Allow for styles to persist over lines (#123)

* Allow for styles to persist over lines

* Update to pass by reference for style attributes

srt.go

@@ -42,6 +42,7 @@ func ReadFromSRT(i io.Reader) (o *Subtitles, err error) {
 	var line string
 	var lineNum int
 	var s = &Item{}
+	var sa = &StyleAttributes{}
 	for scanner.Scan() {
 		// Fetch line
 		line = strings.TrimSpace(scanner.Text())
@@ -58,6 +59,9 @@ func ReadFromSRT(i io.Reader) (o *Subtitles, err error) {
 
 		// Line contains time boundaries
 		if strings.Contains(line, srtTimeBoundariesSeparator) {
+			// Reset style attributes
+			sa = &StyleAttributes{}
+
 			// Remove last item of previous subtitle since it should be the index.
 			// If the last line is empty then the item is missing an index.
 			var index string
@@ -118,7 +122,7 @@ func ReadFromSRT(i io.Reader) (o *Subtitles, err error) {
 			o.Items = append(o.Items, s)
 		} else {
 			// Add text
-			if l := parseTextSrt(strings.TrimSpace(line)); len(l.Items) > 0 {
+			if l := parseTextSrt(line, sa); len(l.Items) > 0 {
 				s.Lines = append(s.Lines, l)
 			}
 		}
@@ -127,7 +131,7 @@ func ReadFromSRT(i io.Reader) (o *Subtitles, err error) {
 }
 
 // parseTextSrt parses the input line to fill the Line
-func parseTextSrt(i string) (o Line) {
+func parseTextSrt(i string, sa *StyleAttributes) (o Line) {
 	// special handling needed for empty line
 	if strings.TrimSpace(i) == "" {
 		o.Items = []LineItem{{Text: ""}}
@@ -138,13 +142,6 @@ func parseTextSrt(i string) (o Line) {
 	tr := html.NewTokenizer(strings.NewReader(i))
 
 	// Loop
-	var (
-		bold      bool
-		italic    bool
-		underline bool
-		color     *string
-		pos       byte
-	)
 	for {
 		// Get next tag
 		t := tr.Next()
@@ -164,46 +161,45 @@ func parseTextSrt(i string) (o Line) {
 			// Parse italic/bold/underline
 			switch token.Data {
 			case "b":
-				bold = false
+				sa.SRTBold = false
 			case "i":
-				italic = false
+				sa.SRTItalics = false
 			case "u":
-				underline = false
+				sa.SRTUnderline = false
 			case "font":
-				color = nil
+				sa.SRTColor = nil
 			}
 		case html.StartTagToken:
 			// Parse italic/bold/underline
 			switch token.Data {
 			case "b":
-				bold = true
+				sa.SRTBold = true
 			case "i":
-				italic = true
+				sa.SRTItalics = true
 			case "u":
-				underline = true
+				sa.SRTUnderline = true
 			case "font":
 				if c := htmlTokenAttribute(&token, "color"); c != nil {
-					color = c
+					sa.SRTColor = c
 				}
 			}
 		case html.TextToken:
 			if s := strings.TrimSpace(raw); s != "" {
 				// Get style attribute
-				var sa *StyleAttributes
-				if bold || italic || underline || color != nil || pos != 0 {
-					sa = &StyleAttributes{
-						SRTBold:      bold,
-						SRTColor:     color,
-						SRTItalics:   italic,
-						SRTPosition:  pos,
-						SRTUnderline: underline,
+				var styleAttributes *StyleAttributes
+				if sa.SRTBold || sa.SRTColor != nil || sa.SRTItalics || sa.SRTUnderline {
+					styleAttributes = &StyleAttributes{
+						SRTBold:      sa.SRTBold,
+						SRTColor:     sa.SRTColor,
+						SRTItalics:   sa.SRTItalics,
+						SRTUnderline: sa.SRTUnderline,
 					}
-					sa.propagateSRTAttributes()
+					styleAttributes.propagateSRTAttributes()
 				}
 
 				// Append item
 				o.Items = append(o.Items, LineItem{
-					InlineStyle: sa,
+					InlineStyle: styleAttributes,
 					Text:        unescapeHTML(s),
 				})
 			}

srt_test.go

@@ -60,7 +60,7 @@ func TestSRTStyled(t *testing.T) {
 	assert.NoError(t, err)
 
 	// assert the items are properly parsed
-	assert.Len(t, s.Items, 6)
+	assert.Len(t, s.Items, 9)
 	assert.Equal(t, 17*time.Second+985*time.Millisecond, s.Items[0].StartAt)
 	assert.Equal(t, 20*time.Second+521*time.Millisecond, s.Items[0].EndAt)
 	assert.Equal(t, "[instrumental music]", s.Items[0].Lines[0].String())
@@ -80,36 +80,47 @@ func TestSRTStyled(t *testing.T) {
 	assert.Equal(t, 1*time.Minute+31*time.Second+992*time.Millisecond, s.Items[5].EndAt)
 	assert.Equal(t, "[automated]", s.Items[5].Lines[0].String())
 	assert.Equal(t, "'The time is 7:35.'", s.Items[5].Lines[1].String())
+	assert.Equal(t, "Test with multi line italics", s.Items[6].Lines[0].String())
+	assert.Equal(t, "Terminated on the next line", s.Items[6].Lines[1].String())
+	assert.Equal(t, "Unterminated styles", s.Items[7].Lines[0].String())
+	assert.Equal(t, "Do no fall to the next item", s.Items[8].Lines[0].String())
 
 	// assert the styles of the items
-	assert.Len(t, s.Items, 6)
 	assert.Equal(t, "#00ff00", *s.Items[0].Lines[0].Items[0].InlineStyle.SRTColor)
-	assert.Zero(t, s.Items[0].Lines[0].Items[0].InlineStyle.SRTPosition)
 	assert.True(t, s.Items[0].Lines[0].Items[0].InlineStyle.SRTBold)
 	assert.False(t, s.Items[0].Lines[0].Items[0].InlineStyle.SRTItalics)
 	assert.False(t, s.Items[0].Lines[0].Items[0].InlineStyle.SRTUnderline)
 	assert.Equal(t, "#ff00ff", *s.Items[1].Lines[0].Items[0].InlineStyle.SRTColor)
-	assert.Zero(t, s.Items[1].Lines[0].Items[0].InlineStyle.SRTPosition)
 	assert.False(t, s.Items[1].Lines[0].Items[0].InlineStyle.SRTBold)
 	assert.False(t, s.Items[1].Lines[0].Items[0].InlineStyle.SRTItalics)
 	assert.False(t, s.Items[1].Lines[0].Items[0].InlineStyle.SRTUnderline)
 	assert.Equal(t, "#00ff00", *s.Items[2].Lines[0].Items[0].InlineStyle.SRTColor)
-	assert.Zero(t, s.Items[2].Lines[0].Items[0].InlineStyle.SRTPosition)
 	assert.False(t, s.Items[2].Lines[0].Items[0].InlineStyle.SRTBold)
 	assert.False(t, s.Items[2].Lines[0].Items[0].InlineStyle.SRTItalics)
 	assert.False(t, s.Items[2].Lines[0].Items[0].InlineStyle.SRTUnderline)
 	assert.Nil(t, s.Items[3].Lines[0].Items[0].InlineStyle.SRTColor)
-	assert.Zero(t, s.Items[3].Lines[0].Items[0].InlineStyle.SRTPosition)
 	assert.True(t, s.Items[3].Lines[0].Items[0].InlineStyle.SRTBold)
 	assert.False(t, s.Items[3].Lines[0].Items[0].InlineStyle.SRTItalics)
 	assert.True(t, s.Items[3].Lines[0].Items[0].InlineStyle.SRTUnderline)
 	assert.Nil(t, s.Items[4].Lines[0].Items[0].InlineStyle)
 	assert.Nil(t, s.Items[5].Lines[0].Items[0].InlineStyle)
 	assert.Nil(t, s.Items[5].Lines[1].Items[0].InlineStyle.SRTColor)
-	assert.Zero(t, s.Items[5].Lines[1].Items[0].InlineStyle.SRTPosition)
 	assert.False(t, s.Items[5].Lines[1].Items[0].InlineStyle.SRTBold)
 	assert.True(t, s.Items[5].Lines[1].Items[0].InlineStyle.SRTItalics)
 	assert.False(t, s.Items[5].Lines[1].Items[0].InlineStyle.SRTUnderline)
+	assert.True(t, s.Items[6].Lines[0].Items[0].InlineStyle.SRTItalics)
+	assert.False(t, s.Items[6].Lines[0].Items[0].InlineStyle.SRTUnderline)
+	assert.False(t, s.Items[6].Lines[0].Items[0].InlineStyle.SRTBold)
+	assert.Nil(t, s.Items[6].Lines[0].Items[0].InlineStyle.SRTColor)
+	assert.True(t, s.Items[6].Lines[1].Items[0].InlineStyle.SRTItalics)
+	assert.False(t, s.Items[6].Lines[1].Items[0].InlineStyle.SRTUnderline)
+	assert.False(t, s.Items[6].Lines[1].Items[0].InlineStyle.SRTBold)
+	assert.Nil(t, s.Items[6].Lines[1].Items[0].InlineStyle.SRTColor)
+	assert.True(t, s.Items[7].Lines[0].Items[0].InlineStyle.SRTItalics)
+	assert.False(t, s.Items[7].Lines[0].Items[0].InlineStyle.SRTUnderline)
+	assert.False(t, s.Items[7].Lines[0].Items[0].InlineStyle.SRTBold)
+	assert.Nil(t, s.Items[7].Lines[0].Items[0].InlineStyle.SRTColor)
+	assert.Nil(t, s.Items[8].Lines[0].Items[0].InlineStyle)
 
 	// Write to srt
 	w := &bytes.Buffer{}

testdata/example-in-styled.srt

@@ -22,3 +22,16 @@
 00:01:29,590 --> 00:01:31,992
 [automated]
 <i>'The time is 7:35.'</i>
+
+7
+00:08:00,000 --> 00:09:00,000
+<i>Test with multi line italics
+Terminated on the next line</i>
+
+8
+00:09:00,000 --> 00:10:00,000
+<i>Unterminated styles
+
+9
+00:10:00,000 --> 00:11:00,000
+Do no fall to the next item

testdata/example-out-styled.srt

@@ -22,3 +22,16 @@
 00:01:29,590 --> 00:01:31,992
 [automated]
 <i>'The time is 7:35.'</i>
+
+7
+00:08:00,000 --> 00:09:00,000
+<i>Test with multi line italics</i>
+<i>Terminated on the next line</i>
+
+8
+00:09:00,000 --> 00:10:00,000
+<i>Unterminated styles</i>
+
+9
+00:10:00,000 --> 00:11:00,000
+Do no fall to the next item

testdata/example-out-styled.vtt

@@ -24,3 +24,16 @@ WEBVTT
 00:01:29.590 --> 00:01:31.992
 [automated]
 <i>'The time is 7:35.'</i>
+
+7
+00:08:00.000 --> 00:09:00.000
+<i>Test with multi line italics</i>
+<i>Terminated on the next line</i>
+
+8
+00:09:00.000 --> 00:10:00.000
+<i>Unterminated styles</i>
+
+9
+00:10:00.000 --> 00:11:00.000
+Do no fall to the next item

webvtt.go

@@ -141,7 +141,7 @@ func ReadFromWebVTT(i io.Reader) (o *Subtitles, err error) {
 	var blockName string
 	var comments []string
 	var index int
-	var webVTTStyles *StyleAttributes
+	var sa = &StyleAttributes{}
 
 	for scanner.Scan() {
 		// Fetch line
@@ -162,11 +162,15 @@ func ReadFromWebVTT(i io.Reader) (o *Subtitles, err error) {
 			// Reset block name, if we are not in the middle of CSS.
 			// If we are in STYLE block and the CSS is empty or we meet the right brace at the end of last line,
 			// then we are not in CSS and can switch to parse next WebVTT block.
-			if blockName != webvttBlockNameStyle || webVTTStyles == nil ||
-				len(webVTTStyles.WebVTTStyles) == 0 ||
-				strings.HasSuffix(webVTTStyles.WebVTTStyles[len(webVTTStyles.WebVTTStyles)-1], "}") {
+			if blockName != webvttBlockNameStyle || sa == nil ||
+				len(sa.WebVTTStyles) == 0 ||
+				strings.HasSuffix(sa.WebVTTStyles[len(sa.WebVTTStyles)-1], "}") {
 				blockName = ""
 			}
+
+			// Reset WebVTTTags
+			sa.WebVTTTags = []WebVTTTag{}
+
 		// Region
 		case strings.HasPrefix(line, "Region: "):
 			// Add region styles
@@ -207,9 +211,9 @@ func ReadFromWebVTT(i io.Reader) (o *Subtitles, err error) {
 			blockName = webvttBlockNameStyle
 
 			if _, ok := o.Styles[webvttDefaultStyleID]; !ok {
-				webVTTStyles = &StyleAttributes{}
+				sa = &StyleAttributes{}
 				o.Styles[webvttDefaultStyleID] = &Style{
-					InlineStyle: webVTTStyles,
+					InlineStyle: sa,
 					ID:          webvttDefaultStyleID,
 				}
 			}
@@ -314,10 +318,10 @@ func ReadFromWebVTT(i io.Reader) (o *Subtitles, err error) {
 			case webvttBlockNameComment:
 				comments = append(comments, line)
 			case webvttBlockNameStyle:
-				webVTTStyles.WebVTTStyles = append(webVTTStyles.WebVTTStyles, line)
+				sa.WebVTTStyles = append(sa.WebVTTStyles, line)
 			case webvttBlockNameText:
 				// Parse line
-				if l := parseTextWebVTT(line); len(l.Items) > 0 {
+				if l := parseTextWebVTT(line, sa); len(l.Items) > 0 {
 					item.Lines = append(item.Lines, l)
 				}
 			default:
@@ -330,12 +334,10 @@ func ReadFromWebVTT(i io.Reader) (o *Subtitles, err error) {
 }
 
 // parseTextWebVTT parses the input line to fill the Line
-func parseTextWebVTT(i string) (o Line) {
+func parseTextWebVTT(i string, sa *StyleAttributes) (o Line) {
 	// Create tokenizer
 	tr := html.NewTokenizer(strings.NewReader(i))
 
-	webVTTTagStack := make([]WebVTTTag, 0, 16)
-
 	// Loop
 	for {
 		// Get next tag
@@ -348,8 +350,8 @@ func parseTextWebVTT(i string) (o Line) {
 		switch t {
 		case html.EndTagToken:
 			// Pop the top of stack if we meet end tag
-			if len(webVTTTagStack) > 0 {
-				webVTTTagStack = webVTTTagStack[:len(webVTTTagStack)-1]
+			if len(sa.WebVTTTags) > 0 {
+				sa.WebVTTTags = sa.WebVTTTags[:len(sa.WebVTTTags)-1]
 			}
 		case html.StartTagToken:
 			if matches := webVTTRegexpTag.FindStringSubmatch(string(tr.Raw())); len(matches) > 4 {
@@ -377,7 +379,7 @@ func parseTextWebVTT(i string) (o Line) {
 				}
 
 				// Push the tag to stack
-				webVTTTagStack = append(webVTTTagStack, WebVTTTag{
+				sa.WebVTTTags = append(sa.WebVTTTags, WebVTTTag{
 					Name:       tagName,
 					Classes:    classes,
 					Annotation: annotation,
@@ -386,18 +388,18 @@ func parseTextWebVTT(i string) (o Line) {
 
 		case html.TextToken:
 			// Get style attribute
-			var sa *StyleAttributes
-			if len(webVTTTagStack) > 0 {
-				tags := make([]WebVTTTag, len(webVTTTagStack))
-				copy(tags, webVTTTagStack)
-				sa = &StyleAttributes{
+			var styleAttributes *StyleAttributes
+			if len(sa.WebVTTTags) > 0 {
+				tags := make([]WebVTTTag, len(sa.WebVTTTags))
+				copy(tags, sa.WebVTTTags)
+				styleAttributes = &StyleAttributes{
 					WebVTTTags: tags,
 				}
-				sa.propagateWebVTTAttributes()
+				styleAttributes.propagateWebVTTAttributes()
 			}
 
 			// Append items
-			o.Items = append(o.Items, parseTextWebVTTTextToken(sa, string(tr.Raw()))...)
+			o.Items = append(o.Items, parseTextWebVTTTextToken(styleAttributes, string(tr.Raw()))...)
 		}
 	}
 	return

webvtt_internal_test.go

@@ -13,7 +13,7 @@ func TestParseTextWebVTT(t *testing.T) {
 	t.Run("When both voice tags are available", func(t *testing.T) {
 		testData := `<v Bob>Correct tag</v>`
 
-		s := parseTextWebVTT(testData)
+		s := parseTextWebVTT(testData, &StyleAttributes{})
 		assert.Equal(t, "Bob", s.VoiceName)
 		assert.Equal(t, 1, len(s.Items))
 		assert.Equal(t, "Correct tag", s.Items[0].Text)
@@ -22,7 +22,7 @@ func TestParseTextWebVTT(t *testing.T) {
 	t.Run("When there is no end tag", func(t *testing.T) {
 		testData := `<v Bob> Text without end tag`
 
-		s := parseTextWebVTT(testData)
+		s := parseTextWebVTT(testData, &StyleAttributes{})
 		assert.Equal(t, "Bob", s.VoiceName)
 		assert.Equal(t, 1, len(s.Items))
 		assert.Equal(t, "Text without end tag", s.Items[0].Text)
@@ -31,7 +31,7 @@ func TestParseTextWebVTT(t *testing.T) {
 	t.Run("When the end tag is correct", func(t *testing.T) {
 		testData := `<v Bob>Incorrect end tag</vi>`
 
-		s := parseTextWebVTT(testData)
+		s := parseTextWebVTT(testData, &StyleAttributes{})
 		assert.Equal(t, "Bob", s.VoiceName)
 		assert.Equal(t, 1, len(s.Items))
 		assert.Equal(t, "Incorrect end tag", s.Items[0].Text)
@@ -40,7 +40,7 @@ func TestParseTextWebVTT(t *testing.T) {
 	t.Run("When inline timestamps are included", func(t *testing.T) {
 		testData := `<00:01:01.000>With inline <00:01:02.000>timestamps`
 
-		s := parseTextWebVTT(testData)
+		s := parseTextWebVTT(testData, &StyleAttributes{})
 		assert.Equal(t, 2, len(s.Items))
 		assert.Equal(t, "With inline", s.Items[0].Text)
 		assert.Equal(t, time.Minute+time.Second, s.Items[0].StartAt)
@@ -51,7 +51,7 @@ func TestParseTextWebVTT(t *testing.T) {
 	t.Run("When inline timestamps together", func(t *testing.T) {
 		testData := `<00:01:01.000><00:01:02.000>With timestamp tags together`
 
-		s := parseTextWebVTT(testData)
+		s := parseTextWebVTT(testData, &StyleAttributes{})
 		assert.Equal(t, 1, len(s.Items))
 		assert.Equal(t, "With timestamp tags together", s.Items[0].Text)
 		assert.Equal(t, time.Minute+2*time.Second, s.Items[0].StartAt)
@@ -60,7 +60,7 @@ func TestParseTextWebVTT(t *testing.T) {
 	t.Run("When inline timestamps is at end", func(t *testing.T) {
 		testData := `With end timestamp<00:01:02.000>`
 
-		s := parseTextWebVTT(testData)
+		s := parseTextWebVTT(testData, &StyleAttributes{})
 		assert.Equal(t, 1, len(s.Items))
 		assert.Equal(t, "With end timestamp", s.Items[0].Text)
 		assert.Equal(t, time.Duration(0), s.Items[0].StartAt)