Setup parsing srt stylings for webvtt

srt.go

@@ -8,6 +8,8 @@ import (
 	"strings"
 	"time"
 	"unicode/utf8"
+
+	"golang.org/x/net/html"
 )
 
 // Constants
@@ -116,7 +118,95 @@ func ReadFromSRT(i io.Reader) (o *Subtitles, err error) {
 			o.Items = append(o.Items, s)
 		} else {
 			// Add text
-			s.Lines = append(s.Lines, Line{Items: []LineItem{{Text: strings.TrimSpace(line)}}})
+			if l := parseTextSrt(strings.TrimSpace(line)); len(l.Items) > 0 {
+				s.Lines = append(s.Lines, l)
+			}
+		}
+	}
+	return
+}
+
+// parseTextSrt parses the input line to fill the Line
+func parseTextSrt(i string) (o Line) {
+	// special handling needed for empty line
+	if strings.TrimSpace(i) == "" {
+		o.Items = []LineItem{{Text: ""}}
+		return
+	}
+
+	// Create tokenizer
+	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()
+
+		// Process error
+		if err := tr.Err(); err != nil {
+			break
+		}
+
+		// Get unmodified text
+		raw := string(tr.Raw())
+		// Get current token
+		token := tr.Token()
+
+		switch t {
+		case html.EndTagToken:
+			// Parse italic/bold/underline
+			switch token.Data {
+			case "b":
+				bold = false
+			case "i":
+				italic = false
+			case "u":
+				underline = false
+			case "font":
+				color = nil
+			}
+		case html.StartTagToken:
+			// Parse italic/bold/underline
+			switch token.Data {
+			case "b":
+				bold = true
+			case "i":
+				italic = true
+			case "u":
+				underline = true
+			case "font":
+				if c := htmlTokenAttribute(&token, "color"); c != nil {
+					color = 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,
+					}
+					sa.propagateSRTAttributes()
+				}
+
+				// Append item
+				o.Items = append(o.Items, LineItem{
+					InlineStyle: sa,
+					Text:        s,
+				})
+			}
 		}
 	}
 	return
@@ -151,8 +241,7 @@ func (s Subtitles) WriteToSRT(o io.Writer) (err error) {
 
 		// Loop through lines
 		for _, l := range v.Lines {
-			c = append(c, []byte(l.String())...)
-			c = append(c, bytesLineSeparator...)
+			c = append(c, []byte(l.srtBytes())...)
 		}
 
 		// Add new line
@@ -169,3 +258,65 @@ func (s Subtitles) WriteToSRT(o io.Writer) (err error) {
 	}
 	return
 }
+
+func (l Line) srtBytes() (c []byte) {
+	for idx, li := range l.Items {
+		c = append(c, li.srtBytes()...)
+		// condition to avoid adding space as the last character.
+		if idx < len(l.Items)-1 {
+			c = append(c, []byte(" ")...)
+		}
+	}
+	c = append(c, bytesLineSeparator...)
+	return
+}
+
+func (li LineItem) srtBytes() (c []byte) {
+	// Get color
+	var color string
+	if li.InlineStyle != nil && li.InlineStyle.SRTColor != nil {
+		color = *li.InlineStyle.SRTColor
+	}
+
+	// Get bold/italics/underline
+	b := li.InlineStyle != nil && li.InlineStyle.SRTBold
+	i := li.InlineStyle != nil && li.InlineStyle.SRTItalics
+	u := li.InlineStyle != nil && li.InlineStyle.SRTUnderline
+
+	// Get position
+	var pos byte
+	if li.InlineStyle != nil {
+		pos = li.InlineStyle.SRTPosition
+	}
+
+	// Append
+	if color != "" {
+		c = append(c, []byte("<font color=\""+color+"\">")...)
+	}
+	if b {
+		c = append(c, []byte("<b>")...)
+	}
+	if i {
+		c = append(c, []byte("<i>")...)
+	}
+	if u {
+		c = append(c, []byte("<u>")...)
+	}
+	if pos != 0 {
+		c = append(c, []byte(fmt.Sprintf(`{\an%d}`, pos))...)
+	}
+	c = append(c, []byte(escapeWebVTT(li.Text))...)
+	if u {
+		c = append(c, []byte("</u>")...)
+	}
+	if i {
+		c = append(c, []byte("</i>")...)
+	}
+	if b {
+		c = append(c, []byte("</b>")...)
+	}
+	if color != "" {
+		c = append(c, []byte("</font>")...)
+	}
+	return
+}

subtitles.go

@@ -12,6 +12,7 @@ import (
 	"time"
 
 	"github.com/asticode/go-astikit"
+	"golang.org/x/net/html"
 )
 
 // Bytes
@@ -173,6 +174,11 @@ var (
 
 // StyleAttributes represents style attributes
 type StyleAttributes struct {
+	SRTBold              bool
+	SRTColor             *string
+	SRTItalics           bool
+	SRTPosition          byte // 1-9 numpad layout
+	SRTUnderline         bool
 	SSAAlignment         *int
 	SSAAlphaLevel        *float64
 	SSAAngle             *float64 // degrees
@@ -236,6 +242,8 @@ type StyleAttributes struct {
 	TTMLWritingMode      *string
 	TTMLZIndex           *int
 	WebVTTAlign          string
+	WebVTTBold           bool
+	WebVTTItalics        bool
 	WebVTTLine           string
 	WebVTTLines          int
 	WebVTTPosition       string
@@ -244,6 +252,7 @@ type StyleAttributes struct {
 	WebVTTSize           string
 	WebVTTStyles         []string
 	WebVTTTags           []WebVTTTag
+	WebVTTUnderline      bool
 	WebVTTVertical       string
 	WebVTTViewportAnchor string
 	WebVTTWidth          string
@@ -279,6 +288,56 @@ func (t WebVTTTag) endTag() string {
 	return "</" + t.Name + ">"
 }
 
+func (sa *StyleAttributes) propagateSRTAttributes() {
+	// copy relevant attrs to WebVTT ones
+	if sa.SRTColor != nil {
+		// TODO: handle non-default colors that need custom styles
+		sa.TTMLColor = sa.SRTColor
+	}
+
+	switch sa.SRTPosition {
+	case 7: // top-left
+		sa.WebVTTAlign = "left"
+		sa.WebVTTPosition = "10%"
+	case 8: // top-center
+		sa.WebVTTPosition = "10%"
+	case 9: // top-right
+		sa.WebVTTAlign = "right"
+		sa.WebVTTPosition = "10%"
+	case 4: // middle-left
+		sa.WebVTTAlign = "left"
+		sa.WebVTTPosition = "50%"
+	case 5: // middle-center
+		sa.WebVTTPosition = "50%"
+	case 6: // middle-right
+		sa.WebVTTAlign = "right"
+		sa.WebVTTPosition = "50%"
+	case 1: // bottom-left
+		sa.WebVTTAlign = "left"
+		sa.WebVTTPosition = "90%"
+	case 2: // bottom-center
+		sa.WebVTTPosition = "90%"
+	case 3: // bottom-right
+		sa.WebVTTAlign = "right"
+		sa.WebVTTPosition = "90%"
+	}
+
+	sa.WebVTTBold = sa.SRTBold
+	sa.WebVTTItalics = sa.SRTItalics
+	sa.WebVTTUnderline = sa.SRTUnderline
+
+	sa.WebVTTTags = make([]WebVTTTag, 0)
+	if sa.WebVTTBold {
+		sa.WebVTTTags = append(sa.WebVTTTags, WebVTTTag{Name: "b"})
+	}
+	if sa.WebVTTItalics {
+		sa.WebVTTTags = append(sa.WebVTTTags, WebVTTTag{Name: "i"})
+	}
+	if sa.WebVTTUnderline {
+		sa.WebVTTTags = append(sa.WebVTTTags, WebVTTTag{Name: "u"})
+	}
+}
+
 func (sa *StyleAttributes) propagateSSAAttributes() {}
 
 func (sa *StyleAttributes) propagateSTLAttributes() {
@@ -352,7 +411,15 @@ func (sa *StyleAttributes) propagateTTMLAttributes() {
 	}
 }
 
-func (sa *StyleAttributes) propagateWebVTTAttributes() {}
+func (sa *StyleAttributes) propagateWebVTTAttributes() {
+	// copy relevant attrs to SRT ones
+	if sa.TTMLColor != nil {
+		sa.SRTColor = sa.TTMLColor
+	}
+	sa.SRTBold = sa.WebVTTBold
+	sa.SRTItalics = sa.WebVTTItalics
+	sa.SRTUnderline = sa.WebVTTUnderline
+}
 
 // Metadata represents metadata
 // TODO Merge attributes
@@ -835,3 +902,14 @@ func appendStringToBytesWithNewLine(i []byte, s string) (o []byte) {
 	o = append(o, bytesLineSeparator...)
 	return
 }
+
+func htmlTokenAttribute(t *html.Token, key string) *string {
+
+	for _, attr := range t.Attr {
+		if attr.Key == key {
+			return &attr.Val
+		}
+	}
+
+	return nil
+}

webvtt.go

@@ -31,13 +31,11 @@ const (
 
 // Vars
 var (
-	bytesWebVTTItalicEndTag            = []byte("</i>")
-	bytesWebVTTItalicStartTag          = []byte("<i>")
 	bytesWebVTTTimeBoundariesSeparator = []byte(webvttTimeBoundariesSeparator)
 	webVTTRegexpInlineTimestamp        = regexp.MustCompile(`<((?:\d{2,}:)?\d{2}:\d{2}\.\d{3})>`)
 	webVTTRegexpTag                    = regexp.MustCompile(`(</*\s*([^\.\s]+)(\.[^\s/]*)*\s*([^/]*)\s*/*>)`)
 	webVTTEscaper                      = strings.NewReplacer("&", "&amp;", "<", "&lt;")
-	webVTTUnescaper                    = strings.NewReplacer("&amp;", "&", "&lt;", "<")
+	webVTTUnescaper                    = strings.NewReplacer("&nbsp;", " ", "&amp;", "&", "&lt;", "<")
 )
 
 // parseDurationWebVTT parses a .vtt duration