Extract formatter input parsing policy

tools/prompt_smoke.py

@@ -29,6 +29,7 @@ import character_profile  # noqa: E402
 import category_cast_config  # noqa: E402
 import category_library  # noqa: E402
 import filter_config  # noqa: E402
+import formatter_input  # noqa: E402
 import hardcore_position_config  # noqa: E402
 import __init__ as sxcp_nodes  # noqa: E402
 import generation_profile_config  # noqa: E402
@@ -847,6 +848,57 @@ def smoke_row_normalization_policy() -> None:
     _expect_no_duplicate_comma_items("row_normalization.pair.hard_row_negative", pair["hardcore_row"].get("negative_prompt"))
 
 
+def smoke_formatter_input_policy() -> None:
+    source_row = {
+        "prompt": "A simple adult portrait. Setting: quiet studio. Pose: standing calmly. Avoid: low quality.",
+        "caption": "adult portrait, quiet studio",
+        "negative_prompt": "low quality",
+        "subject_type": "woman",
+        "primary_subject": "woman",
+        "age": "25-year-old adult",
+        "body_phrase": "average figure",
+        "skin": "warm skin",
+        "hair": "dark hair",
+        "eyes": "brown eyes",
+        "item": "black dress",
+        "scene_text": "quiet studio",
+        "pose": "standing calmly",
+        "composition": "centered portrait",
+        "trigger": Trigger,
+    }
+    source_json = _json(source_row)
+
+    row, method = formatter_input.row_from_inputs(source_json, "", "auto")
+    _expect(method == "source_json", "Formatter input parser should read source JSON when metadata is empty")
+    _expect(row == source_row, "Formatter input parser changed parsed JSON row")
+    _expect(formatter_input.split_avoid("Prompt body. Avoid: blur, watermark") == ("Prompt body", "blur, watermark"), "Avoid split changed")
+    _expect(
+        formatter_input.prompt_field(source_row["prompt"], "Setting") == "quiet studio",
+        "Prompt field extraction changed",
+    )
+    _expect(
+        formatter_input.row_value({"prompt": source_row["prompt"]}, "scene_text", ("Setting",)) == "quiet studio",
+        "Row value prompt fallback changed",
+    )
+
+    _expect(krea_formatter._clean("a  b , c") == formatter_input.clean_text("a  b , c"), "Krea clean helper is not delegated")
+    _expect(sdxl_formatter._clean("a  b , c") == formatter_input.clean_text("a  b , c"), "SDXL clean helper is not delegated")
+    _expect(caption_naturalizer._clean_text("a  b , c") == formatter_input.clean_text("a  b , c"), "Caption clean helper is not delegated")
+    _expect(krea_formatter._strip_trigger(f"{Trigger}, prompt text", False) == "prompt text", "Krea trigger stripping changed")
+    _expect(sdxl_formatter._strip_trigger(f"{SdxlTrigger}, prompt text", False) == "prompt text", "SDXL trigger stripping changed")
+    _expect(caption_naturalizer._remove_trigger(Trigger, Trigger) == "", "Caption exact-trigger removal changed")
+
+    krea = krea_formatter.format_krea2_prompt(source_json, input_hint="auto")
+    sdxl = sdxl_formatter.format_sdxl_prompt(source_json, input_hint="auto", trigger=SdxlTrigger, prepend_trigger=True)
+    caption, caption_method = caption_naturalizer.naturalize_caption(source_json, input_hint="auto", trigger=Trigger)
+    _expect(krea.get("method", "").startswith("source_json:krea2("), "Krea formatter did not use shared source JSON parsing")
+    _expect(sdxl.get("method", "").startswith("source_json:sdxl("), "SDXL formatter did not use shared source JSON parsing")
+    _expect(caption_method.startswith("source_json:metadata("), "Caption naturalizer did not use shared source JSON parsing")
+    _expect_text("formatter_input.krea_prompt", krea.get("krea_prompt"), 20)
+    _expect_text("formatter_input.sdxl_prompt", sdxl.get("sdxl_prompt"), 20)
+    _expect_text("formatter_input.caption", caption, 20)
+
+
 def smoke_hardcore_position_config_policy() -> None:
     _expect(
         pb.HARDCORE_POSITION_FAMILY_CHOICES is hardcore_position_config.HARDCORE_POSITION_FAMILY_CHOICES,
@@ -2818,6 +2870,7 @@ SMOKE_CASES: list[tuple[str, Callable[[], None]]] = [
     ("character_config_policy", smoke_character_config_policy),
     ("character_profile_policy", smoke_character_profile_policy),
     ("row_normalization_policy", smoke_row_normalization_policy),
+    ("formatter_input_policy", smoke_formatter_input_policy),
     ("hardcore_position_config_policy", smoke_hardcore_position_config_policy),
     ("category_library_route", smoke_category_library_route),
     ("hardcore_category_routes", smoke_hardcore_category_routes),