feat(归并文件): 归并文件下沉通用能力
build.yaml / build (push) Successful in 53s

This commit is contained in:
xiaoqidun committed 2026-10-14 10:40:08 +08:00
1 parent d07972346b
commit 70003486b8
37 files changed
+2386 -2509

No files matched your search

+742
View File
@@ -15,13 +15,147 @@
package font
import (
"bytes"
"path"
"slices"
"sort"
"strings"
)
// 字体匹配按精确、部分和模糊匹配排列,零值表示未匹配
const (
MatchNone = iota
MatchExact
MatchPartial
MatchFuzzy
)
// matchRules 保存常用字体的名称、文件名及系统字体匹配规则
var matchRules = normalizeMatchRules([]matchRule{
{
Keys: []string{"小标宋", "方正小标宋", "xiaobiaosong", "fzxiaobiaosong", "fzxbs"},
Names: []string{"小标宋体", "方正小标宋简体", "FZXiaoBiaoSong-B05", "FZXiaoBiaoSong-B05S"},
Files: []string{"FZXBSJW.TTF", "fzxbs*.ttf", "fzxiaobiaosong*.ttf", "xiaobiaosong*.ttf"},
System: "FZXiaoBiaoSong-B05S",
NoSyntheticBold: true,
},
{
Keys: []string{"宋体", "新宋体", "simsun", "nsimsun", "simsunextb", "song", "songti", "stsong", "书宋", "方正书宋"},
Names: []string{"宋体", "新宋体", "SimSun", "NSimSun", "SongTi", "STSong"},
Files: []string{"simsun.ttc", "simsun.ttf", "nsimsun.ttf", "STSONG.TTF"},
System: "SimSun",
},
{
Keys: []string{"黑体", "hei", "simhei", "heiti", "sthei", "方正黑体"},
Names: []string{"黑体", "SimHei", "HeiTi", "STHeiti", "方正黑体简体"},
Files: []string{"simhei.ttf", "STXIHEI.TTF", "FZHT_GB18030.TTF", "fzht*.ttf"},
System: "SimHei",
},
{
Keys: []string{"楷体", "kai", "kaiti", "simkai", "stkaiti", "方正楷体"},
Names: []string{"楷体", "KaiTi", "SimKai", "STKaiti", "方正楷体简体"},
Files: []string{"simkai.ttf", "kaiti.ttf", "STKAITI.TTF", "FZKTJ.TTF", "fzkai*.ttf", "fzkt*.ttf"},
System: "KaiTi",
},
{
Keys: []string{"仿宋", "fang", "fangsong", "simfang", "stfangsong", "方正仿宋"},
Names: []string{"仿宋", "FangSong", "SimFang", "STFangsong", "方正仿宋简体"},
Files: []string{"simfang.ttf", "fangsong.ttf", "STFANGSO.TTF", "FZFSJ.TTF", "fzfs*.ttf"},
System: "FangSong",
},
{
Keys: []string{"微软雅黑", "microsoftyahei", "yahei", "msyh"},
Names: []string{"微软雅黑", "Microsoft YaHei", "YaHei", "MSYH"},
Files: []string{"msyh.ttc", "msyh.ttf"},
System: "Microsoft YaHei",
},
{
Keys: []string{"arialblack"},
Names: []string{"Arial Black", "ArialBlack"},
Files: []string{"ariblk.ttf", "arial black*.ttf", "arialblack*.ttf"},
System: "Arial Black",
},
{
Keys: []string{"arial"},
Names: []string{"Arial"},
Files: []string{"arial.ttf", "arial*.ttf"},
System: "Arial",
},
{
Keys: []string{"helvetica"},
Names: []string{"Helvetica", "Arial"},
Files: []string{"arial.ttf", "arial*.ttf"},
System: "Arial",
},
{
Keys: []string{"couriernew", "courier"},
Names: []string{"Courier New", "CourierNew", "Courier"},
Files: []string{"cour.ttf", "courier new*.ttf", "courier*.ttf"},
System: "Courier New",
},
{
Keys: []string{"segoeui"},
Names: []string{"Segoe UI", "SegoeUI"},
Files: []string{"segoeui.ttf", "segoe*.ttf"},
System: "Segoe UI",
},
{
Keys: []string{"timesnewroman", "times"},
Names: []string{"Times New Roman", "TimesNewRoman"},
Files: []string{"times.ttf", "times new roman*.ttf", "times*.ttf"},
System: "Times New Roman",
},
})
// fontNameReplacer 移除字体名称匹配时忽略的分隔符
var fontNameReplacer = strings.NewReplacer(" ", "", "-", "", "_", "", "(", "", ")", "", "(", "", ")", "")
// matchRule 字体匹配规则
type matchRule struct {
Keys []string
Names []string
Files []string
System string
NoSyntheticBold bool
}
// FileCandidate 字体文件候选
type FileCandidate struct {
Name string
Face int
base string
lowerBase string
filename string
normalized string
names []string
fold string
}
// patternMatcher 字体文件匹配器
type patternMatcher struct {
pattern string
lowerPattern string
stem string
aliases []string
priority int
}
// FileMatch 字体文件匹配结果
type FileMatch struct {
Name string
Face int
Priority int
Rank int
styleRank int
sortName string
}
// fileKey 字体文件与集合索引的去重键
type fileKey struct {
Name string
Face int
}
// NormalizeName 规范化字体名称
// 入参: name 字体名称
// 返回: string 规范化后的字体名称
@@ -62,3 +196,611 @@ func AppendFontName(names []string, seen map[string]bool, name string) []string
seen[key] = true
return append(names, name)
}
// normalizeMatchRules 预处理固定字体规则的匹配名称,保留候选名称及顺序
// 入参: rules 字体匹配规则
// 返回: []matchRule 预处理后的规则
func normalizeMatchRules(rules []matchRule) []matchRule {
for i := range rules {
rule := &rules[i]
keys := append(append(rule.Keys, rule.Names...), rule.System)
rule.Keys = nil
for _, key := range keys {
key = NormalizeName(key)
if key != "" && !slices.Contains(rule.Keys, key) {
rule.Keys = append(rule.Keys, key)
}
}
}
return rules
}
// CandidateNames 获取字体候选名称
// 入参: names 字体名称列表
// 返回: []string 字体候选名称
func CandidateNames(names ...string) []string {
return candidateNamesByLevel(MatchFuzzy, names...)
}
// ExactCandidateNames 获取精确字体候选名称
// 入参: names 字体名称列表
// 返回: []string 精确字体候选名称
func ExactCandidateNames(names ...string) []string {
return candidateNamesByLevel(MatchExact, names...)
}
// qualifiedNames 获取保留名称后缀的字体候选
// 入参: name 字体名称
// 返回: []string 保留后缀的候选名称
func qualifiedNames(name string) []string {
name = NormalizeName(name)
key := NormalizeName(name)
var names []string
seen := make(map[string]bool)
for _, rule := range matchRules {
if ruleMatchLevel(rule, key) != MatchPartial {
continue
}
prefix := ""
for _, alias := range rule.Keys {
if len(alias) > len(prefix) && strings.HasPrefix(name, alias) {
prefix = alias
}
}
suffix := strings.TrimPrefix(name, prefix)
for _, alias := range append([]string{rule.System}, rule.Names...) {
names = AppendFontName(names, seen, NormalizeName(alias)+suffix)
}
}
return names
}
// candidateNamesByLevel 获取指定等级内的字体候选名称
// 入参: maxLevel 最大匹配等级, names 字体名称列表
// 返回: []string 字体候选名称
func candidateNamesByLevel(maxLevel int, names ...string) []string {
var result []string
seen := make(map[string]bool)
var bases []string
for _, name := range names {
name = strings.TrimSpace(name)
if name == "" {
continue
}
result = AppendFontName(result, seen, name)
bases = append(bases, name)
}
levels := matchLevels(bases)
for level := MatchExact; level <= maxLevel; level++ {
for i := range bases {
for j, rule := range matchRules {
if levels[i*len(matchRules)+j] == level {
result = AppendFontName(result, seen, rule.System)
for _, item := range rule.Names {
result = AppendFontName(result, seen, item)
}
}
}
}
}
return result
}
// patternStem 获取字体匹配模式主干
// 入参: pattern 字体文件匹配模式
// 返回: string 字体匹配模式主干
func patternStem(pattern string) string {
stem := strings.TrimSuffix(pattern, "*")
if IsFontFileName(stem) {
stem = strings.TrimSuffix(stem, path.Ext(stem))
}
if index := strings.IndexAny(stem, "*?["); index >= 0 {
stem = stem[:index]
}
return NormalizeName(stem)
}
// FilePatterns 获取字体文件匹配模式
// 入参: names 字体名称列表
// 返回: []string 字体文件匹配模式
func FilePatterns(names ...string) []string {
var result []string
seen := make(map[string]bool)
levels := matchLevels(names)
for level := MatchExact; level <= MatchFuzzy; level++ {
for _, name := range candidateNamesAtLevel(level, names, levels) {
result = appendPattern(result, seen, name+"*")
}
for i := range names {
for j, rule := range matchRules {
if levels[i*len(matchRules)+j] == level {
for _, item := range rule.Files {
result = appendPattern(result, seen, item)
}
}
}
}
}
return result
}
// SystemNames 获取系统字体名称
// 入参: names 字体名称列表
// 返回: []string 系统字体名称
func SystemNames(names ...string) []string {
var result []string
seen := make(map[string]bool)
for _, name := range names {
for _, item := range candidateNamesByLevel(MatchPartial, name) {
result = AppendFontName(result, seen, item)
}
}
return result
}
// candidateNamesAtLevel 获取指定等级的字体候选名称
// 入参: level 匹配等级, names 字体名称列表, levels 名称与规则的匹配等级
// 返回: []string 字体候选名称
func candidateNamesAtLevel(level int, names []string, levels []int) []string {
var result []string
seen := make(map[string]bool)
for i, name := range names {
name = strings.TrimSpace(name)
if name == "" {
continue
}
if level == MatchExact {
result = AppendFontName(result, seen, name)
}
for j, rule := range matchRules {
if levels[i*len(matchRules)+j] == level {
result = AppendFontName(result, seen, rule.System)
for _, item := range rule.Names {
result = AppendFontName(result, seen, item)
}
}
}
}
return result
}
// DefaultSystemNames 获取默认系统字体名称
// 返回: []string 默认系统字体名称
func DefaultSystemNames() []string {
return []string{
"SimSun", "Microsoft YaHei", "SimHei", "KaiTi", "FangSong",
"Noto Sans CJK SC", "Source Han Sans SC", "Arial", "Segoe UI", "Times New Roman",
}
}
// NoSyntheticBold 判断字体是否禁用合成粗体
// 入参: names 字体名称列表
// 返回: bool 是否禁用合成粗体
func NoSyntheticBold(names ...string) bool {
for _, name := range names {
name = NormalizeName(name)
for _, rule := range matchRules {
if !rule.NoSyntheticBold {
continue
}
level := ruleMatchLevel(rule, name)
if level >= MatchExact && level <= MatchPartial {
return true
}
}
}
return false
}
// matchLevels 按名称和规则顺序计算匹配等级
// 入参: names 字体名称列表
// 返回: []int 名称与规则的匹配等级
func matchLevels(names []string) []int {
levels := make([]int, len(names)*len(matchRules))
for i, name := range names {
name = NormalizeName(name)
for j, rule := range matchRules {
levels[i*len(matchRules)+j] = ruleMatchLevel(rule, name)
}
}
return levels
}
// ruleMatchLevel 获取字体规则匹配等级
// 入参: rule 字体匹配规则, name 规范化后的字体名称
// 返回: int 匹配等级
func ruleMatchLevel(rule matchRule, name string) int {
if name == "" {
return MatchNone
}
level := MatchNone
for _, item := range rule.Keys {
if name == item {
return MatchExact
}
if strings.HasPrefix(name, item) {
level = MatchPartial
} else if level == MatchNone && strings.Contains(name, item) {
level = MatchFuzzy
}
}
return level
}
// appendPattern 追加字体文件匹配模式
// 入参: patterns 匹配模式列表, seen 去重映射, pattern 匹配模式
// 返回: []string 匹配模式列表
func appendPattern(patterns []string, seen map[string]bool, pattern string) []string {
pattern = strings.TrimSpace(pattern)
if pattern == "" {
return patterns
}
key := strings.ToLower(pattern)
if seen[key] {
return patterns
}
seen[key] = true
return append(patterns, pattern)
}
// AppendFileNames 建立各字体的内部名称与文件别名匹配候选
// 入参: candidates 字体候选, index 文件候选索引, faces 各字体内部名称
// 返回: []FileCandidate 字体候选列表
func AppendFileNames(candidates []FileCandidate, index int, faces [][]string) []FileCandidate {
file := candidates[index]
replaced := false
for face, names := range faces {
file.Face = face
file.names = make([]string, len(names))
for i, name := range names {
file.names[i] = NormalizeName(name)
}
for _, name := range file.names {
file.normalized = name
if !replaced {
candidates[index] = file
replaced = true
} else {
candidates = append(candidates, file)
}
}
}
return candidates
}
// CollectionIndex 按内部名称与样式选择集合项,未匹配时保留首项
// 入参: data 集合数据, names 字体名称, bold 是否粗体, italic 是否斜体
// 返回: int 零起始字体索引
func CollectionIndex(data []byte, names []string, bold, italic bool) int {
return NameIndex(FileNames(bytes.NewReader(data)), names, bold, italic)
}
// NameIndex 按名称与样式选择字体索引,未匹配时保留首项
// 入参: faces 各字体名称, names 目标名称, bold 是否粗体, italic 是否斜体
// 返回: int 零起始字体索引
func NameIndex(faces [][]string, names []string, bold, italic bool) int {
candidates := AppendFileNames([]FileCandidate{{}}, 0, faces)
if matches := FileMatches(candidates, names, bold, italic); len(matches) > 0 {
return matches[0].Face
}
return 0
}
// FileMatches 匹配字体文件候选
// 入参: candidates 字体候选, names 字体名称, bold 是否粗体, italic 是否斜体
// 返回: []FileMatch 按优先级排列的字体匹配
func FileMatches(candidates []FileCandidate, names []string, bold, italic bool) []FileMatch {
matches := make([]FileMatch, 0, len(candidates))
seen := make(map[fileKey]int, len(candidates))
if len(names) == 0 {
for _, name := range DefaultSystemNames() {
matches = append(matches, FileMatches(candidates, []string{name}, bold, italic)...)
}
sort.SliceStable(matches, func(i, j int) bool {
return matches[i].Rank < matches[j].Rank
})
unique := matches[:0]
for _, match := range matches {
key := fileKey{strings.ToLower(match.Name), match.Face}
if _, ok := seen[key]; !ok {
seen[key] = len(unique)
unique = append(unique, match)
}
}
return unique
}
for i, name := range names {
if name == "" {
continue
}
for j, alias := range append([]string{name}, qualifiedNames(name)...) {
matcher := patternMatcher{stem: NormalizeName(alias), priority: (i-len(names))*2 + min(j, 1)}
if matcher.stem == "" {
continue
}
for _, file := range candidates {
if rank := matcher.rankCandidate(file); rank == MatchExact {
appendFileMatch(&matches, seen, matcher, file, rank, bold, italic)
}
}
}
}
for _, matcher := range newPatternMatchers(FilePatterns(names...)) {
for _, file := range candidates {
rank := matcher.rankCandidate(file)
appendFileMatch(&matches, seen, matcher, file, rank, bold, italic)
}
}
sortFileMatches(matches)
return matches
}
// FileCandidates 筛选字体文件候选
// 入参: names 文件名列表, base 文件名提取函数
// 返回: []FileCandidate 字体文件候选列表
func FileCandidates(names []string, base func(string) string) []FileCandidate {
files := make([]FileCandidate, 0, len(names))
for _, name := range names {
if IsFontFileName(name) {
baseName := base(name)
normalized := NormalizeName(strings.TrimSuffix(baseName, path.Ext(baseName)))
files = append(files, FileCandidate{
Name: name,
base: baseName,
lowerBase: strings.ToLower(baseName),
filename: normalized,
normalized: normalized,
fold: strings.ToLower(name),
})
}
}
return files
}
// newPatternMatcher 创建字体文件匹配器
// 入参: pattern 匹配模式
// 返回: patternMatcher 字体文件匹配器
func newPatternMatcher(pattern string) patternMatcher {
matcher := patternMatcher{
pattern: pattern,
lowerPattern: strings.ToLower(pattern),
stem: patternStem(pattern),
}
if matcher.stem != "" {
for _, alias := range ExactCandidateNames(matcher.stem) {
alias = NormalizeName(alias)
if alias != "" {
matcher.aliases = append(matcher.aliases, alias)
}
}
}
return matcher
}
// newPatternMatchers 创建字体文件匹配器列表
// 入参: patterns 匹配模式列表
// 返回: []patternMatcher 字体文件匹配器列表
func newPatternMatchers(patterns []string) []patternMatcher {
matchers := make([]patternMatcher, 0, len(patterns))
for priority, pattern := range patterns {
matcher := newPatternMatcher(pattern)
matcher.priority = priority
matchers = append(matchers, matcher)
}
return matchers
}
// rankCandidate 获取字体文件候选匹配等级
// 入参: file 字体文件候选
// 返回: int 匹配等级
func (m patternMatcher) rankCandidate(file FileCandidate) int {
exactPath, _ := path.Match(m.pattern, file.base)
lowerPath, _ := path.Match(m.lowerPattern, file.lowerBase)
if m.stem == "" {
if exactPath || lowerPath {
return MatchFuzzy
}
return MatchNone
}
name := m.candidateName(file)
if name == m.stem {
return MatchExact
}
for _, alias := range m.aliases {
if name == alias {
return MatchExact
}
}
if fileKnownStyleSuffix(m.styleSuffixNormalized(name)) {
return MatchExact
}
if strings.HasPrefix(name, m.stem) {
return MatchPartial
}
for _, alias := range m.aliases {
if strings.HasPrefix(name, alias) {
return MatchPartial
}
}
if exactPath || lowerPath || strings.Contains(name, m.stem) {
return MatchFuzzy
}
for _, alias := range m.aliases {
if strings.Contains(name, alias) {
return MatchFuzzy
}
}
return MatchNone
}
// candidateName 保留同族内部名称的区分,只有无关联时使用调用方文件别名
// 入参: file 字体文件候选
// 返回: string 用于统一匹配名称与样式的规范名称
func (m patternMatcher) candidateName(file FileCandidate) string {
if len(file.names) == 0 || file.base == "" || m.stem == "" {
return file.normalized
}
for _, name := range file.names {
if strings.HasPrefix(name, m.stem) {
return file.normalized
}
for _, alias := range m.aliases {
if strings.HasPrefix(name, alias) {
return file.normalized
}
}
}
return file.filename
}
// styleSuffixNormalized 获取规范字体名称的样式后缀
// 入参: name 规范字体名称
// 返回: string 样式后缀
func (m patternMatcher) styleSuffixNormalized(name string) string {
if m.stem == "" || name == "" {
return ""
}
if strings.HasPrefix(name, m.stem) {
return strings.TrimPrefix(name, m.stem)
}
for _, alias := range m.aliases {
if strings.HasPrefix(name, alias) {
return strings.TrimPrefix(name, alias)
}
}
return ""
}
// appendFileMatch 追加字体文件匹配结果
// 入参: matches 匹配结果, seen 已匹配文件, matcher 匹配器, file 字体文件候选, rank 匹配等级, bold 是否粗体, italic 是否斜体
func appendFileMatch(matches *[]FileMatch, seen map[fileKey]int, matcher patternMatcher, file FileCandidate, rank int, bold, italic bool) {
if rank == MatchNone {
return
}
suffix := matcher.styleSuffixNormalized(matcher.candidateName(file))
if rank == MatchFuzzy {
suffix = matcher.styleSuffixNormalized(NormalizeName(file.base))
}
next := FileMatch{
Name: file.Name,
Face: file.Face,
Priority: matcher.priority,
Rank: rank,
styleRank: fileStyleRank(suffix, bold, italic),
sortName: file.lowerBase,
}
key := fileKey{file.fold, file.Face}
if index, ok := seen[key]; ok {
if fileMatchLess(next, (*matches)[index]) {
(*matches)[index] = next
}
return
}
seen[key] = len(*matches)
*matches = append(*matches, next)
}
// sortFileMatches 排序字体文件匹配结果
// 入参: matches 匹配结果
func sortFileMatches(matches []FileMatch) {
sort.SliceStable(matches, func(i, j int) bool {
return fileMatchLess(matches[i], matches[j])
})
}
// fileMatchLess 判断字体文件匹配结果优先级
// 入参: left 左侧匹配结果, right 右侧匹配结果
// 返回: bool 左侧是否优先
func fileMatchLess(left, right FileMatch) bool {
if left.Priority != right.Priority {
return left.Priority < right.Priority
}
if left.Rank != right.Rank {
return left.Rank < right.Rank
}
if left.styleRank != right.styleRank {
return left.styleRank < right.styleRank
}
return left.sortName < right.sortName
}
// FileMatchNames 获取字体文件匹配名称
// 入参: matches 匹配结果
// 返回: []string 字体文件列表
func FileMatchNames(matches []FileMatch) []string {
names := make([]string, 0, len(matches))
seen := make(map[string]bool, len(matches))
for _, match := range matches {
key := strings.ToLower(match.Name)
if !seen[key] {
seen[key] = true
names = append(names, match.Name)
}
}
return names
}
// fileStyleRank 获取字体文件样式匹配等级
// 入参: suffix 样式后缀, bold 是否粗体, italic 是否斜体
// 返回: int 样式匹配等级
func fileStyleRank(suffix string, bold, italic bool) int {
fileBold, fileItalic := fileStyleFromSuffix(suffix)
rank := 0
if fileBold != bold {
rank += 2
}
if fileItalic != italic {
rank += 2
}
if !bold && !italic && !fileBold && !fileItalic && suffix != "" && !fileRegularSuffix(suffix) {
rank++
}
return rank
}
// fileStyleFromSuffix 获取样式后缀对应的字体样式
// 入参: suffix 样式后缀
// 返回: bool 是否粗体, bool 是否斜体
func fileStyleFromSuffix(suffix string) (bool, bool) {
bold := suffix == "b" ||
suffix == "bd" ||
suffix == "bold" ||
suffix == "bi" ||
suffix == "bolditalic" ||
suffix == "boldoblique"
italic := suffix == "i" ||
suffix == "it" ||
suffix == "italic" ||
suffix == "oblique" ||
suffix == "bi" ||
suffix == "bolditalic" ||
suffix == "boldoblique"
return bold, italic
}
// fileRegularSuffix 判断是否为常规样式后缀
// 入参: suffix 样式后缀
// 返回: bool 是否为常规样式后缀
func fileRegularSuffix(suffix string) bool {
switch suffix {
case "r", "regular", "normal", "常规":
return true
default:
return false
}
}
// fileKnownStyleSuffix 判断是否为已知样式后缀
// 入参: suffix 样式后缀
// 返回: bool 是否为已知样式后缀
func fileKnownStyleSuffix(suffix string) bool {
switch suffix {
case "r", "regular", "normal", "常规",
"b", "bd", "bold",
"i", "it", "italic", "oblique",
"bi", "bolditalic", "boldoblique":
return true
default:
return false
}
}