Update iso-prototype.py

src/iso-prototype.py

@@ -477,6 +477,73 @@ def sync_tree(iso, tree: Path, logical_root: PurePosixPath) -> None:
             add_file_all_namespaces(iso, root_path / name, logical)
 
 
+def rebuild_windows_eltorito(iso, tree: Path) -> None:
+    """Replace imported hidden boot entries with complete extracted images.
+
+    Microsoft installation media commonly stores the El Torito boot images as
+    hidden extents.  When PyCdlib opens such an ISO, the catalog's 512-byte load
+    count is not enough to recover the real size of the hidden UEFI FAT image.
+    Writing the imported entry back can therefore truncate it to one sector.
+
+    Rebuild the catalog from the files extracted by the normal Windows image
+    preparation flow instead.  Short root-level ISO9660 names are intentional:
+    the boot images only need stable Directory Records for PyCdlib and do not
+    need to be exposed through the UDF namespace used by Windows Setup.
+    """
+    bios_source = tree / "boot/etfsboot.com"
+    uefi_source = tree / "efi/microsoft/boot/efisys_noprompt.bin"
+
+    missing = [str(path) for path in (bios_source, uefi_source) if not path.is_file()]
+    if missing:
+        raise PrototypeError(
+            "Cannot rebuild Windows El Torito catalog; missing boot image(s): "
+            + ", ".join(missing)
+        )
+
+    bios_iso_path = "/PYBIOS.BIN;1"
+    uefi_iso_path = "/PYUEFI.BIN;1"
+
+    print("Rebuilding El Torito catalog from extracted boot images:")
+    print(f"  BIOS: {bios_source} ({bios_source.stat().st_size} bytes)")
+    print(f"  UEFI: {uefi_source} ({uefi_source.stat().st_size} bytes)")
+
+    # These private root-level records avoid depending on the source ISO's very
+    # small ISO9660 tree.  The original Microsoft files remain available via UDF.
+    for iso_path in (bios_iso_path, uefi_iso_path):
+        try:
+            iso.get_record(iso_path=iso_path)
+        except Exception:
+            continue
+        iso.rm_file(iso_path=iso_path)
+
+    iso.add_file(str(bios_source), iso_path=bios_iso_path)
+    iso.add_file(str(uefi_source), iso_path=uefi_iso_path)
+
+    # Match the production genisoimage settings: no-emulation BIOS boot, eight
+    # 512-byte sectors loaded, and a boot info table patched into etfsboot.com.
+    iso.add_eltorito(
+        bios_iso_path,
+        boot_load_size=8,
+        platform_id=0x00,
+        boot_info_table=True,
+        media_name="noemul",
+        bootable=True,
+        boot_load_seg=0,
+    )
+
+    # A second add_eltorito() call creates an alternate catalog section.  Leaving
+    # boot_load_size unset makes PyCdlib use the complete EFI FAT image rather
+    # than the misleading one-sector count imported from Microsoft media.
+    iso.add_eltorito(
+        uefi_iso_path,
+        platform_id=0xEF,
+        efi=True,
+        media_name="noemul",
+        bootable=True,
+        boot_load_seg=0,
+    )
+
+
 def build_pycdlib(args: argparse.Namespace) -> None:
     try:
         import pycdlib  # type: ignore
@@ -495,6 +562,14 @@ def build_pycdlib(args: argparse.Namespace) -> None:
     try:
         print(f"Opening source ISO: {source}")
         iso.open(str(source))
+
+        # Microsoft media may import hidden El Torito records with an invalid
+        # index_in_parent. The Dockerfile applies a temporary PyCdlib fix that
+        # recovers the real child index, allowing this removal to succeed.
+        if getattr(iso, "eltorito_boot_catalog", None) is not None:
+            print("Removing imported El Torito catalog...")
+            iso.rm_eltorito()
+
         for value in args.replace:
             logical = PurePosixPath("/") / value.lstrip("/")
             print(f"Replacing: {logical}")
@@ -503,6 +578,9 @@ def build_pycdlib(args: argparse.Namespace) -> None:
             logical = PurePosixPath("/") / value.lstrip("/")
             print(f"Synchronizing tree: {logical}")
             sync_tree(iso, tree, logical)
+
+        rebuild_windows_eltorito(iso, tree)
+
         print("Checking PyCdlib consistency...")
         iso.force_consistency()
         print(f"Writing output ISO: {output}")