cgen,pref: add -coverage support + vcover tool

vlib/v/gen/c/assert.v

@@ -106,7 +106,7 @@ fn (mut g Gen) assert_subexpression_to_ctemp(expr ast.Expr, expr_type ast.Type)
 }
 
 fn (mut g Gen) gen_assert_postfailure_mode(node ast.AssertStmt) {
-	g.write_v_source_line_info(node.pos)
+	g.write_v_source_line_info_stmt(node)
 	if g.pref.assert_failure_mode == .continues
 		|| g.fn_decl.attrs.any(it.name == 'assert_continues') {
 		return

vlib/v/gen/c/cgen.v

@@ -42,6 +42,15 @@ fn string_array_to_map(a []string) map[string]bool {
 	return res
 }
 
+// V coverage info
+@[heap]
+struct CoverageInfo {
+pub mut:
+	idx    int   // index
+	points []u64 // code point line nr
+	file   &ast.File = unsafe { nil }
+}
+
 pub struct Gen {
 	pref                &pref.Preferences = unsafe { nil }
 	field_data_type     ast.Type // cache her to avoid map lookups
@@ -69,6 +78,7 @@ mut:
 	auto_str_funcs            strings.Builder // function bodies of all auto generated _str funcs
 	dump_funcs                strings.Builder // function bodies of all auto generated _str funcs
 	pcs_declarations          strings.Builder // -prof profile counter declarations for each function
+	cov_declarations          strings.Builder // -cov coverage
 	embedded_data             strings.Builder // data to embed in the executable/binary
 	shared_types              strings.Builder // shared/lock types
 	shared_functions          strings.Builder // shared constructors
@@ -119,6 +129,7 @@ mut:
 	labeled_loops             map[string]&ast.Stmt
 	inner_loop                &ast.Stmt = unsafe { nil }
 	shareds                   map[int]string // types with hidden mutex for which decl has been emitted
+	coverage_files            map[u64]&CoverageInfo
 	inside_ternary            int  // ?: comma separated statements on a single line
 	inside_map_postfix        bool // inside map++/-- postfix expr
 	inside_map_infix          bool // inside map<</+=/-= infix expr
@@ -299,6 +310,7 @@ pub fn gen(files []&ast.File, mut table ast.Table, pref_ &pref.Preferences) (str
 		auto_str_funcs: strings.new_builder(100)
 		dump_funcs: strings.new_builder(100)
 		pcs_declarations: strings.new_builder(100)
+		cov_declarations: strings.new_builder(100)
 		embedded_data: strings.new_builder(1000)
 		out_options_forward: strings.new_builder(100)
 		out_options: strings.new_builder(100)
@@ -379,6 +391,7 @@ pub fn gen(files []&ast.File, mut table ast.Table, pref_ &pref.Preferences) (str
 			global_g.dump_funcs.write(g.auto_str_funcs) or { panic(err) }
 			global_g.comptime_definitions.write(g.comptime_definitions) or { panic(err) }
 			global_g.pcs_declarations.write(g.pcs_declarations) or { panic(err) }
+			global_g.cov_declarations.write(g.cov_declarations) or { panic(err) }
 			global_g.hotcode_definitions.write(g.hotcode_definitions) or { panic(err) }
 			global_g.embedded_data.write(g.embedded_data) or { panic(err) }
 			global_g.shared_types.write(g.shared_types) or { panic(err) }
@@ -411,6 +424,9 @@ pub fn gen(files []&ast.File, mut table ast.Table, pref_ &pref.Preferences) (str
 			for k, v in g.sumtype_definitions {
 				global_g.sumtype_definitions[k] = v
 			}
+			for k, v in g.coverage_files {
+				global_g.coverage_files[k] = v
+			}
 			global_g.json_forward_decls.write(g.json_forward_decls) or { panic(err) }
 			global_g.enum_typedefs.write(g.enum_typedefs) or { panic(err) }
 			global_g.channel_definitions.write(g.channel_definitions) or { panic(err) }
@@ -505,6 +521,15 @@ pub fn gen(files []&ast.File, mut table ast.Table, pref_ &pref.Preferences) (str
 		g.write_init_function()
 	}
 
+	if g.pref.is_coverage {
+		mut total_code_points := 0
+		for k, cov in g.coverage_files {
+			g.cheaders.writeln('#define _v_cov_file_offset_${k} ${total_code_points}')
+			total_code_points += cov.points.len
+		}
+		g.cheaders.writeln('char _v_cov[${total_code_points}] = {0};')
+	}
+
 	// insert for options forward
 	if g.out_options_forward.len > 0 || g.out_results_forward.len > 0 {
 		tail := g.type_definitions.cut_to(g.options_pos_forward)
@@ -540,6 +565,10 @@ pub fn gen(files []&ast.File, mut table ast.Table, pref_ &pref.Preferences) (str
 		b.writeln('\n// V profile counters:')
 		b.write_string(g.pcs_declarations.str())
 	}
+	if g.pref.is_coverage {
+		b.writeln('\n// V coverage:')
+		b.write_string(g.cov_declarations.str())
+	}
 	b.writeln('\n// V includes:')
 	b.write_string(g.includes.str())
 	b.writeln('\n// Enum definitions:')
@@ -654,6 +683,7 @@ fn cgen_process_one_file_cb(mut p pool.PoolProcessor, idx int, wid int) &Gen {
 		auto_str_funcs: strings.new_builder(100)
 		comptime_definitions: strings.new_builder(100)
 		pcs_declarations: strings.new_builder(100)
+		cov_declarations: strings.new_builder(100)
 		hotcode_definitions: strings.new_builder(100)
 		embedded_data: strings.new_builder(1000)
 		out_options_forward: strings.new_builder(100)
@@ -724,6 +754,7 @@ pub fn (mut g Gen) free_builders() {
 		g.dump_funcs.free()
 		g.comptime_definitions.free()
 		g.pcs_declarations.free()
+		g.cov_declarations.free()
 		g.hotcode_definitions.free()
 		g.embedded_data.free()
 		g.shared_types.free()
@@ -2049,7 +2080,7 @@ fn (mut g Gen) expr_with_tmp_var(expr ast.Expr, expr_typ ast.Type, ret_typ ast.T
 }
 
 @[inline]
-fn (mut g Gen) write_v_source_line_info(pos token.Pos) {
+fn (mut g Gen) write_v_source_line_info_pos(pos token.Pos) {
 	if g.inside_ternary == 0 && g.pref.is_vlines && g.is_vlines_enabled {
 		nline := pos.line_nr + 1
 		lineinfo := '\n#line ${nline} "${g.vlines_path}"'
@@ -2060,6 +2091,31 @@ fn (mut g Gen) write_v_source_line_info(pos token.Pos) {
 	}
 }
 
+@[inline]
+fn (mut g Gen) write_v_source_line_info(node ast.Node) {
+	g.write_v_source_line_info_pos(node.pos())
+	if g.inside_ternary == 0 && g.pref.is_coverage
+		&& node !in [ast.MatchBranch, ast.IfBranch, ast.InfixExpr] {
+		g.write_coverage_point(node.pos())
+	}
+}
+
+@[inline]
+fn (mut g Gen) write_v_source_line_info_stmt(stmt ast.Stmt) {
+	g.write_v_source_line_info_pos(stmt.pos)
+	if g.inside_ternary == 0 && g.pref.is_coverage && !g.inside_for_c_stmt
+		&& stmt !in [ast.FnDecl, ast.ForCStmt, ast.ForInStmt, ast.ForStmt] {
+		if stmt is ast.ExprStmt {
+			if !g.inside_assign
+				&& stmt.expr !in [ast.StringInterLiteral, ast.StringLiteral, ast.Ident] {
+				g.write_coverage_point(stmt.pos)
+			}
+		} else {
+			g.write_coverage_point(stmt.pos)
+		}
+	}
+}
+
 fn (mut g Gen) stmt(node ast.Stmt) {
 	$if trace_cgen_stmt ? {
 		ntype := typeof(node).replace('v.ast.', '')
@@ -2075,19 +2131,19 @@ fn (mut g Gen) stmt(node ast.Stmt) {
 	}
 	match node {
 		ast.AsmStmt {
-			g.write_v_source_line_info(node.pos)
+			g.write_v_source_line_info_stmt(node)
 			g.asm_stmt(node)
 		}
 		ast.AssertStmt {
-			g.write_v_source_line_info(node.pos)
+			g.write_v_source_line_info_stmt(node)
 			g.assert_stmt(node)
 		}
 		ast.AssignStmt {
-			g.write_v_source_line_info(node.pos)
+			g.write_v_source_line_info_stmt(node)
 			g.assign_stmt(node)
 		}
 		ast.Block {
-			g.write_v_source_line_info(node.pos)
+			g.write_v_source_line_info_stmt(node)
 			if node.is_unsafe {
 				g.writeln('{ // Unsafe block')
 			} else {
@@ -2097,11 +2153,11 @@ fn (mut g Gen) stmt(node ast.Stmt) {
 			g.writeln('}')
 		}
 		ast.BranchStmt {
-			g.write_v_source_line_info(node.pos)
+			g.write_v_source_line_info_stmt(node)
 			g.branch_stmt(node)
 		}
 		ast.ConstDecl {
-			g.write_v_source_line_info(node.pos)
+			g.write_v_source_line_info_stmt(node)
 			g.const_decl(node)
 		}
 		ast.ComptimeFor {
@@ -2121,7 +2177,7 @@ fn (mut g Gen) stmt(node ast.Stmt) {
 			g.enum_decl(node)
 		}
 		ast.ExprStmt {
-			g.write_v_source_line_info(node.pos)
+			g.write_v_source_line_info_stmt(node)
 			// af := g.autofree && node.expr is ast.CallExpr && !g.is_builtin_mod
 			// if af {
 			// g.autofree_call_pregen(node.expr as ast.CallExpr)
@@ -2161,7 +2217,7 @@ fn (mut g Gen) stmt(node ast.Stmt) {
 					g.labeled_loops[node.label] = &node
 				}
 			}
-			g.write_v_source_line_info(node.pos)
+			g.write_v_source_line_info_stmt(node)
 			g.for_c_stmt(node)
 			g.branch_parent_pos = prev_branch_parent_pos
 			g.labeled_loops.delete(node.label)
@@ -2177,7 +2233,7 @@ fn (mut g Gen) stmt(node ast.Stmt) {
 					g.labeled_loops[node.label] = &node
 				}
 			}
-			g.write_v_source_line_info(node.pos)
+			g.write_v_source_line_info_stmt(node)
 			g.for_in_stmt(node)
 			g.branch_parent_pos = prev_branch_parent_pos
 			g.labeled_loops.delete(node.label)
@@ -2193,7 +2249,7 @@ fn (mut g Gen) stmt(node ast.Stmt) {
 					g.labeled_loops[node.label] = &node
 				}
 			}
-			g.write_v_source_line_info(node.pos)
+			g.write_v_source_line_info_stmt(node)
 			g.for_stmt(node)
 			g.branch_parent_pos = prev_branch_parent_pos
 			g.labeled_loops.delete(node.label)
@@ -2206,7 +2262,7 @@ fn (mut g Gen) stmt(node ast.Stmt) {
 			g.writeln('${c_name(node.name)}: {}')
 		}
 		ast.GotoStmt {
-			g.write_v_source_line_info(node.pos)
+			g.write_v_source_line_info_stmt(node)
 			g.writeln('goto ${c_name(node.name)};')
 		}
 		ast.HashStmt {
@@ -5266,7 +5322,7 @@ fn (mut g Gen) branch_stmt(node ast.BranchStmt) {
 
 fn (mut g Gen) return_stmt(node ast.Return) {
 	g.set_current_pos_as_last_stmt_pos()
-	g.write_v_source_line_info(node.pos)
+	g.write_v_source_line_info_stmt(node)
 
 	g.inside_return = true
 	defer {
@@ -6410,6 +6466,10 @@ fn (mut g Gen) write_init_function() {
 	if g.pref.use_coroutines {
 		g.writeln('\tdelete_photon_work_pool();')
 	}
+	if g.pref.is_coverage {
+		g.write_coverage_stats()
+		g.writeln('\tvprint_coverage_stats();')
+	}
 	g.writeln('}')
 	if g.pref.printfn_list.len > 0 && '_vcleanup' in g.pref.printfn_list {
 		println(g.out.after(fn_vcleanup_start_pos))

vlib/v/gen/c/coverage.v

@@ -0,0 +1,67 @@
+// Copyright (c) 2024 Felipe Pena. All rights reserved.
+// Use of this source code is governed by an MIT license
+// that can be found in the LICENSE file.
+module c
+
+import v.token
+
+@[inline]
+fn (mut g Gen) write_coverage_point(pos token.Pos) {
+	if g.unique_file_path_hash !in g.coverage_files {
+		g.coverage_files[g.unique_file_path_hash] = &CoverageInfo{
+			points: []
+			file: g.file
+		}
+	}
+	if g.fn_decl != unsafe { nil } {
+		curr_line := u64(pos.line_nr)
+		mut curr_cov := unsafe { g.coverage_files[g.unique_file_path_hash] }
+		if curr_line !in curr_cov.points {
+			curr_cov.points << curr_line
+		}
+		g.writeln('_v_cov[_v_cov_file_offset_${g.unique_file_path_hash}+${curr_cov.points.len - 1}] = 1;')
+	}
+}
+
+fn (mut g Gen) write_coverage_stats() {
+	is_stdout := g.pref.coverage_file == '-'
+
+	g.cov_declarations.writeln('void vprint_coverage_stats() {')
+	if is_stdout {
+		g.cov_declarations.writeln('\tprintf("V coverage\\n");')
+		g.cov_declarations.writeln('\tprintf("${'-':50r}\\n");')
+	} else {
+		g.cov_declarations.writeln('\tFILE *fp = stdout;')
+		g.cov_declarations.writeln('\tfp = fopen ("${g.pref.coverage_file}", "w+");')
+		g.cov_declarations.writeln('\tfprintf(fp, "V coverage\\n");')
+		g.cov_declarations.writeln('\tfprintf(fp, "${'-':50r}\\n");')
+	}
+	g.cov_declarations.writeln('\tint t_counter = 0;')
+	mut last_offset := 0
+	mut t_points := 0
+	for k, cov in g.coverage_files {
+		nr_points := cov.points.len
+		t_points += nr_points
+		g.cov_declarations.writeln('\t{')
+		g.cov_declarations.writeln('\t\tint counter = 0;')
+		g.cov_declarations.writeln('\t\tfor (int i = 0, offset = ${last_offset}; i < ${nr_points}; ++i)')
+		g.cov_declarations.writeln('\t\t\tif (_v_cov[_v_cov_file_offset_${k}+i]) counter++;')
+		g.cov_declarations.writeln('\t\tt_counter += counter;')
+		if is_stdout {
+			g.cov_declarations.writeln('\t\tprintf("[%7.2f%%] %4d / ${nr_points:4d} | ${cov.file.path}\\n", ${nr_points} > 0 ? ((double)counter/${nr_points})*100 : 0, counter);')
+		} else {
+			g.cov_declarations.writeln('\t\tfprintf(fp, "[%7.2f%%] %4d / ${nr_points:4d} | ${cov.file.path}\\n", ${nr_points} > 0 ? ((double)counter/${nr_points})*100 : 0, counter);')
+		}
+		g.cov_declarations.writeln('\t}')
+		last_offset += nr_points
+	}
+	if is_stdout {
+		g.cov_declarations.writeln('\tprintf("${'-':50r}\\n");')
+		g.cov_declarations.writeln('\tprintf("Total coverage: ${g.coverage_files.len} files, %.2f%% coverage\\n", ((double)t_counter/${t_points})*100);')
+	} else {
+		g.cov_declarations.writeln('\tfprintf(fp, "${'-':50r}\\n");')
+		g.cov_declarations.writeln('\tfprintf(fp, "Total coverage: ${g.coverage_files.len} files, %.2f%% coverage\\n", ((double)t_counter/${t_points})*100);')
+		g.cov_declarations.writeln('\tfclose(fp);')
+	}
+	g.cov_declarations.writeln('}')
+}

vlib/v/gen/c/fn.v

@@ -212,7 +212,7 @@ fn (mut g Gen) gen_fn_decl(node &ast.FnDecl, skip bool) {
 		g.definitions.writeln(';')
 	}
 
-	g.write_v_source_line_info(node.pos)
+	g.write_v_source_line_info_stmt(node)
 	fn_attrs := g.write_fn_attrs(node.attrs)
 	// Live
 	is_livefn := node.attrs.contains('live')

vlib/v/gen/c/match.v

@@ -183,15 +183,15 @@ fn (mut g Gen) match_expr_sumtype(node ast.MatchExpr, is_expr bool, cond_var str
 					g.write(' : ')
 				} else {
 					g.writeln('')
-					g.write_v_source_line_info(branch.pos)
+					g.write_v_source_line_info(branch)
 					g.writeln('else {')
 				}
 			} else {
 				if j > 0 || sumtype_index > 0 {
 					if use_ternary {
 						g.write(' : ')
 					} else {
-						g.write_v_source_line_info(branch.pos)
+						g.write_v_source_line_info(branch)
 						g.write('else ')
 					}
 				}
@@ -201,7 +201,7 @@ fn (mut g Gen) match_expr_sumtype(node ast.MatchExpr, is_expr bool, cond_var str
 					if j == 0 && sumtype_index == 0 {
 						g.empty_line = true
 					}
-					g.write_v_source_line_info(branch.pos)
+					g.write_v_source_line_info(branch)
 					g.write('if (')
 				}
 				g.write(cond_var)
@@ -428,7 +428,7 @@ fn (mut g Gen) match_expr_classic(node ast.MatchExpr, is_expr bool, cond_var str
 					g.write(' : ')
 				} else {
 					g.writeln('')
-					g.write_v_source_line_info(branch.pos)
+					g.write_v_source_line_info(branch)
 					g.writeln('else {')
 				}
 			}
@@ -438,7 +438,7 @@ fn (mut g Gen) match_expr_classic(node ast.MatchExpr, is_expr bool, cond_var str
 					g.write(' : ')
 				} else {
 					g.writeln('')
-					g.write_v_source_line_info(branch.pos)
+					g.write_v_source_line_info(branch)
 					g.write('else ')
 				}
 			}
@@ -448,7 +448,7 @@ fn (mut g Gen) match_expr_classic(node ast.MatchExpr, is_expr bool, cond_var str
 				if j == 0 {
 					g.writeln('')
 				}
-				g.write_v_source_line_info(branch.pos)
+				g.write_v_source_line_info(branch)
 				g.write('if (')
 			}
 			for i, expr in branch.exprs {