Update app.py

app.py

@@ -8,6 +8,7 @@ import requests
 import gradio as gr
 import bencodepy
 import py7zr
+from py7zr.exceptions import CrcError  # for granular handling
 
 # =========================
 # Helpers
@@ -247,6 +248,41 @@ def test_7z_integrity(archive_path: str) -> bool:
     except Exception:
         return False
 
+def safe_extract_7z(archive_path: str, dest_dir: str) -> Tuple[int, List[str]]:
+    """
+    Extract an archive. If a CRC error occurs, fall back to per-member extraction,
+    skipping only the bad members. Returns (#extracted, skipped_list).
+    """
+    extracted_count = 0
+    skipped: List[str] = []
+    dest = pathlib.Path(dest_dir)
+    dest.mkdir(parents=True, exist_ok=True)
+
+    # First try normal extraction (fast path).
+    try:
+        with py7zr.SevenZipFile(archive_path, mode="r") as z:
+            z.extract(path=str(dest))
+        # We don't know exact count from here; return -1 to mean "unknown but success"
+        return -1, skipped
+    except CrcError:
+        # Fall back to per-member extraction, skipping corrupted ones.
+        pass
+
+    # Per-member pass
+    with py7zr.SevenZipFile(archive_path, mode="r") as z:
+        members = [info.filename for info in z.list() if not info.is_directory]
+        for name in members:
+            try:
+                # Extract only this member; py7zr streams it to disk
+                z.extract(targets=[name], path=str(dest))
+                extracted_count += 1
+            except CrcError:
+                skipped.append(name)
+            except Exception:
+                skipped.append(name)
+
+    return extracted_count, skipped
+
 # =========================
 # Pipeline
 # =========================
@@ -312,16 +348,24 @@ def run_pipeline(torrent_url: str):
             logs.append(f"CRC test failed for {dest.name}, retrying download fresh…")
             download_file_exact(final_url, dest, expected_size, max_attempts=2)
             if not test_7z_integrity(str(dest)):
-                raise gr.Error(f"Archive still fails CRC after re-download: {dest.name}")
-
+                logs.append(f"Archive still reports CRC problems: {dest.name}. Will try per-file extraction and skip corrupt members.")
         saved_archives.append(str(dest))
 
-    # Extract all .7z archives (after passing CRC test)
+    # Extract all .7z archives (with resilient per-member fallback)
     for apath in saved_archives:
         logs.append(f"Extracting: {apath}")
-        with py7zr.SevenZipFile(apath, mode="r") as z:
-            z.extract(path=str(ex_dir))
-        logs.append(f"Extracted to: {ex_dir}")
+        count, skipped = safe_extract_7z(apath, str(ex_dir))
+        if count == -1:
+            logs.append(f"Extracted OK → {ex_dir}")
+        else:
+            logs.append(f"Extracted {count} members to {ex_dir}")
+            if skipped:
+                logs.append(f"Skipped {len(skipped)} corrupted member(s):")
+                # show up to a few to keep log readable
+                show = skipped[:10]
+                logs += [f"  - {s}" for s in show]
+                if len(skipped) > 10:
+                    logs.append(f"  … and {len(skipped) - 10} more")
 
     # List extracted files
     extracted = list_files_recursive(ex_dir)