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 output oc buf 0 bytes
+ done
+ with
+ End_of_file -> ()
+ );
+ close_in ic; close_out oc
+ with
+ Sys_error s ->
+ Http_getter_logger.log s;
+ close_in ic
+ | e ->
+ Http_getter_logger.log (Printexc.to_string e);
+ close_in ic;
+ raise e
+ with
+ Sys_error s ->
+ Http_getter_logger.log s
+ | e ->
+ Http_getter_logger.log (Printexc.to_string e);
+ raise e
let wget ?output url =
- debug_print
+ Http_getter_logger.log
(sprintf "wgetting %s (output: %s)" url
(match output with None -> "default" | Some f -> f));
match url with
(let oc =
open_out (match output with Some f -> f | None -> Filename.basename url)
in
- http_get_iter_buf ~callback:(fun data -> output_string oc data) url;
+ Http_user_agent.get_iter (fun data -> output_string oc data) url;
close_out oc)
| scheme -> (* unsupported scheme *)
failwith ("Http_getter_misc.wget: unsupported scheme: " ^ scheme)
let gzip ?(keep = false) ?output fname =
let output = match output with None -> fname ^ ".gz" | Some fname -> fname in
- debug_print (sprintf "gzipping %s (keep: %b, output: %s)" fname keep output);
+ Http_getter_logger.log ~level:3
+ (sprintf "gzipping %s (keep: %b, output: %s)" fname keep output);
let (ic, oc) = (open_in fname, Gzip.open_out output) in
let buf = String.create bufsiz in
(try
"Http_getter_misc.gunzip: unable to determine output file name"
| Some fname -> fname
in
- debug_print (sprintf "gunzipping %s (keep: %b, output: %s)"
- fname keep output);
- let (ic, oc) = (Gzip.open_in fname, open_out output) 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 Pervasives.output oc buf 0 bytes
- done
- with End_of_file -> ());
- Gzip.close_in ic; close_out oc;
+ Http_getter_logger.log ~level:3
+ (sprintf "gunzipping %s (keep: %b, output: %s)" fname keep output);
+ (* Open the zipped file manually since Gzip.open_in may
+ * leak the descriptor if it raises an exception *)
+ let zic = open_in fname in
+ begin
+ try
+ let ic = Gzip.open_in_chan zic in
+ let oc = open_out output 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 Pervasives.output oc buf 0 bytes
+ done
+ with End_of_file -> ());
+ close_out oc;
+ Gzip.close_in ic
+ with
+ e -> close_in zic ; raise e
+ end ;
if not keep then Sys.remove fname
;;
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;
+ 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 Http_client *)
+ end else (* other URL, pass it to Http_user_agent *)
try
- Some (Http_client.http_get url)
+ Some (Http_user_agent.get url)
with e ->
- prerr_endline (sprintf
- "Warning: Http_client failed on url %s with exception: %s"
+ Http_getter_logger.log (sprintf
+ "Warning: Http_user_agent failed on url %s with exception: %s"
url (Printexc.to_string e));
None