feat(文件拆分): 拆分字体图像基础能力

2026-10-30 15:31:49 +08:00 · 2026-10-14 00:12:00 +08:00
1 parent ee3fd02a0a
commit 64f9ec4241
@@ -0,0 +1,239 @@
+// Copyright 2025-2026 肖其顿 (XIAO QI DUN)
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package font
+
+import (
+	"bytes"
+	"encoding/binary"
+	"fmt"
+	"math"
+)
+
+// cidSubsetFont 保存CID字体裁剪所需的字典、映射和执行资源
+type cidSubsetFont struct {
+	prefix  []byte
+	strings [][]byte
+	top     cffDict
+	dicts   []cffDict
+	private []cffDict
+	fonts   []*type2Font
+	selects []int
+	charset []byte
+}
+
+// readCIDSubsetFont 读取CID字体的原始字典及逐字形归属，不合并字体字典
+// 入参: data CFF数据
+// 返回: *cidSubsetFont 裁剪资源, error 结构错误或不支持的字典
+func readCIDSubsetFont(data []byte) (*cidSubsetFont, error) {
+	if !isBareCFFData(data) {
+		return nil, fmt.Errorf("invalid CFF header")
+	}
+	names, end, err := readType2Index(data, int(data[2]))
+	if err != nil || len(names) != 1 {
+		return nil, fmt.Errorf("invalid CFF name index")
+	}
+	tops, end, err := readType2Index(data, end)
+	if err != nil || len(tops) != 1 {
+		return nil, fmt.Errorf("invalid CFF top index")
+	}
+	strings, end, err := readType2Index(data, end)
+	if err != nil {
+		return nil, err
+	}
+	globals, _, err := readType2Index(data, end)
+	if err != nil {
+		return nil, err
+	}
+	top, err := readCFFDict(tops[0])
+	if err != nil {
+		return nil, err
+	}
+	ros := top[1230]
+	if len(ros) != 3 {
+		return nil, fmt.Errorf("invalid CFF ROS")
+	}
+	for index, value := range ros {
+		limit := float64(math.MaxInt32)
+		if index < 2 {
+			limit = float64(min(64999, 390+len(strings)))
+		}
+		if !finite(value) || value < 0 || value > limit || value != math.Trunc(value) {
+			return nil, fmt.Errorf("invalid CFF ROS operand")
+		}
+	}
+	for op := range top {
+		switch op {
+		case 0, 1, 2, 3, 4, 5, 13, 14, 15, 17, 1200, 1201, 1202, 1203, 1204, 1205, 1206, 1207, 1208, 1230, 1231, 1232, 1233, 1234, 1236, 1237:
+		default:
+			return nil, fmt.Errorf("unsupported CID CFF top operator %d", op)
+		}
+	}
+	if value, ok := top[1206]; ok && (len(value) != 1 || value[0] != 2) {
+		return nil, fmt.Errorf("unsupported CFF charstring type")
+	}
+	if _, err := readCFFMatrix(top, [6]float64{.001, 0, 0, .001, 0, 0}); err != nil {
+		return nil, err
+	}
+	charOffset, err := cffDictOffset(data, top, 17)
+	if err != nil {
+		return nil, err
+	}
+	chars, _, err := readType2Index(data, charOffset)
+	if err != nil || len(chars) == 0 {
+		return nil, fmt.Errorf("invalid CFF charstrings")
+	}
+	charsetOffset, err := cffDictOffset(data, top, 15)
+	if err != nil || charsetOffset <= 2 {
+		return nil, fmt.Errorf("invalid CID CFF charset offset")
+	}
+	cids, err := parseCFFCharset(data, charsetOffset, len(chars))
+	if err != nil {
+		return nil, err
+	}
+	fdOffset, err := cffDictOffset(data, top, 1236)
+	if err != nil {
+		return nil, err
+	}
+	fds, _, err := readType2Index(data, fdOffset)
+	if err != nil || len(fds) == 0 || len(fds) > 256 {
+		return nil, fmt.Errorf("invalid CFF font dictionary count")
+	}
+	selectOffset, err := cffDictOffset(data, top, 1237)
+	if err != nil {
+		return nil, err
+	}
+	selects, err := readType2FDSelect(data, selectOffset, len(chars), len(fds))
+	if err != nil {
+		return nil, err
+	}
+	if data[selectOffset] == 3 && binary.BigEndian.Uint16(data[selectOffset+3:]) != 0 {
+		return nil, fmt.Errorf("CFF FDSelect must include glyph zero")
+	}
+	font := &cidSubsetFont{
+		prefix:  append(bytes.Clone(data[:int(data[2])]), encodeCFFIndex(names)...),
+		strings: strings, top: top, selects: selects, charset: encodeCFFSubsetCharset(cids),
+		dicts: make([]cffDict, len(fds)), private: make([]cffDict, len(fds)), fonts: make([]*type2Font, len(fds)),
+	}
+	for index, fd := range fds {
+		dict, err := readCFFDict(fd)
+		if err != nil {
+			return nil, err
+		}
+		for op := range dict {
+			switch op {
+			case 18, 1207, 1238:
+			default:
+				return nil, fmt.Errorf("unsupported CFF font dictionary operator %d", op)
+			}
+		}
+		if _, err := readCFFMatrix(dict, [6]float64{1, 0, 0, 1, 0, 0}); err != nil {
+			return nil, err
+		}
+		if len(dict[18]) != 2 {
+			return nil, fmt.Errorf("missing CFF private dictionary")
+		}
+		resource := &type2Font{data: data, chars: chars, globals: globals, seed: 1, cid: true}
+		private, err := readType2Private(data, dict[18], resource)
+		if err != nil {
+			return nil, err
+		}
+		if _, err := encodeCFFSubsetPrivate(private); err != nil {
+			return nil, err
+		}
+		font.dicts[index], font.private[index], font.fonts[index] = dict, private, resource
+	}
+	return font, nil
+}
+
+// write 重建CID字体偏移，保留CID映射、字体矩阵及各字典的私有参数
+// 入参: chars 裁剪后的字形程序
+// 返回: []byte CFF数据, error 偏移或字典错误
+func (f *cidSubsetFont) write(chars [][]byte) ([]byte, error) {
+	suffix := append(encodeCFFIndex(f.strings), encodeCFFIndex(f.fonts[0].globals)...)
+	selects := encodeCFFSubsetFDSelect(f.selects)
+	programs := encodeCFFIndex(chars)
+	private := make([][]byte, len(f.fonts))
+	locals := make([][]byte, len(f.fonts))
+	fds := make([][]byte, len(f.fonts))
+	for index, font := range f.fonts {
+		var err error
+		private[index], err = encodeCFFSubsetPrivate(f.private[index])
+		if err != nil {
+			return nil, err
+		}
+		if _, ok := f.private[index][19]; ok {
+			locals[index] = encodeCFFIndex(font.locals)
+		}
+		fds[index] = encodeCFFDict(f.dicts[index])
+	}
+	for range 16 {
+		top := encodeCFFIndex([][]byte{encodeCFFDict(f.top)})
+		array := encodeCFFIndex(fds)
+		offset := len(f.prefix) + len(top) + len(suffix)
+		f.top[15] = []float64{float64(offset)}
+		offset += len(f.charset)
+		f.top[1237] = []float64{float64(offset)}
+		offset += len(selects)
+		f.top[17] = []float64{float64(offset)}
+		offset += len(programs)
+		f.top[1236] = []float64{float64(offset)}
+		offset += len(array)
+		for index := range f.fonts {
+			f.dicts[index][18] = []float64{float64(len(private[index])), float64(offset)}
+			fds[index] = encodeCFFDict(f.dicts[index])
+			offset += len(private[index]) + len(locals[index])
+		}
+		updatedTop := encodeCFFIndex([][]byte{encodeCFFDict(f.top)})
+		updatedArray := encodeCFFIndex(fds)
+		if len(updatedTop) != len(top) || len(updatedArray) != len(array) {
+			continue
+		}
+		result := make([]byte, 0, offset)
+		for _, part := range [][]byte{f.prefix, updatedTop, suffix, f.charset, selects, programs, updatedArray} {
+			result = append(result, part...)
+		}
+		for index := range f.fonts {
+			result = append(result, private[index]...)
+			result = append(result, locals[index]...)
+		}
+		return result, nil
+	}
+	return nil, fmt.Errorf("CID CFF offsets did not converge")
+}
+
+// encodeCFFSubsetFDSelect 选择较短的逐字形或分段格式保存字体字典归属
+// 入参: selects 字形所属字典
+// 返回: []byte FDSelect数据
+func encodeCFFSubsetFDSelect(selects []int) []byte {
+	ranges := []byte{3, 0, 0}
+	count := 0
+	for gid, fd := range selects {
+		if gid == 0 || fd != selects[gid-1] {
+			ranges = binary.BigEndian.AppendUint16(ranges, uint16(gid))
+			ranges = append(ranges, byte(fd))
+			count++
+		}
+	}
+	binary.BigEndian.PutUint16(ranges[1:], uint16(count))
+	ranges = binary.BigEndian.AppendUint16(ranges, uint16(len(selects)))
+	if len(ranges) < len(selects)+1 {
+		return ranges
+	}
+	result := make([]byte, len(selects)+1)
+	for gid, fd := range selects {
+		result[gid+1] = byte(fd)
+	}
+	return result
+}