nc picsort in python

bundles/nextcloud-picsort/files/nextcloud-picsort

@@ -1,110 +1,209 @@
-#!/bin/bash
+#!/usr/bin/env python3
-set -euo pipefail
+import argparse
+import base64
+import hashlib
+import os
+import shutil
+import subprocess
+from concurrent.futures import ThreadPoolExecutor, as_completed
+from datetime import datetime
+from pathlib import Path
 
-if [[ $# -ne 4 ]]; then
-  echo "Usage: $0 <nc_user> <source_subdir> <dest_subdir> <unsortable_subdir>" >&2
-  exit 1
-fi
 
-NC_USER="$1"
+ALLOWED_EXTS = {
-SOURCE_SUBDIR="$2"
+    ".png", ".jpg", ".jpeg", ".heic", ".cr2", ".cr3", ".mp4", ".mov",
-DEST_SUBDIR="$3"
+    ".webp", ".avif", ".gif",
-UNSORTABLE_SUBDIR="$4"
-
-REL_SOURCE_PATH="/$NC_USER/files/$SOURCE_SUBDIR"
-ABS_SOURCE_PATH="/var/lib/nextcloud/$NC_USER/files/$SOURCE_SUBDIR"
-
-REL_DEST_PATH="/$NC_USER/files/$DEST_SUBDIR"
-ABS_DEST_PATH="/var/lib/nextcloud/$NC_USER/files/$DEST_SUBDIR"
-
-REL_UNSORTABLE_PATH="/$NC_USER/files/$UNSORTABLE_SUBDIR"
-ABS_UNSORTABLE_PATH="/var/lib/nextcloud/$NC_USER/files/$UNSORTABLE_SUBDIR"
-
-echo "STARTING..."
-
-chown -R www-data:www-data "$ABS_SOURCE_PATH"
-chmod -R 770 "$ABS_SOURCE_PATH"
-
-process_file() {
-  local f="$1"
-  local DATETIME DATE TIME YEAR MONTH DAY HOUR MINUTE SECOND HASH EXT RAW FILE RELPATH DIRNAME
-
-  echo "PROCESSING: $f"
-
-  DATETIME="$(
-    exiftool -s -s -s -CreateDate "$f" 2>/dev/null | head -n1
-  )"
-
-  if [[ -z "$DATETIME" ]]; then
-    DATETIME="$(
-      exiftool -s -s -s -FileModifyDate "$f" 2>/dev/null | head -n1 | cut -d'+' -f1 | cut -d'-' -f1
-    )"
-  fi
-
-  if [[ -z "$DATETIME" ]]; then
-    RELPATH="$(realpath --relative-to="$ABS_SOURCE_PATH" "$f")"
-    DIRNAME="$(dirname "$ABS_UNSORTABLE_PATH/$RELPATH")"
-    echo "UNSORTABLE: $f"
-    mkdir -p "$DIRNAME"
-    mv -n -- "$f" "$DIRNAME/"
-    return 0
-  fi
-
-  DATE="$(cut -d' ' -f1 <<< "$DATETIME")"
-  TIME="$(cut -d' ' -f2 <<< "$DATETIME" | cut -d'+' -f1)"
-
-  YEAR="$(cut -d':' -f1 <<< "$DATE")"
-  MONTH="$(cut -d':' -f2 <<< "$DATE")"
-  DAY="$(cut -d':' -f3 <<< "$DATE")"
-  HOUR="$(cut -d':' -f1 <<< "$TIME")"
-  MINUTE="$(cut -d':' -f2 <<< "$TIME")"
-  SECOND="$(cut -d':' -f3 <<< "$TIME")"
-
-  HASH="$(sha256sum "$f" | awk '{print $1}' | xxd -r -p | base64 | head -c 6 | tr '/+' '_-')"
-  EXT="$(tr '[:upper:]' '[:lower:]' <<< "${f##*.}")"
-
-  if [[ "$EXT" == "cr2" || "$EXT" == "cr3" ]]; then
-    RAW="raw/"
-  else
-    RAW=""
-  fi
-
-  FILE="$ABS_DEST_PATH/$YEAR-$MONTH/${RAW}${YEAR}${MONTH}${DAY}-${HOUR}${MINUTE}${SECOND}_${HASH}.${EXT}"
-  echo "DESTINATION: $FILE"
-  mkdir -p "$(dirname "$FILE")"
-  mv -- "$f" "$FILE"
 }
 
-mapfile -d '' -t FILES < <(
+DATETIME_KEYS = [
-  find "$ABS_SOURCE_PATH" -type f \( \
+    ("Composite", "SubSecDateTimeOriginal"),
-    -iname '*.PNG' -o \
+    ("Composite", "SubSecCreateDate"),
-    -iname '*.JPG' -o \
+    ("ExifIFD", "DateTimeOriginal"),
-    -iname '*.JPEG' -o \
+    ("ExifIFD", "CreateDate"),
-    -iname '*.HEIC' -o \
+    ("XMP-xmp", "CreateDate"),
-    -iname '*.CR2' -o \
+    ("Keys", "CreationDate"),
-    -iname '*.CR3' -o \
+    ("QuickTime", "CreateDate"),
-    -iname '*.MP4' -o \
+    ("XMP-photoshop", "DateCreated"),
-    -iname '*.MOV' \
+]
-  \) -print0
-)
 
-if ((${#FILES[@]})); then
-  export -f process_file
-  export ABS_SOURCE_PATH ABS_DEST_PATH ABS_UNSORTABLE_PATH
 
-  printf '%s\0' "${FILES[@]}" |
+def run(command: list[str], check: bool = True) -> subprocess.CompletedProcess:
-    xargs -0 -n1 -P"$(nproc)" bash -c 'process_file "$1"' _
+    return subprocess.run(command, text=True, capture_output=True, check=check)
 
-  echo "SCANNING..."
-  chown -R www-data:www-data "$ABS_DEST_PATH"
-  chown -R www-data:www-data "$ABS_UNSORTABLE_PATH"
-  chmod -R 770 "$ABS_DEST_PATH"
-  chmod -R 770 "$ABS_UNSORTABLE_PATH"
-  sudo -u www-data php /opt/nextcloud/occ files:scan --path "$REL_SOURCE_PATH"
-  sudo -u www-data php /opt/nextcloud/occ files:scan --path "$REL_UNSORTABLE_PATH"
-  sudo -u www-data php /opt/nextcloud/occ files:scan --path "$REL_DEST_PATH"
-else
-  echo "NO MATCHING FILES FOUND."
-fi
 
-echo "FINISH."
+def exiftool_data(file: Path) -> dict | None:
+    result = run([
+        "exiftool",
+        "-j",
+        "-a",
+        "-u",
+        "-g1",
+        "-time:all",
+        "-api", "QuickTimeUTC=1",
+        "-d", "%Y-%m-%dT%H:%M:%S%z",
+        str(file),
+    ], check=False)
+    if result.returncode != 0:
+        return None
+    try:
+        data = __import__("json").loads(result.stdout)
+        return data[0] if data else None
+    except Exception:
+        return None
+
+
+def exiftool_timestamp(file: Path) -> datetime | None:
+    data = exiftool_data(file)
+    if not data:
+        return None
+
+    for category, key in DATETIME_KEYS:
+        try:
+            value = data[category][key]
+        except (KeyError, TypeError):
+            continue
+        try:
+            return datetime.strptime(value, "%Y-%m-%dT%H:%M:%S%z")
+        except ValueError:
+            continue
+
+    return None
+
+
+def short_hash(file: Path) -> str:
+    h = hashlib.sha256()
+    with file.open("rb") as fh:
+        for chunk in iter(lambda: fh.read(1024 * 1024), b""):
+            h.update(chunk)
+    digest = h.digest()
+    b64 = base64.b64encode(digest).decode("ascii")
+    return b64[:3].replace("/", "_").replace("+", "-")
+
+
+def build_destination(dest_root: Path, file: Path, ts: datetime) -> Path:
+    ext = file.suffix.lower().lstrip(".")
+    year = ts.strftime("%Y")
+    month = ts.strftime("%m")
+    day = ts.strftime("%d")
+    hour = ts.strftime("%H")
+    minute = ts.strftime("%M")
+    second = ts.strftime("%S")
+    hash_part = short_hash(file)
+
+    raw_subdir = "raw" if ext in {"cr2", "cr3"} else None
+    month_dir = dest_root / f"{year}-{month}"
+    if raw_subdir:
+        month_dir = month_dir / raw_subdir
+
+    filename = f"{year}{month}{day}-{hour}{minute}{second}_{hash_part}.{ext}"
+    return month_dir / filename
+
+
+def move_unsortable(file: Path, source_root: Path, unsortable_root: Path) -> None:
+    relpath = file.relative_to(source_root)
+    target_dir = (unsortable_root / relpath).parent
+    target_dir.mkdir(parents=True, exist_ok=True)
+    shutil.chown(str(target_dir), user="www-data", group="www-data")
+    target = target_dir / file.name
+    if target.exists():
+        return
+    shutil.move(str(file), str(target))
+    shutil.chown(str(target), user="www-data", group="www-data")
+
+
+def move_sorted(file: Path, target: Path) -> None:
+    target.parent.mkdir(parents=True, exist_ok=True)
+    shutil.chown(str(target.parent), user="www-data", group="www-data")
+    shutil.move(str(file), str(target))
+    shutil.chown(str(target), user="www-data", group="www-data")
+
+def process_file(file: Path, source_root: Path, dest_root: Path, unsortable_root: Path) -> tuple[Path, str]:
+    print(f"PROCESSING: {file}")
+    ts = exiftool_timestamp(file)
+
+    if ts is None:
+        print(f"UNSORTABLE: {file}")
+        move_unsortable(file, source_root, unsortable_root)
+        return file, "unsortable"
+
+    target = build_destination(dest_root, file, ts)
+    print(f"DESTINATION: {target}")
+    move_sorted(file, target)
+    return file, "sorted"
+
+
+def scan_nextcloud(rel_source: str, rel_unsortable: str, rel_dest: str) -> None:
+    print("SCANNING...")
+    # run(["chown", "-R", "www-data:www-data", abs_source_path], check=True)
+    # run(["chmod", "-R", "770", abs_source_path], check=True)
+
+    # run(["chown", "-R", "www-data:www-data", abs_dest_path], check=True)
+    # run(["chown", "-R", "www-data:www-data", abs_unsortable_path], check=True)
+    # run(["chmod", "-R", "770", abs_dest_path], check=True)
+    # run(["chmod", "-R", "770", abs_unsortable_path], check=True)
+
+    run(["sudo", "-u", "www-data", "php", "/opt/nextcloud/occ", "files:scan", "--path", rel_source], check=True)
+    run(["sudo", "-u", "www-data", "php", "/opt/nextcloud/occ", "files:scan", "--path", rel_unsortable], check=True)
+    run(["sudo", "-u", "www-data", "php", "/opt/nextcloud/occ", "files:scan", "--path", rel_dest], check=True)
+
+    run(["systemctl", "start", "nextcloud-generate-new-previews.service"], check=True)
+
+
+def iter_files(source_root: Path):
+    for path in source_root.rglob("*"):
+        if path.is_file() and path.suffix.lower() in ALLOWED_EXTS:
+            yield path
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(
+        description="Sort Nextcloud media files by embedded timestamp."
+    )
+    parser.add_argument("nc_user")
+    parser.add_argument("source_subdir")
+    parser.add_argument("dest_subdir")
+    parser.add_argument("unsortable_subdir")
+    parser.add_argument("--workers", type=int, default=os.cpu_count() or 1)
+    args = parser.parse_args()
+
+    nc_user = args.nc_user
+    source_subdir = args.source_subdir
+    dest_subdir = args.dest_subdir
+    unsortable_subdir = args.unsortable_subdir
+
+    rel_source_path = f"/{nc_user}/files/{source_subdir}"
+    abs_source_path = f"/var/lib/nextcloud/{nc_user}/files/{source_subdir}"
+
+    rel_dest_path = f"/{nc_user}/files/{dest_subdir}"
+    abs_dest_path = f"/var/lib/nextcloud/{nc_user}/files/{dest_subdir}"
+
+    rel_unsortable_path = f"/{nc_user}/files/{unsortable_subdir}"
+    abs_unsortable_path = f"/var/lib/nextcloud/{nc_user}/files/{unsortable_subdir}"
+
+    source_root = Path(abs_source_path)
+    dest_root = Path(abs_dest_path)
+    unsortable_root = Path(abs_unsortable_path)
+
+    print("STARTING...")
+
+    run(["chown", "-R", "www-data:www-data", str(source_root)], check=True)
+    run(["chmod", "-R", "770", str(source_root)], check=True)
+
+    files = list(iter_files(source_root))
+
+    if not files:
+        print("NO MATCHING FILES FOUND.")
+        print("FINISH.")
+        raise SystemExit(0)
+
+    with ThreadPoolExecutor(max_workers=max(1, args.workers)) as executor:
+        futures = {
+            executor.submit(process_file, file, source_root, dest_root, unsortable_root): file
+            for file in files
+        }
+        for future in as_completed(futures):
+            future.result()
+
+    scan_nextcloud(rel_source_path, rel_unsortable_path, rel_dest_path)
+
+    print("FINISH.")