WIP enhance language detection

cmd/root.go

@@ -77,6 +77,9 @@ func cmdMain(cmd *cobra.Command, args []string) (err error) {
 			cmd.PrintErrln("Error:", err)
 			return err
 		}
+		if lexer == nil {
+			lexer = detectShebang(data)
+		}
 		if lexer == nil {
 			lexer = lexers.Analyse(string(data))
 		}
@@ -91,6 +94,10 @@ func cmdMain(cmd *cobra.Command, args []string) (err error) {
 			}
 			if language == "" {
 				lexer = lexers.Match(filename)
+				if lexer == nil {
+					lexer = lexers.Analyse(string(data))
+					// lexer = detectShebang((data))
+				}
 			}
 			printData(&data, cmd, lexer)
 		}
@@ -101,6 +108,65 @@ func cmdMain(cmd *cobra.Command, args []string) (err error) {
 	return
 }
 
+func detectShebang(data []byte) chroma.Lexer {
+	if len(data) < 2 || data[0] != '#' || data[1] != '!' {
+		return nil
+	}
+
+	// Find the end of the first line
+	endIdx := bytes.IndexByte(data, '\n')
+	if endIdx == -1 {
+		endIdx = len(data)
+	}
+
+	shebangLine := bytes.TrimSpace(data[2:endIdx]) // Skip "#!"
+
+	// Parse interpreter from shebang
+	var interpreter string
+	if bytes.Contains(shebangLine, []byte("/env ")) {
+		// Handle "/usr/bin/env python3" format
+		parts := bytes.Fields(shebangLine)
+		if len(parts) >= 2 {
+			interpreter = string(parts[1])
+		}
+	} else {
+		// Handle "/bin/bash" format
+		parts := bytes.Split(shebangLine, []byte("/"))
+		if len(parts) > 0 {
+			lastPart := parts[len(parts)-1]
+			// Remove any arguments after space
+			if idx := bytes.IndexByte(lastPart, ' '); idx != -1 {
+				lastPart = lastPart[:idx]
+			}
+			interpreter = string(lastPart)
+		}
+	}
+
+	// Map interpreter to lexer
+	interpreterMap := map[string]string{
+		"bash":    "bash",
+		"sh":      "bash",
+		"zsh":     "bash",
+		"ksh":     "bash",
+		"python":  "python",
+		"python3": "python",
+		"python2": "python",
+		"ruby":    "ruby",
+		"perl":    "perl",
+		"node":    "javascript",
+		"nodejs":  "javascript",
+		"php":     "php",
+		"lua":     "lua",
+		"awk":     "awk",
+	}
+
+	if lexerName, ok := interpreterMap[interpreter]; ok {
+		return lexers.Get(lexerName)
+	}
+
+	return nil
+}
+
 func printData(data *[]byte, cmd *cobra.Command, lexer chroma.Lexer) {
 	out := cmd.OutOrStdout()
 	if number {

cmd/root_test.go

@@ -353,3 +353,132 @@ func _unhighlightedGoCode() string {
 	}
 	return unhighlightedGoCode
 }
+
+func TestDetectShebang(t *testing.T) {
+	t.Run("Bash shebang", func(t *testing.T) {
+		data := []byte("#!/bin/bash\necho 'hello'")
+		lexer := detectShebang(data)
+		assert.NotNil(t, lexer)
+		assert.Contains(t, lexer.Config().Name, "Bash")
+	})
+
+	t.Run("Python3 with env", func(t *testing.T) {
+		data := []byte("#!/usr/bin/env python3\nprint('hello')")
+		lexer := detectShebang(data)
+		assert.NotNil(t, lexer)
+		assert.Contains(t, lexer.Config().Name, "Python")
+	})
+
+	t.Run("Ruby shebang", func(t *testing.T) {
+		data := []byte("#!/usr/bin/ruby\nputs 'hello'")
+		lexer := detectShebang(data)
+		assert.NotNil(t, lexer)
+		assert.Contains(t, lexer.Config().Name, "Ruby")
+	})
+
+	t.Run("Node with env", func(t *testing.T) {
+		data := []byte("#!/usr/bin/env node\nconsole.log('hello')")
+		lexer := detectShebang(data)
+		assert.NotNil(t, lexer)
+		assert.Contains(t, lexer.Config().Name, "JavaScript")
+	})
+
+	t.Run("Sh shebang", func(t *testing.T) {
+		data := []byte("#!/bin/sh\necho 'hello'")
+		lexer := detectShebang(data)
+		assert.NotNil(t, lexer)
+		assert.Contains(t, lexer.Config().Name, "Bash")
+	})
+
+	t.Run("Perl shebang", func(t *testing.T) {
+		data := []byte("#!/usr/bin/perl\nprint 'hello'")
+		lexer := detectShebang(data)
+		assert.NotNil(t, lexer)
+		assert.Contains(t, lexer.Config().Name, "Perl")
+	})
+
+	t.Run("No shebang", func(t *testing.T) {
+		data := []byte("echo 'hello'")
+		lexer := detectShebang(data)
+		assert.Nil(t, lexer)
+	})
+
+	t.Run("Comment but not shebang", func(t *testing.T) {
+		data := []byte("# This is a comment\necho 'hello'")
+		lexer := detectShebang(data)
+		assert.Nil(t, lexer)
+	})
+
+	t.Run("Unknown interpreter", func(t *testing.T) {
+		data := []byte("#!/bin/unknowninterpreter\necho 'hello'")
+		lexer := detectShebang(data)
+		assert.Nil(t, lexer)
+	})
+
+	t.Run("Shebang with arguments", func(t *testing.T) {
+		data := []byte("#!/bin/bash -e\necho 'hello'")
+		lexer := detectShebang(data)
+		assert.NotNil(t, lexer)
+		assert.Contains(t, lexer.Config().Name, "Bash")
+	})
+
+	t.Run("Empty file", func(t *testing.T) {
+		data := []byte("")
+		lexer := detectShebang(data)
+		assert.Nil(t, lexer)
+	})
+}
+
+func TestShebangFileDetection(t *testing.T) {
+	setupTerminalMock(t)
+
+	t.Run("Bash script without extension", func(t *testing.T) {
+		var o, e bytes.Buffer
+		rootCmd.SetArgs([]string{"testdata/bashscript"})
+		rootCmd.SetOut(&o)
+		rootCmd.SetErr(&e)
+		err := rootCmd.Execute()
+
+		assert.NoError(t, err)
+		assert.Empty(t, e.String())
+		assert.NotEmpty(t, o.String())
+		// Should be syntax highlighted as bash
+		assert.Contains(t, o.String(), "[38;5;")
+	})
+
+	t.Run("Python script without extension", func(t *testing.T) {
+		t.Cleanup(resetStrings)
+		var o, e bytes.Buffer
+		rootCmd.SetArgs([]string{"testdata/pythonscript"})
+		rootCmd.SetOut(&o)
+		rootCmd.SetErr(&e)
+		err := rootCmd.Execute()
+
+		assert.NoError(t, err)
+		assert.Empty(t, e.String())
+		assert.NotEmpty(t, o.String())
+		// Should be syntax highlighted as python
+		assert.Contains(t, o.String(), "[38;5;")
+	})
+}
+
+func TestShebangFromStdin(t *testing.T) {
+	setupTerminalMock(t)
+
+	t.Run("Bash script from stdin", func(t *testing.T) {
+		t.Cleanup(resetStrings)
+		i := bytes.NewBufferString("#!/bin/bash\necho 'hello'")
+		var o, e bytes.Buffer
+		rootCmd.SetArgs([]string{})
+		rootCmd.SetIn(i)
+		rootCmd.SetOut(&o)
+		rootCmd.SetErr(&e)
+		err := rootCmd.Execute()
+
+		assert.NoError(t, err)
+		assert.Empty(t, e.String())
+		assert.NotEmpty(t, o.String())
+		// Should be syntax highlighted
+		assert.Contains(t, o.String(), "[38;5;")
+	})
+}

cmd/testdata/bashscript

@@ -0,0 +1,2 @@
+#!/bin/bash
+echo "hello from bash"

cmd/testdata/pythonscript

@@ -0,0 +1,2 @@
+#!/usr/bin/env python3
+print("hello from python")

testscript/bashscript

@@ -0,0 +1,2 @@
+#!/bin/bash
+echo "hello from bash"

testscript/pythonscript

@@ -0,0 +1,2 @@
+#!/usr/bin/env python3
+print("hello from python")