Support glob syntax for input paths

Author: nicklockwood

Sources/Arguments.swift

@@ -235,7 +235,7 @@ func parseRules(_ rules: String) throws -> [String] {
     }
 }
 
-// Parse single file path
+// Parse single file path, disallowing globs or commas
 func parsePath(_ path: String, for argument: String, in directory: String) throws -> URL {
     let expandedPath = expandPath(path, in: directory)
     if !FileManager.default.fileExists(atPath: expandedPath.path) {
@@ -249,11 +249,9 @@ func parsePath(_ path: String, for argument: String, in directory: String) throw
     return expandedPath
 }
 
-// Parse one or more comma-delimited file paths
-func parsePaths(_ paths: String, for argument: String, in directory: String) throws -> [URL] {
-    return try parseCommaDelimitedList(paths).map {
-        try parsePath($0, for: argument, in: directory)
-    }
+// Parse one or more comma-delimited file paths, expanding globs as required
+func parsePaths(_ paths: String, in directory: String) throws -> [URL] {
+    return try matchGlobs(expandGlobs(paths, in: directory), in: directory)
 }
 
 // Merge two dictionaries of arguments

Sources/CommandLine.swift

@@ -398,16 +398,18 @@ func processArguments(_ args: [String], in directory: String) -> ExitCode {
             let fileListURL = try parsePath(fileListPath, for: "filelist", in: directory)
             do {
                 let source = try String(contentsOf: fileListURL)
-                inputURLs += parseFileList(source, in: fileListURL.deletingLastPathComponent().path)
+                inputURLs += try parseFileList(source, in: fileListURL.deletingLastPathComponent().path)
             } catch {
                 throw FormatError.options("Failed to read file list at \(fileListPath)")
             }
         }
         var useStdin = false
         while let inputPath = args[String(inputURLs.count + 1)] {
-            inputURLs += try parsePaths(inputPath, for: "input", in: directory)
             if inputPath.lowercased() == "stdin" {
                 useStdin = true
+                inputURLs.append(URL(string: "stdin")!)
+            } else {
+                inputURLs += try parsePaths(inputPath, in: directory)
             }
         }
         if useStdin {
@@ -447,7 +449,7 @@ func processArguments(_ args: [String], in directory: String) -> ExitCode {
             if arg.lowercased() == "stdin" {
                 useStdin = true
             } else {
-                inputURLs += try parsePaths(arg, for: "input", in: directory)
+                inputURLs += try parsePaths(arg, in: directory)
             }
         }
         try addInputPaths(for: "quiet")
@@ -715,12 +717,11 @@ func processArguments(_ args: [String], in directory: String) -> ExitCode {
     }
 }
 
-func parseFileList(_ source: String, in directory: String) -> [URL] {
-    return source
+func parseFileList(_ source: String, in directory: String) throws -> [URL] {
+    return try source
         .components(separatedBy: .newlines)
         .map { $0.components(separatedBy: "#")[0].trimmingCharacters(in: .whitespaces) }
-        .filter { !$0.isEmpty }
-        .map { expandPath($0, in: directory) }
+        .flatMap { try parsePaths($0, in: directory) }
 }
 
 func printResult(_ dryrun: Bool, _ lint: Bool, _ lenient: Bool, _ flags: OutputFlags) -> ExitCode {

Sources/Globs.swift

@@ -15,14 +15,18 @@ func pathContainsGlobSyntax(_ path: String) -> Bool {
 /// Glob type represents either an exact path or wildcard
 public enum Glob: CustomStringConvertible {
     case path(String)
-    case regex(NSRegularExpression)
+    case regex(String, NSRegularExpression)
 
     public func matches(_ path: String) -> Bool {
         switch self {
         case let .path(_path):
             return _path == path
-        case let .regex(regex):
-            let range = NSRange(location: 0, length: path.utf16.count)
+        case let .regex(prefix, regex):
+            guard path.hasPrefix(prefix) else {
+                return false
+            }
+            let count = prefix.utf16.count
+            let range = NSRange(location: count, length: path.utf16.count - count)
             return regex.firstMatch(in: path, options: [], range: range) != nil
         }
     }
@@ -31,7 +35,7 @@ public enum Glob: CustomStringConvertible {
         switch self {
         case let .path(path):
             return path
-        case let .regex(regex):
+        case let .regex(prefix, regex):
             var result = regex.pattern.dropFirst().dropLast()
                 .replacingOccurrences(of: "([^/]+)?", with: "*")
                 .replacingOccurrences(of: "(.+/)?", with: "**/")
@@ -42,7 +46,7 @@ public enum Glob: CustomStringConvertible {
                 let options = result[range].dropFirst().dropLast().components(separatedBy: "|")
                 result.replaceSubrange(range, with: "{\(options.joined(separator: ","))}")
             }
-            return result
+            return prefix + result
         }
     }
 }
@@ -70,7 +74,16 @@ public func expandGlobs(_ paths: String, in directory: String) -> [Glob] {
             // TODO: should we also handle cases where path includes tokens?
             return .path(path)
         }
-        var regex = "^\(path)$"
+        var prefix = "", regex = ""
+        let parts = path.components(separatedBy: "/")
+        for (i, part) in parts.enumerated() {
+            if pathContainsGlobSyntax(part) || part.contains("<<<") {
+                regex = parts[i...].joined(separator: "/")
+                break
+            }
+            prefix += "\(part)/"
+        }
+        regex = "^\(regex)$"
             .replacingOccurrences(of: "[.+(){\\\\|]", with: "\\\\$0", options: .regularExpression)
             .replacingOccurrences(of: "?", with: "[^/]")
             .replacingOccurrences(of: "**/", with: "(.+/)?")
@@ -79,16 +92,15 @@ public func expandGlobs(_ paths: String, in directory: String) -> [Glob] {
         for (token, replacement) in tokens {
             regex = regex.replacingOccurrences(of: token, with: replacement)
         }
-        return try! .regex(NSRegularExpression(pattern: regex, options: []))
+        return try! .regex(prefix, NSRegularExpression(pattern: regex, options: []))
     }
 }
 
-// NOTE: currently only used for testing
-func matchGlobs(_ globs: [Glob], in directory: String) -> [URL] {
+func matchGlobs(_ globs: [Glob], in directory: String) throws -> [URL] {
     var urls = [URL]()
     let keys: [URLResourceKey] = [.isDirectoryKey]
     let manager = FileManager.default
-    func enumerate(_ directory: URL) {
+    func enumerate(_ directory: URL, with glob: Glob) {
         guard let files = try? manager.contentsOfDirectory(
             at: directory, includingPropertiesForKeys: keys, options: []
         ) else {
@@ -97,15 +109,34 @@ func matchGlobs(_ globs: [Glob], in directory: String) -> [URL] {
         for url in files {
             let path = url.path
             var isDirectory: ObjCBool = false
-            if globs.contains(where: { $0.matches(path) }) {
+            if glob.matches(path) {
                 urls.append(url)
             } else if manager.fileExists(atPath: path, isDirectory: &isDirectory),
                       isDirectory.boolValue
             {
-                enumerate(url)
+                enumerate(url, with: glob)
+            }
+        }
+    }
+    for glob in globs {
+        switch glob {
+        case let .path(path):
+            if manager.fileExists(atPath: path) {
+                urls.append(URL(fileURLWithPath: path))
+            } else {
+                throw FormatError.options("File not found at \(glob)")
+            }
+        case let .regex(path, _):
+            let count = urls.count
+            if directory.hasPrefix(path) {
+                enumerate(URL(fileURLWithPath: directory).standardized, with: glob)
+            } else if path.hasPrefix(directory) {
+                enumerate(URL(fileURLWithPath: path).standardized, with: glob)
+            }
+            if count == urls.count {
+                throw FormatError.options("Glob did not match any files at \(glob)")
             }
         }
     }
-    enumerate(URL(fileURLWithPath: directory).standardized)
     return urls
 }

Tests/CommandLineTests.swift

@@ -211,14 +211,16 @@ class CommandLineTests: XCTestCase {
     func testParseFileList() {
         let source = """
         #foo
-        foo.swift #bar
+        Package.swift #bar
 
         #baz
-        bar/baz.swift
+        Sources/Rules.swift
+        CommandLineTool/*.swift
         """
-        XCTAssertEqual(parseFileList(source, in: projectDirectory.path), [
-            URL(fileURLWithPath: "\(projectDirectory.path)/foo.swift"),
-            URL(fileURLWithPath: "\(projectDirectory.path)/bar/baz.swift"),
+        XCTAssertEqual(try parseFileList(source, in: projectDirectory.path), [
+            URL(fileURLWithPath: "\(projectDirectory.path)/Package.swift"),
+            URL(fileURLWithPath: "\(projectDirectory.path)/Sources/Rules.swift"),
+            URL(fileURLWithPath: "\(projectDirectory.path)/CommandLineTool/main.swift"),
         ])
     }