From 867f46f78e5777d7bafe83cda3244c1001ce3e60 Mon Sep 17 00:00:00 2001 From: Norwin Date: Thu, 15 Aug 2019 22:09:50 +0000 Subject: 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 --- modules/markup/csv/csv.go | 61 ++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 58 insertions(+), 3 deletions(-) (limited to 'modules/markup/csv/csv.go') diff --git a/modules/markup/csv/csv.go b/modules/markup/csv/csv.go index 077947e774..1e3acc9b47 100644 --- 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(``) for { @@ -50,9 +57,57 @@ func (Parser) Render(rawBytes []byte, urlPrefix string, metas map[string]string, tmpBlock.WriteString(html.EscapeString(field)) tmpBlock.WriteString("") } - tmpBlock.WriteString("") + tmpBlock.WriteString("") } tmpBlock.WriteString("
") 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))) +} -- cgit v1.2.3