Start work on gitignore parsing

This commit is contained in:
Kovid Goyal 2025-07-08 09:43:50 +05:30
parent b19b38f99e
commit bde347457c
No known key found for this signature in database
GPG key ID: 06BC317B515ACE7C
2 changed files with 255 additions and 0 deletions

View file

@ -0,0 +1,174 @@
package ignorefiles
import (
"fmt"
"io/fs"
"os"
"path/filepath"
"slices"
"strings"
)
var _ = fmt.Print
type GitPattern struct {
only_dirs bool
negated bool
parts []string
matcher func(path string) bool
}
func (p GitPattern) Match(path string, ftype fs.FileMode) bool {
if p.only_dirs && ftype&fs.ModeDir == 0 {
return false
}
if os.PathSeparator != '/' {
path = strings.ReplaceAll(path, string(os.PathSeparator), "/")
}
return p.matcher(path)
}
func anchored_single_match(path string, pattern string) bool {
name, _, _ := strings.Cut(path, "/")
matches, err := filepath.Match(pattern, name)
return err == nil && matches
}
func unanchored_single_match(path string, pattern string) bool {
for path != "" {
var name string
name, path, _ = strings.Cut(path, "/")
matches, err := filepath.Match(pattern, name)
if err != nil {
return false
}
if matches {
return true
}
}
return false
}
func anchored_simple_match(path string, parts []string) bool {
for ; path != "" && len(parts) > 0; parts = parts[1:] {
var name string
name, path, _ = strings.Cut(path, "/")
if matches, err := filepath.Match(parts[0], name); err != nil || !matches {
return false
}
}
return path == "" && len(parts) == 0
}
func anchored_full_match(path string, parts []string) bool {
pos, last := 0, len(parts)-1
for pos <= last && path != "" {
var name string
name, path, _ = strings.Cut(path, "/")
switch name {
case "**":
for pos+1 < len(parts) && parts[pos+1] == "**" {
pos++
}
if pos == last {
return true
}
for {
matches, err := filepath.Match(parts[pos], name)
if err != nil {
return false
}
if matches {
return anchored_full_match(path, parts[pos+1:])
}
if path == "" {
return false
}
name, path, _ = strings.Cut(path, "/")
}
default:
if matches, err := filepath.Match(parts[pos], name); err != nil || !matches {
return false
}
pos++
}
}
return path == "" && pos > last
}
// Parse a line from a .gitignore file, see man gitignore for the syntax
func CompileGitIgnoreLine(line string) (ans GitPattern, skipped_line bool) {
// Strip comments
if strings.HasPrefix(line, `#`) {
skipped_line = true
return
}
// Trim OS-specific carriage returns.
line = strings.TrimRight(line, "\r")
// Trim trailing spaces unless backslash escaped
for strings.HasSuffix(line, " ") {
if strings.HasSuffix(line, `\ `) {
line = line[:len(line)-2] + " "
break
}
line = line[:len(line)-1]
}
// Empty lines are ignored
if line == "" {
skipped_line = true
return
}
// Handle negated (accept) patterns
if line[0] == '!' {
line = line[1:]
ans.negated = true
}
// Handle leading slash used to escape leading # or !
if line[0] == '\\' && len(line) > 1 && (line[1] == '#' || line[1] == '!') {
line = line[1:]
}
if strings.HasSuffix(line, "/") {
ans.only_dirs = true
line = strings.TrimRight(line, "/")
if line == "" {
skipped_line = true
return
}
}
starts_with_slash := strings.HasPrefix(line, "/")
if starts_with_slash {
line = strings.TrimLeft(line, "/")
}
ans.parts = strings.Split(line, "/")
if slices.Contains(ans.parts, "") {
ans.parts = slices.DeleteFunc(ans.parts, func(x string) bool { return x == "" })
}
if len(ans.parts) == 0 {
skipped_line = true
return
}
if len(ans.parts) == 1 {
pattern := ans.parts[0]
if pattern == "**" {
ans.matcher = func(string) bool { return true }
} else {
if starts_with_slash {
ans.matcher = func(path string) bool { return anchored_single_match(path, pattern) }
} else {
ans.matcher = func(path string) bool { return unanchored_single_match(path, pattern) }
}
}
} else {
if slices.Contains(ans.parts, "**") {
ans.matcher = func(path string) bool { return anchored_full_match(path, ans.parts) }
} else {
ans.matcher = func(path string) bool { return anchored_simple_match(path, ans.parts) }
}
}
return
}

View file

@ -0,0 +1,81 @@
package ignorefiles
import (
"fmt"
"io/fs"
"strings"
"testing"
"github.com/google/go-cmp/cmp"
"github.com/kovidgoyal/kitty/tools/utils"
)
var _ = fmt.Print
func TestGitignore(t *testing.T) {
for line, expected := range map[string]struct {
skipped, negated, only_dirs bool
parts []string
}{
"": {skipped: true},
" ": {skipped: true},
" ": {skipped: true},
"/": {skipped: true},
"//": {skipped: true},
"# abc": {skipped: true},
`\!moose \ `: {parts: []string{`!moose `}},
`\#m\oose `: {parts: []string{`#m\oose`}},
} {
p, skipped := CompileGitIgnoreLine(line)
if skipped != expected.skipped {
t.Fatalf("skipped: %v != %v for line: %s", expected.skipped, skipped, line)
}
if !skipped {
if p.negated != expected.negated {
t.Fatalf("negated: %v != %v for line: %s", expected.negated, p.negated, line)
}
if p.only_dirs != expected.only_dirs {
t.Fatalf("only_dirs: %v != %v for line: %s", expected.only_dirs, p.only_dirs, line)
}
if diff := cmp.Diff(expected.parts, p.parts); diff != "" {
t.Fatalf("parts not equal for line: %s\n%s", line, diff)
}
}
}
type ptest struct {
path string
expected bool
}
for _, x := range []struct {
line string
tests []ptest
}{
{"foo", []ptest{
{"foo", true}, {"x/foo", true}, {"foo/x", true},
}},
{"/foo", []ptest{
{"foo", true}, {"x/foo", false},
}},
{"doc/frotz/", []ptest{
{"doc/frotz/", true}, {"a/doc/frotz/", false}, {"doc/frotz", false},
}},
{"frotz/", []ptest{
{"frotz/", true}, {"a/doc/frotz/", true}, {"doc/frotz", false}, {"frotz/", true},
}},
{"foo.*", []ptest{
{"foo.txt", true}, {"foo", false}, {"a/foo.x", true}, {"foo.", true},
}},
} {
p, skipped := CompileGitIgnoreLine(x.line)
if skipped {
t.Fatalf("Unexpectedly failed to compile: %#v", x.line)
}
for _, test := range x.tests {
path := strings.TrimRight(test.path, "/")
ftype := utils.IfElse(len(path) < len(test.path), fs.ModeDir, 0)
if actual := p.Match(path, ftype); actual != test.expected {
t.Fatalf("matched: %v != %v for pattern: %#v and path: %#v", test.expected, actual, x.line, test.path)
}
}
}
}