Create dump-file-stats command (#115)

Author: liam923

src/ocaml-index/bin/ocaml_index.ml

@@ -15,11 +15,12 @@ let rewrite_root = ref false
 let store_shapes = ref false
 let do_not_use_cmt_loadpath = ref false
 
-type command = Aggregate | Dump | Stats | Gather_shapes
+type command = Aggregate | Dump | Dump_file_stats | Stats | Gather_shapes
 
 let parse_command = function
   | "aggregate" -> Some Aggregate
   | "dump" -> Some Dump
+  | "dump-file-stats" -> Some Dump_file_stats
   | "stats" -> Some Stats
   | "gather-shapes" -> Some Gather_shapes
   | _ -> None
@@ -90,6 +91,19 @@ let () =
     List.iter
       (fun file -> Index_format.(read_exn ~file |> pp Format.std_formatter))
       (List.rev !input_files_rev)
+  | Some Dump_file_stats ->
+    List.iter
+      (fun file ->
+        let open Merlin_index_format.Index_format in
+        let index = read_exn ~file in
+        Printf.printf "File stats for index %S:\n" file;
+        Stats.iter
+          (fun file { mtime; size; source_digest } ->
+            Printf.printf "  %S: { mtime=%f; size=%d; source_digest=%S }\n" file
+              mtime size
+              (Option.value source_digest ~default:"none"))
+          index.stats)
+      (List.rev !input_files_rev)
   | Some Gather_shapes ->
     Index.gather_shapes ~output_file:!output_file (List.rev !input_files_rev)
   | Some Stats ->

src/ocaml-index/tests/tests-dirs/dump-file-stats.t

@@ -0,0 +1,25 @@
+Test using the dump-file-stats subcommand
+
+mtime will not be consistent between tests, so don't include it
+  $ normalize () {
+  >   cat | sed -E "s/mtime=[0-9\.]+/mtime=<mtime>/g"
+  > }
+
+Create a small project
+  $ mkdir foo
+  $ cat > foo/a.ml << EOF
+  > let hello = "hello"
+  > EOF
+  $ cat > foo/b.ml << EOF
+  > let world = "world"
+  > EOF
+
+Compile the project and create an index file
+  $ $OCAMLC -c -bin-annot -bin-annot-occurrences foo/a.ml foo/b.ml
+  $ ocaml-index aggregate foo/a.cmt foo/b.cmt -o foo.merlin-index
+
+Dump the file stats from the index
+  $ ocaml-index dump-file-stats foo.merlin-index | normalize
+  File stats for index "foo.merlin-index":
+    "foo/a.ml": { mtime=<mtime>; size=20; source_digest="\147<\155xp\000:M2\170\163\134K`\235\226" }
+    "foo/b.ml": { mtime=<mtime>; size=20; source_digest="erv\218\000\233\177\190e\189\199\026q\156\195#" }