fix: address code review findings
This commit is contained in:
@@ -18,10 +18,11 @@ var (
|
||||
parenRe = regexp.MustCompile(`\([^)]*\)`)
|
||||
yearRe = regexp.MustCompile(`\b(1[0-9]{3}|2[0-9]{3})\b`)
|
||||
spaceRe = regexp.MustCompile(`\s+`)
|
||||
// wordRe matches any alphabetic character. Used to decide whether a title
|
||||
// that collapses entirely to a year actually had other words worth keeping
|
||||
// (e.g. "1989 (Deluxe)") versus being a bare year title (e.g. "1989").
|
||||
wordRe = regexp.MustCompile(`[a-z]`)
|
||||
// bareYearRe matches a title that is *only* a single year (with optional
|
||||
// surrounding whitespace), e.g. "1989" or "2112". Used to decide whether a
|
||||
// title that collapses entirely to a year should keep it (so it matches
|
||||
// itself) or be treated as a distinct reissue that must collapse to empty.
|
||||
bareYearRe = regexp.MustCompile(`^\s*(1[0-9]{3}|2[0-9]{3})\s*$`)
|
||||
)
|
||||
|
||||
// NormalizeString normalizes a string for fuzzy matching by:
|
||||
@@ -46,20 +47,22 @@ func NormalizeString(s string) string {
|
||||
s = parenRe.ReplaceAllString(s, "")
|
||||
|
||||
// Remove years (4-digit numbers between 1000-2999). If stripping the year
|
||||
// would empty the entire string, we must decide what to keep:
|
||||
// empties the entire string, decide what to keep:
|
||||
// - A bare year title (e.g. "1989", "2112") has no other words, so keep
|
||||
// the year so it can still match itself (the user owns that album).
|
||||
// - A title that had OTHER words alongside the year (e.g. "1989 (Deluxe)")
|
||||
// collapses to empty on purpose: it is a distinct release group that
|
||||
// must NOT be considered already-present just because the user owns the
|
||||
// standard "1989". Collapsing to empty makes it score 0.0 against a
|
||||
// plain "1989", correctly reporting the reissue as missing.
|
||||
// plain "1989", correctly reporting the reissue as missing. The check
|
||||
// is against the original (brackets intact) so a title like "1989
|
||||
// [2020]" is correctly NOT treated as a bare year.
|
||||
stripped := yearRe.ReplaceAllString(s, "")
|
||||
if strings.TrimSpace(stripped) == "" {
|
||||
if wordRe.MatchString(strings.ToLower(original)) {
|
||||
s = ""
|
||||
} else {
|
||||
if bareYearRe.MatchString(strings.TrimSpace(original)) {
|
||||
s = strings.TrimSpace(s)
|
||||
} else {
|
||||
s = ""
|
||||
}
|
||||
} else {
|
||||
s = stripped
|
||||
|
||||
@@ -44,6 +44,12 @@ func TestNormalizeString_Basic(t *testing.T) {
|
||||
{"1989 (Deluxe)", ""},
|
||||
{"1989 [Deluxe Edition]", ""},
|
||||
{"2112 (Remastered)", ""},
|
||||
// Regression: a year with a bracketed/suffixed year must NOT collapse to
|
||||
// the bare year (it falsely matched "1989" before). It collapses to empty.
|
||||
{"1989 [2020]", ""},
|
||||
{"1989 2020", ""},
|
||||
{"3000 2000", "3000"},
|
||||
{"1989 RMX", "rmx"},
|
||||
}
|
||||
|
||||
for _, tt := range tests {
|
||||
|
||||
Reference in New Issue
Block a user