claude: fold in old sow-tools

internal/music/naming.go

@@ -0,0 +1,197 @@
+package music
+
+import (
+	"crypto/sha1"
+	"fmt"
+	"path/filepath"
+	"strings"
+)
+
+func GenerateStem(fileName, prefix string, used map[string]struct{}) (string, error) {
+	return GenerateStemWithMax(fileName, prefix, MaxStemLen, used)
+}
+
+func GenerateStemWithMax(fileName, prefix string, maxStemLen int, used map[string]struct{}) (string, error) {
+	if maxStemLen <= 0 {
+		maxStemLen = MaxStemLen
+	}
+	maxCore := maxStemLen - len(prefix)
+	if maxCore < 3 {
+		return "", fmt.Errorf("prefix too long: %q", prefix)
+	}
+	base := strings.TrimSuffix(fileName, filepath.Ext(fileName))
+	words := SlugWords(base)
+	core := MakeMusicCore(words, maxCore)
+	if core == "" {
+		sum := sha1.Sum([]byte(fileName))
+		core = fmt.Sprintf("%x", sum[:])[:maxCore]
+	}
+	stem := strings.TrimRight((prefix + core)[:Min(maxStemLen, len(prefix)+len(core))], "_")
+	if stem == "" {
+		return "", fmt.Errorf("could not derive music stem for %s", fileName)
+	}
+	return UniqueNameWithMax(stem, maxStemLen, used), nil
+}
+
+func SlugWords(value string) []string {
+	value = BracketRE.ReplaceAllString(value, " ")
+	value = strings.ReplaceAll(value, "&", " and ")
+	value = strings.ReplaceAll(value, "'", "")
+	value = strings.ReplaceAll(value, "\u2019", "")
+	value = strings.ToLower(value)
+	var ascii strings.Builder
+	for _, r := range value {
+		switch {
+		case r >= 'a' && r <= 'z':
+			ascii.WriteRune(r)
+		case r >= '0' && r <= '9':
+			ascii.WriteRune(r)
+		default:
+			ascii.WriteRune(' ')
+		}
+	}
+	rawWords := SplitRE.Split(ascii.String(), -1)
+	words := make([]string, 0, len(rawWords))
+	for _, word := range rawWords {
+		if word == "" {
+			continue
+		}
+		if _, drop := DropWords[word]; drop {
+			continue
+		}
+		if YearRE.MatchString(word) {
+			continue
+		}
+		words = append(words, word)
+	}
+	return words
+}
+
+func MakeMusicCore(words []string, maxCore int) string {
+	core := ""
+	for _, word := range words {
+		candidate := word
+		if core != "" {
+			candidate = core + "_" + word
+		}
+		if len(candidate) <= maxCore {
+			core = candidate
+			continue
+		}
+		break
+	}
+	if core != "" {
+		return core
+	}
+	abbrev := make([]string, 0, Min(4, len(words)))
+	for _, word := range words {
+		if len(abbrev) == 4 {
+			break
+		}
+		abbrev = append(abbrev, AbbreviateWord(word, 5))
+	}
+	for _, word := range abbrev {
+		candidate := word
+		if core != "" {
+			candidate = core + "_" + word
+		}
+		if len(candidate) <= maxCore {
+			core = candidate
+			continue
+		}
+		break
+	}
+	if core != "" {
+		return core
+	}
+	if len(words) > 0 {
+		return strings.Trim(words[0][:Min(maxCore, len(words[0]))], "_")
+	}
+	return ""
+}
+
+func AbbreviateWord(word string, maxLen int) string {
+	if len(word) <= maxLen {
+		return word
+	}
+	consonants := VowelRE.ReplaceAllString(word, "")
+	candidate := word[:1]
+	if len(consonants) > 1 {
+		candidate += consonants[1:]
+	}
+	if len(candidate) >= 3 {
+		return candidate[:Min(maxLen, len(candidate))]
+	}
+	return word[:Min(maxLen, len(word))]
+}
+
+func UniqueName(stem string, used map[string]struct{}) string {
+	return UniqueNameWithMax(stem, MaxStemLen, used)
+}
+
+func UniqueNameWithMax(stem string, maxStemLen int, used map[string]struct{}) string {
+	if maxStemLen <= 0 {
+		maxStemLen = MaxStemLen
+	}
+	if _, exists := used[stem]; !exists {
+		used[stem] = struct{}{}
+		return stem
+	}
+	for index := 1; ; index++ {
+		suffix := fmt.Sprintf("_%d", index)
+		prefixLen := Min(len(stem), Max(0, maxStemLen-len(suffix)))
+		candidate := strings.TrimRight(stem[:prefixLen], "_") + suffix
+		if _, exists := used[candidate]; exists {
+			continue
+		}
+		used[candidate] = struct{}{}
+		return candidate
+	}
+}
+
+func ReserveName(stem string, used map[string]struct{}) error {
+	stem = strings.ToLower(strings.TrimSpace(stem))
+	if stem == "" {
+		return fmt.Errorf("output filename is empty")
+	}
+	if _, exists := used[stem]; exists {
+		return fmt.Errorf("output filename %s collides with an existing asset", stem)
+	}
+	used[stem] = struct{}{}
+	return nil
+}
+
+func ValidateManualOutputFile(name string) error {
+	return ValidateManualOutputFileWithRules(name, ".bmu", MaxStemLen)
+}
+
+func ValidateManualOutputFileWithRules(name, extension string, maxStemLen int) error {
+	name = strings.ToLower(strings.TrimSpace(name))
+	extension = strings.ToLower(strings.TrimSpace(extension))
+	if extension == "" {
+		extension = ".bmu"
+	}
+	if maxStemLen <= 0 {
+		maxStemLen = MaxStemLen
+	}
+	if !strings.HasSuffix(name, extension) {
+		return fmt.Errorf("output file must end with %s", extension)
+	}
+	stem := strings.TrimSuffix(name, extension)
+	if stem == "" {
+		return fmt.Errorf("output file stem is empty")
+	}
+	if len(stem) > maxStemLen {
+		return fmt.Errorf("output file stem %q exceeds %d characters", stem, maxStemLen)
+	}
+	for _, r := range stem {
+		switch {
+		case r >= 'a' && r <= 'z':
+		case r >= '0' && r <= '9':
+		case r == '_':
+		default:
+			return fmt.Errorf("output file stem %q contains unsupported character %q", stem, string(r))
+		}
+	}
+	return nil
+}