[flow] Add per-error flow_typing_errors Scuba logging for recheck events

Summary: Changelog: [internal]

Reviewed By: SamChou19815

Differential Revision: D98086794

fbshipit-source-id: 5186f4ccf4752ce0add8d4b86f60587c9bcef7a6

Author: panagosg7

src/commands/commandUtils.ml

@@ -1555,6 +1555,7 @@ let make_options
     opt_lazy_mode;
     opt_lint_severities = lint_severities;
     opt_llm_context_include_imports = FlowConfig.llm_context_include_imports flowconfig;
+    opt_log_per_error_typing_telemetry = FlowConfig.log_per_error_typing_telemetry flowconfig;
     opt_log_file;
     opt_log_saving = FlowConfig.log_saving flowconfig;
     opt_long_lived_workers =

src/commands/config/flowConfig.ml

@@ -101,6 +101,7 @@ module Opts = struct
     jest_integration: bool;
     lazy_mode: lazy_mode option;
     llm_context_include_imports: bool;
+    log_per_error_typing_telemetry: bool;
     log_saving: Options.log_saving SMap.t;
     long_lived_workers: bool;
     max_files_checked_per_worker: int;
@@ -262,6 +263,7 @@ module Opts = struct
       jest_integration = false;
       lazy_mode = None;
       llm_context_include_imports = false;
+      log_per_error_typing_telemetry = false;
       log_saving = SMap.empty;
       long_lived_workers = false;
       max_files_checked_per_worker = 100;
@@ -1142,6 +1144,9 @@ module Opts = struct
       ( "experimental.llm_context.include_imports",
         boolean (fun opts v -> Ok { opts with llm_context_include_imports = v })
       );
+      ( "experimental.log_per_error_typing_telemetry",
+        boolean (fun opts v -> Ok { opts with log_per_error_typing_telemetry = v })
+      );
       ("experimental.module.automatic_require_default", automatic_require_default_parser);
       ( "experimental.multi_platform",
         boolean (fun opts v -> Ok { opts with multi_platform = Some v })
@@ -2041,6 +2046,8 @@ let lazy_mode c = c.options.Opts.lazy_mode
 
 let llm_context_include_imports c = c.options.Opts.llm_context_include_imports
 
+let log_per_error_typing_telemetry c = c.options.Opts.log_per_error_typing_telemetry
+
 (* global defaults for lint severities and strict mode *)
 let lint_severities c = c.lint_severities
 

src/commands/config/flowConfig.mli

@@ -179,6 +179,8 @@ val lazy_mode : config -> lazy_mode option
 
 val llm_context_include_imports : config -> bool
 
+val log_per_error_typing_telemetry : config -> bool
+
 (* global defaults for lint suppressions and strict mode *)
 val lint_severities : config -> Severity.severity LintSettings.t
 

src/common/options.ml

@@ -151,6 +151,7 @@ type t = {
   opt_lazy_mode: bool;
   opt_lint_severities: Severity.severity LintSettings.t;
   opt_llm_context_include_imports: bool;
+  opt_log_per_error_typing_telemetry: bool;
   opt_log_file: File_path.t;
   opt_log_saving: log_saving SMap.t;
   opt_long_lived_workers: bool;
@@ -333,6 +334,8 @@ let lint_severities opts = opts.opt_lint_severities
 
 let llm_context_include_imports opts = opts.opt_llm_context_include_imports
 
+let log_per_error_typing_telemetry opts = opts.opt_log_per_error_typing_telemetry
+
 let log_file opts = opts.opt_log_file
 
 let log_saving opts = opts.opt_log_saving

src/server/error_collator/errorCollator.ml

@@ -264,6 +264,62 @@ type error_resolution_stat = {
   time_to_resolve_all_subtyping_errors_in_one_file: float option;
 }
 
+type per_error_info = {
+  error_code: string;
+  line_agnostic_hash: string;
+  error_message: string option;
+}
+
+type per_file_errors = {
+  file_path: string;
+  errors: per_error_info list;
+}
+
+let compute_per_file_errors ~with_context_limit collated_errors =
+  let open Flow_errors_utils in
+  let context_count = ref 0 in
+  let process_errors filename errors acc =
+    if ConcreteLocPrintableErrorSet.is_empty errors then
+      acc
+    else
+      let file_path =
+        match filename with
+        | File_key.SourceFile s
+        | File_key.JsonFile s
+        | File_key.ResourceFile s
+        | File_key.LibFile s ->
+          s
+      in
+      let error_infos =
+        ConcreteLocPrintableErrorSet.fold
+          (fun error acc ->
+            let lsp = Lsp_output.lsp_of_error ~has_detailed_diagnostics:false error in
+            let code_str = lsp.Lsp_output.code in
+            let hash_input = code_str ^ ":" ^ lsp.Lsp_output.message in
+            let hash = Printf.sprintf "%x" (Hashtbl.hash hash_input) in
+            let error_message =
+              if !context_count < with_context_limit then begin
+                context_count := !context_count + 1;
+                let msg = lsp.Lsp_output.message in
+                let msg =
+                  if String.length msg > 400 then
+                    String.sub msg 0 400 ^ "..."
+                  else
+                    msg
+                in
+                Some msg
+              end else
+                None
+            in
+            { error_code = code_str; line_agnostic_hash = hash; error_message } :: acc)
+          errors
+          []
+      in
+      { file_path; errors = error_infos } :: acc
+  in
+  let acc = FilenameMap.fold process_errors collated_errors.collated_merge_errors [] in
+  FilenameMap.fold process_errors collated_errors.collated_local_errors acc
+
 let update_error_state_timestamps collated_errors =
   let current_time = Unix.gettimeofday () in
   let init_timestamps = collated_errors.error_state_timestamps in

src/server/error_collator/errorCollator.mli

@@ -43,11 +43,24 @@ type error_resolution_stat = {
   time_to_resolve_all_subtyping_errors_in_one_file: float option;
 }
 
+type per_error_info = {
+  error_code: string;
+  line_agnostic_hash: string;
+  error_message: string option;
+}
+
+type per_file_errors = {
+  file_path: string;
+  errors: per_error_info list;
+}
+
 (* Update error_state_timestamps,
  * and return a collection of times to resolve different kinds of errors
  * under different initial conditions *)
 val update_error_state_timestamps : Collated_errors.t -> Collated_errors.t * error_resolution_stat
 
+val compute_per_file_errors : with_context_limit:int -> Collated_errors.t -> per_file_errors list
+
 val get_without_suppressed :
   ServerEnv.env ->
   Flow_errors_utils.ConcreteLocPrintableErrorSet.t

src/services/inference/types_js.ml

@@ -1694,6 +1694,49 @@ let recheck_impl
       ~first_internal_error
       ~scm_changed_mergebase:changed_mergebase
       ~profiling;
+    if Options.log_per_error_typing_telemetry options then begin
+      let per_file_errors =
+        ErrorCollator.compute_per_file_errors ~with_context_limit:10 collated_errors
+      in
+      let typing_errors_data =
+        let open Hh_json in
+        let error_info_to_json (info : ErrorCollator.per_error_info) =
+          let props =
+            [
+              ("error_code", JSON_String info.ErrorCollator.error_code);
+              ("line_agnostic_hash", JSON_String info.ErrorCollator.line_agnostic_hash);
+            ]
+          in
+          let props =
+            match info.ErrorCollator.error_message with
+            | Some msg -> ("error_context", JSON_String msg) :: props
+            | None -> props
+          in
+          JSON_Object props
+        in
+        let by_file =
+          List.map
+            (fun ({ ErrorCollator.file_path; errors } : ErrorCollator.per_file_errors) ->
+              (file_path, JSON_Array (List.map error_info_to_json errors)))
+            per_file_errors
+        in
+        let total_count =
+          List.fold_left
+            (fun acc ({ ErrorCollator.errors; _ } : ErrorCollator.per_file_errors) ->
+              acc + List.length errors)
+            0
+            per_file_errors
+        in
+        json_to_string
+          (JSON_Object
+             [
+               ("total_count", JSON_Number (string_of_int total_count));
+               ("by_file", JSON_Object by_file);
+             ]
+          )
+      in
+      FlowEventLogger.log_typing_errors ~data:typing_errors_data
+    end;
     record_recheck_time ()
   in
 

src/stubs/flowEventLogger.ml

@@ -111,6 +111,8 @@ let recheck
     ~scm_changed_mergebase:_ =
   ()
 
+let log_typing_errors ~data:_ = ()
+
 let recheck_canceled
     ~priority:_ ~num_files_to_prioritize:_ ~num_files_to_recheck:_ ~num_files_to_force:_ =
   ()