From 8292f145d9b066863a2a275ba5f1bda7c3f91dd4 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Tue, 21 Jul 2026 18:55:58 +0000
Subject: [PATCH] Integrate S2 File Identifier as first analysis step

Inline the shared S2 File Identifier module into file_repair_tool_v5.html,
keeping it a single self-contained distributable, and run it as the FIRST
step when a file is loaded. The existing detectFormat() magic-number badge
is kept as-is; the module extends it with whole-buffer multi-file
separation:

- identification panel renders above the log on file load
- when multiple concatenated/embedded files are found, each extra piece
  gets a download button before repair runs ('universal' handles every
  type, so renderPanel's foreign list is empty and a small glue lists
  report.segments[1..] instead)
- the repair engine now operates on report.proceedBytes (segment 0)
  instead of the raw buffer

Also add an "Identify a corrupt file" widget (file input + inline module,
fully client-side) to the web/index.html landing page.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01M4ZC9wFPKLzRg1HRgGTPPJ
---
 file_repair_tool_v5.html | 690 +++++++++++++++++++++++++++++++++++++-
 web/index.html           | 697 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 1386 insertions(+), 1 deletion(-)

diff --git a/file_repair_tool_v5.html b/file_repair_tool_v5.html
index b7e45c7..e384131 100644
--- a/file_repair_tool_v5.html
+++ b/file_repair_tool_v5.html
@@ -119,6 +119,8 @@ <h1>🔧 Universal File Repair Tool v5.0</h1>
             <button id="btnDownload" style="display:none; background: var(--ok); color: #000;">Download Result</button>
         </div>
     </div>
+    <!-- S2 File Identifier panel: identification + separated-file downloads, shown before any repair -->
+    <div id="s2FileIdPanel"></div>
     <div id="log">Ready. Load a file to begin repair...</div>
 </div>
 
@@ -144,6 +146,643 @@ <h2>⚠️ Recovery Options</h2>
     </div>
 </div>
 
+<!-- S2 File Identifier — shared S2-suite module, inlined so this file stays a
+     single self-contained distributable. 100% client-side; no uploads. -->
+<script>
+/*!
+ * S2 File Identifier — shared first-step module for the S2 recovery tool suite.
+ *
+ * Before any repair runs, this module:
+ *   1. Reads the file's magic numbers and identifies what kind of file it is.
+ *   2. Scans the whole buffer for additional embedded / concatenated files
+ *      (e.g. a PDF glued onto a DOCX, two files fused by a disk error).
+ *   3. Separates the pieces, offers every "foreign" piece for download, and
+ *      recommends which S2 program repairs that kind of file — before the
+ *      host program tries to fix whatever is left over.
+ *
+ * 100% client-side: nothing is uploaded anywhere.
+ * Works as a browser global (window.S2FileID) and as a CommonJS module.
+ */
+(function (global) {
+  'use strict';
+
+  /* ------------------------------------------------------------------ *
+   * Program registry — the S2 tool suite and what each program handles *
+   * ------------------------------------------------------------------ */
+
+  var PROGRAMS = [
+    { key: 'universal',        name: 'Universal File Repair Tool',               url: 'https://socrtwo.github.io/Universal-File-Repair-Tool/', handles: ['*'] },
+    { key: 'corruptdocxrepair',name: 'CorruptDOCXRepair',                        url: 'https://socrtwo.github.io/CorruptDOCXRepair/',          handles: ['docx'] },
+    { key: 'quickwordrecovr',  name: 'Quick Word Recovr',                        url: 'https://socrtwo.github.io/quickwordrecovr-SF/',         handles: ['docx'] },
+    { key: 'wordrecovery',     name: 'Word Recovery',                            url: 'https://socrtwo.github.io/wordrecovery-SF/',            handles: ['doc', 'docx'] },
+    { key: 'damageddocx2txt',  name: 'Damageddocx2txt',                          url: 'https://socrtwo.github.io/damageddocx2txt-SF/',         handles: ['docx'] },
+    { key: 'corruptexcelrec',  name: 'S2 Recovery Tools for Microsoft Excel',    url: 'https://socrtwo.github.io/corruptexcelrec-SF/',         handles: ['xls', 'xlsx'] },
+    { key: 'excelrcvryaddin',  name: 'Excel Recovery Addin',                     url: 'https://socrtwo.github.io/excelrcvryaddin-SF/',         handles: ['xls'] },
+    { key: 'pptxrecovery',     name: 'PPTX Recovery',                            url: 'https://socrtwo.github.io/pptxrecovery-SF/',            handles: ['pptx'] },
+    { key: 'oorecovery',       name: 'OO Recovery (OpenDocument)',               url: 'https://socrtwo.github.io/oorecovery-SF/',              handles: ['odt', 'ods', 'odp', 'odg'] },
+    { key: 'immortalzip',      name: 'Immortal Zip',                             url: 'https://socrtwo.github.io/Immortal-Zip/',               handles: ['zip', 'jar', 'apk', 'epub'] },
+    { key: 'xmltrncatorfixr',  name: 'XML Truncator-Fixer',                      url: 'https://socrtwo.github.io/xmltrncatorfixr-SF/app/',     handles: ['xml', 'docx', 'xlsx', 'pptx'] },
+    { key: 'savvyoffice',      name: 'Savvy Repair for Microsoft Office',        url: 'https://socrtwo.github.io/savvyoffice-SF/',             handles: ['doc', 'xls', 'ppt', 'docx', 'xlsx', 'pptx'] },
+    { key: 'coffice2txt',      name: 'Coffice2txt',                              url: 'https://socrtwo.github.io/coffice2txt-SF/',             handles: ['doc', 'xls', 'ppt', 'docx', 'xlsx', 'pptx'] },
+    { key: 'crrptoffcxtrctr',  name: 'Crrptoffcxtrctr (Office extractor)',       url: 'https://socrtwo.github.io/crrptoffcxtrctr-SF/',         handles: ['docx', 'xlsx', 'pptx'] },
+    { key: 'saveofficedata',   name: 'Saveofficedata',                           url: 'https://socrtwo.github.io/saveofficedata-SF/',          handles: ['doc', 'xls', 'ppt'] },
+    { key: 'filefixerbot',     name: 'File Fixer Bot',                           url: 'https://socrtwo.github.io/file-fixer-bot/',             handles: ['*'] },
+    { key: 'datarecoverfree',  name: 'Datarecoverfree',                          url: 'https://socrtwo.github.io/datarecoverfree-SF/',         handles: [] },
+    { key: 'vistaprevrsrcvr',  name: 'Previous Version Explorer',                url: 'https://socrtwo.github.io/vistaprevrsrcvr-SF/',         handles: [] }
+  ];
+
+  /* Preferred repair order per detected extension (best tool first). */
+  var RECOMMEND_ORDER = {
+    docx: ['corruptdocxrepair', 'quickwordrecovr', 'wordrecovery', 'xmltrncatorfixr', 'savvyoffice', 'crrptoffcxtrctr', 'damageddocx2txt', 'coffice2txt'],
+    doc:  ['wordrecovery', 'savvyoffice', 'coffice2txt', 'saveofficedata'],
+    xls:  ['corruptexcelrec', 'savvyoffice', 'excelrcvryaddin', 'coffice2txt', 'saveofficedata'],
+    xlsx: ['corruptexcelrec', 'xmltrncatorfixr', 'savvyoffice', 'crrptoffcxtrctr', 'coffice2txt'],
+    ppt:  ['savvyoffice', 'coffice2txt', 'saveofficedata'],
+    pptx: ['pptxrecovery', 'xmltrncatorfixr', 'savvyoffice', 'crrptoffcxtrctr', 'coffice2txt'],
+    odt:  ['oorecovery'], ods: ['oorecovery'], odp: ['oorecovery'], odg: ['oorecovery'],
+    zip:  ['immortalzip'], jar: ['immortalzip'], apk: ['immortalzip'], epub: ['immortalzip'],
+    xml:  ['xmltrncatorfixr']
+  };
+
+  function programByKey(key) {
+    for (var i = 0; i < PROGRAMS.length; i++) if (PROGRAMS[i].key === key) return PROGRAMS[i];
+    return null;
+  }
+
+  /* Ordered list of suite programs that can repair `ext`, excluding the
+     program identified by excludeKey. Falls back to the universal tool. */
+  function recommend(ext, excludeKey) {
+    var keys = RECOMMEND_ORDER[ext] || [];
+    var out = [];
+    for (var i = 0; i < keys.length; i++) {
+      if (keys[i] === excludeKey) continue;
+      var p = programByKey(keys[i]);
+      if (p) out.push(p);
+    }
+    if (!out.length && excludeKey !== 'universal') out.push(programByKey('universal'));
+    return out;
+  }
+
+  function programHandles(programKey, ext) {
+    var p = programByKey(programKey);
+    if (!p) return false;
+    if (p.handles.indexOf('*') !== -1) return true;
+    return p.handles.indexOf(ext) !== -1;
+  }
+
+  /* --------------------------- byte helpers --------------------------- */
+
+  function toU8(input) {
+    if (input instanceof Uint8Array) return input;
+    if (typeof ArrayBuffer !== 'undefined' && input instanceof ArrayBuffer) return new Uint8Array(input);
+    if (input && input.buffer) return new Uint8Array(input.buffer, input.byteOffset || 0, input.byteLength);
+    throw new Error('S2FileID: expected Uint8Array or ArrayBuffer');
+  }
+
+  function u16le(b, o) { return b[o] | (b[o + 1] << 8); }
+  function u32le(b, o) { return (b[o] | (b[o + 1] << 8) | (b[o + 2] << 16) | (b[o + 3] << 24)) >>> 0; }
+
+  function matches(b, off, sig) {
+    if (off + sig.length > b.length) return false;
+    for (var i = 0; i < sig.length; i++) if (b[off + i] !== sig[i]) return false;
+    return true;
+  }
+
+  function ascii(str) {
+    var a = new Uint8Array(str.length);
+    for (var i = 0; i < str.length; i++) a[i] = str.charCodeAt(i) & 0xff;
+    return a;
+  }
+
+  /* Find every occurrence of a byte pattern. */
+  function findAll(b, sig, start, end) {
+    var hits = [];
+    var last = (end === undefined ? b.length : end) - sig.length;
+    for (var i = start || 0; i <= last; i++) {
+      if (b[i] === sig[0] && matches(b, i, sig)) hits.push(i);
+    }
+    return hits;
+  }
+
+  function indexOfBytes(b, sig, start, end) {
+    var last = (end === undefined ? b.length : end) - sig.length;
+    for (var i = start || 0; i <= last; i++) {
+      if (b[i] === sig[0] && matches(b, i, sig)) return i;
+    }
+    return -1;
+  }
+
+  /* UTF-16LE encode an ASCII string (for OLE2 directory entry names). */
+  function utf16le(str) {
+    var a = new Uint8Array(str.length * 2);
+    for (var i = 0; i < str.length; i++) { a[i * 2] = str.charCodeAt(i) & 0xff; a[i * 2 + 1] = 0; }
+    return a;
+  }
+
+  /* ----------------------- signature definitions ----------------------- */
+
+  var SIG = {
+    ZIP_LOCAL:  [0x50, 0x4b, 0x03, 0x04],
+    ZIP_EOCD:   [0x50, 0x4b, 0x05, 0x06],
+    ZIP_CDIR:   [0x50, 0x4b, 0x01, 0x02],
+    ZIP_SPAN:   [0x50, 0x4b, 0x07, 0x08],
+    OLE2:       [0xd0, 0xcf, 0x11, 0xe0, 0xa1, 0xb1, 0x1a, 0xe1],
+    PDF:        [0x25, 0x50, 0x44, 0x46, 0x2d],                   // %PDF-
+    PNG:        [0x89, 0x50, 0x4e, 0x47, 0x0d, 0x0a, 0x1a, 0x0a],
+    GIF87:      [0x47, 0x49, 0x46, 0x38, 0x37, 0x61],
+    GIF89:      [0x47, 0x49, 0x46, 0x38, 0x39, 0x61],
+    JPEG:       [0xff, 0xd8, 0xff],
+    RTF:        [0x7b, 0x5c, 0x72, 0x74, 0x66],                   // {\rtf
+    SEVENZ:     [0x37, 0x7a, 0xbc, 0xaf, 0x27, 0x1c],
+    RAR:        [0x52, 0x61, 0x72, 0x21, 0x1a, 0x07],
+    GZIP:       [0x1f, 0x8b, 0x08],
+    XMLDECL:    [0x3c, 0x3f, 0x78, 0x6d, 0x6c],                   // <?xml
+    TIFF_LE:    [0x49, 0x49, 0x2a, 0x00],
+    TIFF_BE:    [0x4d, 0x4d, 0x00, 0x2a],
+    SQLITE:     [0x53, 0x51, 0x4c, 0x69, 0x74, 0x65, 0x20, 0x66], // "SQLite f"
+    RIFF:       [0x52, 0x49, 0x46, 0x46],
+    OGG:        [0x4f, 0x67, 0x67, 0x53],
+    FLAC:       [0x66, 0x4c, 0x61, 0x43],
+    ID3:        [0x49, 0x44, 0x33],
+    PST:        [0x21, 0x42, 0x44, 0x4e]                          // !BDN
+  };
+
+  var DESCRIPTIONS = {
+    docx: 'Microsoft Word document (DOCX, Office Open XML)',
+    xlsx: 'Microsoft Excel workbook (XLSX, Office Open XML)',
+    pptx: 'Microsoft PowerPoint presentation (PPTX, Office Open XML)',
+    ooxml:'Microsoft Office Open XML document (type unclear)',
+    doc:  'Microsoft Word 97-2003 document (DOC)',
+    xls:  'Microsoft Excel 97-2003 workbook (XLS)',
+    ppt:  'Microsoft PowerPoint 97-2003 presentation (PPT)',
+    msg:  'Microsoft Outlook message (MSG)',
+    vsd:  'Microsoft Visio drawing (VSD)',
+    ole:  'Microsoft Office compound (OLE2) file — exact type unclear',
+    odt:  'OpenDocument text (ODT)',
+    ods:  'OpenDocument spreadsheet (ODS)',
+    odp:  'OpenDocument presentation (ODP)',
+    odg:  'OpenDocument drawing (ODG)',
+    epub: 'EPUB e-book',
+    jar:  'Java archive (JAR)',
+    apk:  'Android package (APK)',
+    zip:  'ZIP archive',
+    pdf:  'PDF document',
+    png:  'PNG image',
+    gif:  'GIF image',
+    jpg:  'JPEG image',
+    tif:  'TIFF image',
+    bmp:  'BMP image',
+    rtf:  'Rich Text Format document (RTF)',
+    '7z': '7-Zip archive',
+    rar:  'RAR archive',
+    gz:   'GZIP compressed data',
+    xml:  'XML document',
+    html: 'HTML document',
+    sqlite:'SQLite database',
+    pst:  'Outlook data file (PST/OST)',
+    wav:  'WAV audio', avi: 'AVI video', webp: 'WebP image',
+    ogg:  'OGG media', flac: 'FLAC audio', mp3: 'MP3 audio',
+    mp4:  'MP4 / QuickTime media',
+    exe:  'Windows executable (EXE/DLL)',
+    txt:  'Plain text',
+    csv:  'Comma-separated values (CSV)',
+    bin:  'Unrecognized binary data'
+  };
+
+  /* ------------------------- ZIP refinement ------------------------- */
+
+  /* Distinguish DOCX / XLSX / PPTX / ODF / EPUB / JAR / APK / plain ZIP.
+     Uses raw byte search for well-known entry names — this keeps working
+     when the archive structure itself is corrupt, because entry names are
+     stored uncompressed in local headers and the central directory. */
+  function refineZip(b, start, end) {
+    function has(s) { return indexOfBytes(b, ascii(s), start, end) !== -1; }
+    if (has('mimetypeapplication/vnd.oasis.opendocument.text')) return 'odt';
+    if (has('mimetypeapplication/vnd.oasis.opendocument.spreadsheet')) return 'ods';
+    if (has('mimetypeapplication/vnd.oasis.opendocument.presentation')) return 'odp';
+    if (has('mimetypeapplication/vnd.oasis.opendocument.graphics')) return 'odg';
+    if (has('mimetypeapplication/epub+zip')) return 'epub';
+    var isOoxml = has('[Content_Types].xml');
+    if (has('word/document.xml') || has('word/_rels/')) return 'docx';
+    if (has('xl/workbook.xml') || has('xl/_rels/')) return 'xlsx';
+    if (has('ppt/presentation.xml') || has('ppt/slides/')) return 'pptx';
+    if (isOoxml) return 'ooxml';
+    if (has('AndroidManifest.xml')) return 'apk';
+    if (has('META-INF/MANIFEST.MF')) return 'jar';
+    // ODF whose mimetype entry was compressed or lost:
+    if (has('content.xml') && has('META-INF/manifest.xml')) {
+      if (has('office:spreadsheet')) return 'ods';
+      if (has('office:presentation')) return 'odp';
+      return 'odt';
+    }
+    return 'zip';
+  }
+
+  /* Best-effort walk of ZIP local file headers starting at `start`.
+     Returns { end, clean } where `end` is the last offset proven to be
+     inside this archive's entry data (so signature hits below it are
+     internal, not a second file) and `clean` says the walk never broke. */
+  function walkZipEntries(b, start) {
+    var pos = start;
+    // Everything up to `verified` is proven to belong to this archive:
+    // each size-based jump that lands on another PK header confirms the
+    // previous entry's declared size was right. The final entry's data is
+    // only confirmed once the walk reaches the central directory / EOCD.
+    var verified = start;
+    while (pos + 30 <= b.length && matches(b, pos, SIG.ZIP_LOCAL)) {
+      verified = pos;
+      var flags = u16le(b, pos + 6);
+      var csize = u32le(b, pos + 18);
+      var nameLen = u16le(b, pos + 26);
+      var extraLen = u16le(b, pos + 28);
+      var dataStart = pos + 30 + nameLen + extraLen;
+      if (nameLen === 0 || dataStart > b.length) break;
+      verified = dataStart;
+      if (flags & 0x08) {
+        // Data descriptor: sizes unknown up front — find the next
+        // descriptor signature; if absent the walk cannot continue.
+        var dd = indexOfBytes(b, SIG.ZIP_SPAN, dataStart);
+        if (dd === -1) { pos = dataStart; break; }
+        pos = dd + 16;
+      } else {
+        var next = dataStart + csize;
+        if (next > b.length) { pos = dataStart; break; }
+        pos = next;
+      }
+    }
+    // Central directory + EOCD follow the entries.
+    if (matches(b, pos, SIG.ZIP_CDIR)) verified = pos;
+    while (pos + 46 <= b.length && matches(b, pos, SIG.ZIP_CDIR)) {
+      var cn = u16le(b, pos + 28), ce = u16le(b, pos + 30), cc = u16le(b, pos + 32);
+      pos = pos + 46 + cn + ce + cc;
+    }
+    if (pos + 22 <= b.length && matches(b, pos, SIG.ZIP_EOCD)) {
+      var comment = u16le(b, pos + 20);
+      return { end: Math.min(pos + 22 + comment, b.length), clean: true };
+    }
+    // No EOCD: the archive is corrupt or truncated. Trust only the
+    // verified prefix, so appended foreign files are still found even when
+    // a truncated entry's declared size would have swallowed them.
+    return { end: Math.min(verified, b.length), clean: false };
+  }
+
+  /* ------------------------- OLE2 refinement ------------------------- */
+
+  function refineOle(b, start, end) {
+    function has16(s) { return indexOfBytes(b, utf16le(s), start, end) !== -1; }
+    if (has16('WordDocument')) return 'doc';
+    if (has16('Workbook') || has16('Book')) return 'xls';
+    if (has16('PowerPoint Document')) return 'ppt';
+    if (has16('VisioDocument')) return 'vsd';
+    if (has16('__properties_version1.0') || has16('__recip_version1.0')) return 'msg';
+    return 'ole';
+  }
+
+  /* --------------------------- identification --------------------------- */
+
+  function looksLikeText(b, start, end) {
+    var n = Math.min(end - start, 4096);
+    if (n === 0) return false;
+    var printable = 0;
+    for (var i = 0; i < n; i++) {
+      var c = b[start + i];
+      if (c === 9 || c === 10 || c === 13 || (c >= 32 && c < 127) || c >= 128) printable++;
+      else if (c === 0) return false;
+    }
+    return printable / n > 0.95;
+  }
+
+  /* Identify the file that starts at `start` (default 0). */
+  function identifyAt(b, start, end) {
+    b = toU8(b);
+    start = start || 0;
+    end = end === undefined ? b.length : end;
+    var res = function (ext, confidence) {
+      return { ext: ext, description: DESCRIPTIONS[ext] || ext, confidence: confidence || 'high' };
+    };
+
+    if (matches(b, start, SIG.ZIP_LOCAL) || matches(b, start, SIG.ZIP_EOCD)) return res(refineZip(b, start, end));
+    if (matches(b, start, SIG.OLE2)) return res(refineOle(b, start, end));
+    if (matches(b, start, SIG.PDF)) return res('pdf');
+    if (matches(b, start, SIG.PNG)) return res('png');
+    if (matches(b, start, SIG.GIF87) || matches(b, start, SIG.GIF89)) return res('gif');
+    if (matches(b, start, SIG.JPEG)) return res('jpg');
+    if (matches(b, start, SIG.RTF)) return res('rtf');
+    if (matches(b, start, SIG.SEVENZ)) return res('7z');
+    if (matches(b, start, SIG.RAR)) return res('rar');
+    if (matches(b, start, SIG.GZIP)) return res('gz');
+    if (matches(b, start, SIG.SQLITE)) return res('sqlite');
+    if (matches(b, start, SIG.PST)) return res('pst');
+    if (matches(b, start, SIG.TIFF_LE) || matches(b, start, SIG.TIFF_BE)) return res('tif');
+    if (matches(b, start, SIG.OGG)) return res('ogg');
+    if (matches(b, start, SIG.FLAC)) return res('flac');
+    if (matches(b, start, SIG.ID3)) return res('mp3', 'medium');
+    if (matches(b, start, SIG.RIFF)) {
+      if (matches(b, start + 8, ascii('WAVE'))) return res('wav');
+      if (matches(b, start + 8, ascii('AVI '))) return res('avi');
+      if (matches(b, start + 8, ascii('WEBP'))) return res('webp');
+      return res('bin', 'low');
+    }
+    if (matches(b, start + 4, ascii('ftyp'))) return res('mp4');
+    if (b[start] === 0x4d && b[start + 1] === 0x5a) return res('exe', 'medium');
+    if (b[start] === 0x42 && b[start + 1] === 0x4d && start === 0) return res('bmp', 'medium');
+    if (matches(b, start, SIG.XMLDECL)) {
+      // <?xml ... — could be plain XML or a Flat ODF / saved OOXML part.
+      return res('xml');
+    }
+    if (looksLikeText(b, start, end)) {
+      var head = '';
+      for (var i = start; i < Math.min(end, start + 512); i++) head += String.fromCharCode(b[i]);
+      var lower = head.toLowerCase();
+      if (lower.indexOf('<!doctype html') !== -1 || lower.indexOf('<html') !== -1) return res('html');
+      if (lower.indexOf('<?xml') !== -1 || /^\s*</.test(head)) return res('xml', 'medium');
+      return res('txt', 'medium');
+    }
+    return res('bin', 'low');
+  }
+
+  function identify(bytes) { return identifyAt(toU8(bytes), 0); }
+
+  /* ------------------- embedded / concatenated file scan ------------------- */
+
+  /* Signatures strong enough to mark the start of a NEW file mid-buffer.
+     minRun guards against random matches in compressed data. */
+  var BOUNDARY_SIGS = [
+    { ext: 'zip', sig: SIG.ZIP_LOCAL },
+    { ext: 'ole', sig: SIG.OLE2 },
+    { ext: 'pdf', sig: SIG.PDF },
+    { ext: 'png', sig: SIG.PNG },
+    { ext: 'gif', sig: SIG.GIF87 }, { ext: 'gif', sig: SIG.GIF89 },
+    // JPEG start-of-image plus a plausible first marker (4 bytes) — the bare
+    // 3-byte SOI is too easy to hit inside compressed data.
+    { ext: 'jpg', sig: [0xff, 0xd8, 0xff, 0xe0] }, { ext: 'jpg', sig: [0xff, 0xd8, 0xff, 0xe1] },
+    { ext: 'jpg', sig: [0xff, 0xd8, 0xff, 0xdb] }, { ext: 'jpg', sig: [0xff, 0xd8, 0xff, 0xee] },
+    { ext: 'jpg', sig: [0xff, 0xd8, 0xff, 0xc0] }, { ext: 'jpg', sig: [0xff, 0xd8, 0xff, 0xe2] },
+    { ext: 'rtf', sig: SIG.RTF },
+    { ext: '7z',  sig: SIG.SEVENZ },
+    { ext: 'rar', sig: SIG.RAR },
+    { ext: 'sqlite', sig: SIG.SQLITE },
+    { ext: 'pst', sig: SIG.PST }
+  ];
+
+  /* Per-segment scan state: how far the file starting at `start` provably
+     extends (`guard`), and which acceptance rules apply inside it. */
+  function segState(b, start) {
+    if (matches(b, start, SIG.ZIP_LOCAL)) {
+      var w = walkZipEntries(b, start);
+      return { start: start, kind: 'zip', walkClean: w.clean, guard: Math.max(start + 4, w.end) };
+    }
+    if (matches(b, start, SIG.OLE2)) return { start: start, kind: 'ole', guard: start + 512 };
+    if (matches(b, start, SIG.JPEG)) return { start: start, kind: 'jpg', guard: start + 4 };
+    return { start: start, kind: 'other', guard: start + 1 };
+  }
+
+  /* Scan the buffer and split it into segments, one per distinct file.
+     Heuristic by design: it never destroys data — segments are offered as
+     extra downloads, and segment 0 is always the file the host program
+     was asked to repair. */
+  function scan(bytes) {
+    var b = toU8(bytes);
+    var segStarts = [0];
+    var cur = segState(b, 0);
+    var pos = Math.max(1, cur.guard);
+
+    while (pos < b.length) {
+      var hit = null;
+      for (var s = 0; s < BOUNDARY_SIGS.length && !hit; s++) {
+        if (matches(b, pos, BOUNDARY_SIGS[s].sig)) hit = BOUNDARY_SIGS[s];
+      }
+      if (hit) {
+        var accept = true;
+        if (cur.kind === 'zip') {
+          if (hit.ext === 'zip') {
+            // A PK header inside a zip we could not walk cleanly is almost
+            // always a member of the SAME corrupt archive, not a new file.
+            accept = cur.walkClean;
+          } else {
+            // Inside a corrupt zip, short signatures are likely deflate
+            // noise; demand 5+ signature bytes.
+            accept = cur.walkClean || hit.sig.length >= 5;
+          }
+        } else if (cur.kind === 'ole') {
+          // A file concatenated after an OLE2 doc always lands on a
+          // 512-byte boundary (OLE size is a multiple of 512).
+          var aligned = ((pos - cur.start) % 512) === 0;
+          accept = aligned || hit.sig.length >= 6;
+        } else if (cur.kind === 'jpg' && hit.ext === 'jpg') {
+          // Embedded EXIF thumbnails also start with SOI — only split if an
+          // end-of-image marker was seen before this point.
+          accept = indexOfBytes(b, [0xff, 0xd9], cur.start, pos) !== -1;
+        }
+        if (accept) {
+          segStarts.push(pos);
+          cur = segState(b, pos);
+          pos = Math.max(pos + 1, cur.guard);
+          continue;
+        }
+      }
+      pos++;
+    }
+
+    var segments = [];
+    for (var i = 0; i < segStarts.length; i++) {
+      var start = segStarts[i];
+      var end = i + 1 < segStarts.length ? segStarts[i + 1] : b.length;
+      var info = identifyAt(b, start, end);
+      segments.push({
+        index: i,
+        offset: start,
+        length: end - start,
+        ext: info.ext,
+        description: info.description,
+        confidence: info.confidence,
+        bytes: b.subarray(start, end)
+      });
+    }
+    return { segments: segments, multiple: segments.length > 1 };
+  }
+
+  /* ------------------------------ analyze ------------------------------ */
+
+  /*
+   * The one call host programs make first.
+   *   report = S2FileID.analyze(bytes, { programKey: 'pptxrecovery', fileName: f.name })
+   *
+   * report.primary            — identification of the file as a whole
+   * report.segments           — every separated piece (bytes included)
+   * report.matching           — segments this program is meant to repair
+   * report.foreign            — segments of a different kind, each with
+   *                             .recommended (ordered S2 programs) and
+   *                             .suggestedName for download
+   * report.mismatch           — true when even the main piece is not a type
+   *                             this program handles
+   * report.proceedBytes       — bytes the host should run its repair on
+   *                             (first matching segment, else segment 0)
+   */
+  function analyze(bytes, opts) {
+    opts = opts || {};
+    var programKey = opts.programKey || null;
+    var fileName = opts.fileName || 'recovered-file';
+    var base = fileName.replace(/\.[A-Za-z0-9]{1,6}$/, '');
+    var b = toU8(bytes);
+    var result = scan(b);
+    var segments = result.segments;
+    var matching = [], foreign = [];
+
+    for (var i = 0; i < segments.length; i++) {
+      var seg = segments[i];
+      var handled = programKey ? programHandles(programKey, seg.ext) : true;
+      seg.handledHere = handled;
+      seg.suggestedName = segments.length > 1
+        ? base + '.part' + (i + 1) + '.' + seg.ext
+        : base + '.' + seg.ext;
+      if (handled) matching.push(seg);
+      else {
+        seg.recommended = recommend(seg.ext, programKey);
+        foreign.push(seg);
+      }
+    }
+
+    var proceedSeg = matching.length ? matching[0] : segments[0];
+    return {
+      fileName: fileName,
+      primary: { ext: segments[0].ext, description: segments[0].description, confidence: segments[0].confidence },
+      segments: segments,
+      matching: matching,
+      foreign: foreign,
+      multiple: result.multiple,
+      mismatch: programKey ? !programHandles(programKey, segments[0].ext) : false,
+      proceedBytes: proceedSeg ? proceedSeg.bytes : b,
+      proceedSegment: proceedSeg
+    };
+  }
+
+  /* ------------------------------ UI panel ------------------------------ */
+
+  var MIME = {
+    docx: 'application/vnd.openxmlformats-officedocument.wordprocessingml.document',
+    xlsx: 'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet',
+    pptx: 'application/vnd.openxmlformats-officedocument.presentationml.presentation',
+    doc: 'application/msword', xls: 'application/vnd.ms-excel', ppt: 'application/vnd.ms-powerpoint',
+    pdf: 'application/pdf', png: 'image/png', gif: 'image/gif', jpg: 'image/jpeg',
+    zip: 'application/zip', xml: 'application/xml', txt: 'text/plain', html: 'text/html'
+  };
+
+  /*
+   * Render the identification report into `container` (a DOM element).
+   * Shows: what the file is, any separated foreign files with download
+   * buttons and program recommendations. Returns the report untouched.
+   */
+  function renderPanel(container, report) {
+    if (!container || typeof document === 'undefined') return report;
+    container.innerHTML = '';
+
+    var panel = document.createElement('div');
+    panel.setAttribute('data-s2fileid', '1');
+    panel.style.cssText = 'margin:12px 0;padding:12px 14px;border:1px solid rgba(128,128,128,.45);' +
+      'border-radius:8px;font:14px/1.5 system-ui,sans-serif;text-align:left;';
+
+    var head = document.createElement('div');
+    head.innerHTML = '<strong>File identification (magic numbers):</strong> ' +
+      escapeHtml(report.primary.description) +
+      (report.primary.confidence !== 'high' ? ' <em>(confidence: ' + report.primary.confidence + ')</em>' : '');
+    panel.appendChild(head);
+
+    if (report.mismatch) {
+      var warn = document.createElement('div');
+      warn.style.cssText = 'margin-top:8px;padding:8px 10px;border-radius:6px;background:rgba(255,165,0,.15);';
+      var recs = recommend(report.primary.ext, null);
+      warn.innerHTML = '&#9888;&#65039; This program is not designed to repair this kind of file.' +
+        (recs.length ? ' Recommended tool: ' + linkHtml(recs[0]) : '');
+      panel.appendChild(warn);
+    }
+
+    if (report.foreign.length) {
+      var note = document.createElement('div');
+      note.style.cssText = 'margin-top:8px;';
+      note.innerHTML = '<strong>' + report.foreign.length + ' embedded/concatenated file(s) of a different type ' +
+        'were found and separated.</strong> Download them below and repair each with the recommended program ' +
+        '<em>before</em> continuing with the remaining data:';
+      panel.appendChild(note);
+
+      for (var i = 0; i < report.foreign.length; i++) {
+        (function (seg) {
+          var row = document.createElement('div');
+          row.style.cssText = 'display:flex;flex-wrap:wrap;gap:8px;align-items:center;margin-top:6px;' +
+            'padding:6px 8px;border:1px dashed rgba(128,128,128,.45);border-radius:6px;';
+          var label = document.createElement('span');
+          label.innerHTML = escapeHtml(seg.description) + ' — ' + fmtSize(seg.length) +
+            ' at offset ' + seg.offset;
+          row.appendChild(label);
+
+          var btn = document.createElement('button');
+          btn.type = 'button';
+          btn.textContent = 'Download ' + seg.suggestedName;
+          btn.style.cssText = 'padding:4px 10px;border-radius:6px;cursor:pointer;';
+          btn.addEventListener('click', function () {
+            var blob = new Blob([seg.bytes], { type: MIME[seg.ext] || 'application/octet-stream' });
+            var a = document.createElement('a');
+            a.href = URL.createObjectURL(blob);
+            a.download = seg.suggestedName;
+            document.body.appendChild(a);
+            a.click();
+            setTimeout(function () { URL.revokeObjectURL(a.href); a.remove(); }, 1500);
+          });
+          row.appendChild(btn);
+
+          if (seg.recommended && seg.recommended.length) {
+            var rec = document.createElement('span');
+            rec.innerHTML = 'Repair with: ' + linkHtml(seg.recommended[0]);
+            row.appendChild(rec);
+          }
+          panel.appendChild(row);
+        })(report.foreign[i]);
+      }
+
+      if (report.matching.length || report.segments.length > report.foreign.length) {
+        var rest = document.createElement('div');
+        rest.style.cssText = 'margin-top:8px;';
+        rest.textContent = 'The remaining data will now be processed by this program.';
+        panel.appendChild(rest);
+      }
+    }
+
+    container.appendChild(panel);
+    return report;
+  }
+
+  function linkHtml(p) {
+    return '<a href="' + p.url + '" target="_blank" rel="noopener">' + escapeHtml(p.name) + '</a>';
+  }
+
+  function escapeHtml(s) {
+    return String(s).replace(/&/g, '&amp;').replace(/</g, '&lt;').replace(/>/g, '&gt;');
+  }
+
+  function fmtSize(n) {
+    if (n < 1024) return n + ' B';
+    if (n < 1048576) return (n / 1024).toFixed(1) + ' KB';
+    return (n / 1048576).toFixed(1) + ' MB';
+  }
+
+  /* ------------------------------ exports ------------------------------ */
+
+  var S2FileID = {
+    PROGRAMS: PROGRAMS,
+    identify: identify,
+    identifyAt: identifyAt,
+    scan: scan,
+    analyze: analyze,
+    recommend: recommend,
+    programHandles: programHandles,
+    renderPanel: renderPanel
+  };
+
+  if (typeof module !== 'undefined' && module.exports) module.exports = S2FileID;
+  global.S2FileID = S2FileID;
+})(typeof self !== 'undefined' ? self : typeof window !== 'undefined' ? window : this);
+</script>
+
 <script>
 // ============================================================================
 // DUMMY ASSETS FOR OFFICE FORMATS
@@ -817,6 +1456,7 @@ <h3>📕 For PDF Files:</h3>
 // ============================================================================
 const logEl = document.getElementById('log');
 let fileBuf = null, fileName = '', detectedFormat = null, contentCandidates = [], recoveredAssets = {};
+let s2Report = null; // S2 File Identifier report (first analysis step)
 
 function log(msg, cls='') {
     const d = document.createElement('div');
@@ -918,6 +1558,15 @@ <h3>📕 For PDF Files:</h3>
             typeEl.className = `detected-type detected-unknown`;
         }
         typeEl.style.display = 'block';
+
+        // S2 File Identifier: FIRST analysis step — extends the existing
+        // magic-number detection above with whole-buffer multi-file
+        // separation. Extra concatenated/embedded files are offered for
+        // download before any repair runs.
+        s2Report = S2FileID.analyze(u8, { programKey: 'universal', fileName: f.name });
+        S2FileID.renderPanel(document.getElementById('s2FileIdPanel'), s2Report);
+        s2ListExtraSegments(document.getElementById('s2FileIdPanel'), s2Report);
+        if (s2Report.multiple) log(`${s2Report.segments.length} concatenated/embedded files detected — repair will run on the first piece; download the other pieces from the identification panel.`, 'warn');
     };
     r.readAsArrayBuffer(f);
 });
@@ -936,7 +1585,10 @@ <h3>📕 For PDF Files:</h3>
     contentCandidates = [];
     recoveredAssets = {};
     
-    const u8 = new Uint8Array(fileBuf);
+    // Repair operates on the piece the S2 File Identifier selected
+    // (segment 0), not necessarily the raw loaded buffer.
+    let u8 = s2Report ? s2Report.proceedBytes : new Uint8Array(fileBuf);
+    if (u8.byteOffset !== 0 || u8.byteLength !== u8.buffer.byteLength) u8 = u8.slice();
     detectedFormat = detectFormat(u8, fileName);
     
     if (!detectedFormat) {
@@ -1278,6 +1930,42 @@ <h4>💡 Can't Open the Repaired PDF?</h4>
 // ============================================================================
 // UTILITY FUNCTIONS
 // ============================================================================
+// S2FileID.renderPanel only lists FOREIGN segments; the 'universal' program
+// handles every type, so list the extra separated segments for download here
+// (segment 0 is the piece this tool repairs).
+function s2ListExtraSegments(container, report) {
+    if (!report.multiple) return;
+    const extras = report.segments.slice(1).filter(s => report.foreign.indexOf(s) === -1);
+    if (!extras.length) return;
+    const panel = container.querySelector('[data-s2fileid]') || container;
+    const note = document.createElement('div');
+    note.style.marginTop = '8px';
+    note.innerHTML = `<strong>${report.segments.length} concatenated files were found and separated.</strong> Download the extra piece(s) below; this program will repair the first piece:`;
+    panel.appendChild(note);
+    extras.forEach(seg => {
+        const row = document.createElement('div');
+        row.style.cssText = 'display:flex;flex-wrap:wrap;gap:8px;align-items:center;margin-top:6px;padding:6px 8px;border:1px dashed rgba(128,128,128,.45);border-radius:6px;';
+        const label = document.createElement('span');
+        label.textContent = `${seg.description} — ${seg.length} bytes at offset ${seg.offset}`;
+        row.appendChild(label);
+        const btn = document.createElement('button');
+        btn.type = 'button';
+        btn.textContent = `Download ${seg.suggestedName}`;
+        btn.style.cssText = 'padding:4px 10px;border-radius:6px;cursor:pointer;';
+        btn.addEventListener('click', () => {
+            const blob = new Blob([seg.bytes], { type: 'application/octet-stream' });
+            const a = document.createElement('a');
+            a.href = URL.createObjectURL(blob);
+            a.download = seg.suggestedName;
+            document.body.appendChild(a);
+            a.click();
+            setTimeout(() => { URL.revokeObjectURL(a.href); a.remove(); }, 1500);
+        });
+        row.appendChild(btn);
+        panel.appendChild(row);
+    });
+}
+
 function rateQuality(u8, name) {
     let valid = 0;
     for (let i = 0; i < Math.min(u8.length, 500); i++) if ((u8[i] >= 32 && u8[i] <= 126) || u8[i] === 10 || u8[i] === 13) valid++;
diff --git a/web/index.html b/web/index.html
index bf167e4..f6e73c6 100644
--- a/web/index.html
+++ b/web/index.html
@@ -104,6 +104,14 @@ <h1 id="title">Loading…</h1>
   <main class="card" id="readme">
     <p class="loading">Fetching README…</p>
   </main>
+  <section class="card" id="identify" style="margin-top:24px">
+    <h2>Identify a corrupt file</h2>
+    <p>Pick a file to identify its real type from its magic numbers and separate
+       any concatenated or embedded files. Runs 100% in your browser &mdash;
+       nothing is uploaded anywhere.</p>
+    <p style="margin-top:10px"><input type="file" id="s2idInput"></p>
+    <div id="s2idResult"></div>
+  </section>
   <footer>
     <p><a id="src" href="#">View source on GitHub</a></p>
   </footer>
@@ -170,5 +178,694 @@ <h1 id="title">Loading…</h1>
   }
 })();
 </script>
+<!-- S2 File Identifier — shared S2-suite module, inlined so this file stays a
+     single self-contained distributable. 100% client-side; no uploads. -->
+<script>
+/*!
+ * S2 File Identifier — shared first-step module for the S2 recovery tool suite.
+ *
+ * Before any repair runs, this module:
+ *   1. Reads the file's magic numbers and identifies what kind of file it is.
+ *   2. Scans the whole buffer for additional embedded / concatenated files
+ *      (e.g. a PDF glued onto a DOCX, two files fused by a disk error).
+ *   3. Separates the pieces, offers every "foreign" piece for download, and
+ *      recommends which S2 program repairs that kind of file — before the
+ *      host program tries to fix whatever is left over.
+ *
+ * 100% client-side: nothing is uploaded anywhere.
+ * Works as a browser global (window.S2FileID) and as a CommonJS module.
+ */
+(function (global) {
+  'use strict';
+
+  /* ------------------------------------------------------------------ *
+   * Program registry — the S2 tool suite and what each program handles *
+   * ------------------------------------------------------------------ */
+
+  var PROGRAMS = [
+    { key: 'universal',        name: 'Universal File Repair Tool',               url: 'https://socrtwo.github.io/Universal-File-Repair-Tool/', handles: ['*'] },
+    { key: 'corruptdocxrepair',name: 'CorruptDOCXRepair',                        url: 'https://socrtwo.github.io/CorruptDOCXRepair/',          handles: ['docx'] },
+    { key: 'quickwordrecovr',  name: 'Quick Word Recovr',                        url: 'https://socrtwo.github.io/quickwordrecovr-SF/',         handles: ['docx'] },
+    { key: 'wordrecovery',     name: 'Word Recovery',                            url: 'https://socrtwo.github.io/wordrecovery-SF/',            handles: ['doc', 'docx'] },
+    { key: 'damageddocx2txt',  name: 'Damageddocx2txt',                          url: 'https://socrtwo.github.io/damageddocx2txt-SF/',         handles: ['docx'] },
+    { key: 'corruptexcelrec',  name: 'S2 Recovery Tools for Microsoft Excel',    url: 'https://socrtwo.github.io/corruptexcelrec-SF/',         handles: ['xls', 'xlsx'] },
+    { key: 'excelrcvryaddin',  name: 'Excel Recovery Addin',                     url: 'https://socrtwo.github.io/excelrcvryaddin-SF/',         handles: ['xls'] },
+    { key: 'pptxrecovery',     name: 'PPTX Recovery',                            url: 'https://socrtwo.github.io/pptxrecovery-SF/',            handles: ['pptx'] },
+    { key: 'oorecovery',       name: 'OO Recovery (OpenDocument)',               url: 'https://socrtwo.github.io/oorecovery-SF/',              handles: ['odt', 'ods', 'odp', 'odg'] },
+    { key: 'immortalzip',      name: 'Immortal Zip',                             url: 'https://socrtwo.github.io/Immortal-Zip/',               handles: ['zip', 'jar', 'apk', 'epub'] },
+    { key: 'xmltrncatorfixr',  name: 'XML Truncator-Fixer',                      url: 'https://socrtwo.github.io/xmltrncatorfixr-SF/app/',     handles: ['xml', 'docx', 'xlsx', 'pptx'] },
+    { key: 'savvyoffice',      name: 'Savvy Repair for Microsoft Office',        url: 'https://socrtwo.github.io/savvyoffice-SF/',             handles: ['doc', 'xls', 'ppt', 'docx', 'xlsx', 'pptx'] },
+    { key: 'coffice2txt',      name: 'Coffice2txt',                              url: 'https://socrtwo.github.io/coffice2txt-SF/',             handles: ['doc', 'xls', 'ppt', 'docx', 'xlsx', 'pptx'] },
+    { key: 'crrptoffcxtrctr',  name: 'Crrptoffcxtrctr (Office extractor)',       url: 'https://socrtwo.github.io/crrptoffcxtrctr-SF/',         handles: ['docx', 'xlsx', 'pptx'] },
+    { key: 'saveofficedata',   name: 'Saveofficedata',                           url: 'https://socrtwo.github.io/saveofficedata-SF/',          handles: ['doc', 'xls', 'ppt'] },
+    { key: 'filefixerbot',     name: 'File Fixer Bot',                           url: 'https://socrtwo.github.io/file-fixer-bot/',             handles: ['*'] },
+    { key: 'datarecoverfree',  name: 'Datarecoverfree',                          url: 'https://socrtwo.github.io/datarecoverfree-SF/',         handles: [] },
+    { key: 'vistaprevrsrcvr',  name: 'Previous Version Explorer',                url: 'https://socrtwo.github.io/vistaprevrsrcvr-SF/',         handles: [] }
+  ];
+
+  /* Preferred repair order per detected extension (best tool first). */
+  var RECOMMEND_ORDER = {
+    docx: ['corruptdocxrepair', 'quickwordrecovr', 'wordrecovery', 'xmltrncatorfixr', 'savvyoffice', 'crrptoffcxtrctr', 'damageddocx2txt', 'coffice2txt'],
+    doc:  ['wordrecovery', 'savvyoffice', 'coffice2txt', 'saveofficedata'],
+    xls:  ['corruptexcelrec', 'savvyoffice', 'excelrcvryaddin', 'coffice2txt', 'saveofficedata'],
+    xlsx: ['corruptexcelrec', 'xmltrncatorfixr', 'savvyoffice', 'crrptoffcxtrctr', 'coffice2txt'],
+    ppt:  ['savvyoffice', 'coffice2txt', 'saveofficedata'],
+    pptx: ['pptxrecovery', 'xmltrncatorfixr', 'savvyoffice', 'crrptoffcxtrctr', 'coffice2txt'],
+    odt:  ['oorecovery'], ods: ['oorecovery'], odp: ['oorecovery'], odg: ['oorecovery'],
+    zip:  ['immortalzip'], jar: ['immortalzip'], apk: ['immortalzip'], epub: ['immortalzip'],
+    xml:  ['xmltrncatorfixr']
+  };
+
+  function programByKey(key) {
+    for (var i = 0; i < PROGRAMS.length; i++) if (PROGRAMS[i].key === key) return PROGRAMS[i];
+    return null;
+  }
+
+  /* Ordered list of suite programs that can repair `ext`, excluding the
+     program identified by excludeKey. Falls back to the universal tool. */
+  function recommend(ext, excludeKey) {
+    var keys = RECOMMEND_ORDER[ext] || [];
+    var out = [];
+    for (var i = 0; i < keys.length; i++) {
+      if (keys[i] === excludeKey) continue;
+      var p = programByKey(keys[i]);
+      if (p) out.push(p);
+    }
+    if (!out.length && excludeKey !== 'universal') out.push(programByKey('universal'));
+    return out;
+  }
+
+  function programHandles(programKey, ext) {
+    var p = programByKey(programKey);
+    if (!p) return false;
+    if (p.handles.indexOf('*') !== -1) return true;
+    return p.handles.indexOf(ext) !== -1;
+  }
+
+  /* --------------------------- byte helpers --------------------------- */
+
+  function toU8(input) {
+    if (input instanceof Uint8Array) return input;
+    if (typeof ArrayBuffer !== 'undefined' && input instanceof ArrayBuffer) return new Uint8Array(input);
+    if (input && input.buffer) return new Uint8Array(input.buffer, input.byteOffset || 0, input.byteLength);
+    throw new Error('S2FileID: expected Uint8Array or ArrayBuffer');
+  }
+
+  function u16le(b, o) { return b[o] | (b[o + 1] << 8); }
+  function u32le(b, o) { return (b[o] | (b[o + 1] << 8) | (b[o + 2] << 16) | (b[o + 3] << 24)) >>> 0; }
+
+  function matches(b, off, sig) {
+    if (off + sig.length > b.length) return false;
+    for (var i = 0; i < sig.length; i++) if (b[off + i] !== sig[i]) return false;
+    return true;
+  }
+
+  function ascii(str) {
+    var a = new Uint8Array(str.length);
+    for (var i = 0; i < str.length; i++) a[i] = str.charCodeAt(i) & 0xff;
+    return a;
+  }
+
+  /* Find every occurrence of a byte pattern. */
+  function findAll(b, sig, start, end) {
+    var hits = [];
+    var last = (end === undefined ? b.length : end) - sig.length;
+    for (var i = start || 0; i <= last; i++) {
+      if (b[i] === sig[0] && matches(b, i, sig)) hits.push(i);
+    }
+    return hits;
+  }
+
+  function indexOfBytes(b, sig, start, end) {
+    var last = (end === undefined ? b.length : end) - sig.length;
+    for (var i = start || 0; i <= last; i++) {
+      if (b[i] === sig[0] && matches(b, i, sig)) return i;
+    }
+    return -1;
+  }
+
+  /* UTF-16LE encode an ASCII string (for OLE2 directory entry names). */
+  function utf16le(str) {
+    var a = new Uint8Array(str.length * 2);
+    for (var i = 0; i < str.length; i++) { a[i * 2] = str.charCodeAt(i) & 0xff; a[i * 2 + 1] = 0; }
+    return a;
+  }
+
+  /* ----------------------- signature definitions ----------------------- */
+
+  var SIG = {
+    ZIP_LOCAL:  [0x50, 0x4b, 0x03, 0x04],
+    ZIP_EOCD:   [0x50, 0x4b, 0x05, 0x06],
+    ZIP_CDIR:   [0x50, 0x4b, 0x01, 0x02],
+    ZIP_SPAN:   [0x50, 0x4b, 0x07, 0x08],
+    OLE2:       [0xd0, 0xcf, 0x11, 0xe0, 0xa1, 0xb1, 0x1a, 0xe1],
+    PDF:        [0x25, 0x50, 0x44, 0x46, 0x2d],                   // %PDF-
+    PNG:        [0x89, 0x50, 0x4e, 0x47, 0x0d, 0x0a, 0x1a, 0x0a],
+    GIF87:      [0x47, 0x49, 0x46, 0x38, 0x37, 0x61],
+    GIF89:      [0x47, 0x49, 0x46, 0x38, 0x39, 0x61],
+    JPEG:       [0xff, 0xd8, 0xff],
+    RTF:        [0x7b, 0x5c, 0x72, 0x74, 0x66],                   // {\rtf
+    SEVENZ:     [0x37, 0x7a, 0xbc, 0xaf, 0x27, 0x1c],
+    RAR:        [0x52, 0x61, 0x72, 0x21, 0x1a, 0x07],
+    GZIP:       [0x1f, 0x8b, 0x08],
+    XMLDECL:    [0x3c, 0x3f, 0x78, 0x6d, 0x6c],                   // <?xml
+    TIFF_LE:    [0x49, 0x49, 0x2a, 0x00],
+    TIFF_BE:    [0x4d, 0x4d, 0x00, 0x2a],
+    SQLITE:     [0x53, 0x51, 0x4c, 0x69, 0x74, 0x65, 0x20, 0x66], // "SQLite f"
+    RIFF:       [0x52, 0x49, 0x46, 0x46],
+    OGG:        [0x4f, 0x67, 0x67, 0x53],
+    FLAC:       [0x66, 0x4c, 0x61, 0x43],
+    ID3:        [0x49, 0x44, 0x33],
+    PST:        [0x21, 0x42, 0x44, 0x4e]                          // !BDN
+  };
+
+  var DESCRIPTIONS = {
+    docx: 'Microsoft Word document (DOCX, Office Open XML)',
+    xlsx: 'Microsoft Excel workbook (XLSX, Office Open XML)',
+    pptx: 'Microsoft PowerPoint presentation (PPTX, Office Open XML)',
+    ooxml:'Microsoft Office Open XML document (type unclear)',
+    doc:  'Microsoft Word 97-2003 document (DOC)',
+    xls:  'Microsoft Excel 97-2003 workbook (XLS)',
+    ppt:  'Microsoft PowerPoint 97-2003 presentation (PPT)',
+    msg:  'Microsoft Outlook message (MSG)',
+    vsd:  'Microsoft Visio drawing (VSD)',
+    ole:  'Microsoft Office compound (OLE2) file — exact type unclear',
+    odt:  'OpenDocument text (ODT)',
+    ods:  'OpenDocument spreadsheet (ODS)',
+    odp:  'OpenDocument presentation (ODP)',
+    odg:  'OpenDocument drawing (ODG)',
+    epub: 'EPUB e-book',
+    jar:  'Java archive (JAR)',
+    apk:  'Android package (APK)',
+    zip:  'ZIP archive',
+    pdf:  'PDF document',
+    png:  'PNG image',
+    gif:  'GIF image',
+    jpg:  'JPEG image',
+    tif:  'TIFF image',
+    bmp:  'BMP image',
+    rtf:  'Rich Text Format document (RTF)',
+    '7z': '7-Zip archive',
+    rar:  'RAR archive',
+    gz:   'GZIP compressed data',
+    xml:  'XML document',
+    html: 'HTML document',
+    sqlite:'SQLite database',
+    pst:  'Outlook data file (PST/OST)',
+    wav:  'WAV audio', avi: 'AVI video', webp: 'WebP image',
+    ogg:  'OGG media', flac: 'FLAC audio', mp3: 'MP3 audio',
+    mp4:  'MP4 / QuickTime media',
+    exe:  'Windows executable (EXE/DLL)',
+    txt:  'Plain text',
+    csv:  'Comma-separated values (CSV)',
+    bin:  'Unrecognized binary data'
+  };
+
+  /* ------------------------- ZIP refinement ------------------------- */
+
+  /* Distinguish DOCX / XLSX / PPTX / ODF / EPUB / JAR / APK / plain ZIP.
+     Uses raw byte search for well-known entry names — this keeps working
+     when the archive structure itself is corrupt, because entry names are
+     stored uncompressed in local headers and the central directory. */
+  function refineZip(b, start, end) {
+    function has(s) { return indexOfBytes(b, ascii(s), start, end) !== -1; }
+    if (has('mimetypeapplication/vnd.oasis.opendocument.text')) return 'odt';
+    if (has('mimetypeapplication/vnd.oasis.opendocument.spreadsheet')) return 'ods';
+    if (has('mimetypeapplication/vnd.oasis.opendocument.presentation')) return 'odp';
+    if (has('mimetypeapplication/vnd.oasis.opendocument.graphics')) return 'odg';
+    if (has('mimetypeapplication/epub+zip')) return 'epub';
+    var isOoxml = has('[Content_Types].xml');
+    if (has('word/document.xml') || has('word/_rels/')) return 'docx';
+    if (has('xl/workbook.xml') || has('xl/_rels/')) return 'xlsx';
+    if (has('ppt/presentation.xml') || has('ppt/slides/')) return 'pptx';
+    if (isOoxml) return 'ooxml';
+    if (has('AndroidManifest.xml')) return 'apk';
+    if (has('META-INF/MANIFEST.MF')) return 'jar';
+    // ODF whose mimetype entry was compressed or lost:
+    if (has('content.xml') && has('META-INF/manifest.xml')) {
+      if (has('office:spreadsheet')) return 'ods';
+      if (has('office:presentation')) return 'odp';
+      return 'odt';
+    }
+    return 'zip';
+  }
+
+  /* Best-effort walk of ZIP local file headers starting at `start`.
+     Returns { end, clean } where `end` is the last offset proven to be
+     inside this archive's entry data (so signature hits below it are
+     internal, not a second file) and `clean` says the walk never broke. */
+  function walkZipEntries(b, start) {
+    var pos = start;
+    // Everything up to `verified` is proven to belong to this archive:
+    // each size-based jump that lands on another PK header confirms the
+    // previous entry's declared size was right. The final entry's data is
+    // only confirmed once the walk reaches the central directory / EOCD.
+    var verified = start;
+    while (pos + 30 <= b.length && matches(b, pos, SIG.ZIP_LOCAL)) {
+      verified = pos;
+      var flags = u16le(b, pos + 6);
+      var csize = u32le(b, pos + 18);
+      var nameLen = u16le(b, pos + 26);
+      var extraLen = u16le(b, pos + 28);
+      var dataStart = pos + 30 + nameLen + extraLen;
+      if (nameLen === 0 || dataStart > b.length) break;
+      verified = dataStart;
+      if (flags & 0x08) {
+        // Data descriptor: sizes unknown up front — find the next
+        // descriptor signature; if absent the walk cannot continue.
+        var dd = indexOfBytes(b, SIG.ZIP_SPAN, dataStart);
+        if (dd === -1) { pos = dataStart; break; }
+        pos = dd + 16;
+      } else {
+        var next = dataStart + csize;
+        if (next > b.length) { pos = dataStart; break; }
+        pos = next;
+      }
+    }
+    // Central directory + EOCD follow the entries.
+    if (matches(b, pos, SIG.ZIP_CDIR)) verified = pos;
+    while (pos + 46 <= b.length && matches(b, pos, SIG.ZIP_CDIR)) {
+      var cn = u16le(b, pos + 28), ce = u16le(b, pos + 30), cc = u16le(b, pos + 32);
+      pos = pos + 46 + cn + ce + cc;
+    }
+    if (pos + 22 <= b.length && matches(b, pos, SIG.ZIP_EOCD)) {
+      var comment = u16le(b, pos + 20);
+      return { end: Math.min(pos + 22 + comment, b.length), clean: true };
+    }
+    // No EOCD: the archive is corrupt or truncated. Trust only the
+    // verified prefix, so appended foreign files are still found even when
+    // a truncated entry's declared size would have swallowed them.
+    return { end: Math.min(verified, b.length), clean: false };
+  }
+
+  /* ------------------------- OLE2 refinement ------------------------- */
+
+  function refineOle(b, start, end) {
+    function has16(s) { return indexOfBytes(b, utf16le(s), start, end) !== -1; }
+    if (has16('WordDocument')) return 'doc';
+    if (has16('Workbook') || has16('Book')) return 'xls';
+    if (has16('PowerPoint Document')) return 'ppt';
+    if (has16('VisioDocument')) return 'vsd';
+    if (has16('__properties_version1.0') || has16('__recip_version1.0')) return 'msg';
+    return 'ole';
+  }
+
+  /* --------------------------- identification --------------------------- */
+
+  function looksLikeText(b, start, end) {
+    var n = Math.min(end - start, 4096);
+    if (n === 0) return false;
+    var printable = 0;
+    for (var i = 0; i < n; i++) {
+      var c = b[start + i];
+      if (c === 9 || c === 10 || c === 13 || (c >= 32 && c < 127) || c >= 128) printable++;
+      else if (c === 0) return false;
+    }
+    return printable / n > 0.95;
+  }
+
+  /* Identify the file that starts at `start` (default 0). */
+  function identifyAt(b, start, end) {
+    b = toU8(b);
+    start = start || 0;
+    end = end === undefined ? b.length : end;
+    var res = function (ext, confidence) {
+      return { ext: ext, description: DESCRIPTIONS[ext] || ext, confidence: confidence || 'high' };
+    };
+
+    if (matches(b, start, SIG.ZIP_LOCAL) || matches(b, start, SIG.ZIP_EOCD)) return res(refineZip(b, start, end));
+    if (matches(b, start, SIG.OLE2)) return res(refineOle(b, start, end));
+    if (matches(b, start, SIG.PDF)) return res('pdf');
+    if (matches(b, start, SIG.PNG)) return res('png');
+    if (matches(b, start, SIG.GIF87) || matches(b, start, SIG.GIF89)) return res('gif');
+    if (matches(b, start, SIG.JPEG)) return res('jpg');
+    if (matches(b, start, SIG.RTF)) return res('rtf');
+    if (matches(b, start, SIG.SEVENZ)) return res('7z');
+    if (matches(b, start, SIG.RAR)) return res('rar');
+    if (matches(b, start, SIG.GZIP)) return res('gz');
+    if (matches(b, start, SIG.SQLITE)) return res('sqlite');
+    if (matches(b, start, SIG.PST)) return res('pst');
+    if (matches(b, start, SIG.TIFF_LE) || matches(b, start, SIG.TIFF_BE)) return res('tif');
+    if (matches(b, start, SIG.OGG)) return res('ogg');
+    if (matches(b, start, SIG.FLAC)) return res('flac');
+    if (matches(b, start, SIG.ID3)) return res('mp3', 'medium');
+    if (matches(b, start, SIG.RIFF)) {
+      if (matches(b, start + 8, ascii('WAVE'))) return res('wav');
+      if (matches(b, start + 8, ascii('AVI '))) return res('avi');
+      if (matches(b, start + 8, ascii('WEBP'))) return res('webp');
+      return res('bin', 'low');
+    }
+    if (matches(b, start + 4, ascii('ftyp'))) return res('mp4');
+    if (b[start] === 0x4d && b[start + 1] === 0x5a) return res('exe', 'medium');
+    if (b[start] === 0x42 && b[start + 1] === 0x4d && start === 0) return res('bmp', 'medium');
+    if (matches(b, start, SIG.XMLDECL)) {
+      // <?xml ... — could be plain XML or a Flat ODF / saved OOXML part.
+      return res('xml');
+    }
+    if (looksLikeText(b, start, end)) {
+      var head = '';
+      for (var i = start; i < Math.min(end, start + 512); i++) head += String.fromCharCode(b[i]);
+      var lower = head.toLowerCase();
+      if (lower.indexOf('<!doctype html') !== -1 || lower.indexOf('<html') !== -1) return res('html');
+      if (lower.indexOf('<?xml') !== -1 || /^\s*</.test(head)) return res('xml', 'medium');
+      return res('txt', 'medium');
+    }
+    return res('bin', 'low');
+  }
+
+  function identify(bytes) { return identifyAt(toU8(bytes), 0); }
+
+  /* ------------------- embedded / concatenated file scan ------------------- */
+
+  /* Signatures strong enough to mark the start of a NEW file mid-buffer.
+     minRun guards against random matches in compressed data. */
+  var BOUNDARY_SIGS = [
+    { ext: 'zip', sig: SIG.ZIP_LOCAL },
+    { ext: 'ole', sig: SIG.OLE2 },
+    { ext: 'pdf', sig: SIG.PDF },
+    { ext: 'png', sig: SIG.PNG },
+    { ext: 'gif', sig: SIG.GIF87 }, { ext: 'gif', sig: SIG.GIF89 },
+    // JPEG start-of-image plus a plausible first marker (4 bytes) — the bare
+    // 3-byte SOI is too easy to hit inside compressed data.
+    { ext: 'jpg', sig: [0xff, 0xd8, 0xff, 0xe0] }, { ext: 'jpg', sig: [0xff, 0xd8, 0xff, 0xe1] },
+    { ext: 'jpg', sig: [0xff, 0xd8, 0xff, 0xdb] }, { ext: 'jpg', sig: [0xff, 0xd8, 0xff, 0xee] },
+    { ext: 'jpg', sig: [0xff, 0xd8, 0xff, 0xc0] }, { ext: 'jpg', sig: [0xff, 0xd8, 0xff, 0xe2] },
+    { ext: 'rtf', sig: SIG.RTF },
+    { ext: '7z',  sig: SIG.SEVENZ },
+    { ext: 'rar', sig: SIG.RAR },
+    { ext: 'sqlite', sig: SIG.SQLITE },
+    { ext: 'pst', sig: SIG.PST }
+  ];
+
+  /* Per-segment scan state: how far the file starting at `start` provably
+     extends (`guard`), and which acceptance rules apply inside it. */
+  function segState(b, start) {
+    if (matches(b, start, SIG.ZIP_LOCAL)) {
+      var w = walkZipEntries(b, start);
+      return { start: start, kind: 'zip', walkClean: w.clean, guard: Math.max(start + 4, w.end) };
+    }
+    if (matches(b, start, SIG.OLE2)) return { start: start, kind: 'ole', guard: start + 512 };
+    if (matches(b, start, SIG.JPEG)) return { start: start, kind: 'jpg', guard: start + 4 };
+    return { start: start, kind: 'other', guard: start + 1 };
+  }
+
+  /* Scan the buffer and split it into segments, one per distinct file.
+     Heuristic by design: it never destroys data — segments are offered as
+     extra downloads, and segment 0 is always the file the host program
+     was asked to repair. */
+  function scan(bytes) {
+    var b = toU8(bytes);
+    var segStarts = [0];
+    var cur = segState(b, 0);
+    var pos = Math.max(1, cur.guard);
+
+    while (pos < b.length) {
+      var hit = null;
+      for (var s = 0; s < BOUNDARY_SIGS.length && !hit; s++) {
+        if (matches(b, pos, BOUNDARY_SIGS[s].sig)) hit = BOUNDARY_SIGS[s];
+      }
+      if (hit) {
+        var accept = true;
+        if (cur.kind === 'zip') {
+          if (hit.ext === 'zip') {
+            // A PK header inside a zip we could not walk cleanly is almost
+            // always a member of the SAME corrupt archive, not a new file.
+            accept = cur.walkClean;
+          } else {
+            // Inside a corrupt zip, short signatures are likely deflate
+            // noise; demand 5+ signature bytes.
+            accept = cur.walkClean || hit.sig.length >= 5;
+          }
+        } else if (cur.kind === 'ole') {
+          // A file concatenated after an OLE2 doc always lands on a
+          // 512-byte boundary (OLE size is a multiple of 512).
+          var aligned = ((pos - cur.start) % 512) === 0;
+          accept = aligned || hit.sig.length >= 6;
+        } else if (cur.kind === 'jpg' && hit.ext === 'jpg') {
+          // Embedded EXIF thumbnails also start with SOI — only split if an
+          // end-of-image marker was seen before this point.
+          accept = indexOfBytes(b, [0xff, 0xd9], cur.start, pos) !== -1;
+        }
+        if (accept) {
+          segStarts.push(pos);
+          cur = segState(b, pos);
+          pos = Math.max(pos + 1, cur.guard);
+          continue;
+        }
+      }
+      pos++;
+    }
+
+    var segments = [];
+    for (var i = 0; i < segStarts.length; i++) {
+      var start = segStarts[i];
+      var end = i + 1 < segStarts.length ? segStarts[i + 1] : b.length;
+      var info = identifyAt(b, start, end);
+      segments.push({
+        index: i,
+        offset: start,
+        length: end - start,
+        ext: info.ext,
+        description: info.description,
+        confidence: info.confidence,
+        bytes: b.subarray(start, end)
+      });
+    }
+    return { segments: segments, multiple: segments.length > 1 };
+  }
+
+  /* ------------------------------ analyze ------------------------------ */
+
+  /*
+   * The one call host programs make first.
+   *   report = S2FileID.analyze(bytes, { programKey: 'pptxrecovery', fileName: f.name })
+   *
+   * report.primary            — identification of the file as a whole
+   * report.segments           — every separated piece (bytes included)
+   * report.matching           — segments this program is meant to repair
+   * report.foreign            — segments of a different kind, each with
+   *                             .recommended (ordered S2 programs) and
+   *                             .suggestedName for download
+   * report.mismatch           — true when even the main piece is not a type
+   *                             this program handles
+   * report.proceedBytes       — bytes the host should run its repair on
+   *                             (first matching segment, else segment 0)
+   */
+  function analyze(bytes, opts) {
+    opts = opts || {};
+    var programKey = opts.programKey || null;
+    var fileName = opts.fileName || 'recovered-file';
+    var base = fileName.replace(/\.[A-Za-z0-9]{1,6}$/, '');
+    var b = toU8(bytes);
+    var result = scan(b);
+    var segments = result.segments;
+    var matching = [], foreign = [];
+
+    for (var i = 0; i < segments.length; i++) {
+      var seg = segments[i];
+      var handled = programKey ? programHandles(programKey, seg.ext) : true;
+      seg.handledHere = handled;
+      seg.suggestedName = segments.length > 1
+        ? base + '.part' + (i + 1) + '.' + seg.ext
+        : base + '.' + seg.ext;
+      if (handled) matching.push(seg);
+      else {
+        seg.recommended = recommend(seg.ext, programKey);
+        foreign.push(seg);
+      }
+    }
+
+    var proceedSeg = matching.length ? matching[0] : segments[0];
+    return {
+      fileName: fileName,
+      primary: { ext: segments[0].ext, description: segments[0].description, confidence: segments[0].confidence },
+      segments: segments,
+      matching: matching,
+      foreign: foreign,
+      multiple: result.multiple,
+      mismatch: programKey ? !programHandles(programKey, segments[0].ext) : false,
+      proceedBytes: proceedSeg ? proceedSeg.bytes : b,
+      proceedSegment: proceedSeg
+    };
+  }
+
+  /* ------------------------------ UI panel ------------------------------ */
+
+  var MIME = {
+    docx: 'application/vnd.openxmlformats-officedocument.wordprocessingml.document',
+    xlsx: 'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet',
+    pptx: 'application/vnd.openxmlformats-officedocument.presentationml.presentation',
+    doc: 'application/msword', xls: 'application/vnd.ms-excel', ppt: 'application/vnd.ms-powerpoint',
+    pdf: 'application/pdf', png: 'image/png', gif: 'image/gif', jpg: 'image/jpeg',
+    zip: 'application/zip', xml: 'application/xml', txt: 'text/plain', html: 'text/html'
+  };
+
+  /*
+   * Render the identification report into `container` (a DOM element).
+   * Shows: what the file is, any separated foreign files with download
+   * buttons and program recommendations. Returns the report untouched.
+   */
+  function renderPanel(container, report) {
+    if (!container || typeof document === 'undefined') return report;
+    container.innerHTML = '';
+
+    var panel = document.createElement('div');
+    panel.setAttribute('data-s2fileid', '1');
+    panel.style.cssText = 'margin:12px 0;padding:12px 14px;border:1px solid rgba(128,128,128,.45);' +
+      'border-radius:8px;font:14px/1.5 system-ui,sans-serif;text-align:left;';
+
+    var head = document.createElement('div');
+    head.innerHTML = '<strong>File identification (magic numbers):</strong> ' +
+      escapeHtml(report.primary.description) +
+      (report.primary.confidence !== 'high' ? ' <em>(confidence: ' + report.primary.confidence + ')</em>' : '');
+    panel.appendChild(head);
+
+    if (report.mismatch) {
+      var warn = document.createElement('div');
+      warn.style.cssText = 'margin-top:8px;padding:8px 10px;border-radius:6px;background:rgba(255,165,0,.15);';
+      var recs = recommend(report.primary.ext, null);
+      warn.innerHTML = '&#9888;&#65039; This program is not designed to repair this kind of file.' +
+        (recs.length ? ' Recommended tool: ' + linkHtml(recs[0]) : '');
+      panel.appendChild(warn);
+    }
+
+    if (report.foreign.length) {
+      var note = document.createElement('div');
+      note.style.cssText = 'margin-top:8px;';
+      note.innerHTML = '<strong>' + report.foreign.length + ' embedded/concatenated file(s) of a different type ' +
+        'were found and separated.</strong> Download them below and repair each with the recommended program ' +
+        '<em>before</em> continuing with the remaining data:';
+      panel.appendChild(note);
+
+      for (var i = 0; i < report.foreign.length; i++) {
+        (function (seg) {
+          var row = document.createElement('div');
+          row.style.cssText = 'display:flex;flex-wrap:wrap;gap:8px;align-items:center;margin-top:6px;' +
+            'padding:6px 8px;border:1px dashed rgba(128,128,128,.45);border-radius:6px;';
+          var label = document.createElement('span');
+          label.innerHTML = escapeHtml(seg.description) + ' — ' + fmtSize(seg.length) +
+            ' at offset ' + seg.offset;
+          row.appendChild(label);
+
+          var btn = document.createElement('button');
+          btn.type = 'button';
+          btn.textContent = 'Download ' + seg.suggestedName;
+          btn.style.cssText = 'padding:4px 10px;border-radius:6px;cursor:pointer;';
+          btn.addEventListener('click', function () {
+            var blob = new Blob([seg.bytes], { type: MIME[seg.ext] || 'application/octet-stream' });
+            var a = document.createElement('a');
+            a.href = URL.createObjectURL(blob);
+            a.download = seg.suggestedName;
+            document.body.appendChild(a);
+            a.click();
+            setTimeout(function () { URL.revokeObjectURL(a.href); a.remove(); }, 1500);
+          });
+          row.appendChild(btn);
+
+          if (seg.recommended && seg.recommended.length) {
+            var rec = document.createElement('span');
+            rec.innerHTML = 'Repair with: ' + linkHtml(seg.recommended[0]);
+            row.appendChild(rec);
+          }
+          panel.appendChild(row);
+        })(report.foreign[i]);
+      }
+
+      if (report.matching.length || report.segments.length > report.foreign.length) {
+        var rest = document.createElement('div');
+        rest.style.cssText = 'margin-top:8px;';
+        rest.textContent = 'The remaining data will now be processed by this program.';
+        panel.appendChild(rest);
+      }
+    }
+
+    container.appendChild(panel);
+    return report;
+  }
+
+  function linkHtml(p) {
+    return '<a href="' + p.url + '" target="_blank" rel="noopener">' + escapeHtml(p.name) + '</a>';
+  }
+
+  function escapeHtml(s) {
+    return String(s).replace(/&/g, '&amp;').replace(/</g, '&lt;').replace(/>/g, '&gt;');
+  }
+
+  function fmtSize(n) {
+    if (n < 1024) return n + ' B';
+    if (n < 1048576) return (n / 1024).toFixed(1) + ' KB';
+    return (n / 1048576).toFixed(1) + ' MB';
+  }
+
+  /* ------------------------------ exports ------------------------------ */
+
+  var S2FileID = {
+    PROGRAMS: PROGRAMS,
+    identify: identify,
+    identifyAt: identifyAt,
+    scan: scan,
+    analyze: analyze,
+    recommend: recommend,
+    programHandles: programHandles,
+    renderPanel: renderPanel
+  };
+
+  if (typeof module !== 'undefined' && module.exports) module.exports = S2FileID;
+  global.S2FileID = S2FileID;
+})(typeof self !== 'undefined' ? self : typeof window !== 'undefined' ? window : this);
+</script>
+<script>
+(function () {
+  // renderPanel lists only FOREIGN segments; also offer any extra separated
+  // segments of handled types for download.
+  function listExtraSegments(container, report) {
+    if (!report.multiple) return;
+    var extras = [];
+    for (var i = 1; i < report.segments.length; i++) {
+      if (report.foreign.indexOf(report.segments[i]) === -1) extras.push(report.segments[i]);
+    }
+    if (!extras.length) return;
+    var panel = container.querySelector('[data-s2fileid]') || container;
+    var note = document.createElement('div');
+    note.style.marginTop = '8px';
+    note.innerHTML = '<strong>' + report.segments.length +
+      ' concatenated files were found and separated.</strong> Download the extra piece(s) below:';
+    panel.appendChild(note);
+    extras.forEach(function (seg) {
+      var row = document.createElement('div');
+      row.style.cssText = 'display:flex;flex-wrap:wrap;gap:8px;align-items:center;margin-top:6px;' +
+        'padding:6px 8px;border:1px dashed rgba(128,128,128,.45);border-radius:6px;';
+      var label = document.createElement('span');
+      label.textContent = seg.description + ' \u2014 ' + seg.length + ' bytes at offset ' + seg.offset;
+      row.appendChild(label);
+      var btn = document.createElement('button');
+      btn.type = 'button';
+      btn.textContent = 'Download ' + seg.suggestedName;
+      btn.style.cssText = 'padding:4px 10px;border-radius:6px;cursor:pointer;';
+      btn.addEventListener('click', function () {
+        var blob = new Blob([seg.bytes], { type: 'application/octet-stream' });
+        var a = document.createElement('a');
+        a.href = URL.createObjectURL(blob);
+        a.download = seg.suggestedName;
+        document.body.appendChild(a);
+        a.click();
+        setTimeout(function () { URL.revokeObjectURL(a.href); a.remove(); }, 1500);
+      });
+      row.appendChild(btn);
+      panel.appendChild(row);
+    });
+  }
+
+  document.getElementById('s2idInput').addEventListener('change', async function () {
+    var f = this.files[0];
+    if (!f) return;
+    var buf = await f.arrayBuffer();
+    var report = S2FileID.analyze(buf, { programKey: 'universal', fileName: f.name });
+    var container = document.getElementById('s2idResult');
+    S2FileID.renderPanel(container, report);
+    listExtraSegments(container, report);
+  });
+})();
+</script>
 </body>
 </html>