Documentation
¶
Index ¶
Constants ¶
This section is empty.
Variables ¶
var DefaultNormalizer = NewDefaultNormalizer()
DefaultNormalizer is a statically allocated default normalizer for strings.
Functions ¶
func NormalizeForMatching ¶ added in v1.10.0
NormalizeForMatching applies cached full normalization for cross-seed matching:
- Unicode normalization (removes diacritics, decomposes ligatures)
- Lowercase
- Strip apostrophes (including Unicode variants)
- Strip colons, exclamation and question marks
- Convert commas to spaces
- Convert ampersand to "and"
- Convert hyphens to spaces
- Replace decorative anime title symbols via animeTitleSymbolReplacer, e.g. "Classic★Stars" to "classic stars"
- Collapse multiple spaces to single space
Results are cached per input string (5 minute TTL) to avoid repeated expensive transformations.
Examples:
- "Shōgun S01" → "shogun s01"
- "Bob's Burgers" → "bobs burgers"
- "CSI: Miami" → "csi miami"
- "Title, With Comma" → "title with comma"
- "Spider-Man" → "spider man"
- "His & Hers" → "his and hers"
func NormalizeUnicode ¶ added in v1.10.0
NormalizeUnicode removes diacritics and decomposes ligatures with caching. Results are cached per input string (5 minute TTL) to avoid repeated expensive transformations. For the full normalization with additional punctuation handling, use NormalizeForMatching instead. Examples:
- "Shōgun" → "Shogun"
- "Amélie" → "Amelie"
- "naïve" → "naive"
- "Björk" → "Bjork"
- "æ" → "ae"
- "fi" → "fi"
func SanitizeUTF8 ¶ added in v1.24.0
SanitizeUTF8 replaces invalid UTF-8 byte sequences in s with U+FFFD, returning valid UTF-8.
The BitTorrent spec requires all string fields in a torrent file (torrent name, file paths, comment, etc.) to be UTF-8, but torrents with raw legacy-encoded bytes (e.g. "á" as Latin-1 0xe1) exist in the wild and break code that assumes valid UTF-8, for example regexp.Compile rejects invalid-UTF-8 patterns and panics via MustCompile. Invalid bytes are replaced with U+FFFD rather than dropped because the other lossy decoders in the pipeline (encoding/json, which delivers all qBittorrent API strings, and the NFKD transform in NormalizeForMatching) coerce invalid bytes to U+FFFD too, so sanitized strings still compare equal to those forms. This is a no-op on well-formed input, so legitimate values are unaffected.
Types ¶
type Normalizer ¶
type Normalizer[K comparable, V any] struct { // contains filtered or unexported fields }
Normalizer caches transformed results so we do not repeatedly transform the same inputs.
func NewDefaultNormalizer ¶
func NewDefaultNormalizer() *Normalizer[string, string]
NewDefaultNormalizer returns a normalizer using the default TTL and default transform (ToLower + TrimSpace).
func NewNormalizer ¶
func NewNormalizer[K comparable, V any](ttl time.Duration, transform TransformFunc[K, V]) *Normalizer[K, V]
NewNormalizer returns a normalizer with the provided TTL and transform function for cached entries.
func (*Normalizer[K, V]) Clear ¶
func (n *Normalizer[K, V]) Clear(key K)
Clear removes a cached entry.
func (*Normalizer[K, V]) Normalize ¶
func (n *Normalizer[K, V]) Normalize(key K) V
Normalize returns the transformed value.
type TransformFunc ¶
type TransformFunc[K, V any] func(K) V
TransformFunc is a function that transforms K to V.
Source Files