feat(graph): C/C++ file detection and tree-sitter parsing foundation

sast-engine/graph/clike/detection.go

@@ -0,0 +1,120 @@
+// Package clike contains shared helpers for parsing C and C++ source files.
+//
+// The parsing pipeline treats C and C++ as two distinct languages with separate
+// tree-sitter grammars but a large amount of shared structure (declarations,
+// statements, type strings). The helpers in this package live alongside the
+// language-specific siblings (graph/golang, graph/python, graph/java) and
+// provide the cross-cutting primitives — language detection today, AST and
+// type extraction in subsequent PRs.
+package clike
+
+import (
+	"bufio"
+	"os"
+	"path/filepath"
+	"strings"
+	"sync"
+)
+
+// headerLanguageCache stores the C/C++ classification of .h files keyed by
+// absolute path. Values are bool: true => C++, false => C.
+//
+// .h is shared between C and C++ grammars and the correct grammar can only be
+// chosen by inspecting the file contents. Reading the file once per AST node
+// would be unacceptable, so the worker that drives the parsing pipeline calls
+// CacheHeaderLanguage exactly once per .h file before AST traversal begins;
+// IsCSourceFile and IsCppSourceFile then read the cache without performing I/O.
+var headerLanguageCache sync.Map
+
+// CacheHeaderLanguage records that filename is a C++ header (isCpp == true) or
+// a C header (isCpp == false). Must be called once per .h file in the parsing
+// worker before any IsCSourceFile / IsCppSourceFile lookup for that file.
+func CacheHeaderLanguage(filename string, isCpp bool) {
+	headerLanguageCache.Store(filename, isCpp)
+}
+
+// IsCSourceFile reports whether filename should be parsed with the C grammar.
+//
+// Source-extension cases (.c) are answered directly. For .h files the answer
+// comes from the header cache populated by CacheHeaderLanguage; an uncached
+// .h falls back to C as the safe default (the grammar overlap means a
+// misclassified C++ header still parses as a structurally-valid translation
+// unit, just with reduced fidelity).
+func IsCSourceFile(filename string) bool {
+	ext := filepath.Ext(filename)
+	if ext == ".c" {
+		return true
+	}
+	if ext == ".h" {
+		if v, ok := headerLanguageCache.Load(filename); ok {
+			return !v.(bool)
+		}
+		return true
+	}
+	return false
+}
+
+// IsCppSourceFile reports whether filename should be parsed with the C++
+// grammar.
+//
+// .cpp/.cc/.cxx and the C++-only header extensions (.hpp/.hh/.hxx) are
+// answered directly. For .h files the answer comes from the header cache
+// populated by CacheHeaderLanguage; an uncached .h falls back to "not C++"
+// so that IsCSourceFile and IsCppSourceFile remain mutually exclusive.
+func IsCppSourceFile(filename string) bool {
+	ext := filepath.Ext(filename)
+	switch ext {
+	case ".cpp", ".cc", ".cxx", ".hpp", ".hh", ".hxx":
+		return true
+	case ".h":
+		if v, ok := headerLanguageCache.Load(filename); ok {
+			return v.(bool)
+		}
+		return false
+	}
+	return false
+}
+
+// cppHeaderIndicators are byte sequences whose presence in a header file is a
+// strong signal that the header should be parsed as C++ rather than C. The
+// list is intentionally small and conservative — every entry is invalid in
+// pure C, so a positive match is high-confidence.
+var cppHeaderIndicators = []string{
+	"class ", "namespace ", "template<", "template <",
+	"public:", "private:", "protected:", "::",
+}
+
+// detectCppHeaderScanLines bounds how far DetectCppInHeader reads into a file.
+// 100 lines covers the include guards, license header, and the first few
+// declarations of every real-world header surveyed during the design phase.
+const detectCppHeaderScanLines = 100
+
+// DetectCppInHeader scans the first detectCppHeaderScanLines lines of filename
+// for C++-only indicators (see cppHeaderIndicators).
+//
+// This is a best-effort heuristic, not a full preprocessor or parser. It is
+// invoked from the worker exactly once per file and its result is stored via
+// CacheHeaderLanguage; AST-traversal code reads the cache and never calls
+// this function on the hot path.
+//
+// A missing or unreadable file returns false (treated as C) so that downstream
+// parsing fails on an obviously-broken input rather than mislabeling the
+// language.
+func DetectCppInHeader(filename string) bool {
+	f, err := os.Open(filename)
+	if err != nil {
+		return false
+	}
+	defer f.Close()
+
+	scanner := bufio.NewScanner(f)
+	for lineCount := 0; lineCount < detectCppHeaderScanLines && scanner.Scan(); lineCount++ {
+		line := scanner.Text()
+		for _, ind := range cppHeaderIndicators {
+			if strings.Contains(line, ind) {
+				return true
+			}
+		}
+	}
+	return false
+}

sast-engine/graph/clike/detection_test.go

@@ -0,0 +1,158 @@
+package clike
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+)
+
+func TestIsCSourceFile(t *testing.T) {
+	defer headerLanguageCache.Delete("/tmp/cache_h_c.h")
+	defer headerLanguageCache.Delete("/tmp/cache_h_cpp.h")
+
+	CacheHeaderLanguage("/tmp/cache_h_c.h", false)
+	CacheHeaderLanguage("/tmp/cache_h_cpp.h", true)
+
+	tests := []struct {
+		name     string
+		filename string
+		want     bool
+	}{
+		{".c file", "main.c", true},
+		{".cpp file", "main.cpp", false},
+		{".cc file", "main.cc", false},
+		{".cxx file", "main.cxx", false},
+		{".hpp file", "main.hpp", false},
+		{".hh file", "main.hh", false},
+		{".hxx file", "main.hxx", false},
+		{".java file", "Main.java", false},
+		{".py file", "main.py", false},
+		{".go file", "main.go", false},
+		{"no extension", "Makefile", false},
+		{".h cached as C", "/tmp/cache_h_c.h", true},
+		{".h cached as C++", "/tmp/cache_h_cpp.h", false},
+		{".h not cached defaults to C", "/tmp/cache_h_unknown.h", true},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := IsCSourceFile(tt.filename); got != tt.want {
+				t.Errorf("IsCSourceFile(%q) = %v, want %v", tt.filename, got, tt.want)
+			}
+		})
+	}
+}
+
+func TestIsCppSourceFile(t *testing.T) {
+	defer headerLanguageCache.Delete("/tmp/cppcache_c.h")
+	defer headerLanguageCache.Delete("/tmp/cppcache_cpp.h")
+
+	CacheHeaderLanguage("/tmp/cppcache_c.h", false)
+	CacheHeaderLanguage("/tmp/cppcache_cpp.h", true)
+
+	tests := []struct {
+		name     string
+		filename string
+		want     bool
+	}{
+		{".cpp file", "main.cpp", true},
+		{".cc file", "main.cc", true},
+		{".cxx file", "main.cxx", true},
+		{".hpp file", "main.hpp", true},
+		{".hh file", "main.hh", true},
+		{".hxx file", "main.hxx", true},
+		{".c file", "main.c", false},
+		{".java file", "Main.java", false},
+		{".py file", "main.py", false},
+		{".go file", "main.go", false},
+		{".h cached as C", "/tmp/cppcache_c.h", false},
+		{".h cached as C++", "/tmp/cppcache_cpp.h", true},
+		{".h not cached defaults to C", "/tmp/cppcache_unknown.h", false},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := IsCppSourceFile(tt.filename); got != tt.want {
+				t.Errorf("IsCppSourceFile(%q) = %v, want %v", tt.filename, got, tt.want)
+			}
+		})
+	}
+}
+
+func TestDetectCppInHeader(t *testing.T) {
+	tests := []struct {
+		name    string
+		content string
+		want    bool
+	}{
+		{
+			name: "pure C header with typedef and struct",
+			content: `#ifndef UTILS_H
+#define UTILS_H
+typedef struct Point { int x; int y; } Point;
+int add(int a, int b);
+#endif
+`,
+			want: false,
+		},
+		{
+			name: "C++ class header",
+			content: `#pragma once
+class Foo {
+public:
+    int bar();
+};
+`,
+			want: true,
+		},
+		{
+			name:    "namespace header",
+			content: "namespace mylib {\nint compute();\n}\n",
+			want:    true,
+		},
+		{
+			name:    "template header",
+			content: "template<typename T>\nT identity(T v) { return v; }\n",
+			want:    true,
+		},
+		{
+			name:    "qualified call uses ::",
+			content: "void f() { std::cout << 1; }\n",
+			want:    true,
+		},
+		{
+			name:    "empty file",
+			content: "",
+			want:    false,
+		},
+		{
+			name:    "extern C block (no C++ indicator on first lines)",
+			content: "#ifdef __cplusplus\nextern \"C\" {\n#endif\nint plain_c(void);\n",
+			want:    false,
+		},
+	}
+
+	dir, err := os.MkdirTemp("", "detect_cpp_header")
+	if err != nil {
+		t.Fatalf("temp dir: %v", err)
+	}
+	defer os.RemoveAll(dir)
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			path := filepath.Join(dir, "h_"+tt.name+".h")
+			if err := os.WriteFile(path, []byte(tt.content), 0644); err != nil {
+				t.Fatalf("write: %v", err)
+			}
+			if got := DetectCppInHeader(path); got != tt.want {
+				t.Errorf("DetectCppInHeader(%q) = %v, want %v", tt.name, got, tt.want)
+			}
+		})
+	}
+
+	t.Run("missing file returns false", func(t *testing.T) {
+		if DetectCppInHeader(filepath.Join(dir, "does_not_exist.h")) {
+			t.Error("expected false for missing file")
+		}
+	})
+}

sast-engine/graph/initialize.go

@@ -7,7 +7,10 @@ import (
 	"sync"
 	"time"
 
+	"github.com/shivasurya/code-pathfinder/sast-engine/graph/clike"
 	sitter "github.com/smacker/go-tree-sitter"
+	clang "github.com/smacker/go-tree-sitter/c"
+	cpplang "github.com/smacker/go-tree-sitter/cpp"
 	"github.com/smacker/go-tree-sitter/golang"
 	"github.com/smacker/go-tree-sitter/java"
 	"github.com/smacker/go-tree-sitter/python"
@@ -90,18 +93,27 @@ func Initialize(directory string, callbacks *ProgressCallbacks) *CodeGraph {
 				continue
 			}
 
-			// Handle tree-sitter based parsing for Java and Python
-			switch fileExt {
-			case ".java":
+			// For .h files, classify as C vs C++ once and cache the result so
+			// per-AST-node language checks remain zero-I/O during traversal.
+			if fileExt == ".h" {
+				clike.CacheHeaderLanguage(file, clike.DetectCppInHeader(file))
+			}
+
+			// Handle tree-sitter based parsing for Java, Python, Go, C, and C++.
+			// C/C++ cases come first because .h is shared across both grammars and
+			// must route via the cached heuristic, not a simple extension match.
+			switch {
+			case clike.IsCSourceFile(file):
+				parser.SetLanguage(clang.GetLanguage())
+			case clike.IsCppSourceFile(file):
+				parser.SetLanguage(cpplang.GetLanguage())
+			case fileExt == ".java":
 				parser.SetLanguage(java.GetLanguage())
-			case ".py":
+			case fileExt == ".py":
 				parser.SetLanguage(python.GetLanguage())
-			case ".go":
+			case fileExt == ".go":
 				parser.SetLanguage(golang.GetLanguage())
 			default:
-				// NOTE: This case is currently unreachable because getFiles() only returns
-				// .java, .py, Dockerfile*, and docker-compose* files. This exists as defensive
-				// programming in case getFiles() is modified to include additional file types.
 				Log("Unsupported file type:", file)
 				if callbacks != nil && callbacks.OnProgress != nil {
 					callbacks.OnProgress()

sast-engine/graph/utils.go

@@ -252,8 +252,9 @@ func extractMethodName(node *sitter.Node, sourceCode []byte, filepath string) (s
 	return methodName, methodID
 }
 
-// getFiles walks through a directory and returns all source files (Java, Python, Go, Dockerfile, docker-compose).
-// It skips vendor/, testdata/, node_modules/, .git/, and directories starting with "_".
+// getFiles walks through a directory and returns all source files (Java, Python, Go, C/C++, Dockerfile, docker-compose).
+// It skips vendor/, testdata/, node_modules/, .git/, common C/C++ build artifact directories,
+// and directories starting with "_".
 func getFiles(directory string) ([]string, error) {
 	var files []string
 	err := filepath.Walk(directory, func(path string, info os.FileInfo, err error) error {
@@ -264,22 +265,29 @@ func getFiles(directory string) ([]string, error) {
 		if info.IsDir() {
 			name := info.Name()
 			switch name {
-			case "vendor", "testdata", "node_modules", ".git":
+			case "vendor", "testdata", "node_modules", ".git",
+				"build", "cmake-build-debug", "cmake-build-release",
+				"third_party", "external", "obj", "bin", "dist", ".cache":
 				return filepath.SkipDir
 			}
-			if strings.HasPrefix(name, "_") {
+			if strings.HasPrefix(name, "_") || strings.HasPrefix(name, "cmake-build-") {
 				return filepath.SkipDir
 			}
 			return nil
 		}
-		// append java, python, go, dockerfile, and docker-compose files
+		// append java, python, go, c/c++, dockerfile, and docker-compose files
 		ext := filepath.Ext(path)
 		base := filepath.Base(path)
 		baseLower := strings.ToLower(base)
 
 		switch {
 		case ext == ".java" || ext == ".py" || ext == ".go":
 			files = append(files, path)
+		case ext == ".c" || ext == ".h":
+			files = append(files, path)
+		case ext == ".cpp" || ext == ".cc" || ext == ".cxx" ||
+			ext == ".hpp" || ext == ".hh" || ext == ".hxx":
+			files = append(files, path)
 		case strings.HasPrefix(baseLower, "dockerfile"):
 			// Match Dockerfile, Dockerfile.dev, dockerfile, etc.
 			files = append(files, path)