Test Release (#80)

* Fix typo in env

* Change model to gemma3

* Revert wardrobe prompt change

* Bump versions server

* Fix ollama multimodal issue #78

* Add batch update script

* ImageTool use png

* Change tool arg name

Author: otjamin

server/agents/outfit_crew/pyproject.toml

@@ -5,7 +5,7 @@ description = "outfit_crew using crewAI"
 authors = [{ name = "Your Name", email = "you@example.com" }]
 requires-python = ">=3.10,<3.14"
 dependencies = [
-    "crewai[tools]>=0.126.0,<1.0.0"
+    "crewai[tools]>=0.140.0,<1.0.0"
 ]
 
 [project.scripts]

server/agents/quiz_crew/pyproject.toml

@@ -5,7 +5,7 @@ description = "quiz_crew using crewAI"
 authors = [{ name = "Your Name", email = "you@example.com" }]
 requires-python = ">=3.10,<3.14"
 dependencies = [
-    "crewai[tools]>=0.126.0,<1.0.0"
+    "crewai[tools]>=0.140.0,<1.0.0"
 ]
 
 [project.scripts]

server/agents/trend_crew/pyproject.toml

@@ -5,7 +5,7 @@ description = "trend_crew using crewAI"
 authors = [{ name = "Your Name", email = "you@example.com" }]
 requires-python = ">=3.10,<3.14"
 dependencies = [
-    "crewai[tools]>=0.126.0,<1.0.0",
+    "crewai[tools]>=0.140.0,<1.0.0",
     "ollama>=0.5.1",
 ]
 

server/agents/wardrobe_crew/pyproject.toml

@@ -5,7 +5,7 @@ description = "wardrobe_crew using crewAI"
 authors = [{ name = "Your Name", email = "you@example.com" }]
 requires-python = ">=3.10,<3.14"
 dependencies = [
-    "crewai[tools]>=0.126.0,<1.0.0"
+    "crewai[tools]>=0.140.0,<1.0.0"
 ]
 
 [project.scripts]

server/agents/wardrobe_crew/src/wardrobe_crew/add_image_tool.py

@@ -0,0 +1,58 @@
+import base64
+import requests
+from io import BytesIO
+
+from PIL import Image
+
+from typing import Dict, Optional, Union
+
+from pydantic import BaseModel, Field
+
+from crewai.tools.base_tool import BaseTool
+
+
+class AddImageToolSchema(BaseModel):
+    image_url: str = Field(..., description="The URL of the image to add")
+    action: Optional[str] = Field(
+        default=None, description="Optional context or question about the image"
+    )
+
+
+class AddImageTool(BaseTool):
+    """Tool for adding images to the content"""
+
+    name: str = "Add image to content"
+    description: str = "See image to understand its content, you can optionally ask a question about the image"
+    args_schema: type[BaseModel] = AddImageToolSchema
+
+    def _get_base64_image(self, image_url: str) -> str:
+        response = requests.get(image_url)
+        buffered = BytesIO()
+        pil_image = Image.open(BytesIO(response.content))
+        pil_image.save(buffered, format="PNG")  # You can change the format if needed
+        img_str = base64.b64encode(buffered.getvalue()).decode("utf-8")
+        return img_str
+
+    def _run(
+        self,
+        image_url: str,
+        action: Optional[str] = None,
+        **kwargs,
+    ) -> dict:
+        base64_image = self._get_base64_image(image_url)
+
+        action = (
+            action
+            or "Please provide a detailed description of this image, including all visual elements, context, and any notable details you can observe."
+        )
+        content = [
+            {"type": "text", "text": action},
+            {
+                "type": "image_url",
+                "image_url": {
+                    "url": f"data:image/png;base64,{base64_image}",
+                },
+            },
+        ]
+
+        return {"role": "user", "content": content}

server/agents/wardrobe_crew/src/wardrobe_crew/crew.py

@@ -3,12 +3,17 @@
 from typing import List, Optional
 from pydantic import BaseModel
 
+from wardrobe_crew.add_image_tool import AddImageTool
+
 
 class TagOutput(BaseModel):
     tags: List[str]
     description: Optional[str] = None
 
 
+add_image_tool = AddImageTool()
+
+
 @CrewBase
 class WardrobeCrew:
     """Crew automatisch taggt Kleidung mit lokalem Ollama-LLM"""
@@ -23,17 +28,26 @@ def image_tagger(self) -> Agent:
                 "Material und Schnitt, spezialisiert auf visuelle Analyse."
             ),
             verbose=True,
-            multimodal=True,
+        )
+
+    @task
+    def image_analyzing_task(self) -> Task:
+        return Task(
+            description=(
+                "Analysiere das Bild eines Kleidungsstücks. Bild: {image_url}"
+            ),
+            expected_output="Eine Beschreibung des Bildes mit besonderem Fokus auf das Kleidungsstück, einschließlich Stil, Farbe, Material und Schnitt.",
+            agent=self.image_tagger(),
+            tools=[add_image_tool],
         )
 
     @task
     def tagging_task(self) -> Task:
         return Task(
             description=(
-                "Analysiere das Bild eines Kleidungsstücks hier {image_url} und extrahiere 5–8 "
-                "präzise Tags wie Stil, Farbe, Material und Schnitt."
+                "Finde 5-8 Tags, die das Kleidungsstück beschreiben, basierend auf der Analyse des Bildes. "
             ),
-            expected_output="Ein Dictionary mit 'tags' (Liste) und optional 'description'",
+            expected_output="Eine Liste von Tags, die das Kleidungsstück beschreiben.",
             agent=self.image_tagger(),
             output_pydantic=TagOutput,
         )
@@ -42,7 +56,7 @@ def tagging_task(self) -> Task:
     def crew(self) -> Crew:
         return Crew(
             agents=[self.image_tagger()],
-            tasks=[self.tagging_task()],
+            tasks=[self.image_analyzing_task(), self.tagging_task()],
             process=Process.sequential,
             verbose=True,
         )

server/agents/wardrobe_crew/src/wardrobe_crew/item_listener.py

@@ -13,16 +13,18 @@
 # Replace with inputs you want to test with, it will automatically
 # interpolate any tasks and agents information
 
+
 def run():
     """
     Run the crew.
     """
     inputs = {
-        'item_image': 'http://127.0.0.1:8090/api/files/items/xrfe7x38w1ddk95/cq5dam_web_he7e3da_1200_4vljrgza11.1200-1.avif'
+        "image_url": "http://127.0.0.1:8090/api/files/pbc_710432678/a6g0uob9w61fjsc/bikini_ix3phjsbiy.jpg"
     }
-    
+
     try:
-        WardrobeCrew().crew().kickoff(inputs=inputs)
+        result = WardrobeCrew().crew().kickoff(inputs=inputs)
+        print(result.pydantic)
     except Exception as e:
         raise Exception(f"An error occurred while running the crew: {e}")
 
@@ -31,16 +33,16 @@ def train():
     """
     Train the crew for a given number of iterations.
     """
-    inputs = {
-        "topic": "AI LLMs",
-        'current_year': str(datetime.now().year)
-    }
+    inputs = {"topic": "AI LLMs", "current_year": str(datetime.now().year)}
     try:
-        WardrobeCrew().crew().train(n_iterations=int(sys.argv[1]), filename=sys.argv[2], inputs=inputs)
+        WardrobeCrew().crew().train(
+            n_iterations=int(sys.argv[1]), filename=sys.argv[2], inputs=inputs
+        )
 
     except Exception as e:
         raise Exception(f"An error occurred while training the crew: {e}")
 
+
 def replay():
     """
     Replay the crew execution from a specific task.
@@ -51,17 +53,17 @@ def replay():
     except Exception as e:
         raise Exception(f"An error occurred while replaying the crew: {e}")
 
+
 def test():
     """
     Test the crew execution and returns the results.
     """
-    inputs = {
-        "topic": "AI LLMs",
-        "current_year": str(datetime.now().year)
-    }
-    
+    inputs = {"topic": "AI LLMs", "current_year": str(datetime.now().year)}
+
     try:
-        WardrobeCrew().crew().test(n_iterations=int(sys.argv[1]), eval_llm=sys.argv[2], inputs=inputs)
+        WardrobeCrew().crew().test(
+            n_iterations=int(sys.argv[1]), eval_llm=sys.argv[2], inputs=inputs
+        )
 
     except Exception as e:
         raise Exception(f"An error occurred while testing the crew: {e}")

server/agents/wardrobe_crew/src/wardrobe_crew/main.py

@@ -0,0 +1,30 @@
+import os
+import requests
+import json
+
+from wardrobe_crew.crew import WardrobeCrew
+
+POCKETBASE_URL = os.getenv("PB_URL", "http://localhost:8090")
+
+print(f"Using PocketBase URL: {POCKETBASE_URL}")
+
+
+def tag_item_image(image_url):
+    result = WardrobeCrew().crew().kickoff(inputs={"image_url": image_url})
+    tags = result.pydantic.tags
+    print(f"Tags for image {image_url}: {tags}")
+    return tags
+
+
+res = requests.get(f"{POCKETBASE_URL}/api/collections/items/records")
+items = res.json()["items"]
+for item in items:
+    if item["image"]:
+        image_url = f"{POCKETBASE_URL}/api/files/items/{item['id']}/{item['image']}"
+        tags = tag_item_image(image_url)
+        requests.patch(
+            f"{POCKETBASE_URL}/api/collections/items/records/{item['id']}",
+            data={"tags": json.dumps(tags)},
+        )
+    else:
+        print(f"Item {item['id']} has no image URL.")

server/agents/wardrobe_crew/src/wardrobe_crew/update_images.py

@@ -5,7 +5,7 @@ description = "Add your description here"
 readme = "README.md"
 requires-python = ">=3.13"
 dependencies = [
-    "fastapi[standard]>=0.115.12",
+    "fastapi[standard]>=0.115.14",
     "trend_crew",
     "wardrobe_crew",
     "outfit_crew",
@@ -15,15 +15,15 @@ dependencies = [
 
 [dependency-groups]
 dev = [
-    "ruff>=0.11.11",
+    "ruff>=0.12.2",
 ]
 
 [tool.uv.sources]
 trend_crew = { workspace = true }
 wardrobe_crew = { workspace = true }
 outfit_crew = { workspace = true }
 quiz_crew = { workspace = true }
-litellm = { git = "https://github.com/otjamin/litellm", tag = "v1.72.0-stable.1" }
+litellm = { git = "https://github.com/otjamin/litellm", tag = "v1.72.6.rc.1" }
 
 [tool.uv.workspace]
 members = ["agents/*"]