feat: enhance MetaHub chunk processing and add StylePromptEncoder2 node

metadata_utils.py

@@ -3,7 +3,7 @@ Compatibility shim for MetaHub Save Node metadata helpers.
 
 The implementation is kept in metadata_utils_impl.py; this module exposes the
 same public API expected by the save nodes while pinning the published node
-version to 1.1.3.
+version to 1.1.4.
 """
 
 try:
@@ -11,7 +11,7 @@ try:
 except ImportError:
     import metadata_utils_impl as _impl
 
-METAHUB_SAVE_NODE_VERSION = "1.1.3"
+METAHUB_SAVE_NODE_VERSION = "1.1.4"
 _impl.METAHUB_SAVE_NODE_VERSION = METAHUB_SAVE_NODE_VERSION
 
 for _name in dir(_impl):

pyproject.toml

@@ -1,7 +1,7 @@
 [project]
 name = "imagemetahub-comfyui-save"
 description = "Official companion node for Image MetaHub. Saves A1111/Civitai-compatible parameters plus extended Image MetaHub metadata (workflow JSON, SHA256 model hashes, VRAM peak/total, GPU device, generation time, and steps/sec) for reproducibility and benchmarking."
-version = "1.1.3"
+version = "1.1.4"
 license = {file = "LICENSE"} 
 # classifiers = [
 #     # For OS-independent nodes (works on all operating systems)

tests/test_workflow_extractor.py

@@ -135,6 +135,75 @@ def test_workflow_extractor_detects_lora_alternate_keys():
     assert data["lora_list"] == [{"name": "style.safetensors", "weight": 0.65}]
 
 
+def test_workflow_extractor_detects_rgthree_power_lora_objects():
+    prompt = {
+        "67": {
+            "class_type": "Power Lora Loader (rgthree)",
+            "inputs": {
+                "lora_1": {
+                    "on": True,
+                    "lora": "Z-Detail-Slider.safetensors",
+                    "strength": 0.5325,
+                },
+                "lora_2": {
+                    "on": True,
+                    "lora": "zy_CinematicShot_zit.safetensors",
+                    "strength": 0.6875,
+                },
+                "lora_3": {
+                    "on": False,
+                    "lora": "disabled.safetensors",
+                    "strength": 1.0,
+                },
+            },
+        }
+    }
+
+    extractor = WorkflowExtractor(prompt)
+    data, _missing = extractor.extract()
+
+    assert data["lora_list"] == [
+        {"name": "Z-Detail-Slider.safetensors", "weight": 0.5325},
+        {"name": "zy_CinematicShot_zit.safetensors", "weight": 0.6875},
+    ]
+
+
+def test_workflow_extractor_reads_style_prompt_encoder_through_basic_guider():
+    prompt = {
+        "61": {
+            "class_type": "SamplerCustomAdvanced",
+            "inputs": {
+                "noise": ["63", 0],
+                "guider": ["101", 0],
+                "sampler": ["108", 0],
+                "sigmas": ["62", 0],
+                "latent_image": ["113", 0],
+            },
+        },
+        "101": {
+            "class_type": "BasicGuider",
+            "inputs": {
+                "model": ["100", 0],
+                "conditioning": ["114", 0],
+            },
+        },
+        "114": {
+            "class_type": "StylePromptEncoder2 //ZImagePowerNodes",
+            "inputs": {
+                "style": "\"Production Photo\"",
+                "text": "A white cat on the roof of a brick house",
+                "clip": ["67", 1],
+            },
+        },
+    }
+
+    extractor = WorkflowExtractor(prompt)
+    data, missing = extractor.extract()
+
+    assert data["positive"] == "A white cat on the roof of a brick house"
+    assert "positive" not in missing
+
+
 def test_workflow_extractor_resolves_connected_seed_and_prompt_strings():
     prompt = {
         "1": {

workflow_extractor.py

@@ -274,8 +274,12 @@ class WorkflowExtractor:
             if stacked_loras:
                 loras.extend(stacked_loras)
                 continue
+            dynamic_loras = self._extract_dynamic_lora_entries(inputs)
+            if dynamic_loras:
+                loras.extend(dynamic_loras)
+                continue
             lora_name = self._get_first_literal(inputs, ("lora_name", "lora", "lora_name_1", "lora_1"))
-            if not lora_name:
+            if not isinstance(lora_name, str) or not lora_name.strip():
                 continue
             strength_model = self._get_first_literal(
                 inputs,
@@ -291,6 +295,28 @@ class WorkflowExtractor:
 
         return loras, has_lora_nodes
 
+    def _extract_dynamic_lora_entries(self, inputs: Dict[str, Any]) -> List[Dict[str, Any]]:
+        loras: List[Dict[str, Any]] = []
+        for key, raw_lora in inputs.items():
+            if not key.lower().startswith("lora_") or not isinstance(raw_lora, dict):
+                continue
+            if raw_lora.get("on") is False or raw_lora.get("active") is False:
+                continue
+            name = raw_lora.get("lora") or raw_lora.get("lora_name") or raw_lora.get("name")
+            if not isinstance(name, str) or not name.strip():
+                continue
+            weight = self._coerce_float(
+                raw_lora.get("strength")
+                if raw_lora.get("strength") is not None
+                else raw_lora.get("modelStrength")
+            )
+            if weight is None:
+                weight = self._coerce_float(raw_lora.get("strength_model"))
+            if weight is None:
+                weight = 1.0
+            loras.append({"name": name, "weight": float(weight)})
+        return loras
+
     def _extract_lora_manager_entries(self, inputs: Dict[str, Any]) -> List[Dict[str, Any]]:
         raw_loras = inputs.get("loras")
         if isinstance(raw_loras, dict) and isinstance(raw_loras.get("__value__"), list):
@@ -453,7 +479,7 @@ class WorkflowExtractor:
             class_type = self._class_type(node)
             if class_type == "ConditioningZeroOut":
                 return ""
-            if class_type in self.CLIP_NODES:
+            if self._is_prompt_encoder(class_type):
                 text = self._get_clip_text(node)
                 if text is not None:
                     texts.append(text)
@@ -506,7 +532,7 @@ class WorkflowExtractor:
         if class_type == "ConditioningZeroOut":
             return ""
 
-        if class_type in self.CLIP_NODES:
+        if self._is_prompt_encoder(class_type):
             return self._get_clip_text(node)
 
         if class_lower == "joinstrings" or "join strings" in class_lower:
@@ -663,3 +689,9 @@ class WorkflowExtractor:
         if class_type in cls.LORA_NODES:
             return True
         return "lora" in class_type.lower()
+
+    @classmethod
+    def _is_prompt_encoder(cls, class_type: str) -> bool:
+        if class_type in cls.CLIP_NODES:
+            return True
+        return "promptencoder" in class_type.lower().replace(" ", "")