[utils] Cache results of realpath

Summary: Speed up capture on MacOS by caching results of realpath. Improvements to lint speed on MacOS will be sent later

Reviewed By: jvillard

Differential Revision: D4166902

fbshipit-source-id: 4d065bc

infer/src/base/Config.ml

@@ -236,7 +236,7 @@ let initial_analysis_time = Unix.time ()
 (* Resolve symlinks to get to the real executable. The real executable is located in [bin_dir]
    below, which allows us to find [lib_dir], [models_dir], etc., relative to it. *)
 let real_exe_name =
-  Core.Std.Filename.realpath Sys.executable_name
+  realpath Sys.executable_name
 
 let current_exe =
   if !Sys.interactive then CLOpt.Interactive
@@ -402,7 +402,7 @@ let init_work_dir, is_originator =
           Sys.getcwd ()
       with _ ->
         Sys.getcwd () in
-    let real_cwd = Core.Std.Filename.realpath cwd in
+    let real_cwd = realpath cwd in
     Unix.putenv "INFER_CWD" real_cwd;
     (real_cwd, true)
 
@@ -1465,7 +1465,6 @@ and print_types = !print_types
 and print_using_diff = !print_using_diff
 and procs_csv = !procs_csv
 and procs_xml = !procs_xml
-and project_root_realpath = Core.Std.Filename.realpath project_root
 and quandary = !quandary
 and quiet = !quiet
 and reactive_mode = !reactive

infer/src/base/Config.mli

@@ -228,7 +228,6 @@ val print_using_diff : bool
 val procs_csv : string option
 val procs_xml : string option
 val project_root : string
-val project_root_realpath : string
 val quandary : bool
 val quiet : bool
 val reactive_mode : bool

infer/src/base/Utils.ml

@@ -662,3 +662,12 @@ let rec create_path path =
   | Unix.Unix_error (Unix.ENOENT, _, _) ->
       create_path (Filename.dirname path);
       create_dir path
+
+let realpath_cache = Hashtbl.create 1023
+
+let realpath path =
+  try Hashtbl.find realpath_cache path
+  with Not_found ->
+    let realpath = Core.Std.Filename.realpath path in
+    Hashtbl.add realpath_cache path realpath;
+    realpath

infer/src/base/Utils.mli

@@ -298,3 +298,7 @@ val create_dir : string -> unit
 (** [create_path path] will create a directory at [path], creating all the parent directories if
     non-existing *)
 val create_path : string -> unit
+
+(** [realpath path] returns path with all symbolic links resolved. It caches results of previous
+    calls to avoid expensive system calls *)
+val realpath : string -> string

infer/src/clang/cLocation.ml

@@ -30,21 +30,14 @@ let clang_to_sil_location trans_unit_ctx clang_loc =
   Location.{line; col; file}
 
 let file_in_project file =
-  (* Look at file paths before resolving them to real paths. Do it to
-     improve performance of most likely case when calling realpath is not needed *)
-  let file_in_project, used_project_root, used_file =
-    if string_is_prefix Config.project_root file then
-      true, Config.project_root, file
-    else if string_is_prefix Config.project_root_realpath file then
-      true, Config.project_root_realpath, file
-    else
-      let real_root = Config.project_root_realpath in
-      let real_file = Core.Std.Filename.realpath file in
-      string_is_prefix real_root real_file, real_root, real_file in
+  (* IMPORTANT: results of realpath are cached to not ruin performance *)
+  let real_root = realpath Config.project_root in
+  let real_file = realpath file in
+  let file_in_project = string_is_prefix real_root real_file in
   let paths = Config.skip_translation_headers in
   let file_should_be_skipped =
     IList.exists
-      (fun path -> string_is_prefix (Filename.concat used_project_root path) used_file)
+      (fun path -> string_is_prefix (Filename.concat real_root path) real_file)
       paths in
   file_in_project && not (file_should_be_skipped)