Detect delimiter in CSV rendering (#7869)

* detect csv delimiter in csv rendering fixes #7868 * make linter happy * fix failing testcase & use ints where possible * expose markup type to template previously all markup had the .markdown class, which is incorrect, as it applies markdown CSS & JS logic to CSV rendering * fix build (missing `make css`) * ignore quoted csv content for delimiter scoring also fix html generation
2024-12-25 13:19:37 -05:00 · 2019-08-15 22:09:50 +00:00 · 2019-08-15 22:09:50 +00:00 · 867f46f78e
commit 867f46f78e
parent 0a86d25e6a
6 changed files with 76 additions and 9 deletions
--- a/modules/markup/csv/csv.go
+++ b/modules/markup/csv/csv.go
@ -9,12 +9,18 @@ import (
 	"encoding/csv"
 	"html"
 	"io"
+	"regexp"
+	"strings"

 	"code.gitea.io/gitea/modules/markup"
+	"code.gitea.io/gitea/modules/util"
 )

+var quoteRegexp = regexp.MustCompile(`["'][\s\S]+?["']`)
+
 func init() {
 	markup.RegisterParser(Parser{})
+
 }

 // Parser implements markup.Parser for orgmode
@ -28,12 +34,13 @@ func (Parser) Name() string {

 // Extensions implements markup.Parser
 func (Parser) Extensions() []string {
-	return []string{".csv"}
+	return []string{".csv", ".tsv"}
 }

 // Render implements markup.Parser
-func (Parser) Render(rawBytes []byte, urlPrefix string, metas map[string]string, isWiki bool) []byte {
+func (p Parser) Render(rawBytes []byte, urlPrefix string, metas map[string]string, isWiki bool) []byte {
 	rd := csv.NewReader(bytes.NewReader(rawBytes))
+	rd.Comma = p.bestDelimiter(rawBytes)
 	var tmpBlock bytes.Buffer
 	tmpBlock.WriteString(`<table class="table">`)
 	for {
@ -50,9 +57,57 @@ func (Parser) Render(rawBytes []byte, urlPrefix string, metas map[string]string,
 			tmpBlock.WriteString(html.EscapeString(field))
 			tmpBlock.WriteString("</td>")
 		}
-		tmpBlock.WriteString("<tr>")
+		tmpBlock.WriteString("</tr>")
 	}
 	tmpBlock.WriteString("</table>")

 	return tmpBlock.Bytes()
 }
+
+// bestDelimiter scores the input CSV data against delimiters, and returns the best match.
+// Reads at most 10k bytes & 10 lines.
+func (p Parser) bestDelimiter(data []byte) rune {
+	maxLines := 10
+	maxBytes := util.Min(len(data), 1e4)
+	text := string(data[:maxBytes])
+	text = quoteRegexp.ReplaceAllLiteralString(text, "")
+	lines := strings.SplitN(text, "\n", maxLines+1)
+	lines = lines[:util.Min(maxLines, len(lines))]
+
+	delimiters := []rune{',', ';', '\t', '|'}
+	bestDelim := delimiters[0]
+	bestScore := 0.0
+	for _, delim := range delimiters {
+		score := p.scoreDelimiter(lines, delim)
+		if score > bestScore {
+			bestScore = score
+			bestDelim = delim
+		}
+	}
+
+	return bestDelim
+}
+
+// scoreDelimiter uses a count & regularity metric to evaluate a delimiter against lines of CSV
+func (Parser) scoreDelimiter(lines []string, delim rune) (score float64) {
+	countTotal := 0
+	countLineMax := 0
+	linesNotEqual := 0
+
+	for _, line := range lines {
+		if len(line) == 0 {
+			continue
+		}
+
+		countLine := strings.Count(line, string(delim))
+		countTotal += countLine
+		if countLine != countLineMax {
+			if countLineMax != 0 {
+				linesNotEqual++
+			}
+			countLineMax = util.Max(countLine, countLineMax)
+		}
+	}
+
+	return float64(countTotal) * (1 - float64(linesNotEqual)/float64(len(lines)))
+}
--- a/modules/markup/csv/csv_test.go
+++ b/modules/markup/csv/csv_test.go
@ -13,9 +13,14 @@ import (
 func TestRenderCSV(t *testing.T) {
 	var parser Parser
 	var kases = map[string]string{
-		"a":     "<table class=\"table\"><tr><td>a</td><tr></table>",
-		"1,2":   "<table class=\"table\"><tr><td>1</td><td>2</td><tr></table>",
-		"<br/>": "<table class=\"table\"><tr><td>&lt;br/&gt;</td><tr></table>",
+		"a":                         "<table class=\"table\"><tr><td>a</td></tr></table>",
+		"1,2":                       "<table class=\"table\"><tr><td>1</td><td>2</td></tr></table>",
+		"1;2":                       "<table class=\"table\"><tr><td>1</td><td>2</td></tr></table>",
+		"1\t2":                      "<table class=\"table\"><tr><td>1</td><td>2</td></tr></table>",
+		"1|2":                       "<table class=\"table\"><tr><td>1</td><td>2</td></tr></table>",
+		"1,2,3;4,5,6;7,8,9\na;b;c":  "<table class=\"table\"><tr><td>1,2,3</td><td>4,5,6</td><td>7,8,9</td></tr><tr><td>a</td><td>b</td><td>c</td></tr></table>",
+		"\"1,2,3,4\";\"a\nb\"\nc;d": "<table class=\"table\"><tr><td>1,2,3,4</td><td>a\nb</td></tr><tr><td>c</td><td>d</td></tr></table>",
+		"<br/>":                     "<table class=\"table\"><tr><td>&lt;br/&gt;</td></tr></table>",
 	}

 	for k, v := range kases {
--- a/public/css/index.css
+++ b/public/css/index.css
@ -489,6 +489,7 @@ footer .ui.left,footer .ui.right{line-height:40px}
 .repository.file.list .non-diff-file-content .view-raw img{padding:5px 5px 0 5px}
 .repository.file.list .non-diff-file-content .plain-text{padding:1em 2em 1em 2em}
 .repository.file.list .non-diff-file-content .plain-text pre{word-break:break-word;white-space:pre-wrap}
+.repository.file.list .non-diff-file-content .csv{overflow-x:auto}
 .repository.file.list .non-diff-file-content pre{overflow:auto}
 .repository.file.list .sidebar{padding-left:0}
 .repository.file.list .sidebar .octicon{width:16px}
--- a/public/less/_repository.less
+++ b/public/less/_repository.less
@ -400,6 +400,10 @@
                }
            }

+            .csv {
+                overflow-x: auto;
+            }
+
            pre {
                overflow: auto;
            }
--- a/routers/repo/view.go
+++ b/routers/repo/view.go
@ -162,8 +162,9 @@ func renderDirectory(ctx *context.Context, treeLink string) {
 				d, _ := ioutil.ReadAll(dataRc)
 				buf = charset.ToUTF8WithFallback(append(buf, d...))

-				if markup.Type(readmeFile.Name()) != "" {
+				if markupType := markup.Type(readmeFile.Name()); markupType != "" {
 					ctx.Data["IsMarkup"] = true
+					ctx.Data["MarkupType"] = string(markupType)
 					ctx.Data["FileContent"] = string(markup.Render(readmeFile.Name(), buf, treeLink, ctx.Repo.Repository.ComposeMetas()))
 				} else {
 					ctx.Data["IsRenderedHTML"] = true
@ -282,8 +283,9 @@ func renderFile(ctx *context.Context, entry *git.TreeEntry, treeLink, rawLink st

 		readmeExist := markup.IsReadmeFile(blob.Name())
 		ctx.Data["ReadmeExist"] = readmeExist
-		if markup.Type(blob.Name()) != "" {
+		if markupType := markup.Type(blob.Name()); markupType != "" {
 			ctx.Data["IsMarkup"] = true
+			ctx.Data["MarkupType"] = markupType
 			ctx.Data["FileContent"] = string(markup.Render(blob.Name(), buf, path.Dir(treeLink), ctx.Repo.Repository.ComposeMetas()))
 		} else if readmeExist {
 			ctx.Data["IsRenderedHTML"] = true
--- a/templates/repo/view_file.tmpl
+++ b/templates/repo/view_file.tmpl
@ -45,7 +45,7 @@
 		</div>
 	</h4>
 	<div class="ui attached table unstackable segment">
-		<div class="file-view {{if .IsMarkup}}markdown{{else if .IsRenderedHTML}}plain-text{{else if .IsTextFile}}code-view{{end}} has-emoji">
+		<div class="file-view {{if .IsMarkup}}{{.MarkupType}}{{else if .IsRenderedHTML}}plain-text{{else if .IsTextFile}}code-view{{end}} has-emoji">
 			{{if .IsMarkup}}
 				{{if .FileContent}}{{.FileContent | Safe}}{{end}}
 			{{else if .IsRenderedHTML}}