feat: metadati AI detection + prompt/lyrics + pannello frame grezzi (v1.13.0)

Metadati tab estesa per audit e cleanup file AI-generated:

- Nuovi campi read/write: source_url (WOAS ID3), lyrics (USLT), plus
  \xa9lyr per M4A
- Badge AI con riconoscimento: Suno, Udio, Riffusion, Mubert, AIVA,
  Boomy, SoundRAW, Loudly + presenza C2PA manifest (GEOB frame)
- Pannello "Metadati avanzati" collapsable: tabella di tutti i frame
  ID3 grezzi (chiave, tipo text/binary, preview, size)
- Rimozione selettiva frame custom via bridge remove_metadata_frames:
  ID3.__delitem__() + save() per MP3, MP4.tags per M4A. Frame standard
  protetti (TIT2/TPE1/APIC ecc.) marcati 🔒 nella UI, si svuotano dai
  campi form
- Rilevamento AI riusabile via core.metadata._detect_ai(comment,
  source_url, extra) — case-insensitive, returns (is_ai, source_name)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

api/bridge.py

@@ -446,6 +446,21 @@ class Api:
         except Exception as e:
             return {"ok": False, "error": str(e)}
 
+    def remove_metadata_frames(self, path: str, frame_keys: list) -> dict:
+        """Rimuove i frame ID3/atomi indicati dal file. Usato dal pannello
+        'Metadati avanzati' per pulire TXXX, GEOB (C2PA), USLT extra, ecc."""
+        if not path or not os.path.exists(path):
+            return {"ok": False, "error": "File non trovato"}
+        gate = self._gate("metadata")
+        if gate:
+            return gate
+        try:
+            from core.metadata import remove_frames
+            n = remove_frames(path, frame_keys or [])
+            return {"ok": True, "removed": n}
+        except Exception as e:
+            return {"ok": False, "error": str(e)}
+
     def save_metadata(self, payload: dict) -> dict:
         path = (payload.get("path") or "").strip()
         if not path:

core/config.py

@@ -5,7 +5,7 @@ import os
 import sys
 from pathlib import Path
 
-VERSION = "v1.12.0"
+VERSION = "v1.13.0"
 
 
 APP_NAME = "MusicTools"

core/metadata.py

@@ -14,6 +14,7 @@ from mutagen import File as MutagenFile
 from mutagen.id3 import (
     ID3, ID3NoHeaderError,
     APIC, TIT2, TPE1, TPE2, TALB, TDRC, TRCK, TCON, COMM, TBPM, TKEY,
+    USLT, WOAS,
 )
 from mutagen.mp4 import MP4, MP4Cover
 from mutagen.flac import FLAC, Picture
@@ -84,6 +85,69 @@ def _text(value) -> str:
     return str(value)
 
 
+_AI_MARKERS = {
+    "suno": ["suno.com", "made with suno", "sunoai"],
+    "udio": ["udio.com", "made with udio"],
+    "riffusion": ["riffusion.com", "made with riffusion"],
+    "mubert": ["mubert.com", "made with mubert"],
+    "aiva": ["aiva.ai"],
+    "boomy": ["boomy.com", "made with boomy"],
+    "soundraw": ["soundraw.io"],
+    "loudly": ["loudly.com"],
+}
+
+
+def _detect_ai(comment: str, source_url: str, extra: str = "") -> tuple:
+    """Ritorna (is_ai: bool, source_name: str). Case-insensitive.
+    `extra` = altre stringhe da controllare (es. presenza C2PA manifest)."""
+    blob = f"{comment} {source_url} {extra}".lower()
+    for source, markers in _AI_MARKERS.items():
+        for m in markers:
+            if m in blob:
+                return True, source
+    if "c2pa" in blob:
+        return True, ""
+    return False, ""
+
+
+def _dump_id3_frames(tags) -> list:
+    """Lista compatta di tutti i frame ID3 (esclusa APIC binaria).
+    Ogni voce: {key, preview, size, kind}. Utile per audit AI-generated."""
+    out = []
+    for key in tags.keys():
+        try:
+            frame = tags[key]
+        except Exception:
+            continue
+        kind = "text"
+        preview = ""
+        size = 0
+        if key.startswith("APIC"):
+            kind = "binary"
+            preview = f"{frame.mime or 'image'} ({len(frame.data)} bytes)"
+            size = len(frame.data)
+        elif key.startswith("GEOB"):
+            kind = "binary"
+            data = getattr(frame, "data", b"") or b""
+            desc = getattr(frame, "desc", "") or ""
+            mime = getattr(frame, "mime", "") or ""
+            preview = f"{desc or mime or 'blob'} ({len(data)} bytes)"
+            size = len(data)
+        elif key.startswith("USLT"):
+            txt = getattr(frame, "text", "") or ""
+            preview = txt[:180] + ("…" if len(txt) > 180 else "")
+            size = len(txt)
+        elif key.startswith("WOAS") or key.startswith("WXXX") or key.startswith("WOAR"):
+            preview = getattr(frame, "url", "") or str(frame)
+            size = len(preview)
+        else:
+            txt = str(frame)
+            preview = txt[:180] + ("…" if len(txt) > 180 else "")
+            size = len(txt)
+        out.append({"key": key, "preview": preview, "size": size, "kind": kind})
+    return out
+
+
 def _read_mp3(path: str, result: dict) -> dict:
     result["format"] = "MP3"
     try:
@@ -91,6 +155,8 @@ def _read_mp3(path: str, result: dict) -> dict:
     except ID3NoHeaderError:
         return result
 
+    result["raw_frames"] = _dump_id3_frames(tags)
+
     result["title"] = _text(tags.get("TIT2"))
     result["artist"] = _text(tags.get("TPE1"))
     result["album_artist"] = _text(tags.get("TPE2"))
@@ -100,12 +166,19 @@ def _read_mp3(path: str, result: dict) -> dict:
     result["genre"] = _text(tags.get("TCON"))
     result["bpm"] = _text(tags.get("TBPM"))
     result["key"] = _text(tags.get("TKEY"))
+    result["source_url"] = _text(tags.get("WOAS"))
 
     for k in tags.keys():
         if k.startswith("COMM"):
             result["comment"] = _text(tags[k])
             break
 
+    for k in tags.keys():
+        if k.startswith("USLT"):
+            frame = tags[k]
+            result["lyrics"] = getattr(frame, "text", "") or ""
+            break
+
     for k in tags.keys():
         if k.startswith("APIC"):
             apic = tags[k]
@@ -113,13 +186,44 @@ def _read_mp3(path: str, result: dict) -> dict:
             result["cover_mime"] = apic.mime or "image/jpeg"
             break
 
+    # AI detection: C2PA manifest (GEOB frame) o markers testuali
+    has_c2pa = any("c2pa" in k.lower() for k in tags.keys())
+    result["ai_generated"], result["ai_source"] = _detect_ai(
+        result["comment"], result["source_url"], "c2pa" if has_c2pa else "",
+    )
+
     return result
 
 
+def _dump_mp4_atoms(tags) -> list:
+    out = []
+    for key, val in tags.items():
+        preview = ""
+        kind = "text"
+        size = 0
+        if key == "covr":
+            kind = "binary"
+            data = bytes(val[0]) if val else b""
+            preview = f"cover ({len(data)} bytes)"
+            size = len(data)
+        else:
+            try:
+                preview = str(val[0] if isinstance(val, list) and val else val)
+            except Exception:
+                preview = "<binary>"
+            if isinstance(preview, str):
+                size = len(preview)
+                if len(preview) > 180:
+                    preview = preview[:180] + "…"
+        out.append({"key": key, "preview": preview, "size": size, "kind": kind})
+    return out
+
+
 def _read_mp4(path: str, result: dict) -> dict:
     result["format"] = "M4A/MP4"
     m = MP4(path)
     tags = m.tags or {}
+    result["raw_frames"] = _dump_mp4_atoms(tags)
     result["title"] = _text(tags.get("\xa9nam"))
     result["artist"] = _text(tags.get("\xa9ART"))
     result["album_artist"] = _text(tags.get("aART"))
@@ -132,12 +236,24 @@ def _read_mp4(path: str, result: dict) -> dict:
     result["genre"] = _text(tags.get("\xa9gen"))
     result["comment"] = _text(tags.get("\xa9cmt"))
     result["bpm"] = _text(tags.get("tmpo"))
+    result["lyrics"] = _text(tags.get("\xa9lyr"))
+    # M4A non ha WOAS: cerchiamo in freeform ----:com.apple.iTunes:URL
+    for k in tags.keys():
+        if isinstance(k, str) and k.startswith("----") and "URL" in k.upper():
+            v = tags[k]
+            if v:
+                result["source_url"] = v[0].decode("utf-8", errors="ignore") if isinstance(v[0], bytes) else str(v[0])
+                break
 
     covers = tags.get("covr")
     if covers:
         c = covers[0]
         result["cover_base64"] = base64.b64encode(bytes(c)).decode("ascii")
         result["cover_mime"] = "image/jpeg" if c.imageformat == MP4Cover.FORMAT_JPEG else "image/png"
+
+    result["ai_generated"], result["ai_source"] = _detect_ai(
+        result["comment"], result["source_url"],
+    )
     return result
 
 
@@ -292,8 +408,13 @@ def read_metadata(path: str) -> dict:
         "bpm": "", "key": "",
         "duration": 0, "bitrate": 0,
         "cover_base64": "", "cover_mime": "",
+        "lyrics": "",           # USLT (MP3), \xa9lyr (M4A), UNSYNCEDLYRICS (FLAC)
+        "source_url": "",       # WOAS (MP3), custom (altri) — es. URL Suno
+        "ai_generated": False,  # True se rilevato marker AI (C2PA, "made with suno"...)
+        "ai_source": "",        # es. "suno", "udio", "riffusion" — sorgente dedotta
         "where_from": _read_where_from(str(p)),
         "is_macos": sys.platform == "darwin",
+        "raw_frames": [],       # list of {key, preview, size, kind} — tutti i frame ID3/atomi
     }
 
     audio = MutagenFile(str(p))
@@ -346,6 +467,22 @@ def _write_mp3(path: str, data: dict, cover_path: Optional[str], remove_cover: b
     if comment:
         tags.add(COMM(encoding=3, lang="ita", desc="", text=comment))
 
+    # Lyrics (USLT). Rimuovi tutti gli USLT esistenti, poi (ri)crea se non vuoto
+    for k in list(tags.keys()):
+        if k.startswith("USLT"):
+            del tags[k]
+    lyrics = data.get("lyrics", "")
+    if lyrics:
+        tags.add(USLT(encoding=3, lang="eng", desc="", text=lyrics))
+
+    # Source URL (WOAS)
+    if "source_url" in data:
+        src = data.get("source_url", "")
+        if "WOAS" in tags:
+            del tags["WOAS"]
+        if src:
+            tags.add(WOAS(url=src))
+
     # Cover
     if remove_cover or cover_path:
         for k in list(tags.keys()):
@@ -379,6 +516,7 @@ def _write_mp4(path: str, data: dict, cover_path: Optional[str], remove_cover: b
     set_or_del("\xa9day", data.get("year", ""))
     set_or_del("\xa9gen", data.get("genre", ""))
     set_or_del("\xa9cmt", data.get("comment", ""))
+    set_or_del("\xa9lyr", data.get("lyrics", ""))
 
     track = data.get("track", "")
     if track:
@@ -518,3 +656,38 @@ def write_metadata(path: str, data: dict, cover_path: Optional[str] = None,
     # macOS extended attribute: kMDItemWhereFroms (lista URL/origini)
     if "where_from" in data:
         _write_where_from(path, data.get("where_from") or [])
+
+
+def remove_frames(path: str, frame_keys: list) -> int:
+    """Rimuove specifici frame ID3 (MP3) o atomi (MP4) dal file.
+    Ritorna il numero di frame rimossi. `frame_keys` sono le chiavi
+    esatte così come tornano da `raw_frames` (es. "TXXX:comment",
+    "GEOB:c2pa manifest store", "\xa9lyr")."""
+    if not frame_keys:
+        return 0
+    ext = Path(path).suffix.lower()
+    removed = 0
+    if ext == ".mp3":
+        try:
+            tags = ID3(path)
+        except ID3NoHeaderError:
+            return 0
+        for key in frame_keys:
+            if key in tags:
+                del tags[key]
+                removed += 1
+        if removed:
+            tags.save(path)
+    elif ext in (".m4a", ".mp4", ".aac"):
+        m = MP4(path)
+        if m.tags is None:
+            return 0
+        for key in frame_keys:
+            if key in m.tags:
+                del m.tags[key]
+                removed += 1
+        if removed:
+            m.save()
+    else:
+        raise ValueError(f"Rimozione frame non supportata per: {ext}")
+    return removed

webui/css/style.css

@@ -578,6 +578,87 @@ input[type="number"]::-webkit-inner-spin-button {
 .path-display:empty,
 .path-display.empty { color: var(--text-3); }
 
+.ai-badge {
+  display: inline-flex;
+  align-items: center;
+  gap: 4px;
+  padding: 3px 10px;
+  border-radius: 999px;
+  font-size: 11px;
+  font-weight: 800;
+  letter-spacing: 0.5px;
+  text-transform: uppercase;
+  background: linear-gradient(135deg, var(--purple), var(--pink));
+  color: #fff;
+  box-shadow: 0 2px 6px rgba(168, 85, 247, 0.35);
+  white-space: nowrap;
+  cursor: default;
+}
+.ai-badge[hidden] { display: none !important; }
+.ai-badge::before { content: "🤖"; }
+
+.raw-frames-details {
+  margin-top: 8px;
+  padding: 12px;
+  background: var(--bg-card-2);
+  border-radius: var(--r-sm);
+  border: 1px solid var(--border);
+}
+.raw-frames-details summary {
+  cursor: pointer;
+  font-weight: 700;
+  font-size: 13px;
+  color: var(--text-2);
+  padding: 2px 0;
+  user-select: none;
+}
+.raw-frames-details summary:hover { color: var(--text); }
+.raw-frames-details[open] summary { margin-bottom: 10px; color: var(--text); }
+.raw-frames-table {
+  width: 100%;
+  border-collapse: collapse;
+  font-size: 12px;
+  font-variant-numeric: tabular-nums;
+}
+.raw-frames-table th,
+.raw-frames-table td {
+  padding: 6px 8px;
+  border-bottom: 1px solid var(--divider);
+  text-align: left;
+  vertical-align: top;
+}
+.raw-frames-table th {
+  color: var(--text-3);
+  font-weight: 700;
+  text-transform: uppercase;
+  letter-spacing: 0.5px;
+  font-size: 10px;
+}
+.raw-frames-table td.raw-key {
+  font-family: ui-monospace, SF Mono, Menlo, monospace;
+  color: var(--indigo-text);
+  white-space: nowrap;
+}
+.raw-frames-table td.raw-preview {
+  word-break: break-word;
+  color: var(--text);
+}
+.raw-frames-table td.raw-size {
+  text-align: right;
+  color: var(--text-3);
+  white-space: nowrap;
+}
+.raw-frames-table .kind-binary {
+  display: inline-block;
+  padding: 1px 6px;
+  border-radius: 3px;
+  font-size: 10px;
+  font-weight: 700;
+  background: var(--amber-dim);
+  color: var(--amber-text);
+  text-transform: uppercase;
+}
+
 /* ===============================
    BUTTONS
    =============================== */

webui/index.html

@@ -444,6 +444,7 @@
           <div class="row">
             <button class="btn btn-amber pill" id="metaPickBtn">📂 Apri file audio</button>
             <div class="path-display" id="metaPathDisplay">Nessun file selezionato</div>
+            <span id="metaAiBadge" class="ai-badge" hidden title="Brano generato con AI"></span>
           </div>
           <div class="meta-info" id="metaInfo" hidden></div>
         </div>
@@ -492,6 +493,42 @@
                 <label class="field-label">Commento</label>
                 <textarea id="metaComment" class="input textarea" rows="3"></textarea>
               </div>
+              <div class="field full">
+                <label class="field-label">URL sorgente (WOAS)</label>
+                <input type="text" id="metaSourceUrl" class="input pill" placeholder="https://...">
+                <div class="cover-hint">Link alla pagina d'origine del brano (es. Suno, YouTube, SoundCloud). Sopravvive alle copie del file.</div>
+              </div>
+              <div class="field full">
+                <label class="field-label">Testo / Prompt (USLT)</label>
+                <textarea id="metaLyrics" class="input textarea" rows="8" placeholder="Lyrics del brano o prompt AI utilizzato..."></textarea>
+                <div class="cover-hint">Testo o prompt strutturato. Suno lo compila automaticamente al momento della generazione.</div>
+              </div>
+              <details class="field full raw-frames-details">
+                <summary>Metadati avanzati (tutti i frame ID3 grezzi)</summary>
+                <table class="raw-frames-table" id="metaRawFrames">
+                  <thead>
+                    <tr>
+                      <th style="width:32px;"><input type="checkbox" id="metaRawSelectAll" title="Seleziona tutti"></th>
+                      <th>Frame</th>
+                      <th>Tipo</th>
+                      <th>Contenuto</th>
+                      <th style="text-align:right;">Bytes</th>
+                    </tr>
+                  </thead>
+                  <tbody></tbody>
+                </table>
+                <div class="row" style="margin-top:10px;justify-content:space-between;align-items:center;">
+                  <span id="metaRawSelectedCount" style="font-size:12px;color:var(--text-3);">0 selezionati</span>
+                  <button id="metaRawRemoveBtn" class="btn btn-danger pill btn-sm" disabled>
+                    🗑 Rimuovi selezionati
+                  </button>
+                </div>
+                <div class="cover-hint" style="margin-top:8px;">
+                  I frame standard (TIT2 titolo, TPE1 artista, TALB album, APIC cover…) si gestiscono con i campi sopra.
+                  Qui rimuovi frame custom come TXXX, GEOB (C2PA), USLT (lyrics) o COMM extra.
+                </div>
+              </details>
+
               <div class="field full" id="metaWhereFromField" hidden>
                 <label class="field-label">
                   Origine (macOS · kMDItemWhereFroms)

webui/js/app.js

@@ -1147,6 +1147,51 @@ function populateMetaForm(d) {
   $("#metaBpm").value = d.bpm || "";
   $("#metaKey").value = d.key || "";
   $("#metaComment").value = d.comment || "";
+  $("#metaSourceUrl").value = d.source_url || "";
+  $("#metaLyrics").value = d.lyrics || "";
+
+  // Raw frames table
+  const rawBody = document.querySelector("#metaRawFrames tbody");
+  if (rawBody) {
+    // Frame gestiti dai campi standard: no checkbox (li rimuovi svuotando il campo)
+    const PROTECTED_MP3 = /^(TIT2|TPE1|TPE2|TALB|TDRC|TRCK|TCON|TBPM|TKEY|APIC)/;
+    const PROTECTED_MP4 = new Set(["\xa9nam","\xa9ART","aART","\xa9alb","\xa9day","trkn","\xa9gen","tmpo","covr"]);
+    rawBody.innerHTML = "";
+    (d.raw_frames || []).forEach((f) => {
+      const tr = document.createElement("tr");
+      const kindHtml = f.kind === "binary"
+        ? '<span class="kind-binary">binary</span>'
+        : '<span style="color:var(--text-3);font-size:11px;">text</span>';
+      const isProtected = PROTECTED_MP3.test(f.key) || PROTECTED_MP4.has(f.key);
+      const cbHtml = isProtected
+        ? '<span title="Frame standard — gestito dai campi sopra" style="color:var(--text-3);font-size:14px;">🔒</span>'
+        : `<input type="checkbox" class="meta-raw-check" data-key="${_escapeHtml(f.key)}">`;
+      tr.innerHTML = `
+        <td style="text-align:center;">${cbHtml}</td>
+        <td class="raw-key">${_escapeHtml(f.key)}</td>
+        <td>${kindHtml}</td>
+        <td class="raw-preview">${_escapeHtml(f.preview || "")}</td>
+        <td class="raw-size">${(f.size || 0).toLocaleString()}</td>
+      `;
+      rawBody.appendChild(tr);
+    });
+    // Reset stato selezione
+    const master = $("#metaRawSelectAll");
+    if (master) { master.checked = false; master.indeterminate = false; }
+    _updateRawFramesSelection();
+  }
+
+  // AI badge
+  const aiBadge = $("#metaAiBadge");
+  if (aiBadge) {
+    if (d.ai_generated) {
+      aiBadge.hidden = false;
+      aiBadge.textContent = d.ai_source ? `AI · ${d.ai_source}` : "AI Generated";
+    } else {
+      aiBadge.hidden = true;
+      aiBadge.textContent = "";
+    }
+  }
 
   // WhereFroms (macOS only)
   if (d.is_macos) {
@@ -1204,6 +1249,51 @@ $("#metaPickBtn").addEventListener("click", async () => {
   if (path) await loadMetaFile(path);
 });
 
+function _updateRawFramesSelection() {
+  const boxes = $$("#metaRawFrames .meta-raw-check");
+  const checked = boxes.filter((b) => b.checked);
+  const cnt = $("#metaRawSelectedCount");
+  if (cnt) cnt.textContent = `${checked.length} selezionati`;
+  const btn = $("#metaRawRemoveBtn");
+  if (btn) btn.disabled = checked.length === 0;
+  const master = $("#metaRawSelectAll");
+  if (master) {
+    if (boxes.length === 0) { master.checked = false; master.indeterminate = false; }
+    else if (checked.length === 0) { master.checked = false; master.indeterminate = false; }
+    else if (checked.length === boxes.length) { master.checked = true; master.indeterminate = false; }
+    else { master.checked = false; master.indeterminate = true; }
+  }
+}
+
+document.addEventListener("change", (e) => {
+  const t = e.target;
+  if (t && t.classList && t.classList.contains("meta-raw-check")) {
+    _updateRawFramesSelection();
+  }
+});
+
+$("#metaRawSelectAll")?.addEventListener("change", (e) => {
+  $$("#metaRawFrames .meta-raw-check").forEach((b) => { b.checked = e.target.checked; });
+  _updateRawFramesSelection();
+});
+
+$("#metaRawRemoveBtn")?.addEventListener("click", async () => {
+  const keys = $$("#metaRawFrames .meta-raw-check")
+    .filter((b) => b.checked)
+    .map((b) => b.dataset.key);
+  if (keys.length === 0) return;
+  if (!window.confirm(`Rimuovere ${keys.length} frame dal file?\n\n${keys.join("\n")}\n\nOperazione permanente.`)) return;
+  const path = state.meta?.path;
+  if (!path) return;
+  const res = await window.pywebview.api.remove_metadata_frames(path, keys);
+  if (res && res.ok) {
+    toast(`Rimossi ${res.removed || 0} frame`, "ok");
+    await loadMetaFile(path);  // ricarica per aggiornare tabella
+  } else {
+    toast("Errore: " + (res?.error || "sconosciuto"), "error");
+  }
+});
+
 $("#metaCoverBtn").addEventListener("click", async () => {
   const path = await window.pywebview.api.pick_image_file();
   if (!path) return;
@@ -1254,6 +1344,8 @@ $("#metaSaveBtn").addEventListener("click", async () => {
       bpm: $("#metaBpm").value,
       key: $("#metaKey").value,
       comment: $("#metaComment").value,
+      source_url: $("#metaSourceUrl").value.trim(),
+      lyrics: $("#metaLyrics").value,
       where_from: $("#metaWhereFrom").value
         .split("\n")
         .map(s => s.trim())