gg/main.go

/*
gg is a recursive grep written in Go, with some shortcuts for everyday use.
*/
package main

import (
	"bytes"
	"errors"
	"fmt"
	"os"
	"path/filepath"
	"regexp"
	"strings"
	"unicode"
	"unicode/utf8"

	"github.com/spf13/cobra"
	"golang.org/x/sys/unix"
)

// TODO:
//  - bold of escaped output doesn't work
//  - binary file detection
//  - long-line / minified-file detection
//  - ignore files by extension (or glob?)

func main() {
	if err := rootCmd.Execute(); err != nil {
		os.Exit(1)
	}
}

var rootCmd = &cobra.Command{
	Use:   "gg pattern [path1 [path2 …]]",
	Short: "gg is a recursive grep",
	Long: `gg is a recursive grep. Given a regexp (or fixed pattern) it will search for
the pattern recursively in the current working directory. It will print a
coloured header per file along with the matching line and pattern.

It is possible to scan specific files or directories, rather than the default
current working directory. To do this, simply specify the path(s) as arguments
following the pattern.

It is possible to scan for multiple patterns using the -e (or -Q) argument,
which can be repeated multiple times. -e specifies a regular expression and
-Q a fixed pattern. When using either flag, any non-flag arguments are treated
as paths to scan.

Search defaults to case-sensitive but the -i flag may be passed to make regular
expression searches case-insensitive. Alternatively, the "(?i)" construct may be
added to a regular expression to make that specific expression case insensitive.
Fixed pattern matches are always case-sensitive.`,

	RunE: run,
}

var (
	searchRegexp []string
	regexps      []*regexp.Regexp
	searchFixed  []string
	searchBytes  [][]byte
	searchPath   []string
	ignoreList   []string
	ignoreMap    map[string]struct{}
	ignoreCase   bool
	noColour     bool
	display      *Display
	matchedAny   bool
)

func init() {
	rootCmd.Flags().StringSliceVarP(&searchRegexp, "grep", "e", nil, "pattern to match (regular expression)")
	rootCmd.Flags().StringSliceVarP(&searchFixed, "fixed", "Q", nil, "pattern to match (fixed string)")
	rootCmd.Flags().StringSliceVarP(&ignoreList, "exclude", "x", []string{".git"}, "files/directories to exclude")
	rootCmd.Flags().BoolVarP(&ignoreCase, "ignore-case", "i", false, "make all searches case insensitive")
	rootCmd.Flags().BoolVarP(&noColour, "no-colour", "C", false, "disable colour output")
}

func run(c *cobra.Command, args []string) error {
	display = NewDisplay(noColour)

	if len(searchRegexp) == 0 && len(searchFixed) == 0 {
		if len(args) == 0 {
			return errors.New("no pattern specified")
		}
		searchRegexp = append(searchRegexp, args[0])
		args = args[1:]
	}

	searchPath = args
	if len(searchPath) == 0 {
		searchPath = append(searchPath, ".")
	}

	ignoreMap = make(map[string]struct{}, len(ignoreList))
	for _, i := range ignoreList {
		ignoreMap[i] = struct{}{}
	}

	for _, r := range searchRegexp {
		if ignoreCase {
			r = "(?i)" + r
		}
		re, err := regexp.Compile(r)
		if err != nil {
			return err
		}
		regexps = append(regexps, re)
	}

	for _, s := range searchFixed {
		searchBytes = append(searchBytes, []byte(s))
	}

	var errs []error
	for _, path := range searchPath {
		if err := search(path); err != nil {
			errs = append(errs, err)
		}
	}

	return errors.Join(errs...)
}

func recurse(path string) error {
	d, err := os.ReadDir(path)
	if err != nil {
		return err
	}

	var errs []error
	for _, de := range d {
		name := de.Name()
		fullPath := filepath.Join(path, name)

		if _, ignored := ignoreMap[name]; ignored {
			continue
		}

		if err := search(fullPath); err != nil {
			errs = append(errs, err)
		}
	}
	return errors.Join(errs...)
}

func search(path string) error {
	st, err := os.Stat(path)
	if err != nil {
		return err
	}

	switch {
	case st.IsDir():
		return recurse(path)

	case !st.Mode().IsRegular(),
		st.Size() == 0:
		return nil
	}

	f, err := os.Open(path)
	if err != nil {
		return err
	}
	defer f.Close()
	fullData, err := unix.Mmap(int(f.Fd()), 0, int(st.Size()), unix.PROT_READ, unix.MAP_PRIVATE)
	if err != nil {
		return err
	}
	defer unix.Munmap(fullData)

	var printedHeader bool

	var (
		data    = fullData
		lineNum int
		b, b2   strings.Builder
	)
	for len(data) > 0 {
		eol := bytes.IndexByte(data, '\n')
		lineNum++
		var line []byte
		if eol == -1 {
			line = data
			data = nil
		} else {
			line = data[:eol]
			data = data[eol+1:]
		}

		if len(line) == 0 {
			continue
		}

		loc := matches(line)
		if loc == nil {
			continue
		}

		if !printedHeader {
			printedHeader = true
			if !matchedAny {
				matchedAny = true
			} else {
				fmt.Println("")
			}
			fmt.Println(display.Filename(path))
		}

		b.Reset()
		fmt.Fprintf(&b, "%4d: ", display.LineNumber(lineNum))

		if loc[0] < 128 {
			escape(&b, line[0:loc[0]])
		} else {
			start := loc[0] - 128
			for i := 0; i < 5; i++ {
				if utf8.RuneStart(line[start]) {
					break
				}
				start++
			}

			b.WriteString(display.TruncatedBytes(start).String())
			b.WriteString(display.TruncatedMarker().String())
			escape(&b, line[start:loc[0]])
		}

		if loc[1]-loc[0] < 128 {
			b2.Reset()
			escape(&b2, line[loc[0]:loc[1]])
			b.WriteString(display.Match(b2.String()).String())

			if loc[1]+128 > len(line) {
				escape(&b, line[loc[1]:])
			} else {
				end := loc[1] + 128
				for i := 0; i < 5; i++ {
					if utf8.RuneStart(line[end]) {
						break
					}
					end--
				}
				escape(&b, line[loc[1]:end])
				b.WriteString(display.TruncatedBytes(len(line) - end).String())
				b.WriteString(display.TruncatedMarker().String())
			}

		} else {
			end := loc[1]
			for i := 0; i < 5; i++ {
				if utf8.RuneStart(line[end]) {
					break
				}
				end--
			}

			b2.Reset()
			escape(&b2, line[loc[0]:end])
			b.WriteString(display.Match(b2.String()).String())
			b.WriteString(display.TruncatedMarker().String())
			b.WriteString(display.TruncatedBytes(len(line) - end).String())

		}

		b.WriteRune('\n')
		fmt.Print(b.String())
	}

	return nil
}

func matches(data []byte) (loc []int) {
	for _, re := range regexps {
		loc := re.FindIndex(data)
		if loc != nil {
			return loc
		}
	}
	for _, s := range searchBytes {
		pos := bytes.Index(data, s)
		if pos != -1 {
			return []int{pos, pos + len(s)}
		}
	}
	return nil
}

func escape(b *strings.Builder, s []byte) {
	for len(s) > 0 {
		r, size := utf8.DecodeRune(s)
		s = s[size:]

		switch {
		case r == utf8.RuneError && size == 1:
			b.WriteString(display.BadUTF8Char().String())

		case r == '\r':
			b.WriteString(display.CarriageReturn().String())

		case r == '\t',
			unicode.IsPrint(r):
			b.WriteRune(r)

		default:
			b.WriteString(display.UnprintableChar().String())
		}
	}
}