forgejo/modules/git/grep.go
Shiny Nematoda baac15f316 [FEAT] Support Include/Exclude Filters for Grep (#3058)
fixes `TestRepoSearch` failing occasionally

Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/3058
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Co-authored-by: Shiny Nematoda <snematoda.751k2@aleeas.com>
Co-committed-by: Shiny Nematoda <snematoda.751k2@aleeas.com>
2024-04-06 13:25:39 +00:00

127 lines
3.3 KiB
Go

// Copyright 2024 The Gitea Authors. All rights reserved.
// SPDX-License-Identifier: MIT
package git
import (
"bufio"
"bytes"
"cmp"
"context"
"errors"
"fmt"
"os"
"strconv"
"strings"
"code.gitea.io/gitea/modules/setting"
)
type GrepResult struct {
Filename string
LineNumbers []int
LineCodes []string
}
type GrepOptions struct {
RefName string
MaxResultLimit int
ContextLineNumber int
IsFuzzy bool
}
func GrepSearch(ctx context.Context, repo *Repository, search string, opts GrepOptions) ([]*GrepResult, error) {
stdoutReader, stdoutWriter, err := os.Pipe()
if err != nil {
return nil, fmt.Errorf("unable to create os pipe to grep: %w", err)
}
defer func() {
_ = stdoutReader.Close()
_ = stdoutWriter.Close()
}()
/*
The output is like this ( "^@" means \x00):
HEAD:.air.toml
6^@bin = "gitea"
HEAD:.changelog.yml
2^@repo: go-gitea/gitea
*/
var results []*GrepResult
cmd := NewCommand(ctx, "grep", "--null", "--break", "--heading", "--fixed-strings", "--line-number", "--ignore-case", "--full-name")
cmd.AddOptionValues("--context", fmt.Sprint(opts.ContextLineNumber))
if opts.IsFuzzy {
words := strings.Fields(search)
for _, word := range words {
cmd.AddOptionValues("-e", strings.TrimLeft(word, "-"))
}
} else {
cmd.AddOptionValues("-e", strings.TrimLeft(search, "-"))
}
// pathspec
files := make([]string, 0, len(setting.Indexer.IncludePatterns)+len(setting.Indexer.ExcludePatterns))
for _, expr := range setting.Indexer.IncludePatterns {
files = append(files, expr.Pattern())
}
for _, expr := range setting.Indexer.ExcludePatterns {
files = append(files, ":^"+expr.Pattern())
}
cmd.AddDynamicArguments(cmp.Or(opts.RefName, "HEAD")).AddDashesAndList(files...)
opts.MaxResultLimit = cmp.Or(opts.MaxResultLimit, 50)
stderr := bytes.Buffer{}
err = cmd.Run(&RunOpts{
Dir: repo.Path,
Stdout: stdoutWriter,
Stderr: &stderr,
PipelineFunc: func(ctx context.Context, cancel context.CancelFunc) error {
_ = stdoutWriter.Close()
defer stdoutReader.Close()
isInBlock := false
scanner := bufio.NewScanner(stdoutReader)
var res *GrepResult
for scanner.Scan() {
line := scanner.Text()
if !isInBlock {
if _ /* ref */, filename, ok := strings.Cut(line, ":"); ok {
isInBlock = true
res = &GrepResult{Filename: filename}
results = append(results, res)
}
continue
}
if line == "" {
if len(results) >= opts.MaxResultLimit {
cancel()
break
}
isInBlock = false
continue
}
if line == "--" {
continue
}
if lineNum, lineCode, ok := strings.Cut(line, "\x00"); ok {
lineNumInt, _ := strconv.Atoi(lineNum)
res.LineNumbers = append(res.LineNumbers, lineNumInt)
res.LineCodes = append(res.LineCodes, lineCode)
}
}
return scanner.Err()
},
})
// git grep exits by cancel (killed), usually it is caused by the limit of results
if IsErrorExitCode(err, -1) && stderr.Len() == 0 {
return results, nil
}
// git grep exits with 1 if no results are found
if IsErrorExitCode(err, 1) && stderr.Len() == 0 {
return nil, nil
}
if err != nil && !errors.Is(err, context.Canceled) {
return nil, fmt.Errorf("unable to run git grep: %w, stderr: %s", err, stderr.String())
}
return results, nil
}