speed up serialize_testset_result_file

Author: IanButterworth

test/buildkitetestjson.jl

@@ -104,6 +104,16 @@ function generalize_file_paths(path::AbstractString)
     end
 end
 
+# raw_file_path,line => file,location
+const location_cache = Dict{Tuple{Symbol,Int},Tuple{String,String}}()
+function get_location(file::Symbol, line::Int)
+    return get!(location_cache, (file, line)) do
+        _file = generalize_file_paths(string(file))
+        _location = string(_file, ":", line)
+        return _file, _location
+    end
+end
+
 # passed, failed, skipped, or unknown
 function get_status(result)
     if result isa Test.Pass && result.test_type === :skipped
@@ -140,24 +150,31 @@ end
 
 get_rid(rdata) = (rdata["location"], rdata["result"], haskey(rdata, "failure_expanded") ? hash(rdata["failure_expanded"]) : UInt64(0))
 
-function result_dict(result::Test.Result, result_counts::Dict{Tuple{String,String,UInt64},Int})
+const ResultCountDict = Dict{Tuple{String,String,UInt64},Int}
+
+function is_duplicate_pass(result::Test.Pass, location, status, result_counts::ResultCountDict)
+    rid = (location, status, UInt64(0))
+    count = get(result_counts, rid, nothing)
+    if count !== nothing
+        result_counts[rid] = count + 1
+        return true
+    end
+    return false
+end
+is_duplicate_pass(result::Test.Result, location, status, result_counts::ResultCountDict) = false
+
+function result_dict(result::Test.Result, result_counts::ResultCountDict)
     file, line = if !hasproperty(result, :source) || isnothing(result.source)
-        "unknown", 0
+        :unknown, 0
     else
-        something(result.source.file, "unknown"), result.source.line
+        something(result.source.file, :unknown), result.source.line
     end
-    file = generalize_file_paths(string(file))
-    location = string(file, ':', line)
+    file, location = get_location(file, line)
     status = get_status(result)
 
     # Early exit for passed tests before more expensive operations
-    if result isa Test.Pass
-        rid = (location, status, UInt64(0))
-        duplicate = haskey(result_counts, rid)
-        if duplicate
-            result_counts[rid] += 1
-            return nothing
-        end
+    if is_duplicate_pass(result, location, status, result_counts)
+        return nothing
     end
 
     data = Dict{String,Any}(
@@ -193,23 +210,27 @@ function result_dict(result::Test.Result, result_counts::Dict{Tuple{String,Strin
     end
 end
 
+function collect_results!(results::Vector{Dict{String,Any}}, result::Test.Result, common_data::Dict{String,Any}, result_counts::ResultCountDict)
+    rdata = result_dict(result, result_counts)
+    if rdata !== nothing # nothing if it's a duplicate that's been counted
+        push!(results, merge(common_data, rdata))
+    end
+end
+
+function collect_results!(results::Vector{Dict{String,Any}}, result::Test.DefaultTestSet, common_data::Dict{String,Any}, result_counts::ResultCountDict)
+    collect_results!(results, result, common_data["scope"])
+end
+
 function collect_results!(results::Vector{Dict{String,Any}}, testset::Test.DefaultTestSet, prefix::String="")
     common_data = result_dict(testset, prefix)
     # testset duration is not relevant for individual test results
     common_data["history"]["duration"] = 0.0 # required field
     delete!(common_data["history"], "end_at")
     result_offset = length(results) + 1
-    result_counts = Dict{Tuple{String,String,UInt64},Int}()
+    result_counts = ResultCountDict()
 
-    for (i, result) in enumerate(testset.results)
-        if result isa Test.Result
-            rdata = result_dict(result, result_counts)
-            if rdata !== nothing # nothing if it's a duplicate that's been counted
-                push!(results, merge(common_data, rdata))
-            end
-        elseif result isa Test.DefaultTestSet
-            collect_results!(results, result, common_data["scope"])
-        end
+    for result in testset.results
+        collect_results!(results, result, common_data, result_counts)
     end
     # Add a tag for count of each result
     for result in results[result_offset:end]
@@ -229,7 +250,7 @@ function serialize_testset_result_file(dir::String, testset::Test.DefaultTestSet
     res_file = joinpath(dir, "results_$(name).dat")
     t = @elapsed Serialization.serialize(res_file, data)
     if t > 10
-        @warn "Serializing test result data was slow: $t seconds" file=res_file size=Base.format_bytes(filesize(res_file))
+        @warn "Serializing test result data was slow: $t seconds" file = res_file size = Base.format_bytes(filesize(res_file))
     end
     return res_file
 end