Add option to ignore filepaths matching a pattern

[dups.git] / dups.ml

diff --git a/dups.ml b/dups.ml
index b6782b6..697894d 100644 (file)
--- a/dups.ml
+++ b/dups.ml
@@ -78,17 +78,17 @@ end = struct
 end
 
 type input =
-  | Root_paths of string list
-  | Paths_on_stdin
+  | Stdin
+  | Directories of string list
 
 type output =
   | Stdout
   | Directory of string
 
 let make_input_stream = function
-  | Paths_on_stdin ->
+  | Stdin ->
       In_channel.lines stdin
-  | Root_paths paths ->
+  | Directories paths ->
       let paths = StrSet.elements (StrSet.of_list paths) in
       Stream.concat (List.map paths ~f:Directory_tree.find_files)
 
@@ -108,14 +108,13 @@ let make_output_fun = function
         );
         close_out oc
 
-let main input output =
+let main input output ignore =
   let output = make_output_fun  output in
   let input  = make_input_stream input in
   let paths_by_digest = Hashtbl.create 1_000_000 in
   let path_count = ref 0 in
   let t0 = Sys.time () in
-  Stream.iter input ~f:(fun path ->
-    incr path_count;
+  let process path =
     try
       let digest = Digest.file path in
       let count, paths =
@@ -128,14 +127,23 @@ let main input output =
       Hashtbl.replace paths_by_digest digest (count + 1, StrSet.add path paths)
     with Sys_error e ->
       eprintf "WARNING: Failed to process %S: %S\n%!" path e
+  in
+  Stream.iter input ~f:(fun path ->
+    incr path_count;
+    match ignore with
+    | Some regexp when (Str.string_match regexp path 0) ->
+        ()
+    | Some _ | None ->
+        process path
   );
   Hashtbl.iter (fun d (n, ps) -> if n > 1 then output d n ps) paths_by_digest;
   let t1 = Sys.time () in
   eprintf "Processed %d files in %f seconds.\n%!" !path_count (t1 -. t0)
 
 let () =
-  let input  = ref Paths_on_stdin in
+  let input  = ref Stdin in
   let output = ref Stdout in
+  let ignore = ref None in
   let assert_file_exists path =
     if Sys.file_exists path then
       ()
@@ -161,6 +169,10 @@ let () =
         )
       , " Output to this directory instead of stdout."
       )
+    ; ( "-ignore"
+      , Arg.String (fun regexp -> ignore := Some (Str.regexp regexp))
+      , " Ignore file paths which match this regexp pattern (see Str module)."
+      )
     ]
   in
   Arg.parse
@@ -169,10 +181,10 @@ let () =
       assert_file_exists path;
       assert_file_is_dir path;
       match !input with
-      | Paths_on_stdin ->
-          input := Root_paths [path]
-      | Root_paths paths ->
-          input := Root_paths (path :: paths)
+      | Stdin ->
+          input := Directories [path]
+      | Directories paths ->
+          input := Directories (path :: paths)
     )
     "";
-  main !input !output
+  main !input !output !ignore