[clang] compare file paths in a more principled way

Reviewed By: jeremydubreil

Differential Revision: D4083490

fbshipit-source-id: 82db8f4

infer/src/base/Config.ml

@@ -243,34 +243,6 @@ let version_string =
     Unix.time *)
 let initial_analysis_time = Unix.time ()
 
-(* Splits a path into its parts. For example:
-   - (split "path/to/file") is [".", "path"; "to"; "file"]
-   - (split "/path/to/file") is ["/", "path"; "to"; "file"] *)
-let split path =
-  let rec loop accu p =
-    match Filename.dirname p with
-    | d when d = p -> d :: accu
-    | d -> loop ((Filename.basename p) :: accu) d in
-  loop [] path
-
-(* Recursively resolve symlinks until we get something that is not a link. Executables may be
-   (multiple levels of) symbolic links to the real binary directory, eg after `make install` or
-   packaging. *)
-let real_path path =
-  let rec resolve p =
-    match Unix.readlink p with
-    | link when Filename.is_relative link ->
-        (* [p] is a relative symbolic link *)
-        resolve ((Filename.dirname p) // link)
-    | link ->
-        (* [p] is an absolute symbolic link *)
-        resolve link
-    | exception Unix.Unix_error(Unix.EINVAL, _, _) ->
-        (* [p] is not a symbolic link *)
-        p in
-  IList.fold_left
-    (fun resolved_path p -> resolve (resolved_path // p)) "" (split path)
-
 (* 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 =

infer/src/base/DB.ml

@@ -81,6 +81,11 @@ let source_file_to_abs_path fname =
   | Relative path -> Filename.concat (project_root()) path
   | Absolute path -> path
 
+let inode_equal sf1 sf2 =
+  let stat1 = Unix.stat (source_file_to_abs_path sf1) in
+  let stat2 = Unix.stat (source_file_to_abs_path sf2) in
+  stat1.st_dev = stat2.st_dev && stat1.st_ino = stat2.st_ino
+
 let source_file_to_rel_path fname =
   match fname with
   | Relative path -> path

infer/src/base/DB.mli

@@ -92,6 +92,9 @@ val source_file_compare : source_file -> source_file -> int
 (** equality of source files *)
 val source_file_equal : source_file -> source_file -> bool
 
+(** equality of the files on disk *)
+val inode_equal : source_file -> source_file -> bool
+
 (** empty source file *)
 val source_file_empty : source_file
 

infer/src/base/Utils.ml

@@ -677,3 +677,27 @@ let rec create_path path =
   | Unix.Unix_error (Unix.ENOENT, _, _) ->
       create_path (Filename.dirname path);
       create_dir path
+
+let real_path path =
+  (* Splits a path into its parts. For example:
+     - (split "path/to/file") is [".", "path"; "to"; "file"]
+     - (split "/path/to/file") is ["/", "path"; "to"; "file"] *)
+  let split path =
+    let rec loop accu p =
+      match Filename.dirname p with
+      | d when d = p -> d :: accu
+      | d -> loop ((Filename.basename p) :: accu) d in
+    loop [] path in
+  let rec resolve p =
+    match Unix.readlink p with
+    | link when Filename.is_relative link ->
+        (* [p] is a relative symbolic link *)
+        resolve ((Filename.dirname p) // link)
+    | link ->
+        (* [p] is an absolute symbolic link *)
+        resolve link
+    | exception Unix.Unix_error(Unix.EINVAL, _, _) ->
+        (* [p] is not a symbolic link *)
+        p in
+  IList.fold_left
+    (fun resolved_path p -> resolve (resolved_path // p)) "" (split path)

infer/src/base/Utils.mli

@@ -305,3 +305,8 @@ 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
+
+(** Recursively resolve symlinks until we get something that is not a link. Executables may be
+    (multiple levels of) symbolic links to the real binary directory, eg after `make install` or
+    packaging. *)
+val real_path : string -> string

infer/src/clang/Capture.re

@@ -50,6 +50,7 @@ let init_global_state_for_capture_and_linters source_file => {
   );
   register_perf_stats_report source_file;
   Config.curr_language := Config.Clang;
+  CLocation.curr_file := source_file;
   DB.Results_dir.init source_file;
   Clang_ast_main.reset_cache ();
   CFrontend_config.reset_global_state ()

infer/src/clang/cFrontend_utils.ml

@@ -441,9 +441,10 @@ struct
     let decl_info = Clang_ast_proj.get_decl_tuple decl in
     let file_opt = (fst decl_info.Clang_ast_t.di_source_range).Clang_ast_t.sl_file in
     match file_opt with
-    | None -> false
+    | None ->
+        false
     | Some file ->
-        DB.source_file_equal
+        DB.inode_equal
           (CLocation.source_file_from_path file)
           translation_unit_context.CFrontend_config.source_file
 

infer/src/clang/cLocation.ml

@@ -31,7 +31,7 @@ let source_file_from_path path =
 
 let choose_sloc_to_update_curr_file trans_unit_ctx sloc1 sloc2 =
   match sloc2.Clang_ast_t.sl_file with
-  | Some f when DB.source_file_equal (source_file_from_path f)
+  | Some f when DB.inode_equal (source_file_from_path f)
         trans_unit_ctx.CFrontend_config.source_file ->
       sloc2
   | _ -> sloc1
@@ -62,7 +62,7 @@ let clang_to_sil_location trans_unit_ctx clang_loc procdesc_opt =
         | Some f ->
             let file_db = source_file_from_path f in
             let nloc =
-              if (DB.source_file_equal file_db trans_unit_ctx.CFrontend_config.source_file) then
+              if (DB.inode_equal file_db trans_unit_ctx.CFrontend_config.source_file) then
                 !Config.nLOC
               else -1 in
             file_db, nloc
@@ -72,11 +72,13 @@ let clang_to_sil_location trans_unit_ctx clang_loc procdesc_opt =
 let file_in_project file =
   match Config.project_root with
   | Some root ->
-      let file_in_project = string_is_prefix root file in
+      let real_root = real_path root in
+      let real_file = real_path 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 root path) file)
+          (fun path -> string_is_prefix (Filename.concat real_root path) real_file)
           paths in
       file_in_project && not (file_should_be_skipped)
   | None -> false
@@ -84,7 +86,8 @@ let file_in_project file =
 let should_do_frontend_check trans_unit_ctx (loc_start, _) =
   match loc_start.Clang_ast_t.sl_file with
   | Some file ->
-      let equal_current_source file = DB.source_file_equal (source_file_from_path file)
+      let equal_current_source file =
+        DB.inode_equal (source_file_from_path file)
           trans_unit_ctx.CFrontend_config.source_file in
       equal_current_source file ||
       (file_in_project file &&  not Config.testing_mode)
@@ -104,7 +107,7 @@ let should_translate trans_unit_ctx (loc_start, loc_end) decl_trans_context ~tra
     let path_pred path = pred (source_file_from_path path) in
     map_path_of path_pred loc
   in
-  let equal_current_source = DB.source_file_equal trans_unit_ctx.CFrontend_config.source_file
+  let equal_current_source = DB.inode_equal trans_unit_ctx.CFrontend_config.source_file
   in
   let file_in_project = map_path_of file_in_project loc_end
                         || map_path_of file_in_project loc_start in