Version 1.1.7 fügt PNG to JPEG Funktion hinzu. Ist das komprimierte PNG immer noch >500 KB wird versucht, dies als JPEG mit quality=90 zu konvertieren, das ist oft sehr viel besser komprimiert

- Re-raise worker futures in as_completed to surface thread exceptions
- Replace hardcoded extension set with ALLOWED_EXT constant in compress_with_caesium
- Initialise work_dir/scratch_dir to None before try block to prevent NameError in finally
- Remove unused dead function get_slide_numbers_for_image
- Simplify redundant caesium_threads guard (threads and threads > 1 -> threads > 1)
- Write [Content_Types].xml first in ZIP to satisfy OOXML spec

Co-Authored-By: Abacus.AI CLI <agent@abacus.ai>

.gitignore

@@ -2,4 +2,5 @@ python-3.*-embed-amd64.zip
 python-embed/*
 .vscode/launch.json
 logs/*.log
-__pycache__/*
+__pycache__/*
+.coverage

README.md

@@ -1,6 +1,6 @@
 
 # PPTX Image Compressor (CaesiumCLT only)
-**Version 1.1.4**
+**Version 1.1.7**
 
 Dieses Paket enthält:
 
@@ -11,8 +11,6 @@ PPTX-Image-Compressor/
 ├─ pptx_image_compress.py
 ├─ bin/
 │  └─ caesiumclt.exe
-└─ samples/
-   └─ README.txt
 ```
 
 ## Schnellstart (ohne Admin-Rechte)
@@ -34,13 +32,20 @@ Die Batch lädt bei Bedarf automatisch das **Windows Embeddable Python Package**
 - Entpackt die PPTX in einen Temp‑Ordner
 - Komprimiert **JPG/JPEG, PNG, WebP, GIF** mit **CaesiumCLT** (Default `-q 90`, `-O bigger`)
 - Ersetzt Bilder nur, wenn die komprimierte Datei kleiner ist
+- Versucht bei PNG zusätzlich einen PNG->JPG Wechsel, wenn das Bild nach Kompression noch größer als 500 KB ist
+- Ersetzt Bilder nur, wenn sei mindestens 2% kleiner sind (verhindert *doppelte Komprimierung*)
 - Schreibt ein CSV‑Log (`.log` neben der Output‑PPTX)
-- Baut eine neue PPTX und zeigt eine Summary (Name, Größe vorher/nachher, Ersparnis %, Zeit)
+   Baut eine neue PPTX und zeigt eine Summary (Name, Größe vorher/nachher, Ersparnis %, Zeit)
 - Räumt alle temporären Dateien auf (keine Caesium‑Tempfiles in der finalen PPTX)
 
+## Änderungen in 1.1.7
+- PNG->JPG Fallback für große PNGs (> 500 KB nach Kompression) hinzugefügt
+- CSV-Logging um `image_type_changed` erweitert (`png_jpg` bei Typwechsel)
+
 ## Hinweise
 - `-t` steuert die Parallelität der Python‑Threads; intern wird `caesiumclt --threads 1` gesetzt, sobald `-t > 1`, um Oversubscription zu vermeiden. Default ist 16
 -  `-q` steuert das Qualitätslevel; intern wird `caesiumclt -q` mit diesem Wert von `0..100` benutzt, Default ist 90
+-  `--min-savings` steuert das Mindestmass an Komprimierung zur Verhinderung von doppelter Komprimierunt, Default ist 2%
 - Die Batch **verwendet bevorzugt das Embeddable Python** neben der BAT; ansonsten sucht sie echte `python.exe`/`py.exe` im PATH, **ignoriert** aber die Microsoft‑Store‑Alias‑Pfade (`WindowsApps`).
 
 ## Manuelle Nutzung des .py (falls Python vorhanden)

install_and_run.bat

@@ -12,7 +12,7 @@ set "SELF_DIR=%~dp0"
 set "SCRIPT=%SELF_DIR%pptx_image_compress.py"
 
 rem ---- Python Embeddable config ----
-set "PY_EMBED_VERSION=3.14.4"
+set "PY_EMBED_VERSION=3.14.5"
 set "PY_EMBED_ZIP=python-%PY_EMBED_VERSION%-embed-amd64.zip"
 set "PY_EMBED_URL=https://www.python.org/ftp/python/%PY_EMBED_VERSION%/%PY_EMBED_ZIP%"
 set "PY_DIR=%SELF_DIR%python-embed"
@@ -23,7 +23,7 @@ set "CAE_DIR=%SELF_DIR%bin"
 set "CAE_EXE=caesiumclt.exe"
 if exist "%CAE_DIR%\%CAE_EXE%" (
     set "PATH=%CAE_DIR%;%PATH%"
-) else (
+) else (    
     if exist "%SELF_DIR%%CAE_EXE%" (
         set "PATH=%SELF_DIR%;%PATH%"
     ) else (
@@ -110,11 +110,14 @@ if not exist "%SCRIPT%" (
     exit /b 6
 )
 
+set "RUN_ARGS=%*"
+if "%~1"=="" set "RUN_ARGS=-h"
+
 echo.
 echo [%APP_NAME%] Starte ...
-echo Command: "%PY_CMD%" "%SCRIPT%" %*
+echo Command: "%PY_CMD%" "%SCRIPT%" %RUN_ARGS%
 echo.
-"%PY_CMD%" "%SCRIPT%" %*
+"%PY_CMD%" "%SCRIPT%" %RUN_ARGS%
 set "RC=%ERRORLEVEL%"
 
 echo.

latestversion.ini

@@ -1,6 +1,6 @@
 [DEFAULT]
-python_used_version = 3.14.4
-python_latest_version = 3.14.4
+python_used_version = 3.14.5
+python_latest_version = 3.14.5
 caesiumclt_used_version = v1.3.0
 caesiumclt_latest_version = v1.3.0
 

pptx_image_compress.py

@@ -2,21 +2,26 @@
 # -*- coding: utf-8 -*-
 """
 PPTX Grafik-Komprimier-Tool (nur CaesiumCLT, Multi-Thread, Batch, sauberes Cleanup)
-Version: 1.1.6
+Version: 1.1.7
 
 
 Highlights:
 - Caesium-Scratch außerhalb des PPTX-Arbeitsverzeichnisses -> keine Tempfiles in finaler PPTX
 - Safety-Cleanup: entfernt 'caesium*' Ordner und '*.tmp' in ppt/media, bevor gezippt wird
 - Overwrite Policy: -O bigger
-- Log: image_name,size_before,size_after,saving,saving_percent
+- Log: image_name,size_before,size_after,saving,saving_percent,in_slide_number,image_type_changed
 - Summary inkl. Zeit benötigt
 
+Änderungen in 1.1.7:
+- PNG->JPG Fallback für große PNGs hinzugefügt (wenn nach Kompression weiterhin > 500 KB)
+- Logging erweitert: neue Spalte image_type_changed mit Wert png_jpg bei Typwechsel
+
 Änderungen in 1.1.6:
 - Libcaesium 1.3.0 kann nun auch files ignorieren, wenn die Kompression kleiner als <MIN_SAVING> ist
 """
 
 import argparse
+import inspect
 import os
 import re
 import xml.etree.ElementTree as ET
@@ -36,12 +41,14 @@ from dataclasses import dataclass
 from typing import Callable, List, Optional
 
 
-__version__ = "1.1.6"
+
+__version__ = "1.1.7"
 
 ALLOWED_EXT = {".jpg", ".jpeg", ".png", ".webp", ".gif"}
 PROGRESS_BAR_LEN = 40
 TEMP_PREFIX = "pptx_compress_"
 DEFAULT_MIN_SAVINGS = "2%"
+PNG_TO_JPEG_THRESHOLD_BYTES = 500 * 1024
 
 
 @dataclass
@@ -62,6 +69,7 @@ class ImageProcessResult:
     orig_size: int
     chosen_size: int
     slide_nr: str
+    image_type_changed: str = ""
 
 
 def discover_images(media_dir: Path) -> list[Path]:
@@ -76,7 +84,7 @@ def discover_images(media_dir: Path) -> list[Path]:
 def image_result_to_log_line(image_result: ImageProcessResult) -> str:
     saving = image_result.orig_size - image_result.chosen_size
     saving_percent = round((saving / image_result.orig_size) * 100, 2) if image_result.orig_size > 0 else 0.0
-    return f"{image_result.image_name};{human_kb(image_result.orig_size)};{human_kb(image_result.chosen_size)};{human_kb(saving)};{saving_percent};{image_result.slide_nr}\n"
+    return f"{image_result.image_name};{human_kb(image_result.orig_size)};{human_kb(image_result.chosen_size)};{human_kb(saving)};{saving_percent};{image_result.slide_nr};{image_result.image_type_changed}\n"
 
 
 # -------------------- Utilities --------------------
@@ -116,25 +124,50 @@ def print_progress(i: int, total: int):
     print(f"\rBilder: |{bar}| {i}/{total} ({pct}%)", end="", flush=True)
 
 def zip_dir_to_pptx(src_dir: Path, out_pptx: Path):
+    all_files: list[Path] = []
+    for root, _, files in os.walk(src_dir):
+        for f in files:
+            all_files.append(Path(root) / f)
+
+    content_types = [f for f in all_files if f.name == "[Content_Types].xml"]
+    rest = [f for f in all_files if f.name != "[Content_Types].xml"]
+
     with zipfile.ZipFile(out_pptx, "w", compression=zipfile.ZIP_DEFLATED) as z:
-        for root, _, files in os.walk(src_dir):
-            for f in files:
-                full = Path(root) / f
-                rel = full.relative_to(src_dir)
-                z.write(full, arcname=str(rel))
+        for full in content_types + rest:
+            rel = full.relative_to(src_dir)
+            z.write(full, arcname=str(rel))
 
 def which(cmd: str):
     return shutil.which(cmd)
 
-def compress_with_caesium(original: Path, out_dir: Path, caesium_threads: int | None, quality: int, min_savings: str) -> Path | None:
+def compress_with_caesium(
+    original: Path,
+    out_dir: Path,
+    caesium_threads: int | None,
+    quality: int,
+    min_savings: str,
+    output_format: str = "original",
+) -> Path | None:
     exe = which("caesiumclt")
     if not exe:
         raise RuntimeError("[ERROR] 'caesiumclt' wurde nicht gefunden. Bitte CaesiumCLT installieren und in PATH verfügbar machen.")
     out_dir.mkdir(parents=True, exist_ok=True)
     ext = original.suffix.lower()
-    if ext not in {".jpg", ".jpeg", ".png", ".webp", ".gif"}:
+    if ext not in ALLOWED_EXT:
         return None
-    cmd = [exe, "-q", str(quality), "-O", "bigger", "--min-savings", min_savings, "-o", str(out_dir)]
+    cmd = [
+        exe,
+        "-q",
+        str(quality),
+        "-O",
+        "bigger",
+        "--min-savings",
+        min_savings,
+        "--format",
+        output_format,
+        "-o",
+        str(out_dir),
+    ]
     if caesium_threads is not None:
         cmd += ["--threads", str(caesium_threads)]
     cmd += [str(original)]
@@ -143,12 +176,94 @@ def compress_with_caesium(original: Path, out_dir: Path, caesium_threads: int |
         if r.returncode != 0:
             sys.stderr.write(f"[caesiumclt] Fehler bei {original.name}:{r.stderr}")
             return None
+        if output_format == "jpeg":
+            jpg_out = out_dir / f"{original.stem}.jpg"
+            jpeg_out = out_dir / f"{original.stem}.jpeg"
+            if jpg_out.exists():
+                return jpg_out
+            if jpeg_out.exists():
+                return jpeg_out
+            return None
         out_file = out_dir / original.name
         return out_file if out_file.exists() else None
     except Exception as ex:
         sys.stderr.write(f"[caesiumclt] Ausnahme bei {original.name}: {ex}")
         return None
 
+def compressor_accepts_output_format(compressor: Callable[..., Path | None]) -> bool:
+    if compressor is compress_with_caesium:
+        return True
+    try:
+        signature = inspect.signature(compressor)
+    except (TypeError, ValueError):
+        return False
+    return "output_format" in signature.parameters
+
+
+def run_compressor(
+    compressor: Callable[..., Path | None],
+    original: Path,
+    out_dir: Path,
+    caesium_threads: int | None,
+    quality: int,
+    min_savings: str,
+    output_format: str = "original",
+) -> Path | None:
+    if output_format != "original" and not compressor_accepts_output_format(compressor):
+        return None
+    if compressor is compress_with_caesium:
+        return compressor(original, out_dir, caesium_threads, quality, min_savings, output_format)
+    if compressor_accepts_output_format(compressor):
+        return compressor(original, out_dir, caesium_threads, quality, min_savings, output_format)
+    return compressor(original, out_dir, caesium_threads, quality, min_savings)
+
+
+def update_relationship_targets(work_dir: Path, old_name: str, new_name: str) -> None:
+    rels_namespace = "{http://schemas.openxmlformats.org/package/2006/relationships}Relationship"
+    for rels_file in work_dir.rglob("*.rels"):
+        try:
+            tree = ET.parse(rels_file)
+            root = tree.getroot()
+            changed = False
+            for rel in root.findall(f".//{rels_namespace}"):
+                target = rel.attrib.get("Target", "")
+                if Path(target).name == old_name:
+                    rel.attrib["Target"] = re.sub(r"[^/\\]+$", new_name, target)
+                    changed = True
+            if changed:
+                tree.write(rels_file, encoding="utf-8", xml_declaration=True)
+        except (ET.ParseError, OSError):
+            continue
+
+
+def ensure_jpg_content_type(work_dir: Path) -> None:
+    content_types_path = work_dir / "[Content_Types].xml"
+    if not content_types_path.exists():
+        return
+    content_ns = "{http://schemas.openxmlformats.org/package/2006/content-types}"
+    try:
+        tree = ET.parse(content_types_path)
+        root = tree.getroot()
+        has_jpg_default = False
+        for default in root.findall(f"{content_ns}Default"):
+            ext = default.attrib.get("Extension", "").lower()
+            if ext == "jpg":
+                has_jpg_default = True
+                break
+        if not has_jpg_default:
+            ET.SubElement(
+                root,
+                f"{content_ns}Default",
+                {
+                    "Extension": "jpg",
+                    "ContentType": "image/jpeg",
+                },
+            )
+            tree.write(content_types_path, encoding="utf-8", xml_declaration=True)
+    except (ET.ParseError, OSError):
+        return
+
+
 def format_duration(seconds: float) -> str:
     total_ms = int(round(seconds * 1000))
     td = timedelta(milliseconds=total_ms)
@@ -187,12 +302,6 @@ def build_image_slide_index(rels_dir: Path) -> dict[str, List[int]]:
     return {img: sorted(slides) for img, slides in image_to_slides.items()}
 
 
-def get_slide_numbers_for_image(rels_dir: Path, image_filename: str) -> Optional[List[int]]:
-    image_to_slides = build_image_slide_index(rels_dir)
-    slides = image_to_slides.get(image_filename)
-    return slides if slides else None
-
-
 def process_image_file(
     idx: int,
     img_path: Path,
@@ -201,34 +310,69 @@ def process_image_file(
     caesium_threads: int | None,
     quality: int,
     min_savings: str,
-    compressor: Callable[[Path, Path, int | None, int, str], Path | None],
+    compressor: Callable[..., Path | None],
 ) -> ImageProcessResult:
     orig_size = img_path.stat().st_size
     chosen_size = orig_size
+    chosen_name = img_path.name
+    image_type_changed = ""
     found_in_slide = image_to_slides.get(img_path.name)
     slide_nr = "NOT_USED" if found_in_slide is None else str(found_in_slide)
 
     try:
         out_sub = scratch_dir / f"img_{idx:06d}"
-        caesium_out = compressor(img_path, out_sub, caesium_threads, quality, min_savings)
+        caesium_out = run_compressor(
+            compressor=compressor,
+            original=img_path,
+            out_dir=out_sub,
+            caesium_threads=caesium_threads,
+            quality=quality,
+            min_savings=min_savings,
+        )
         if caesium_out and caesium_out.exists():
-            s = caesium_out.stat().st_size
-            if s < orig_size:
+            compressed_size = caesium_out.stat().st_size
+            if compressed_size < orig_size:
                 tmp_target = img_path.with_suffix(img_path.suffix + ".tmp")
                 shutil.copy2(caesium_out, tmp_target)
                 tmp_target.replace(img_path)
-                chosen_size = s
+                chosen_size = compressed_size
+
+        if img_path.suffix.lower() == ".png" and chosen_size > PNG_TO_JPEG_THRESHOLD_BYTES:
+            jpg_candidate = run_compressor(
+                compressor=compressor,
+                original=img_path,
+                out_dir=out_sub,
+                caesium_threads=caesium_threads,
+                quality=quality,
+                min_savings="0%",
+                output_format="jpeg",
+            )
+            if jpg_candidate and jpg_candidate.exists():
+                jpg_size = jpg_candidate.stat().st_size
+                if jpg_size < chosen_size:
+                    jpg_target = img_path.with_suffix(".jpg")
+                    tmp_target = jpg_target.with_suffix(".jpg.tmp")
+                    shutil.copy2(jpg_candidate, tmp_target)
+                    tmp_target.replace(jpg_target)
+                    img_path.unlink(missing_ok=True)
+                    chosen_size = jpg_size
+                    chosen_name = jpg_target.name
+                    image_type_changed = "png_jpg"
     except Exception:
         chosen_size = orig_size
+        chosen_name = img_path.name
+        image_type_changed = ""
 
     return ImageProcessResult(
-        image_name=img_path.name,
+        image_name=chosen_name,
         orig_size=orig_size,
         chosen_size=chosen_size,
         slide_nr=slide_nr,
+        image_type_changed=image_type_changed,
     )
 
 
+
 # -------------------- Core per-deck processing --------------------
 def process_single_deck(
     input_pptx: Path,
@@ -236,13 +380,15 @@ def process_single_deck(
     threads: int,
     quality: int,
     min_savings: str,
-    compressor: Callable[[Path, Path, int | None, int, str], Path | None] = compress_with_caesium,
+    compressor: Callable[..., Path | None] = compress_with_caesium,
 ) -> DeckResult:
     start_time = time.perf_counter()
     result = DeckResult(
         input=str(input_pptx),
         output=str(output_pptx),
     )
+    work_dir: Optional[Path] = None
+    scratch_dir: Optional[Path] = None
 
     try:
         if not input_pptx.exists() or input_pptx.suffix.lower() != ".pptx":
@@ -253,10 +399,9 @@ def process_single_deck(
 
         work_dir = Path(tempfile.mkdtemp(prefix=TEMP_PREFIX + "work_"))
         scratch_dir = Path(tempfile.mkdtemp(prefix=TEMP_PREFIX + "scratch_"))
-
         log_file = output_pptx.with_suffix(".log.csv")
         ensure_clean_file(log_file)
-        log_lines = ["image_name;size_before(kb);size_after(kb);saving(kb);saving_percent(%);in_slide_number\n"]
+        log_lines = ["image_name;size_before(kb);size_after(kb);saving(kb);saving_percent(%);in_slide_number;image_type_changed\n"]
 
         size_before = input_pptx.stat().st_size
         result.size_before = size_before
@@ -277,10 +422,11 @@ def process_single_deck(
         if not which("caesiumclt") and compressor is compress_with_caesium:
             raise RuntimeError("'caesiumclt' nicht gefunden. Bitte installieren und in PATH verfügbar machen.")
 
-        caesium_threads = 1 if threads and threads > 1 else None
+        caesium_threads = 1 if threads > 1 else None
         lock = Lock()
         done_count = 0
         image_to_slides = build_image_slide_index(rels_dir)
+        renamed_images: list[tuple[str, str]] = []
 
         def worker(idx: int, img_path: Path):
             nonlocal done_count
@@ -296,6 +442,8 @@ def process_single_deck(
             )
 
             with lock:
+                if image_result.image_name != img_path.name:
+                    renamed_images.append((img_path.name, image_result.image_name))
                 log_lines.append(image_result_to_log_line(image_result))
                 done_count += 1
                 print_progress(done_count, total)
@@ -303,8 +451,16 @@ def process_single_deck(
         if total > 0:
             with ThreadPoolExecutor(max_workers=max(1, threads)) as ex:
                 futures = [ex.submit(worker, i, p) for i, p in enumerate(images, start=1)]
-                for _ in as_completed(futures):
-                    pass
+                for fut in as_completed(futures):
+                    try:
+                        fut.result()
+                    except Exception as exc:
+                        sys.stderr.write(f"[worker] Unerwarteter Fehler: {exc}\n")
+
+        if renamed_images:
+            for old_name, new_name in renamed_images:
+                update_relationship_targets(work_dir, old_name, new_name)
+            ensure_jpg_content_type(work_dir)
 
         print()  # newline
 
@@ -351,14 +507,10 @@ def process_single_deck(
     except Exception as e:
         result.error = str(e)
     finally:
-        try:
-            shutil.rmtree(work_dir, ignore_errors=True)  # type: ignore[name-defined]
-        except Exception:
-            pass
-        try:
-            shutil.rmtree(scratch_dir, ignore_errors=True)  # type: ignore[name-defined]
-        except Exception:
-            pass
+        if work_dir is not None:
+            shutil.rmtree(work_dir, ignore_errors=True)
+        if scratch_dir is not None:
+            shutil.rmtree(scratch_dir, ignore_errors=True)
         cleanup_old_temps()
 
     return result
@@ -412,39 +564,8 @@ def collect_from_dir(input_dir: Path, pattern: str, recursive: bool) -> list[Pat
 
 # -------------------- CLI --------------------
 def main():
-    parser = argparse.ArgumentParser(
-        description="PPTX Grafik-Komprimier-Tool (nur CaesiumCLT, Multi-Thread, Batch, sauberes Cleanup)",
-        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
-    )
-    parser.add_argument('-i','--input', nargs='*', help='Input-PPTX (eine oder mehrere, Wildcards erlaubt). Bei mehreren: -O erforderlich.')
-    parser.add_argument('--input-dir', help='Eingabe-Verzeichnis (optional, für Batch)')
-    parser.add_argument('-o','--output', help='Output-PPTX (nur Single-Mode)')
-    parser.add_argument('-O','--output-dir', help='Output-Verzeichnis (erforderlich für Batch)')
-    parser.add_argument('--pattern', default='*.pptx', help='Dateimuster für --input-dir')
-    parser.add_argument('--recursive', action='store_true', help='Rekursiv in --input-dir suchen')
-    #parser.add_argument('-t','--threads', type=int, default=min(32, os.cpu_count() or 4), help='Anzahl paralleler Threads pro Datei')
-    parser.add_argument('-t','--threads', type=int, default=16, help='Anzahl paralleler Threads pro Datei')
-    parser.add_argument('-q','--quality', type=int, default=90, help='Qualität für caesiumclt (0..100), höher = bessere Qualität / größere Datei')
-    parser.add_argument('--min-savings', default=DEFAULT_MIN_SAVINGS, help="Mindestersparnis für caesiumclt (z. B. 2%%, 100KB, 1MB oder Bytes als Zahl)")
-    parser.add_argument('--version', action='version', version=f'%(prog)s {__version__}',  help="Zeigt die Versionsnummer an" )
-    
-    args = parser.parse_args()
-
-    print("Threads used:     ", args.threads," Threads")
-
-    if args.quality < 0 or args.quality > 100:
-        print('[ERROR] Ungültige Qualität. Erlaubt: 0..100')
-        sys.exit(1)
-
-    input_files: list[Path] = []
-    if args.input:
-        input_files.extend(expand_inputs(args.input))
-    if args.input_dir:
-        input_files.extend(collect_from_dir(Path(args.input_dir), args.pattern, args.recursive))
-
-    if len(input_files) == 0:
-        parser.print_help()
-        sys.exit(1)
+    parser, args = extractParserArguments()
+    input_files = validateParserArguments(parser, args)
 
     batch_mode = len(input_files) > 1
 
@@ -511,6 +632,44 @@ def main():
         print(f"Gesamtgröße nachher: {human_mb(overall_after)} MB")
         print(f"Gesamt-Ersparnis:    {pct}%")
 
+def validateParserArguments(parser, args):
+    print("Threads used:     ", args.threads," Threads")
+
+    if args.quality < 0 or args.quality > 100:
+        print('[ERROR] Ungültige Qualität. Erlaubt: 0..100')
+        sys.exit(1)
+
+    input_files: list[Path] = []
+    if args.input:
+        input_files.extend(expand_inputs(args.input))
+    if args.input_dir:
+        input_files.extend(collect_from_dir(Path(args.input_dir), args.pattern, args.recursive))
+
+    if len(input_files) == 0:
+        parser.print_help()
+        sys.exit(1)
+    return input_files
+
+def extractParserArguments():
+    parser = argparse.ArgumentParser(
+        description="PPTX Grafik-Komprimier-Tool (nur CaesiumCLT, Multi-Thread, Batch, sauberes Cleanup)",
+        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
+    )
+    parser.add_argument('-i','--input', nargs='*', help='Input-PPTX (eine oder mehrere, Wildcards erlaubt). Bei mehreren: -O erforderlich.')
+    parser.add_argument('--input-dir', help='Eingabe-Verzeichnis (optional, für Batch)')
+    parser.add_argument('-o','--output', help='Output-PPTX (nur Single-Mode)')
+    parser.add_argument('-O','--output-dir', help='Output-Verzeichnis (erforderlich für Batch)')
+    parser.add_argument('--pattern', default='*.pptx', help='Dateimuster für --input-dir')
+    parser.add_argument('--recursive', action='store_true', help='Rekursiv in --input-dir suchen')
+    #parser.add_argument('-t','--threads', type=int, default=min(32, os.cpu_count() or 4), help='Anzahl paralleler Threads pro Datei')
+    parser.add_argument('-t','--threads', type=int, default=16, help='Anzahl paralleler Threads pro Datei')
+    parser.add_argument('-q','--quality', type=int, default=90, help='Qualität für caesiumclt (0..100), höher = bessere Qualität / größere Datei')
+    parser.add_argument('--min-savings', default=DEFAULT_MIN_SAVINGS, help="Mindestersparnis für caesiumclt (z. B. 2%%, 100KB, 1MB oder Bytes als Zahl)")
+    parser.add_argument('--version', action='version', version=f'%(prog)s {__version__}',  help="Zeigt die Versionsnummer an" )
+    
+    args = parser.parse_args()
+    return parser,args
+
 
 
 

test_pptx_image_compress.py

@@ -23,11 +23,13 @@ class TestPptxImageCompress(unittest.TestCase):
             orig_size=1000,
             chosen_size=800,
             slide_nr="[1, 2]",
+            image_type_changed="",
         )
         line = pic.image_result_to_log_line(image_result)
         self.assertIn("image1.png", line)
         self.assertIn("[1, 2]", line)
         self.assertIn("20.0", line)
+        self.assertTrue(line.endswith(";\n"))
 
     def test_process_image_file_replaces_when_smaller(self):
         with tempfile.TemporaryDirectory() as td:
@@ -57,6 +59,40 @@ class TestPptxImageCompress(unittest.TestCase):
             self.assertEqual(img.stat().st_size, 40)
             self.assertEqual(result.slide_nr, "[1]")
 
+    def test_process_image_file_converts_large_png_to_jpg(self):
+        with tempfile.TemporaryDirectory() as td:
+            root = Path(td)
+            img = root / "image1.png"
+            img.write_bytes(b"A" * 800000)
+            scratch = root / "scratch"
+
+            def fake_compressor(original: Path, out_dir: Path, caesium_threads: int | None, quality: int, min_savings: str, output_format: str = "original"):
+                out_dir.mkdir(parents=True, exist_ok=True)
+                if output_format == "jpeg":
+                    out = out_dir / "image1.jpg"
+                    out.write_bytes(b"C" * 200000)
+                    return out
+                out = out_dir / original.name
+                out.write_bytes(b"B" * 700000)
+                return out
+
+            result = pic.process_image_file(
+                idx=1,
+                img_path=img,
+                scratch_dir=scratch,
+                image_to_slides={"image1.png": [2]},
+                caesium_threads=1,
+                quality=90,
+                min_savings="2%",
+                compressor=fake_compressor,
+            )
+
+            self.assertEqual(result.image_name, "image1.jpg")
+            self.assertEqual(result.chosen_size, 200000)
+            self.assertEqual(result.image_type_changed, "png_jpg")
+            self.assertFalse(img.exists())
+            self.assertTrue((root / "image1.jpg").exists())
+
     def test_process_image_file_keeps_original_when_bigger(self):
         with tempfile.TemporaryDirectory() as td:
             root = Path(td)
@@ -84,6 +120,7 @@ class TestPptxImageCompress(unittest.TestCase):
             self.assertEqual(result.chosen_size, 100)
             self.assertEqual(img.stat().st_size, 100)
             self.assertEqual(result.slide_nr, "NOT_USED")
+            self.assertEqual(result.image_type_changed, "")
 
     def test_process_single_deck_with_injected_compressor(self):
         with tempfile.TemporaryDirectory() as td:
@@ -136,8 +173,75 @@ class TestPptxImageCompress(unittest.TestCase):
             if log_file is None:
                 self.fail("log_file should not be None")
             log_text = Path(log_file).read_text(encoding="utf-8")
+            self.assertIn("image_name;size_before(kb);size_after(kb);saving(kb);saving_percent(%);in_slide_number;image_type_changed", log_text)
             self.assertIn("image1.png", log_text)
             self.assertIn("[1]", log_text)
+    def test_process_single_deck_updates_rels_on_png_to_jpg(self):
+        with tempfile.TemporaryDirectory() as td:
+            root = Path(td)
+            input_pptx = root / "input_convert.pptx"
+            output_pptx = root / "output_convert.pptx"
+            source_tree = root / "src_convert"
+            rels_dir = source_tree / "ppt" / "slides" / "_rels"
+            media_dir = source_tree / "ppt" / "media"
+            rels_dir.mkdir(parents=True, exist_ok=True)
+            media_dir.mkdir(parents=True, exist_ok=True)
+
+            rels_xml = (
+                "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"yes\"?>"
+                "<Relationships xmlns=\"http://schemas.openxmlformats.org/package/2006/relationships\">"
+                "<Relationship Id=\"rId2\" Type=\"http://schemas.openxmlformats.org/officeDocument/2006/relationships/image\" Target=\"../media/image1.png\"/>"
+                "</Relationships>"
+            )
+            content_types_xml = (
+                "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"yes\"?>"
+                "<Types xmlns=\"http://schemas.openxmlformats.org/package/2006/content-types\">"
+                "<Default Extension=\"rels\" ContentType=\"application/vnd.openxmlformats-package.relationships+xml\"/>"
+                "<Default Extension=\"xml\" ContentType=\"application/xml\"/>"
+                "<Default Extension=\"png\" ContentType=\"image/png\"/>"
+                "</Types>"
+            )
+            (source_tree / "[Content_Types].xml").write_text(content_types_xml, encoding="utf-8")
+            (rels_dir / "slide1.xml.rels").write_text(rels_xml, encoding="utf-8")
+            (media_dir / "image1.png").write_bytes(b"A" * 800000)
+
+            with zipfile.ZipFile(input_pptx, "w", compression=zipfile.ZIP_DEFLATED) as z:
+                for p in source_tree.rglob("*"):
+                    if p.is_file():
+                        z.write(p, arcname=str(p.relative_to(source_tree)))
+
+            def fake_compressor(original: Path, out_dir: Path, caesium_threads: int | None, quality: int, min_savings: str, output_format: str = "original"):
+                out_dir.mkdir(parents=True, exist_ok=True)
+                if output_format == "jpeg":
+                    out = out_dir / "image1.jpg"
+                    out.write_bytes(b"C" * 200000)
+                    return out
+                out = out_dir / original.name
+                out.write_bytes(b"B" * 700000)
+                return out
+
+            result = pic.process_single_deck(
+                input_pptx=input_pptx,
+                output_pptx=output_pptx,
+                threads=1,
+                quality=90,
+                min_savings="2%",
+                compressor=fake_compressor,
+            )
+
+            self.assertTrue(result.ok)
+            with zipfile.ZipFile(output_pptx, "r") as z:
+                self.assertIn("ppt/media/image1.jpg", z.namelist())
+                self.assertNotIn("ppt/media/image1.png", z.namelist())
+                rels_out = z.read("ppt/slides/_rels/slide1.xml.rels").decode("utf-8")
+                content_types_out = z.read("[Content_Types].xml").decode("utf-8")
+            self.assertIn("image1.jpg", rels_out)
+            self.assertIn("Extension=\"jpg\"", content_types_out)
+            log_file = result.log_file
+            if log_file is None:
+                self.fail("log_file should not be None")
+            log_text = Path(log_file).read_text(encoding="utf-8")
+            self.assertIn("png_jpg", log_text)
 
 
 if __name__ == "__main__":