Added teletext

subtitles.go

@@ -32,31 +32,30 @@ var Now = func() time.Time {
 
 // Options represents open or write options
 type Options struct {
-	Page int
-	PID  int
-	Src  string
+	Filename string
+	Teletext TeletextOptions
 }
 
-// Open opens a subtitle file based on options
+// Open opens a subtitle reader based on options
 func Open(o Options) (s *Subtitles, err error) {
 	// Open the file
 	var f *os.File
-	if f, err = os.Open(o.Src); err != nil {
-		err = errors.Wrapf(err, "astisub: opening %s failed", o.Src)
+	if f, err = os.Open(o.Filename); err != nil {
+		err = errors.Wrapf(err, "astisub: opening %s failed", o.Filename)
 		return
 	}
 	defer f.Close()
 
 	// Parse the content
-	switch filepath.Ext(o.Src) {
+	switch filepath.Ext(o.Filename) {
 	case ".srt":
 		s, err = ReadFromSRT(f)
 	case ".ssa", ".ass":
 		s, err = ReadFromSSA(f)
 	case ".stl":
 		s, err = ReadFromSTL(f)
 	case ".ts":
-		s, err = ReadFromTeletext(f, o.PID, o.Page)
+		s, err = ReadFromTeletext(f, o.Teletext)
 	case ".ttml":
 		s, err = ReadFromTTML(f)
 	case ".vtt":
@@ -68,8 +67,8 @@ func Open(o Options) (s *Subtitles, err error) {
 }
 
 // OpenFile opens a file regardless of other options
-func OpenFile(src string) (*Subtitles, error) {
-	return Open(Options{Src: src})
+func OpenFile(filename string) (*Subtitles, error) {
+	return Open(Options{Filename: filename})
 }
 
 // Subtitles represents an ordered list of items with formatting

teletext.go

@@ -1,8 +1,303 @@
 package astisub
 
-import "io"
+import (
+	"context"
+	"fmt"
+	"io"
+
+	"github.com/asticode/go-astits"
+	"github.com/pkg/errors"
+)
+
+// Errors
+var (
+	ErrNoValidTeletextPID = errors.New("astisub: no valid teletext PID")
+)
+
+// Teletext PES data types
+const (
+	teletextPESDataTypeEBU     = "EBU"
+	teletextPESDataTypeUnknown = "unknown"
+)
+
+// Teletext PES data unit types
+const (
+	teletextPESDataUnitTypeEBUNonSubtitleData = 0x2
+	teletextPESDataUnitTypeEBUSubtitleData    = 0x3
+	teletextPESDataUnitTypeStuffing           = 0xff
+)
+
+// TeletextOptions represents teletext options
+type TeletextOptions struct {
+	Page int
+	PID  int
+}
 
 // ReadFromTeletext parses a teletext content
-func ReadFromTeletext(r io.ReadSeeker, pid, page int) (o *Subtitles, err error) {
+// http://www.etsi.org/deliver/etsi_en/300400_300499/300472/01.03.01_60/en_300472v010301p.pdf
+// http://www.etsi.org/deliver/etsi_i_ets/300700_300799/300706/01_60/ets_300706e01p.pdf
+func ReadFromTeletext(r io.Reader, o TeletextOptions) (s *Subtitles, err error) {
+	// Init demuxer
+	var dmx = astits.New(context.Background(), r)
+
+	// Get the teletext PID
+	var pid uint16
+	if pid, err = teletextPID(dmx, o); err != nil {
+		if err != ErrNoValidTeletextPID {
+			err = errors.Wrap(err, "astisub: getting teletext PID failed")
+		}
+		return
+	}
+
+	// Loop in data
+	var d *astits.Data
+	for {
+		// Fetch next data
+		if d, err = dmx.NextData(); err != nil {
+			if err == astits.ErrNoMorePackets {
+				err = nil
+				break
+			}
+			err = errors.Wrap(err, "astisub: fetching next data failed")
+			return
+		}
+
+		// This data is not of interest to us
+		if d.PID != pid || d.PES == nil || d.PES.Header.StreamID != astits.StreamIDPrivateStream1 {
+			continue
+		}
+
+		// Parse PES data
+		var td *teletextPESData
+		if td, err = parseTeletextPESData(d.PES.Data); err != nil {
+			err = errors.Wrap(err, "astisub: parsing teletext PES data failed")
+			return
+		}
+		_ = td
+	}
+	return
+}
+
+// teletextPID returns the teletext PID.
+// If the PID teletext option is not indicated, it will walk through the ts data until it reaches a PMT packet to
+// detect the first valid teletext PID
+func teletextPID(dmx *astits.Demuxer, o TeletextOptions) (pid uint16, err error) {
+	// PID is in the options
+	if o.PID > 0 {
+		pid = uint16(o.PID)
+		return
+	}
+
+	// Loop in data
+	var d *astits.Data
+	for {
+		// Fetch next data
+		if d, err = dmx.NextData(); err != nil {
+			if err == astits.ErrNoMorePackets {
+				err = ErrNoValidTeletextPID
+				return
+			}
+			err = errors.Wrap(err, "astisub: fetching next data failed")
+			return
+		}
+
+		// PMT data
+		if d.PMT != nil {
+			// Retrieve valid teletext PIDs
+			var pids []uint16
+			for _, s := range d.PMT.ElementaryStreams {
+				for _, dsc := range s.ElementaryStreamDescriptors {
+					if dsc.Tag == astits.DescriptorTagTeletext || dsc.Tag == astits.DescriptorTagVBITeletext {
+						pids = append(pids, s.ElementaryPID)
+					}
+				}
+			}
+
+			// No valid teletext PIDs
+			if len(pids) == 0 {
+				err = ErrNoValidTeletextPID
+				return
+			}
+
+			// Set pid
+			pid = pids[0]
+
+			// Rewind
+			if _, err = dmx.Rewind(); err != nil {
+				err = errors.Wrap(err, "astisub: rewinding failed")
+				return
+			}
+			return
+		}
+	}
+	return
+}
+
+// teletextPESData represents a teletext PES data
+type teletextPESData struct {
+	dataIdentifier uint8
+	units          []*teletextPESDataUnit
+}
+
+// teletextPESDataUnit represents a teletext PES data unit
+type teletextPESDataUnit struct {
+	data            []byte
+	designationCode uint8
+	fieldParity     bool
+	framingCode     uint8
+	id              uint8
+	length          uint8
+	lineOffset      uint8
+	magazineNumber  uint8
+	packetNumber    uint8
+}
+
+// LineOffsetNumber returns the teletext data unit line offset number
+func (u teletextPESDataUnit) LineOffsetNumber() int {
+	if u.lineOffset < 0x7 || u.lineOffset > 0x16 {
+		return 0
+	}
+	var offset int
+	if !u.fieldParity {
+		offset = 313
+	}
+	return int(u.lineOffset) + offset
+}
+
+// parseTeletextPESData parses a Teletext PES data
+func parseTeletextPESData(i []byte) (d *teletextPESData, err error) {
+	// Init
+	d = &teletextPESData{}
+	var offset int
+
+	// Data identifier
+	d.dataIdentifier = uint8(i[offset])
+	offset += 1
+
+	// Loop until end of data
+	for offset < len(i) {
+		// Parse data unit
+		parseTeletextPESDataUnit(i, &offset, d)
+	}
+	return
+}
+
+// parseTeletextPESData parses a Teletext PES data unit
+func parseTeletextPESDataUnit(i []byte, offset *int, d *teletextPESData) {
+	// Init
+	var u = &teletextPESDataUnit{}
+
+	// ID
+	u.id = uint8(i[*offset])
+	*offset += 1
+
+	// Length
+	u.length = uint8(i[*offset])
+	*offset += 1
+
+	// Make sure we seek at the end of the data unit once everything is done
+	var offsetEnd = *offset + int(u.length)
+	defer func(offset *int) {
+		*offset = offsetEnd
+	}(offset)
+
+	// Unprocessed data unit ids
+	// TODO Should we process other ids?
+	if u.id != teletextPESDataUnitTypeEBUSubtitleData {
+		return
+	}
+
+	// Field parity
+	u.fieldParity = i[*offset]&0x20 > 0
+
+	// Line offset
+	u.lineOffset = uint8(i[*offset] & 0x1f)
+	*offset += 1
+
+	// Framing code
+	u.framingCode = uint8(i[*offset])
+	*offset += 1
+
+	// Framing code must be 11100100
+	if u.framingCode != 0xe4 {
+		return
+	}
+
+	// Magazine number and packet number
+	var h, h1, h2 uint8
+	var errHamming error
+	if h1, errHamming = hamming84Decode(i[*offset]); errHamming != nil {
+		return
+	}
+	if h2, errHamming = hamming84Decode(i[*offset+1]); errHamming != nil {
+		return
+	}
+	h = h2<<4 | h1
+	u.magazineNumber = h & 0x7
+	if u.magazineNumber == 0 {
+		u.magazineNumber = 8
+	}
+	u.packetNumber = h >> 3
+	*offset += 2
+
+	// Designation code
+	if u.packetNumber > 25 {
+		if u.designationCode, errHamming = hamming84Decode(i[*offset]); errHamming != nil {
+			return
+		}
+		*offset += 1
+	}
+
+	// Data
+	u.data = i[*offset:offsetEnd]
+
+	// Append data unit
+	d.units = append(d.units, u)
+	return
+}
+
+// teletextPESDataType returns the teletext PES data type based on the data identifier
+func teletextPESDataType(dataIdentifier uint8) string {
+	switch {
+	case dataIdentifier >= 0x10 && dataIdentifier <= 0x1f:
+		return teletextPESDataTypeEBU
+	}
+	return teletextPESDataTypeUnknown
+}
+
+// hamming84Decode decodes a Hamming 8/4
+func hamming84Decode(i byte) (o uint8, err error) {
+	p1, d1, p2, d2, p3, d3, p4, d4 := i>>7&0x1, i>>6&0x1, i>>5&0x1, i>>4&0x1, i>>3&0x1, i>>2&0x1, i>>1&0x1, i&0x1
+	testA := p1^d1^d3^d4 > 0
+	testB := d1^p2^d2^d4 > 0
+	testC := d1^d2^p3^d3 > 0
+	testD := p1^d1^p2^d2^p3^d3^p4^d4 > 0
+	if testA && testB && testC {
+		// p4 may be incorrect
+	} else if testD && (!testA || !testB || !testC) {
+		err = fmt.Errorf("hamming 8/4 decode of %.8b failed", i)
+		return
+	} else {
+		if !testA && testB && testC {
+			// p1 is incorrect
+		} else if testA && !testB && testC {
+			// p2 is incorrect
+		} else if testA && testB && !testC {
+			// p3 is incorrect
+		} else if !testA && !testB && testC {
+			// d4 is incorrect
+			d4 ^= 1
+		} else if testA && !testB && !testC {
+			// d2 is incorrect
+			d2 ^= 1
+		} else if !testA && testB && !testC {
+			// d3 is incorrect
+			d3 ^= 1
+		} else {
+			// d1 is incorrect
+			d1 ^= 1
+		}
+	}
+	o = uint8(d4<<3 | d3<<2 | d2<<1 | d1)
 	return
 }