Crashcontext ocaml method summary merging

Reviewed By: sblackshear

Differential Revision: D3643101

fbshipit-source-id: eaea332

infer/lib/python/inferlib/analyze.py

@@ -30,10 +30,6 @@ csv.field_size_limit(sys.maxsize)
 
 INFER_ANALYZE_BINARY = 'InferAnalyze'
 
-CRASHCONTEXT_METHOD_FIELD = 'method'
-CRASHCONTEXT_LOCATION_FIELD = 'location'
-CRASHCONTEXT_CALLEES_FIELD = 'callees'
-
 
 def get_infer_version():
     try:
@@ -492,54 +488,6 @@ class AnalyzerWrapper(object):
 
         return exit_status
 
-    def crashcontext_stitch_summaries(self):
-        """Take crashcontext per-method summaries and join them together to
-        produce the final crashcontext.json output file."""
-
-        crashcontext_dir = os.path.join(self.args.infer_out, 'crashcontext')
-        summaries_map_by_frame_id = {}
-        st_json = utils.load_json_from_path(self.args.stacktrace)
-        stacktrace = st_json['stack_trace']
-        k = 1   # Where k is the number of levels of inlined calls.
-        for f in os.listdir(crashcontext_dir):
-            if f.endswith('.json'):
-                path = os.path.join(crashcontext_dir, f)
-                method_summary = utils.load_json_from_path(path)
-                method_signature = method_summary[CRASHCONTEXT_METHOD_FIELD]
-                method_name = method_signature.split('(')[0]
-                src_path = method_summary[CRASHCONTEXT_LOCATION_FIELD]['file']
-                line = method_summary[CRASHCONTEXT_LOCATION_FIELD]['line']
-                frame_id = "{0}({1}:{2})".format(
-                    method_name,
-                    os.path.basename(src_path),
-                    line)
-                summaries_map_by_frame_id[frame_id] = method_summary
-
-        def expand(summary, k):
-            if k == 0:
-                # Make sure leaf nodes have an empty 'callees' field.
-                leaf_nodes = summary[CRASHCONTEXT_CALLEES_FIELD]
-                for leaf_node in leaf_nodes:
-                    if CRASHCONTEXT_CALLEES_FIELD not in leaf_node:
-                        leaf_node[CRASHCONTEXT_CALLEES_FIELD] = []
-                return summary
-            else:
-                NotImplementedError()  # TODO
-
-        json_frames = []
-        for frame in stacktrace:
-            frame_id = frame.strip()
-            if not frame_id:
-                continue
-            assert frame_id.startswith('at ')
-            frame_id = frame_id[3:]
-            assert frame_id in summaries_map_by_frame_id
-            summary = summaries_map_by_frame_id[frame_id]
-            json_frames.append(expand(summary, k - 1))
-        out_json = {}
-        out_json['stack'] = json_frames
-        out_file = os.path.join(crashcontext_dir, 'crashcontext.json')
-        utils.dump_json_to_path(out_json, out_file)
 
     def read_proc_stats(self):
         proc_stats_path = os.path.join(
@@ -573,8 +521,6 @@ class AnalyzerWrapper(object):
         if self.args.analyzer not in [config.ANALYZER_COMPILE,
                                       config.ANALYZER_CAPTURE]:
             if self.analyze() == os.EX_OK:
-                if self.args.analyzer == config.ANALYZER_CRASHCONTEXT:
-                    self.crashcontext_stitch_summaries()
                 reporting_start_time = time.time()
                 report_status = self.create_report()
                 elapsed = utils.elapsed_time(reporting_start_time)

infer/src/backend/crashcontext.ml

@@ -0,0 +1,74 @@
+(*
+ * Copyright (c) 2016 - present Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under the BSD style license found in the
+ * LICENSE file in the root directory of this source tree. An additional grant
+ * of patent rights can be found in the PATENTS file in the same directory.
+ *)
+
+open! Utils
+
+module F = Format
+module L = Logging
+
+let frame_id_of_stackframe frame =
+  F.sprintf
+    "%s.%s(%s:%d)"
+    frame.Stacktrace.class_str
+    frame.Stacktrace.method_str
+    frame.Stacktrace.file_str
+    frame.Stacktrace.line_num
+
+let frame_id_of_summary stacktree =
+  let short_name = IList.hd
+      (Str.split (Str.regexp "(") stacktree.Stacktree_j.method_name) in
+  match stacktree.Stacktree_j.location with
+  | None ->
+      failwith "Attempted to take signature of a frame without location \
+                information. This is undefined."
+  | Some loc ->
+      F.sprintf "%s(%s:%d)" short_name (Filename.basename loc.file) loc.line
+
+let stracktree_of_frame frame =
+  { Stacktree_j.method_name = F.sprintf
+        "%s.%s"
+        frame.Stacktrace.class_str
+        frame.Stacktrace.method_str;
+    location = Some { Stacktree_j.location_type = "call_site";
+                      file = frame.Stacktrace.file_str;
+                      line = frame.Stacktrace.line_num };
+    callees = [];
+  }
+
+(** k = 1 implementation, where k is the number of levels of calls inlined *)
+let stitch_summaries stacktrace_file summary_files out_file =
+  let stacktrace = Stacktrace.of_json_file stacktrace_file in
+  let summaries = IList.map
+      (Ag_util.Json.from_file Stacktree_j.read_stacktree)
+      summary_files in
+  let summary_map = IList.fold_left
+      (fun acc stacktree ->
+         StringMap.add (frame_id_of_summary stacktree) stacktree acc)
+      StringMap.empty
+      summaries in
+  let expand_stack_frame frame =
+    (** TODO: Implement k > 1 case *)
+    let frame_id = frame_id_of_stackframe frame in
+    if StringMap.exists (fun key _ -> key = frame_id) summary_map then
+      StringMap.find frame_id summary_map
+    else
+      stracktree_of_frame frame in
+  let expanded_frames = IList.map expand_stack_frame stacktrace.frames in
+  let crashcontext = { Stacktree_j.stack = expanded_frames} in
+  Ag_util.Json.to_file Stacktree_j.write_crashcontext_t out_file crashcontext
+
+let collect_all_summaries root_out_dir stacktrace_file =
+  let out_dir = Filename.concat root_out_dir "crashcontext" in
+  DB.create_dir out_dir;
+  let out_file = Filename.concat out_dir "crashcontext.json" in
+  let path_regexp = Str.regexp ".*crashcontext/.*\\..*\\.json" in
+  let path_matcher path =  Str.string_match path_regexp path 0 in
+  let method_summaries =
+    DB.paths_matching root_out_dir path_matcher in
+  stitch_summaries stacktrace_file method_summaries out_file;

infer/src/backend/infer.ml

@@ -94,6 +94,37 @@ let () =
     ) in
   let pid = Unix.create_process args_py.(0) args_py Unix.stdin Unix.stdout Unix.stderr in
   let _, status = Unix.waitpid [] pid in
+  (** Collect crashcontext summaries *)
+  let analysis_is_crashcontext = match Config.analyzer with
+    | Some Crashcontext -> true
+    | _ -> false in
+  if analysis_is_crashcontext then
+    (** Check whether this is the top-level infer process *)
+    let top_level_infer =
+      (** if the '--buck' option was passed, then this is the top level process
+          iff the build command starts with 'buck' *)
+      if Config.buck then buck
+      (** otherwise, we assume javac as the build command and thus only
+          one process *)
+      else true in
+    if top_level_infer then
+      (** If we are the top-level process, then find the output directory and
+          collect all crashcontext summaries under it in a single
+          crashcontext.json file *)
+      let root_out_dir = if buck then begin
+          let project_root = match Config.project_root with
+            | Some root -> root
+            | None -> Filename.dirname Config.results_dir in
+          let buck_out = match Config.buck_out with
+            | Some dir -> dir
+            | None -> "buck-out" in
+          Filename.concat project_root buck_out
+        end
+        else Config.results_dir in
+      match Config.stacktrace with
+      | None -> failwith "Detected -a crashcontext without --stacktrace, \
+                          this should have been checked earlier."
+      | Some s -> Crashcontext.collect_all_summaries root_out_dir s;
   if status <> Unix.WEXITED 0 then (
     prerr_endline ("Failed to execute: " ^ (String.concat " " (Array.to_list args_py))) ;
     exit 1

infer/src/checkers/BoundedCallTree.ml

@@ -27,25 +27,22 @@ module TransferFunctions (CFG : ProcCfg.S) = struct
   module Domain = Domain
   type extras = Stacktrace.t
 
-  let json_of_summary caller astate loc loc_type =
+  let stacktree_of_summary caller astate loc location_type =
     let procs = Domain.elements astate in
-    let json = `Assoc [
-        ("method", `String (Procname.to_unique_id caller));
-        ("location", `Assoc [
-            ("type", `String loc_type);
-            ("file", `String (DB.source_file_to_string loc.Location.file));
-            ("line", `Int loc.Location.line);
-          ]);
-        ("callees", `List (IList.map
-                             (fun pn -> `Assoc [
-                                  ("method", `String (Procname.to_unique_id pn))
-                                ])
-                             procs))
-      ] in
-    json
+    let method_name = Procname.to_unique_id caller in
+    let file = DB.source_file_to_string loc.Location.file in
+    let line = loc.Location.line in
+    let location = Some { Stacktree_j.location_type ; file ; line } in
+    let callees = IList.map
+        (fun pn ->
+           { Stacktree_j.method_name = Procname.to_unique_id pn;
+             location = None;
+             callees = [] } )
+        procs in
+    { Stacktree_j.method_name; location; callees }
 
   let output_summary caller astate loc loc_type =
-    let json = json_of_summary caller astate loc loc_type in
+    let stacktree = stacktree_of_summary caller astate loc loc_type in
     let dir = Filename.concat Config.results_dir "crashcontext" in
     let suffix = F.sprintf "%s_%d" loc_type loc.Location.line in
     let fname = F.sprintf "%s.%s.json"
@@ -53,14 +50,26 @@ module TransferFunctions (CFG : ProcCfg.S) = struct
         suffix in
     let fpath = Filename.concat dir fname in
     DB.create_dir dir;
-    Utils.write_json_to_file fpath json
+    Ag_util.Json.to_file Stacktree_j.write_stacktree fpath stacktree
 
   let exec_instr astate proc_data _ = function
     | Sil.Call (_, Const (Const.Cfun pn), _, loc, _) ->
-        (** TODO: Match class. *)
         let caller = Cfg.Procdesc.get_proc_name proc_data.ProcData.pdesc in
         let matches_proc frame =
-          frame.Stacktrace.method_str = (Procname.get_method caller) in
+          let matches_class pname = match pname with
+            | Procname.Java java_proc ->
+                string_equal
+                  frame.Stacktrace.class_str
+                  (Procname.java_get_class_name java_proc)
+            | Procname.ObjC_Cpp objc_cpp_prod ->
+                string_equal
+                  frame.Stacktrace.class_str
+                  (Procname.objc_cpp_get_class_name objc_cpp_prod)
+            | Procname.C _ -> true (** Needed for test code. *)
+            | Procname.Block _ ->
+                failwith "Proc type not supported by crashcontext: block" in
+          frame.Stacktrace.method_str = (Procname.get_method caller) &&
+          matches_class caller in
         let proc_in_trace = IList.exists
             matches_proc
             proc_data.ProcData.extras.Stacktrace.frames in

infer/tests/utils/CrashContextResults.java

@@ -37,12 +37,12 @@ public class CrashContextResults {
 
   public boolean hasStackFrame(String methodSignature, int pos) {
     return methodSignature.equals(
-      json.path("stack").get(pos).path("method").asText());
+      json.path("stack").get(pos).path("method_name").asText());
   }
 
   public boolean hasStackFrame(String methodSignature) {
     for (JsonNode frame : json.path("stack")) {
-      if(methodSignature.equals(frame.path("method").asText())) {
+      if (methodSignature.equals(frame.path("method_name").asText())) {
         return true;
       }
     }
@@ -52,7 +52,7 @@ public class CrashContextResults {
   private List<JsonNode> findNodesForMethod(JsonNode node,
                                             String methodSignature,
                                             List<JsonNode> accumulator) {
-    if(methodSignature.equals(node.path("method").asText())) {
+    if (methodSignature.equals(node.path("method_name").asText())) {
       accumulator.add(node);
     }
     for (JsonNode callee : node.path("callees")) {