erm/app/fontinfo/parse.go

package fontinfo

import (
	"bytes"
	"fmt"
	"io"
	"unicode/utf8"

	"golang.org/x/text/encoding/unicode"
)

func read(r io.Reader, length int) ([]byte, error) {
	buf := make([]byte, length)
	if n, err := r.Read(buf); err != nil {
		return nil, err
	} else if n < length {
		return nil, fmt.Errorf("invalid length")
	}
	return buf, nil
}

func u16(buf []byte) uint16 {
	return (uint16(buf[0]) << 8) + uint16(buf[1])
}

func u32(buf []byte) uint32 {
	return (uint32(buf[0]) << 24) + (uint32(buf[1]) << 16) + (uint32(buf[2]) << 8) + uint32(buf[3])
}

type FontMetadata struct {
	FontFamily string
	FontStyle  string
}

type FontParser interface {
	Parse(r io.ReadSeeker) (*FontMetadata, error)
}

// for ttf and otf files
type TruetypeParser struct {
}

func (t *TruetypeParser) Parse(r io.ReadSeeker) (*FontMetadata, error) {
	buf, err := read(r, 12)
	if err != nil {
		return nil, err
	}
	tableCount := u16(buf[4:6])
	for i := 0; i < int(tableCount); i++ {
		if _, err := r.Seek(12+(int64(i)*16), 0); err != nil {
			return nil, err
		}
		table, err := read(r, 16)
		if err != nil {
			return nil, err
		}
		if string(table[0:4]) != "name" {
			continue
		}
		offset := u32(table[8:12])
		return t.readNameTable(r, offset)
	}
	return nil, fmt.Errorf("name table not found")
}

func (t *TruetypeParser) readNameTable(r io.ReadSeeker, offset uint32) (*FontMetadata, error) {
	if _, err := r.Seek(int64(offset), 0); err != nil {
		return nil, fmt.Errorf("invalid font file")
	}
	nameTable, err := read(r, 6)
	if err != nil {
		return nil, err
	}
	nameCount := u16(nameTable[2:4])
	stringOffset := int64(u16(nameTable[4:6])) + int64(offset)
	var done uint8
	var metadata FontMetadata
	nameRecordStart := offset + 6
	for j := 0; j < int(nameCount); j++ {
		recordOffset := nameRecordStart + uint32(12*j)
		if _, err := r.Seek(int64(recordOffset), 0); err != nil {
			return nil, err
		}
		buf, err := read(r, 12)
		if err != nil {
			return nil, err
		}
		language := u16(buf[4:6])
		if language != 0 && language != 1033 { // not english or english us
			continue
		}
		nameID := u16(buf[6:8])
		switch nameID {
		case 1, 2: //family, style
			if _, err := r.Seek(int64(stringOffset)+int64(u16(buf[10:12])), 0); err != nil {
				return nil, err
			}
			raw, err := read(r, int(u16(buf[8:10])))
			if err != nil {
				return nil, err
			}
			if nameID == 1 {
				done |= 1
				metadata.FontFamily = lazyUnicode(raw)
			} else {
				done |= 2
				metadata.FontStyle = lazyUnicode(raw)
			}
			if done == 3 { // bail early if we have what we need
				return &metadata, nil
			}
		}
	}
	return &metadata, nil
}

func lazyUnicode(r []byte) string {
	var back bool
	if bytes.HasSuffix(r, []byte{0}) {
		r = bytes.TrimSuffix(r, []byte{0})
		back = true
	}
	if utf8.Valid(r) && (!bytes.Contains(r, []byte{0})) {
		return string(r)
	}
	if back {
		r = append(r, 0)
	}
	dec := unicode.UTF16(unicode.BigEndian, unicode.UseBOM).NewDecoder()
	ans, err := dec.String(string(r))
	if err != nil {
		return "Invalid_UTF16_Encoding"
	}
	return ans
}