Add clangd server mode and CppMapper subcommand for C/C++ support

Introduces LSPServerMode to support both sourcekit-lsp (Swift) and clangd directly (C/C++ via CMake). Adds CppMapper subcommand and restructures CLI under CodeMapperCLI with swift/cpp subcommands. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-06-20 08:35:16 -04:00
parent 4aee51762a
commit fa8590048d
4 changed files with 252 additions and 23 deletions
@@ -0,0 +1,119 @@
+import Foundation
+import ArgumentParser
+
+struct CppMapper: ParsableCommand {
+    static let configuration = CommandConfiguration(
+        commandName: "cpp",
+        abstract: "Generate LLM-friendly architectural maps from a C/C++ project via compile_commands.json."
+    )
+
+    @Option(name: .long, help: "Path to compile_commands.json (clang JSON Compilation Database).")
+    var compileCommands: String
+
+    @Option(name: .long, help: "Path to clangd binary.")
+    var clangdPath: String = "/usr/bin/clangd"
+
+    @Option(name: .long, help: "Only write .map files for paths under this folder or matching this file.")
+    var path: String?
+
+    @Option(name: .long, parsing: .upToNextOption, help: "Exclude files under this folder or matching this file. Repeat for multiple exclusions.")
+    var excludePath: [String] = []
+
+    @Flag(name: .long, help: "Omit >> and << call graph lines (faster, fewer tokens).")
+    var noCalls: Bool = false
+
+    @Flag(name: .long, help: "Show only outgoing >> calls, omit << incoming callers.")
+    var outgoingOnly: Bool = false
+
+    @Flag(name: .long, help: "Print map output to stdout instead of writing .map files.")
+    var stdout: Bool = false
+
+    mutating func run() throws {
+        let resolvedCompileCommands = (compileCommands as NSString).standardizingPath
+        let resolvedPath = path.map { ($0 as NSString).standardizingPath }
+        let resolvedExcludePaths = excludePath.map { ($0 as NSString).standardizingPath }
+
+        log("CppMapper starting...")
+        log("compile_commands.json: \(resolvedCompileCommands)")
+        log("clangd: \(clangdPath)")
+
+        let walker = CppSourceWalker(compileCommandsPath: resolvedCompileCommands)
+        let symbolTable = SymbolTable()
+
+        let discoveredFiles = try walker.discoverFiles(symbolTable: symbolTable)
+        let files = discoveredFiles.filter { filePath, _, _ in
+            pathIsIncluded(filePath, includePath: resolvedPath, excludePaths: resolvedExcludePaths)
+        }
+
+        guard !files.isEmpty else {
+            log("No matching C/C++ source files found in compile_commands.json.")
+            return
+        }
+        log("Found \(files.count) C/C++ files.")
+
+        // clangd is pointed directly at the compile_commands.json directory;
+        // it resolves include paths and compiler flags from that database.
+        let lsp = try LSPClient(
+            lspPath: clangdPath,
+            projectRoot: walker.compileCommandsDir,
+            serverMode: .clangd(compileCommandsDir: walker.compileCommandsDir)
+        )
+        log("Initializing clangd...")
+        try lsp.initialize()
+        log("clangd ready.")
+
+        let extractor = SymbolExtractor(lsp: lsp, symbolTable: symbolTable)
+
+        log("Pass 1: extracting symbols...")
+        for (i, (filePath, targetName, language)) in files.enumerated() {
+            if (i + 1) % 20 == 0 { log("  \(i + 1)/\(files.count)") }
+            do {
+                try extractor.process(filePath: filePath, targetName: targetName, language: language)
+            } catch {
+                fputs("Warning: symbol extraction failed for \(filePath): \(error)\n", stderr)
+            }
+        }
+        log("  \(files.count)/\(files.count) done.")
+
+        if !noCalls {
+            let callBuilder = CallGraphBuilder(lsp: lsp, symbolTable: symbolTable)
+            log("Pass 1b: building call graph...")
+            for (i, (filePath, _, _)) in files.enumerated() {
+                if (i + 1) % 10 == 0 { log("  \(i + 1)/\(files.count)") }
+                do {
+                    try callBuilder.process(filePath: filePath)
+                } catch {
+                    fputs("Warning: call graph failed for \(filePath): \(error)\n", stderr)
+                }
+            }
+            log("  \(files.count)/\(files.count) done.")
+        }
+
+        log("Pass 2: building reverse index...")
+        symbolTable.buildReverseIndex()
+        symbolTable.buildImplementorMap()
+
+        let writer = OutputWriter(
+            symbolTable: symbolTable,
+            packageRoot: walker.compileCommandsDir,
+            includeCalls: !noCalls,
+            outgoingOnly: outgoingOnly,
+            pathFilter: resolvedPath,
+            excludePaths: resolvedExcludePaths
+        )
+
+        if stdout {
+            try writer.printAll()
+        } else {
+            log("Writing .map files...")
+            try writer.writeAll()
+            log("Done. Wrote \(symbolTable.fileTargets.count) .map files.")
+        }
+
+        try lsp.shutdownGracefully()
+    }
+
+    private func log(_ msg: String) {
+        fputs(msg + "\n", stderr)
+    }
+}
@@ -0,0 +1,83 @@
+import Foundation
+
+/// Reads a `compile_commands.json` (clang JSON Compilation Database) and
+/// extracts the ordered, deduplicated list of source files to map. Each entry
+/// already carries the full compiler invocation (include paths, defines, std
+/// flags), so clangd can resolve every symbol without any further setup.
+struct CppSourceWalker {
+    let compileCommandsPath: String
+
+    /// Returns the absolute path to the directory containing compile_commands.json,
+    /// used as the clangd `--compile-commands-dir` argument.
+    var compileCommandsDir: String {
+        (compileCommandsPath as NSString).deletingLastPathComponent
+    }
+
+    /// Parses compile_commands.json and returns a deduplicated file list in the
+    /// order they appear in the database. Files with unrecognized extensions are
+    /// silently skipped. Symlinks are resolved via `standardizingPath` so
+    /// duplicates introduced by build-system path aliasing are collapsed.
+    func discoverFiles(symbolTable: SymbolTable) throws -> [(filePath: String, targetName: String, language: SourceLanguage)] {
+        let data = try Data(contentsOf: URL(fileURLWithPath: compileCommandsPath))
+        guard let entries = try JSONSerialization.jsonObject(with: data) as? [[String: Any]] else {
+            throw CppSourceWalkerError.invalidFormat(compileCommandsPath)
+        }
+
+        var files: [(filePath: String, targetName: String, language: SourceLanguage)] = []
+        var seen = Set<String>()
+
+        for entry in entries {
+            guard let rawFile = entry["file"] as? String else { continue }
+
+            // Resolve relative paths against the entry's "directory" field.
+            let absPath: String
+            if (rawFile as NSString).isAbsolutePath {
+                absPath = (rawFile as NSString).standardizingPath
+            } else {
+                let dir = entry["directory"] as? String ?? compileCommandsDir
+                absPath = ((dir + "/" + rawFile) as NSString).standardizingPath
+            }
+
+            guard !seen.contains(absPath) else { continue }
+            seen.insert(absPath)
+
+            let ext = (absPath as NSString).pathExtension
+            guard let language = SourceLanguage(pathExtension: ext),
+                  case .cFamily = language else { continue }
+
+            // Use the directory component as a logical "target" name so the
+            // OutputWriter can group symbols — mirrors how SourceWalker assigns
+            // SwiftPM target names to Swift files.
+            let targetName = inferTargetName(from: absPath)
+            symbolTable.fileTargets[absPath] = targetName
+            files.append((filePath: absPath, targetName: targetName, language: language))
+        }
+
+        return files
+    }
+
+    /// Derives a short logical name from the file's directory path. For a CMake
+    /// project rooted at `compile_commands.json`'s directory, strips that prefix
+    /// and uses the first path component as the target name (e.g. "libs" from
+    /// "src/libs/scanner/ScannerService.cpp"). Falls back to the immediate
+    /// parent directory name.
+    private func inferTargetName(from filePath: String) -> String {
+        let dir = compileCommandsDir
+        let rel = filePath.hasPrefix(dir + "/")
+            ? String(filePath.dropFirst(dir.count + 1))
+            : filePath
+        let components = rel.split(separator: "/")
+        return components.first.map(String.init) ?? (filePath as NSString).lastPathComponent
+    }
+}
+
+enum CppSourceWalkerError: Error, CustomStringConvertible {
+    case invalidFormat(String)
+
+    var description: String {
+        switch self {
+        case .invalidFormat(let path):
+            return "compile_commands.json at \(path) is not a JSON array of objects"
+        }
+    }
+}
@@ -16,6 +16,15 @@ enum LSPError: Error, CustomStringConvertible {
    }
 }

+enum LSPServerMode {
+    /// sourcekit-lsp — used for Swift packages. clangd flags are forwarded
+    /// via `-Xclangd`. Index-store discovery is attempted for faster call graphs.
+    case sourcekitLsp
+    /// clangd directly — used for pure C/C++ projects (e.g. CMake). Flags are
+    /// passed directly; no index-store discovery.
+    case clangd(compileCommandsDir: String)
+}
+
 class LSPClient {
    private let process: Process
    private let stdinPipe: Pipe
@@ -23,6 +32,7 @@ class LSPClient {
    private var nextId = 1
    private var isShutdown = false
    let projectRoot: String
+    private let serverMode: LSPServerMode

    /// Diagnostics pushed by the server via `textDocument/publishDiagnostics`
    /// notifications, keyed by document URI. The server computes these
@@ -33,24 +43,30 @@ class LSPClient {
    /// arrive interleaved with (or before) the next request's response.
    private var diagnosticsByURI: [String: [[String: Any]]] = [:]

-    init(lspPath: String, projectRoot: String, compileCommandsDir: String? = nil) throws {
+    init(lspPath: String, projectRoot: String, compileCommandsDir: String? = nil, serverMode: LSPServerMode = .sourcekitLsp) throws {
        self.projectRoot = projectRoot
+        self.serverMode = serverMode

        process = Process()
        process.executableURL = URL(fileURLWithPath: lspPath)

-        if let compileCommandsDir {
-            // Forwarded through sourcekit-lsp to its internal clangd so
-            // C-family files are parsed with the correct driver language
-            // (see CompilationDatabaseWriter). We deliberately don't force
-            // `--default-workspace-type compilationDatabase` — that bypasses
-            // SwiftPM build-system integration entirely, which would degrade
-            // Swift-side indexing/call-hierarchy in mixed Swift+C-family
-            // packages (the primary use case) for no diagnostic benefit:
-            // sourcekit-lsp doesn't forward clangd's publishDiagnostics for
-            // C-family files in either workspace mode (verified empirically —
-            // not even injected syntax errors produce a notification).
-            process.arguments = ["-Xclangd", "-compile-commands-dir=\(compileCommandsDir)"]
+        switch serverMode {
+        case .sourcekitLsp:
+            if let compileCommandsDir {
+                // Forwarded through sourcekit-lsp to its internal clangd so
+                // C-family files are parsed with the correct driver language
+                // (see CompilationDatabaseWriter). We deliberately don't force
+                // `--default-workspace-type compilationDatabase` — that bypasses
+                // SwiftPM build-system integration entirely, which would degrade
+                // Swift-side indexing/call-hierarchy in mixed Swift+C-family
+                // packages (the primary use case) for no diagnostic benefit:
+                // sourcekit-lsp doesn't forward clangd's publishDiagnostics for
+                // C-family files in either workspace mode (verified empirically —
+                // not even injected syntax errors produce a notification).
+                process.arguments = ["-Xclangd", "-compile-commands-dir=\(compileCommandsDir)"]
+            }
+        case .clangd(let compileCommandsDir):
+            process.arguments = ["--compile-commands-dir=\(compileCommandsDir)"]
        }

        stdinPipe = Pipe()
@@ -71,13 +87,15 @@ class LSPClient {
    }

    func initialize() throws {
-        // Try to find the existing index store for faster call hierarchy
-        let indexStorePath = findIndexStorePath()
-
        var initOptions: [String: Any] = [:]
-        if let isp = indexStorePath {
-            initOptions["indexStorePath"] = isp
-            initOptions["indexDatabasePath"] = "/tmp/codemapper-lsp-index"
+
+        // Index-store discovery only applies to sourcekit-lsp (Swift packages).
+        // clangd builds its own index from compile_commands.json.
+        if case .sourcekitLsp = serverMode {
+            if let isp = findIndexStorePath() {
+                initOptions["indexStorePath"] = isp
+                initOptions["indexDatabasePath"] = "/tmp/codemapper-lsp-index"
+            }
        }

        let params: [String: Any] = [
@@ -94,8 +112,8 @@ class LSPClient {
        _ = try sendRequest("initialize", params: params)
        try sendNotification("initialized", params: [:])

-        if indexStorePath != nil {
-            // Brief pause for LSP to load the index
+        if case .sourcekitLsp = serverMode, !initOptions.isEmpty {
+            // Brief pause for sourcekit-lsp to load the index store
            Thread.sleep(forTimeInterval: 2.0)
        }
    }
@@ -3,9 +3,18 @@ import ArgumentParser

 let buildSignature = "CODEMAPPER-SIGNATURE-izackp"

-struct CodeMapper: ParsableCommand {
+struct CodeMapperCLI: ParsableCommand {
    static let configuration = CommandConfiguration(
        commandName: "CodeMapper",
+        abstract: "Generate LLM-friendly architectural maps from source files.",
+        subcommands: [CodeMapper.self, CppMapper.self],
+        defaultSubcommand: CodeMapper.self
+    )
+}
+
+struct CodeMapper: ParsableCommand {
+    static let configuration = CommandConfiguration(
+        commandName: "swift",
        abstract: "Generate LLM-friendly architectural maps from Swift source files."
    )

@@ -173,7 +182,7 @@ if CommandLine.arguments.contains("--signature") {
    exit(0)
 }

-CodeMapper.main()
+CodeMapperCLI.main()

 private extension JSONEncoder {
    static var pretty: JSONEncoder {