bjoernager/forgejo
bjoernager/forgejo
1
Fork 0
Code Activity

feat: highlighted code search results (#4749)

Browse source 
closes #4534

<details>
<summary>Screenshots</summary>

![](https://codeberg.org/attachments/0ab8a7b0-6485-46dc-a730-c016abb1f287)
</details>

Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/4749
Reviewed-by: 0ko <0ko@noreply.codeberg.org>
Co-authored-by: Shiny Nematoda <snematoda.751k2@aleeas.com>
Co-committed-by: Shiny Nematoda <snematoda.751k2@aleeas.com>
This commit is contained in:
Shiny Nematoda 2024-08-06 05:57:25 +00:00 • committed by Earl Warren
parent 517637137c
commit 06d2e90fa4
10 changed files with 214 additions and 75 deletions
Download patch file Download diff file Expand all files Collapse all files

48
modules/git/grep.go
View file

@ -1,4 +1,5 @@
// Copyright 2024 The Gitea Authors. All rights reserved.
// Copyright 2024 The Forgejo Authors. All rights reserved.
// SPDX-License-Identifier: MIT
package git
@ -19,9 +20,10 @@ import (
)
type GrepResult struct {
Filename string
LineNumbers []int
LineCodes []string
Filename string
LineNumbers []int
LineCodes []string
HighlightedRanges [][3]int
}
type GrepOptions struct {
@ -33,6 +35,13 @@ type GrepOptions struct {
PathSpec []setting.Glob
}
func hasPrefixFold(s, t string) bool {
if len(s) < len(t) {
return false
}
return strings.EqualFold(s[:len(t)], t)
}
func GrepSearch(ctx context.Context, repo *Repository, search string, opts GrepOptions) ([]*GrepResult, error) {
stdoutReader, stdoutWriter, err := os.Pipe()
if err != nil {
@ -53,18 +62,19 @@ func GrepSearch(ctx context.Context, repo *Repository, search string, opts GrepO
2^@repo: go-gitea/gitea
*/
var results []*GrepResult
cmd := NewCommand(ctx, "grep", "--null", "--break", "--heading", "--fixed-strings", "--line-number", "--ignore-case", "--full-name")
cmd := NewCommand(ctx, "grep",
"--null", "--break", "--heading", "--column",
"--fixed-strings", "--line-number", "--ignore-case", "--full-name")
cmd.AddOptionValues("--context", fmt.Sprint(opts.ContextLineNumber))
if opts.MatchesPerFile > 0 {
cmd.AddOptionValues("--max-count", fmt.Sprint(opts.MatchesPerFile))
}
words := []string{search}
if opts.IsFuzzy {
words := strings.Fields(search)
for _, word := range words {
cmd.AddOptionValues("-e", strings.TrimLeft(word, "-"))
}
} else {
cmd.AddOptionValues("-e", strings.TrimLeft(search, "-"))
words = strings.Fields(search)
}
for _, word := range words {
cmd.AddOptionValues("-e", strings.TrimLeft(word, "-"))
}
// pathspec
@ -128,6 +138,24 @@ func GrepSearch(ctx context.Context, repo *Repository, search string, opts GrepO
if lineNum, lineCode, ok := strings.Cut(line, "\x00"); ok {
lineNumInt, _ := strconv.Atoi(lineNum)
res.LineNumbers = append(res.LineNumbers, lineNumInt)
if lineCol, lineCode2, ok := strings.Cut(lineCode, "\x00"); ok {
lineColInt, _ := strconv.Atoi(lineCol)
start := lineColInt - 1
matchLen := len(lineCode2)
for _, word := range words {
if hasPrefixFold(lineCode2[start:], word) {
matchLen = len(word)
break
}
}
res.HighlightedRanges = append(res.HighlightedRanges, [3]int{
len(res.LineCodes),
start,
start + matchLen,
})
res.LineCodes = append(res.LineCodes, lineCode2)
continue
}
res.LineCodes = append(res.LineCodes, lineCode)
}
}

59
modules/git/grep_test.go
View file

@ -20,28 +20,43 @@ func TestGrepSearch(t *testing.T) {
require.NoError(t, err)
defer repo.Close()
res, err := GrepSearch(context.Background(), repo, "void", GrepOptions{})
res, err := GrepSearch(context.Background(), repo, "public", GrepOptions{})
require.NoError(t, err)
assert.Equal(t, []*GrepResult{
{
Filename: "java-hello/main.java",
LineNumbers: []int{3},
LineCodes: []string{" public static void main(String[] args)"},
LineNumbers: []int{1, 3},
LineCodes: []string{
"public class HelloWorld",
" public static void main(String[] args)",
},
HighlightedRanges: [][3]int{{0, 0, 6}, {1, 1, 7}},
},
{
Filename: "main.vendor.java",
LineNumbers: []int{3},
LineCodes: []string{" public static void main(String[] args)"},
LineNumbers: []int{1, 3},
LineCodes: []string{
"public class HelloWorld",
" public static void main(String[] args)",
},
HighlightedRanges: [][3]int{{0, 0, 6}, {1, 1, 7}},
},
}, res)
res, err = GrepSearch(context.Background(), repo, "void", GrepOptions{MaxResultLimit: 1})
res, err = GrepSearch(context.Background(), repo, "void", GrepOptions{MaxResultLimit: 1, ContextLineNumber: 2})
require.NoError(t, err)
assert.Equal(t, []*GrepResult{
{
Filename: "java-hello/main.java",
LineNumbers: []int{3},
LineCodes: []string{" public static void main(String[] args)"},
LineNumbers: []int{1, 2, 3, 4, 5},
LineCodes: []string{
"public class HelloWorld",
"{",
" public static void main(String[] args)",
" {",
" System.out.println(\"Hello world!\");",
},
HighlightedRanges: [][3]int{{2, 15, 19}},
},
}, res)
@ -49,24 +64,28 @@ func TestGrepSearch(t *testing.T) {
require.NoError(t, err)
assert.Equal(t, []*GrepResult{
{
Filename: "i-am-a-python.p",
LineNumbers: []int{1},
LineCodes: []string{"## This is a simple file to do a hello world"},
Filename: "i-am-a-python.p",
LineNumbers: []int{1},
LineCodes: []string{"## This is a simple file to do a hello world"},
HighlightedRanges: [][3]int{{0, 39, 44}},
},
{
Filename: "java-hello/main.java",
LineNumbers: []int{1},
LineCodes: []string{"public class HelloWorld"},
Filename: "java-hello/main.java",
LineNumbers: []int{1},
LineCodes: []string{"public class HelloWorld"},
HighlightedRanges: [][3]int{{0, 18, 23}},
},
{
Filename: "main.vendor.java",
LineNumbers: []int{1},
LineCodes: []string{"public class HelloWorld"},
Filename: "main.vendor.java",
LineNumbers: []int{1},
LineCodes: []string{"public class HelloWorld"},
HighlightedRanges: [][3]int{{0, 18, 23}},
},
{
Filename: "python-hello/hello.py",
LineNumbers: []int{1},
LineCodes: []string{"## This is a simple file to do a hello world"},
Filename: "python-hello/hello.py",
LineNumbers: []int{1},
LineCodes: []string{"## This is a simple file to do a hello world"},
HighlightedRanges: [][3]int{{0, 39, 44}},
},
}, res)

83
modules/indexer/code/search.go
View file

@ -12,6 +12,7 @@ import (
"code.gitea.io/gitea/modules/highlight"
"code.gitea.io/gitea/modules/indexer/code/internal"
"code.gitea.io/gitea/modules/timeutil"
"code.gitea.io/gitea/services/gitdiff"
)
// Result a search result to display
@ -70,11 +71,85 @@ func writeStrings(buf *bytes.Buffer, strs ...string) error {
return nil
}
func HighlightSearchResultCode(filename string, lineNums []int, code string) []ResultLine {
const (
highlightTagStart = "<span class=\"search-highlight\">"
highlightTagEnd = "</span>"
)
func HighlightSearchResultCode(filename string, lineNums []int, highlightRanges [][3]int, code string) []ResultLine {
hcd := gitdiff.NewHighlightCodeDiff()
hcd.CollectUsedRunes(code)
startTag, endTag := hcd.NextPlaceholder(), hcd.NextPlaceholder()
hcd.PlaceholderTokenMap[startTag] = highlightTagStart
hcd.PlaceholderTokenMap[endTag] = highlightTagEnd
// we should highlight the whole code block first, otherwise it doesn't work well with multiple line highlighting
hl, _ := highlight.Code(filename, "", code)
highlightedLines := strings.Split(string(hl), "\n")
conv := hcd.ConvertToPlaceholders(string(hl))
convLines := strings.Split(conv, "\n")
// each highlightRange is of the form [line number, start pos, end pos]
for _, highlightRange := range highlightRanges {
ln, start, end := highlightRange[0], highlightRange[1], highlightRange[2]
line := convLines[ln]
if line == "" || len(line) <= start || len(line) < end {
continue
}
sb := strings.Builder{}
count := -1
isOpen := false
for _, r := range line {
if token, ok := hcd.PlaceholderTokenMap[r];
// token was not found
!ok ||
// token was marked as used
token == "" ||
// the token is not an valid html tag emited by chroma
!(len(token) > 6 && (token[0:5] == "<span" || token[0:6] == "</span")) {
count++
} else if !isOpen {
// open the tag only after all other placeholders
sb.WriteRune(r)
continue
} else if isOpen && count < end {
// if the tag is open, but a placeholder exists in between
// close the tag
sb.WriteRune(endTag)
// write the placeholder
sb.WriteRune(r)
// reopen the tag
sb.WriteRune(startTag)
continue
}
switch count {
case end:
// if tag is not open, no need to close
if !isOpen {
break
}
sb.WriteRune(endTag)
isOpen = false
case start:
// if tag is open, do not open again
if isOpen {
break
}
isOpen = true
sb.WriteRune(startTag)
}
sb.WriteRune(r)
}
if isOpen {
sb.WriteRune(endTag)
}
convLines[ln] = sb.String()
}
conv = strings.Join(convLines, "\n")
highlightedLines := strings.Split(hcd.Recover(conv), "\n")
// The lineNums outputted by highlight.Code might not match the original lineNums, because "highlight" removes the last `\n`
lines := make([]ResultLine, min(len(highlightedLines), len(lineNums)))
for i := 0; i < len(lines); i++ {
@ -92,6 +167,7 @@ func searchResult(result *internal.SearchResult, startIndex, endIndex int) (*Res
contentLines := strings.SplitAfter(result.Content[startIndex:endIndex], "\n")
lineNums := make([]int, 0, len(contentLines))
index := startIndex
var highlightRanges [][3]int
for i, line := range contentLines {
var err error
if index < result.EndIndex &&
@ -99,6 +175,7 @@ func searchResult(result *internal.SearchResult, startIndex, endIndex int) (*Res
result.StartIndex < result.EndIndex {
openActiveIndex := max(result.StartIndex-index, 0)
closeActiveIndex := min(result.EndIndex-index, len(line))
highlightRanges = append(highlightRanges, [3]int{i, openActiveIndex, closeActiveIndex})
err = writeStrings(&formattedLinesBuffer,
line[:openActiveIndex],
line[openActiveIndex:closeActiveIndex],
@ -122,7 +199,7 @@ func searchResult(result *internal.SearchResult, startIndex, endIndex int) (*Res
UpdatedUnix: result.UpdatedUnix,
Language: result.Language,
Color: result.Color,
Lines: HighlightSearchResultCode(result.Filename, lineNums, formattedLinesBuffer.String()),
Lines: HighlightSearchResultCode(result.Filename, lineNums, highlightRanges, formattedLinesBuffer.String()),
}, nil
}