fix: address image CLI review feedback

Author: ningzimu

CHANGELOG.md

@@ -4,6 +4,10 @@ Release notes are generated from this file. Keep changelog entries in English.
 
 ## Unreleased
 
+### Fixes
+
+- Teach `image_gen.py --prompt-file` to read prepared JSON slide jobs from their `prompt` field and restore image `output_compression` and `moderation` CLI options.
+
 ## 0.5.1
 
 ### Improvements

skills/codex-ppt/prompts/slide-worker.md

@@ -26,8 +26,8 @@ Strict input images already prepared by the parent:
 Read the JSON job file, then follow its `prompt` field exactly. Use the selected image backend and the recorded sample generation method only.
 You must produce the final slide candidate by calling the selected image generation backend:
 - Before calling the backend, visually inspect every local image listed in `style_reference_images` and `input_images`.
-- Local CLI mode for normal new slides: use `scripts/image_gen.py generate --prompt-file <job prompt file> --out <deck dir>/drafts/slide_<NN>_candidate.png`.
-- Local CLI mode for slides with strict `input_images`: use `scripts/image_gen.py edit --image <asset> ... --prompt-file <job prompt file> --out <deck dir>/drafts/slide_<NN>_candidate.png`.
+- Local CLI mode for normal new slides: use `scripts/image_gen.py generate --prompt-file <job prompt file> --out <deck dir>/drafts/slide_<NN>_candidate.png`; the CLI reads the JSON job file's `prompt` field.
+- Local CLI mode for slides with strict `input_images`: use `scripts/image_gen.py edit --image <asset> ... --prompt-file <job prompt file> --out <deck dir>/drafts/slide_<NN>_candidate.png`; the CLI reads the JSON job file's `prompt` field.
 - Local CLI mode for repairing an existing generated slide: use `scripts/image_gen.py edit --image <existing slide> --prompt ... --out <deck dir>/drafts/slide_<NN>_revised.png`.
 - Do not pass approved sample/style reference images as `--image` inputs unless the JSON job also lists them under `input_images`.
 - UI built-in mode is allowed only if the parent explicitly selected it and described how to save the result to the target path.

skills/codex-ppt/scripts/image_gen.py

@@ -269,7 +269,18 @@ def _read_prompt(prompt: Optional[str], prompt_file: Optional[str]) -> str:
         path = Path(prompt_file)
         if not path.exists():
             _die(f"Prompt file not found: {path}")
-        return path.read_text(encoding="utf-8").strip()
+        raw = path.read_text(encoding="utf-8").strip()
+        if path.suffix.lower() == ".json" or raw.startswith("{"):
+            try:
+                data = json.loads(raw)
+            except json.JSONDecodeError:
+                return raw
+            if isinstance(data, dict) and isinstance(data.get("prompt"), str):
+                prompt_text = data["prompt"].strip()
+                if prompt_text:
+                    return prompt_text
+                _die(f"Prompt field is empty in JSON prompt file: {path}")
+        return raw
     if prompt:
         return prompt.strip()
     _die("Missing prompt. Use --prompt or --prompt-file.")
@@ -407,6 +418,16 @@ def _validate_generate_payload(payload: Dict[str, Any]) -> None:
     _validate_model_specific_options(model=model, background=background)
 
 
+def _validate_output_compression(
+    output_compression: Optional[int],
+    output_format: Optional[str] = None,
+) -> None:
+    if output_compression is not None and not (0 <= output_compression <= 100):
+        _die("--output-compression must be between 0 and 100")
+    if output_compression is not None and output_format not in (None, "jpeg", "webp"):
+        _die("--output-compression requires --output-format jpeg or webp")
+
+
 def _build_output_paths(
     out: str,
     output_format: str,
@@ -682,6 +703,8 @@ async def _run_generate_batch(args: argparse.Namespace) -> int:
         "quality": args.quality,
         "background": args.background,
         "output_format": args.output_format,
+        "output_compression": args.output_compression,
+        "moderation": args.moderation,
     }
 
     provider = _create_provider(args)
@@ -704,6 +727,10 @@ async def _run_generate_batch(args: argparse.Namespace) -> int:
             _validate_generate_payload(job_payload)
             effective_output_format = _normalize_output_format(job_payload.get("output_format"))
             _validate_transparency(job_payload.get("background"), effective_output_format)
+            _validate_output_compression(
+                job_payload.get("output_compression"),
+                effective_output_format,
+            )
             job_payload["output_format"] = effective_output_format
 
             n = int(job_payload.get("n", 1))
@@ -758,6 +785,10 @@ async def run_job(i: int, job: Dict[str, Any]) -> Tuple[int, Optional[str]]:
         _validate_generate_payload(payload)
         effective_output_format = _normalize_output_format(payload.get("output_format"))
         _validate_transparency(payload.get("background"), effective_output_format)
+        _validate_output_compression(
+            payload.get("output_compression"),
+            effective_output_format,
+        )
         payload["output_format"] = effective_output_format
         outputs = _job_output_paths(
             out_dir=out_dir,
@@ -825,11 +856,14 @@ def _generate(args: argparse.Namespace) -> None:
         "quality": args.quality,
         "background": args.background,
         "output_format": args.output_format,
+        "output_compression": args.output_compression,
+        "moderation": args.moderation,
     }
     payload = {k: v for k, v in payload.items() if v is not None}
 
     output_format = _normalize_output_format(args.output_format)
     _validate_transparency(args.background, output_format)
+    _validate_output_compression(args.output_compression, output_format)
     payload["output_format"] = output_format
     output_paths = _build_output_paths(args.out, output_format, args.n, args.out_dir)
     downscaled = None
@@ -889,12 +923,15 @@ def _edit(args: argparse.Namespace) -> None:
         "quality": args.quality,
         "background": args.background,
         "output_format": args.output_format,
+        "output_compression": args.output_compression,
         "input_fidelity": args.input_fidelity,
+        "moderation": args.moderation,
     }
     payload = {k: v for k, v in payload.items() if v is not None}
 
     output_format = _normalize_output_format(args.output_format)
     _validate_transparency(args.background, output_format)
+    _validate_output_compression(args.output_compression, output_format)
     payload["output_format"] = output_format
     _validate_input_fidelity(args.input_fidelity)
     output_paths = _build_output_paths(args.out, output_format, args.n, args.out_dir)
@@ -953,6 +990,8 @@ def _add_shared_args(parser: argparse.ArgumentParser) -> None:
     parser.add_argument("--quality", default=DEFAULT_QUALITY)
     parser.add_argument("--background")
     parser.add_argument("--output-format")
+    parser.add_argument("--output-compression", type=int)
+    parser.add_argument("--moderation")
     parser.add_argument("--out", default=DEFAULT_OUTPUT_PATH)
     parser.add_argument("--out-dir")
     parser.add_argument("--force", action="store_true")
@@ -1023,6 +1062,7 @@ def main() -> int:
     _validate_size(args.size, args.model)
     _validate_quality(args.quality)
     _validate_background(args.background)
+    _validate_output_compression(args.output_compression)
     _validate_model_specific_options(
         model=args.model,
         background=args.background,

skills/codex-ppt/scripts/image_providers/codex_oauth.py

@@ -69,14 +69,22 @@ def _run_images_endpoint(self, payload: Dict[str, Any], image_paths: List[Path])
 
     def _images_body(self, payload: Dict[str, Any], image_paths: List[Path]) -> Dict[str, Any]:
         output_format = payload.get("output_format")
-        if isinstance(output_format, str) and output_format.lower() != "png":
-            raise RuntimeError("Codex OAuth images endpoint currently supports png output only.")
+        if isinstance(output_format, str) and output_format.lower() not in {"png", "jpeg", "webp"}:
+            raise RuntimeError("Codex OAuth images endpoint supports png, jpeg, or webp output.")
 
         body: Dict[str, Any] = {
             "prompt": str(payload["prompt"]),
             "model": str(payload["model"]),
         }
-        for key in ("n", "size", "quality", "background"):
+        for key in (
+            "n",
+            "size",
+            "quality",
+            "background",
+            "output_format",
+            "output_compression",
+            "moderation",
+        ):
             value = payload.get(key)
             if value is not None:
                 body[key] = value