Initial commit

lib/jav/codeParser.ts

@@ -0,0 +1,50 @@
+/**
+ * Extract a JAV product code from a filename.
+ * Handles common patterns: "SSIS-001.jpg", "[ABC] DEF-123.png", "ssis001 cover.jpg",
+ * "abp-456_cover.jpeg", and pre-fixes like "FHD-MIDV-123.jpg".
+ *
+ * Returns the canonical "PREFIX-NUMBER" form (uppercased), or null if no match.
+ */
+
+// Prefix must start with a letter and may contain digits (e.g. "T28", "ABP",
+// "MIAA"). Total prefix length 2-6 alphanumerics. The numeric tail is 3-5
+// digits, with an optional single-letter suffix for variant releases —
+// "IBW-610z" (z-version / re-cut), "ABP-456a" (part-a), etc. Suffix is
+// preserved in the canonical form so e.g. IBW-610 and IBW-610Z stay
+// distinct (they're different versions of the same release).
+const CODE_RE = /\b([A-Za-z][A-Za-z0-9]{1,5})[-_ ]?(\d{3,5})([a-zA-Z])?\b/;
+const QUALITY_PREFIXES = new Set(["fhd", "hd", "sd", "uhd", "4k", "1080p", "720p", "480p", "hevc", "x265", "x264"]);
+
+/**
+ * Canonicalise an already-known code to "PREFIX-NUMBER" with the number
+ * left-padded to a minimum of 3 digits and the prefix uppercased. Inputs
+ * coming from NFO `<id>` tags can arrive in any of "MIAA-291", "miaa291",
+ * "MIAA-00291", etc.; we want them all to compare equal so the collision
+ * detector isn't fooled by formatting drift.
+ */
+export function normalizeCode(raw: string | null | undefined): string | null {
+  if (!raw) return null;
+  const m = raw.trim().match(CODE_RE);
+  if (!m) return null;
+  const suffix = (m[3] ?? "").toUpperCase();
+  return `${m[1].toUpperCase()}-${m[2].padStart(3, "0")}${suffix}`;
+}
+
+export function extractCode(filename: string): string | null {
+  const base = filename.replace(/\.[^.]+$/, "");
+  // Walk left-to-right, skipping quality prefixes that look like codes (e.g. "FHD-").
+  let cursor = 0;
+  while (cursor < base.length) {
+    const m = base.slice(cursor).match(CODE_RE);
+    if (!m) return null;
+    const prefix = m[1];
+    const num = m[2];
+    const suffix = (m[3] ?? "").toUpperCase();
+    if (QUALITY_PREFIXES.has(prefix.toLowerCase())) {
+      cursor += (m.index ?? 0) + m[0].length;
+      continue;
+    }
+    return `${prefix.toUpperCase()}-${num.padStart(3, "0")}${suffix}`;
+  }
+  return null;
+}