cov: store "is branch" in file coverage segment

Now closer to the function coverage  region storage format. Added tests for
the full  coverage information  in the file  segmentation test  and updated
JSON diff to cover two models of  the superset addition -- "add field" (can
be ignored in most of the cases) and  "add list item" (is not ignore as the
lists are mostly  asserted to contain a specific number  of elements -- not
"starts with")

scripts/cxx_codegen/profdata_merger/profdata_merger.cpp

@@ -24,6 +24,7 @@
 #include <llvm/ADT/Hashing.h>
 #include <hstd/system/Formatter.hpp>
 #include <algorithm>
+#include <absl/hash/hash.h>
 
 #include <perfetto.h>
 
@@ -779,6 +780,7 @@ struct queries {
                     "SegmentIndex", // 11
                     "NestedIn",     // 12
                     "IsLeaf",       // 13
+                    "IsBranch",     // 14
                 }))
         ,
         // ---
@@ -885,6 +887,7 @@ struct CountedRegionHasher {
     }
 };
 
+
 struct CountedRegionComparator {
     size_t operator()(
         const CounterMappingRegion& lhs,
@@ -898,6 +901,35 @@ struct CountedRegionComparator {
     }
 };
 
+
+struct FileSpan {
+    unsigned int LineStart{};
+    unsigned int LineEnd{};
+    unsigned int ColStart{};
+    unsigned int ColEnd{};
+};
+
+struct FileSpanHasher {
+    size_t operator()(const FileSpan& region) const {
+        return hash_combine(
+            region.LineStart,
+            region.LineEnd,
+            region.ColStart,
+            region.ColEnd);
+    }
+};
+
+
+struct FileSpanComparator {
+    size_t operator()(const FileSpan& lhs, const FileSpan& rhs) const {
+        return lhs.LineStart == rhs.LineStart //
+            && lhs.ColStart == rhs.ColStart   //
+            && lhs.LineEnd == rhs.LineEnd     //
+            && lhs.ColEnd == rhs.ColEnd;
+    }
+};
+
+
 struct db_build_ctx {
     int                                  context_id{};
     int                                  instantiation_id{};
@@ -1141,6 +1173,32 @@ void add_file(CoverageData const& file, queries& q, db_build_ctx& ctx) {
             return lhs.first.self_id < rhs.first.self_id;
         });
 
+    std::unordered_map<FileSpan, bool, FileSpanHasher, FileSpanComparator>
+        branch_cache{};
+
+    for (auto it : enumerate(file.getBranches())) {
+        CountedRegion const& r = it.value();
+        branch_cache.insert_or_assign(
+            FileSpan{
+                .LineStart = r.LineStart,
+                .LineEnd   = r.LineEnd,
+                .ColStart  = r.ColumnStart,
+                .ColEnd    = r.ColumnEnd,
+            },
+            true);
+        q.file_branch.bind(1, ctx.context_id);
+        q.file_branch.bind(2, (int64_t)r.ExecutionCount);
+        q.file_branch.bind(3, (int64_t)r.FalseExecutionCount);
+        q.file_branch.bind(4, r.Folded);
+        q.file_branch.bind(5, (int64_t)r.LineStart);
+        q.file_branch.bind(6, (int64_t)r.ColumnStart);
+        q.file_branch.bind(7, (int64_t)r.LineEnd);
+        q.file_branch.bind(8, (int64_t)r.ColumnEnd);
+        q.func_region.bind(9, r.Kind);
+        q.file_branch.exec();
+        q.file_branch.reset();
+    }
+
     for (auto it : enumerate(segment_pairs)) {
         auto const& [nesting, end] = it.value();
         auto const& start          = nesting.segment;
@@ -1162,24 +1220,19 @@ void add_file(CoverageData const& file, queries& q, db_build_ctx& ctx) {
             q.segment.bind(12, nullptr);
         }
         q.segment.bind(13, nesting.is_leaf);
+
+        FileSpan span{
+            .LineStart = start.Line,
+            .LineEnd   = end.Line,
+            .ColStart  = start.Col,
+            .ColEnd    = end.Col,
+        };
+
+        q.segment.bind(14, branch_cache.contains(span));
         q.segment.exec();
         q.segment.reset();
     }
 
-    for (auto it : enumerate(file.getBranches())) {
-        CountedRegion const& r = it.value();
-        q.file_branch.bind(1, ctx.context_id);
-        q.file_branch.bind(2, (int64_t)r.ExecutionCount);
-        q.file_branch.bind(3, (int64_t)r.FalseExecutionCount);
-        q.file_branch.bind(4, r.Folded);
-        q.file_branch.bind(5, (int64_t)r.LineStart);
-        q.file_branch.bind(6, (int64_t)r.ColumnStart);
-        q.file_branch.bind(7, (int64_t)r.LineEnd);
-        q.file_branch.bind(8, (int64_t)r.ColumnEnd);
-        q.func_region.bind(9, r.Kind);
-        q.file_branch.exec();
-        q.file_branch.reset();
-    }
 
     for (ExpansionRecord const& e : file.getExpansions()) {
         int function_id = get_function_id(e.Function, q, ctx);

scripts/cxx_codegen/profdata_merger/profdata_merger.sql

@@ -104,6 +104,7 @@ CREATE TABLE "CovSegment" (
 	"SegmentIndex" INTEGER NOT NULL, 
 	"NestedIn" INTEGER, 
 	"IsLeaf" BOOLEAN NOT NULL, 
+	"IsBranch" BOOLEAN NOT NULL, 
 	PRIMARY KEY ("Id"), 
 	FOREIGN KEY("File") REFERENCES "CovFile" ("Id"), 
 	FOREIGN KEY("Context") REFERENCES "CovContext" ("Id"), 

scripts/py_repository/py_repository/gen_coverage_cxx.py

@@ -113,6 +113,7 @@ class CovSegment(CoverageSchema):
     SegmentIndex = IntColumn()
     NestedIn = ForeignId("CovSegment.Id", nullable=True)
     IsLeaf = BoolColumn()
+    IsBranch = BoolColumn()
 
     def intersects(self, line: int, col: int) -> bool:
         return (self.LineStart <= line <= self.LineEnd) and (self.ColStart <= col <=

scripts/py_scriptutils/py_scriptutils/json_utils.py

@@ -12,7 +12,8 @@
 class Op(Enum):
     Replace = auto()
     Remove = auto()
-    Add = auto()
+    AddField = auto()
+    AppendItem = auto()
 
 
 @beartype
@@ -39,7 +40,7 @@ def json_diff(source: Json,
         result.append(DiffItem(Op.Replace, path, target))
         return result
 
-    if isinstance(source, list):
+    if isinstance(source, (list, tuple)):
         i = 0
         while i < len(source) and i < len(target):
             temp_diff = json_diff(source[i], target[i], path.child(Index(i)), ignore)
@@ -52,9 +53,11 @@ def json_diff(source: Json,
             i += 1
 
         while i < len(target):
-            result.append(DiffItem(Op.Add, path.child(Index(i)), target[i]))
+            result.append(DiffItem(Op.AppendItem, path.child(Index(i)), target[i]))
             i += 1
 
+        assert i == max(len(target), len(source))
+
     elif isinstance(source, dict):
         for key in source:
             if key in target:
@@ -67,7 +70,7 @@ def json_diff(source: Json,
 
         for key in target:
             if key not in source:
-                result.append(DiffItem(Op.Add, path.child(Fields(key)), target[key]))
+                result.append(DiffItem(Op.AddField, path.child(Fields(key)), target[key]))
 
     else:
         result.append(DiffItem(Op.Replace, path, target))
@@ -84,7 +87,8 @@ def get_subset_diff(main_set: Json, expected_subset: Json) -> List[DiffItem]:
     # If some element from expect *sub*set was added, it is an expected behavior. All other operations
     # are returned.
     return [
-        it for it in json_diff(source=expected_subset, target=main_set) if it.op != Op.Add
+        it for it in json_diff(source=expected_subset, target=main_set)
+        if it.op != Op.AddField
     ]
 
 
@@ -101,8 +105,13 @@ def describe_diff(
 
     if it.op == Op.Remove:
         description += f"{source_name} has extra entry"
-    elif it.op == Op.Add:
-        description += f"{target_name} missing entry"
+
+    elif it.op == Op.AddField:
+        description += f"{target_name} missing field"
+
+    elif it.op == Op.AppendItem:
+        description += f"{target_name} missing list item"
+
     elif it.op == Op.Replace:
         description += "changed entry"
 
@@ -115,7 +124,10 @@ def get_path(value: Json) -> Json:
     if it.op == Op.Remove:
         description += f"    {json.dumps(get_path(source), indent=2)}"
 
-    elif it.op == Op.Add:
+    elif it.op == Op.AddField:
+        description += f"    {json.dumps(it.value, indent=2)}"
+
+    elif it.op == Op.AppendItem:
         description += f"    {json.dumps(it.value, indent=2)}"
 
     elif it.op == Op.Replace:

tests/python/repo/coverage_corpus/test_file_segmentation_1.cpp

@@ -31,6 +31,11 @@ void coverage_branches(int arg) {
     } else {
         action();
     }
+
+    if (arg == 0 || arg == 1 || arg == 2
+        || ((arg <= 5 || arg <= 5) && arg < 10)) {
+        action();
+    }
 }
 
 

tests/python/repo/test_code_coverage.py

@@ -326,21 +326,40 @@ def test_file_segmentation_1():
         session = open_sqlite_session(cmd.get_sqlite(), cov.CoverageSchema)
         main_cov = cov.get_coverage_of(session, cmd.get_code("main.cpp"))
         lines = code.split("\n")
+        # print(format_db_all(session))
 
         segtree = cov.CoverageSegmentTree(it[0] for it in session.execute(main_cov))
         df = pd.read_sql(main_cov, session.get_bind())
         add_cov_segment_text(df, lines)
 
         # print(render_rich(dataframe_to_rich_table(df)))
 
-        # Coverage segments only overlay executable blocks and do not 
+        # Coverage segments only overlay executable blocks and do not
         # account for extraneous elements such as function headers etc.
         assert segtree.query(line=1, col=15)
         assert not segtree.query(line=1, col=14)
         assert_frame(df[df["LineStart"] == 1], [
             dict(IsLeaf=True, Text="{}", ColStart=15, ColEnd=17),
         ])
 
+        # print(render_rich(dataframe_to_rich_table(df[df["IsBranch"] == True])))
+        assert_frame(df[df["IsBranch"] == True], [
+            dict(LineStart=3, ColStart=9, LineEnd=3, ColEnd=16, Text="arg > 0"),
+            dict(LineStart=4, ColStart=13, LineEnd=4, ColEnd=21, Text="arg > 10"),
+            dict(LineStart=5, ColStart=17, LineEnd=5, ColEnd=25, Text="arg > 20"),
+            dict(LineStart=11, ColStart=17, LineEnd=11, ColEnd=24, Text="arg > 5"),
+            dict(LineStart=17, ColStart=16, LineEnd=17, ColEnd=23, Text="arg < 0"),
+            dict(LineStart=18, ColStart=13, LineEnd=18, ColEnd=22, Text="arg < -10"),
+            dict(LineStart=19, ColStart=17, LineEnd=19, ColEnd=26, Text="arg < -20"),
+            dict(LineStart=25, ColStart=17, LineEnd=25, ColEnd=25, Text="arg < -5"),
+            dict(LineStart=35, ColStart=9, LineEnd=35, ColEnd=17, Text="arg == 0"),
+            dict(LineStart=35, ColStart=21, LineEnd=35, ColEnd=29, Text="arg == 1"),
+            dict(LineStart=35, ColStart=33, LineEnd=35, ColEnd=41, Text="arg == 2"),
+            dict(LineStart=36, ColStart=14, LineEnd=36, ColEnd=22, Text="arg <= 5"),
+            dict(LineStart=36, ColStart=26, LineEnd=36, ColEnd=34, Text="arg <= 5"),
+            dict(LineStart=36, ColStart=39, LineEnd=36, ColEnd=47, Text="arg < 10"),
+        ])
+
 
 def test_file_segmentation_2():
     with TemporaryDirectory() as tmp: