camel-ai 0.2.69a1__py3-none-any.whl → 0.2.69a2__py3-none-any.whl

camel/__init__.py

@@ -14,7 +14,7 @@
 
 from camel.logger import disable_logging, enable_logging, set_log_level
 
-__version__ = '0.2.69a1'
+__version__ = '0.2.69a2'
 
 __all__ = [
     '__version__',

camel/societies/workforce/single_agent_worker.py

@@ -286,8 +286,8 @@ class SingleAgentWorker(Worker):
             )
         except Exception as e:
             print(
-                f"{Fore.RED}Error occurred while processing task {task.id}:"
-                f"\n{e}{Fore.RESET}"
+                f"{Fore.RED}Error processing task {task.id}: "
+                f"{type(e).__name__}: {e}{Fore.RESET}"
             )
             return TaskState.FAILED
         finally:
@@ -336,8 +336,15 @@ class SingleAgentWorker(Worker):
 
         print(f"======\n{Fore.GREEN}Reply from {self}:{Fore.RESET}")
 
-        result_dict = json.loads(response.msg.content)
-        task_result = TaskResult(**result_dict)
+        try:
+            result_dict = json.loads(response.msg.content)
+            task_result = TaskResult(**result_dict)
+        except json.JSONDecodeError as e:
+            print(
+                f"{Fore.RED}JSON parsing error for task {task.id}: "
+                f"Invalid response format - {e}{Fore.RESET}"
+            )
+            return TaskState.FAILED
 
         color = Fore.RED if task_result.failed else Fore.GREEN
         print(

camel/societies/workforce/workforce.py

@@ -882,11 +882,16 @@ class Workforce(BaseNode):
         self._task = task
         self._state = WorkforceState.RUNNING
         task.state = TaskState.FAILED  # TODO: Add logic for OPEN
-        self._pending_tasks.append(task)
 
         # Decompose the task into subtasks first
         subtasks = self._decompose_task(task)
-        self._pending_tasks.extendleft(reversed(subtasks))
+        if subtasks:
+            # If decomposition happened, the original task becomes a container.
+            # We only execute its subtasks.
+            self._pending_tasks.extendleft(reversed(subtasks))
+        else:
+            # If no decomposition, execute the original task.
+            self._pending_tasks.append(task)
         self.set_channel(TaskChannel())
 
         # Save initial snapshot

camel/toolkits/function_tool.py

@@ -531,9 +531,10 @@ class FunctionTool:
             param_dict = properties[param_name]
             if "description" not in param_dict:
                 warnings.warn(
-                    f"Parameter description is missing "
-                    f"for {param_dict}. This may affect the "
-                    f"quality of tool calling."
+                    f"Parameter description is missing for the "
+                    f"function '{openai_tool_schema['function']['name']}'. "
+                    f"The parameter definition is {param_dict}. "
+                    f"This may affect the quality of tool calling."
                 )
 
     def get_openai_tool_schema(self) -> Dict[str, Any]:

camel/toolkits/video_analysis_toolkit.py

@@ -17,6 +17,7 @@ from __future__ import annotations
 
 import io
 import os
+import re
 import tempfile
 from pathlib import Path
 from typing import List, Optional
@@ -41,6 +42,11 @@ VIDEO_QA_PROMPT = """
 Analyze the provided video frames and corresponding audio transcription to \
 answer the given question(s) thoroughly and accurately.
 
+The transcriptions may come from two sources:
+  1. **Audio Transcription**: The spoken words in the video.
+  2. **Visual Text (OCR)**: Text extracted from the video frames (like \
+  captions, on-screen text, etc.).
+
 Instructions:
     1. Visual Analysis:
         - Examine the video frames to identify visible entities.
@@ -49,11 +55,13 @@ such as size, color, shape, texture, or behavior.
         - Note significant groupings, interactions, or contextual patterns \
 relevant to the analysis.
 
-    2. Audio Integration:
+    2. Audio and Text Integration:
         - Use the audio transcription to complement or clarify your visual \
 observations.
+        - Use the visual text (OCR) to get exact textual information that may \
+not be accurately readable from the images alone.
         - Identify names, descriptions, or contextual hints in the \
-transcription that help confirm or refine your visual analysis.
+transcriptions that help confirm or refine your visual analysis.
 
     3. Detailed Reasoning and Justification:
         - Provide a brief explanation of how you identified and distinguished \
@@ -65,7 +73,7 @@ your reasoning.
         - Specify the total number of distinct species or object types \
 identified in the video.
         - Describe the defining characteristics and any supporting evidence \
-from the video and transcription.
+from the video and transcription sources.
 
     5. Important Considerations:
         - Pay close attention to subtle differences that could distinguish \
@@ -76,6 +84,9 @@ similar-looking species or objects
 **Audio Transcription:**
 {audio_transcription}
 
+**Visual Text (OCR):**
+{visual_text}
+
 **Question:**
 {question}
 """
@@ -96,6 +107,8 @@ class VideoAnalysisToolkit(BaseToolkit):
             transcription using OpenAI's audio models. Requires a valid OpenAI
             API key. When disabled, video analysis will be based solely on
             visual content. (default: :obj:`False`)
+        use_ocr (bool, optional): Whether to enable OCR for extracting text
+            from video frames. (default: :obj:`False`)
         frame_interval (float, optional): Interval in seconds between frames
             to extract from the video. (default: :obj:`4.0`)
         output_language (str, optional): The language for output responses.
@@ -113,6 +126,7 @@ class VideoAnalysisToolkit(BaseToolkit):
         download_directory: Optional[str] = None,
         model: Optional[BaseModelBackend] = None,
         use_audio_transcription: bool = False,
+        use_ocr: bool = False,
         frame_interval: float = 4.0,
         output_language: str = "English",
         cookies_path: Optional[str] = None,
@@ -122,6 +136,7 @@ class VideoAnalysisToolkit(BaseToolkit):
         self._cleanup = download_directory is None
         self._temp_files: list[str] = []  # Track temporary files for cleanup
         self._use_audio_transcription = use_audio_transcription
+        self._use_ocr = use_ocr
         self.output_language = output_language
         self.frame_interval = frame_interval
 
@@ -211,6 +226,53 @@ class VideoAnalysisToolkit(BaseToolkit):
                     f"{self._download_directory}: {e}"
                 )
 
+    @dependencies_required("pytesseract", "cv2", "numpy")
+    def _extract_text_from_frame(self, frame: Image.Image) -> str:
+        r"""Extract text from a video frame using OCR.
+
+        Args:
+            frame (Image.Image): PIL image frame to process.
+
+        Returns:
+            str: Extracted text from the frame.
+        """
+        import cv2
+        import numpy as np
+        import pytesseract
+
+        try:
+            # Convert to OpenCV format for preprocessing
+            cv_image = cv2.cvtColor(np.array(frame), cv2.COLOR_RGB2BGR)
+
+            # Preprocessing for better OCR results
+            gray = cv2.cvtColor(cv_image, cv2.COLOR_BGR2GRAY)
+            blur = cv2.GaussianBlur(gray, (3, 3), 0)
+            _, threshold = cv2.threshold(
+                blur, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU
+            )
+
+            # Convert back to PIL image for OCR
+            preprocessed_frame = Image.fromarray(threshold)
+            return pytesseract.image_to_string(preprocessed_frame).strip()
+        except Exception as e:
+            logger.error(f"OCR failed: {e}")
+            return ""
+
+    def _process_extracted_text(self, text: str) -> str:
+        r"""Clean and format OCR-extracted text.
+
+        Args:
+            text (str): Raw extracted OCR text.
+
+        Returns:
+            str: Cleaned and formatted text.
+        """
+        # Filter irrelevant characters and noise
+        text = re.sub(r'[^\w\s,.?!:;\'"()-]', '', text)
+        # Remove excessive whitespace
+        text = re.sub(r'\s+', ' ', text).strip()
+        return text
+
     def _extract_audio_from_video(
         self, video_path: str, output_format: str = "mp3"
     ) -> str:
@@ -511,9 +573,21 @@ class VideoAnalysisToolkit(BaseToolkit):
                 audio_path = self._extract_audio_from_video(video_path)
                 audio_transcript = self._transcribe_audio(audio_path)
 
+            # Extract visual text with OCR
+            visual_text = ""
             video_frames = self._extract_keyframes(video_path)
+            # Build visual text only if OCR is enabled
+            if self._use_ocr:
+                for frame in video_frames:
+                    text = self._extract_text_from_frame(frame)
+                    processed = self._process_extracted_text(text)
+                    if processed:
+                        visual_text += processed + "\n"
+                visual_text = visual_text.strip() or "No visual text detected."
+
             prompt = VIDEO_QA_PROMPT.format(
                 audio_transcription=audio_transcript,
+                visual_text=visual_text,
                 question=question,
             )
 

{camel_ai-0.2.69a1.dist-info → camel_ai-0.2.69a2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: camel-ai
-Version: 0.2.69a1
+Version: 0.2.69a2
 Summary: Communicative Agents for AI Society Study
 Project-URL: Homepage, https://www.camel-ai.org/
 Project-URL: Repository, https://github.com/camel-ai/camel
@@ -99,6 +99,7 @@ Requires-Dist: pymupdf<2,>=1.22.5; extra == 'all'
 Requires-Dist: pyobvector>=0.1.18; extra == 'all'
 Requires-Dist: pyowm<4,>=3.3.0; extra == 'all'
 Requires-Dist: pytelegrambotapi<5,>=4.18.0; extra == 'all'
+Requires-Dist: pytesseract>=0.3.13; extra == 'all'
 Requires-Dist: pytest-asyncio<0.24,>=0.23.0; extra == 'all'
 Requires-Dist: pytest-cov<5,>=4; extra == 'all'
 Requires-Dist: pytest<8,>=7; extra == 'all'
@@ -226,6 +227,7 @@ Provides-Extra: media-tools
 Requires-Dist: ffmpeg-python<0.3,>=0.2.0; extra == 'media-tools'
 Requires-Dist: imageio[pyav]<3,>=2.34.2; extra == 'media-tools'
 Requires-Dist: pydub<0.26,>=0.25.1; extra == 'media-tools'
+Requires-Dist: pytesseract>=0.3.13; extra == 'media-tools'
 Requires-Dist: scenedetect>=0.6.5.2; extra == 'media-tools'
 Requires-Dist: yt-dlp<2025,>=2024.11.4; extra == 'media-tools'
 Provides-Extra: model-platforms
@@ -252,6 +254,7 @@ Requires-Dist: ffmpeg-python<0.3,>=0.2.0; extra == 'owl'
 Requires-Dist: fpdf>=1.7.2; extra == 'owl'
 Requires-Dist: html2text>=2024.2.26; extra == 'owl'
 Requires-Dist: imageio[pyav]<3,>=2.34.2; extra == 'owl'
+Requires-Dist: markitdown==0.1.1; extra == 'owl'
 Requires-Dist: mcp-server-fetch==2025.1.17; extra == 'owl'
 Requires-Dist: mcp-simple-arxiv==0.2.2; extra == 'owl'
 Requires-Dist: newspaper3k<0.3,>=0.2.8; extra == 'owl'
@@ -266,6 +269,7 @@ Requires-Dist: pyautogui<0.10,>=0.9.54; extra == 'owl'
 Requires-Dist: pydub<0.26,>=0.25.1; extra == 'owl'
 Requires-Dist: pylatex>=1.4.2; extra == 'owl'
 Requires-Dist: pymupdf<2,>=1.22.5; extra == 'owl'
+Requires-Dist: pytesseract>=0.3.13; extra == 'owl'
 Requires-Dist: python-dotenv<2,>=1.0.0; extra == 'owl'
 Requires-Dist: python-pptx>=1.0.2; extra == 'owl'
 Requires-Dist: requests-oauthlib<2,>=1.3.1; extra == 'owl'

{camel_ai-0.2.69a1.dist-info → camel_ai-0.2.69a2.dist-info}/RECORD

@@ -1,4 +1,4 @@
-camel/__init__.py,sha256=1jdrogMF4lEAc0fG6MGwLdfMt_c4luht9mj7KUDk-ts,901
+camel/__init__.py,sha256=WMfYI5vXSsv0CSAsfpSAJsK0QJ-RqKRd4nHzG-3LE7Q,901
 camel/generators.py,sha256=JRqj9_m1PF4qT6UtybzTQ-KBT9MJQt18OAAYvQ_fr2o,13844
 camel/human.py,sha256=Xg8x1cS5KK4bQ1SDByiHZnzsRpvRP-KZViNvmu38xo4,5475
 camel/logger.py,sha256=rZVeOVYuQ9RYJ5Tqyv0usqy0g4zaVEq4qSfZ9nd2640,5755
@@ -273,11 +273,11 @@ camel/societies/workforce/__init__.py,sha256=bkTI-PE-MSK9AQ2V2gR6cR2WY-R7Jqy_NmX
 camel/societies/workforce/base.py,sha256=z2DmbTP5LL5-aCAAqglznQqCLfPmnyM5zD3w6jjtsb8,2175
 camel/societies/workforce/prompts.py,sha256=_l7OUkzH5p7KOd8HMZle9zB9W3jKza_Yb_6elFKiZ2s,11813
 camel/societies/workforce/role_playing_worker.py,sha256=pWPCtkLx-xbc4SWyZBfMwvWr-R5ZpANhN7g6PDuH8Po,7615
-camel/societies/workforce/single_agent_worker.py,sha256=NIP0cHueDNQZxzN-MikmwIe-1LoT3V8Z1LgqdiPidCE,14438
+camel/societies/workforce/single_agent_worker.py,sha256=5lF9zzxDJMCtELh6Y407m_D2bMAXMsY3yvY3_Bssd5A,14704
 camel/societies/workforce/task_channel.py,sha256=uqQQI67Tr4awbR4bjZXdx8_4gL6-ON5IjQk_H_ryqT4,7431
 camel/societies/workforce/utils.py,sha256=Gjlz7pLo4r1b6iNHtlIMxeEuat4d6tEEQMI40JAU3kY,6190
 camel/societies/workforce/worker.py,sha256=36tkOyz4G2wfBdrFjt9NBPXsx4UbE6uL5on8sP2aoqk,6414
-camel/societies/workforce/workforce.py,sha256=gLbjcGeAXwHCyk6jGs7mBHJnpEG7xcvEcwjA2ibi6-0,88526
+camel/societies/workforce/workforce.py,sha256=xiWLSoRHOC1yz9WK0uXImwQkblr7XIJ6BXdNrEUf7Rk,88755
 camel/societies/workforce/workforce_logger.py,sha256=2xGMMYQPphC5WlPbgWWp9SEY2hGpP-GkhgsNC19irns,24237
 camel/storages/__init__.py,sha256=bFpvvAS2QyZoIr-tnwhMWsZRL411kIRq6IMUHcI7KHs,1989
 camel/storages/graph_storages/__init__.py,sha256=G29BNn651C0WTOpjCl4QnVM-4B9tcNh8DdmsCiONH8Y,948
@@ -327,7 +327,7 @@ camel/toolkits/dappier_toolkit.py,sha256=OEHOYXX_oXhgbVtWYAy13nO9uXf9i5qEXSwY4Pe
 camel/toolkits/data_commons_toolkit.py,sha256=aHZUSL1ACpnYGaf1rE2csVKTmXTmN8lMGRUBYhZ_YEk,14168
 camel/toolkits/excel_toolkit.py,sha256=DSjBXl24_LrJubGFFmB_vqliKzzWTbT1TH309YQVUO8,6653
 camel/toolkits/file_write_toolkit.py,sha256=i1CivjvqTpHakYK_EynNvNikjZZYsusxkjT2v-f9EYc,16473
-camel/toolkits/function_tool.py,sha256=045I_vyqvB6PorcoVaEbxYjGe-fD-FBKtO4lfpZ2eVw,33535
+camel/toolkits/function_tool.py,sha256=xCDzjxTRCVj_kmbnMFBuwK-3NuzM4JNe_kv9HLtjnIA,33644
 camel/toolkits/github_toolkit.py,sha256=iUyRrjWGAW_iljZVfNyfkm1Vi55wJxK6PsDAQs9pOag,13099
 camel/toolkits/google_calendar_toolkit.py,sha256=E-sdgdbtNBs_CXbhht9t1dsVr4DsTr5NguHkx4fvSmc,15410
 camel/toolkits/google_maps_toolkit.py,sha256=WTnkURpGri9KcY5OwV7AJJHOzmpu5RNmYE1QCVqvwWM,12023
@@ -365,7 +365,7 @@ camel/toolkits/task_planning_toolkit.py,sha256=Ttw9fHae4omGC1SA-6uaeXVHJ1YkwiVlo
 camel/toolkits/terminal_toolkit.py,sha256=gupuTvNkwnFzcFwDB_irSJ9-dXRr8yEAsYq5ChEkkHg,37230
 camel/toolkits/thinking_toolkit.py,sha256=nZYLvKWIx2BM1DYu69I9B5EISAG7aYcLYXKv9663BVk,8000
 camel/toolkits/twitter_toolkit.py,sha256=Px4N8aUxUzy01LhGSWkdrC2JgwKkrY3cvxgMeJ2XYfU,15939
-camel/toolkits/video_analysis_toolkit.py,sha256=Mf7kZ2UDKFzIq8XjJc6EhL8qXQnEomQ8OBy_eyjD49A,20647
+camel/toolkits/video_analysis_toolkit.py,sha256=Wh08MAVvs3PtgXN88Sk0TXYaGfVmQAol8FPCXMPPpIM,23375
 camel/toolkits/video_download_toolkit.py,sha256=jBb2SQ9OA5HIuGF7FbNQ0KrvvwMWPxUnvUyCHjbHuQQ,7501
 camel/toolkits/weather_toolkit.py,sha256=fs9x9aC38Wsvni6A4PPpbRX6-aBnZiqs2Jix39yoULU,7413
 camel/toolkits/whatsapp_toolkit.py,sha256=udUQXkXyeWsmrUlOJZsGBhHtc_jhB05Axe_TchhibsU,5760
@@ -431,7 +431,7 @@ camel/verifiers/math_verifier.py,sha256=tA1D4S0sm8nsWISevxSN0hvSVtIUpqmJhzqfbuMo
 camel/verifiers/models.py,sha256=GdxYPr7UxNrR1577yW4kyroRcLGfd-H1GXgv8potDWU,2471
 camel/verifiers/physics_verifier.py,sha256=c1grrRddcrVN7szkxhv2QirwY9viIRSITWeWFF5HmLs,30187
 camel/verifiers/python_verifier.py,sha256=ogTz77wODfEcDN4tMVtiSkRQyoiZbHPY2fKybn59lHw,20558
-camel_ai-0.2.69a1.dist-info/METADATA,sha256=jl-Gs6_fECtoyaAusXMrlvSrTx1WvMmSAx39z4biZT8,44837
-camel_ai-0.2.69a1.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-camel_ai-0.2.69a1.dist-info/licenses/LICENSE,sha256=id0nB2my5kG0xXeimIu5zZrbHLS6EQvxvkKkzIHaT2k,11343
-camel_ai-0.2.69a1.dist-info/RECORD,,
+camel_ai-0.2.69a2.dist-info/METADATA,sha256=5FBG-fY28g9tYIZIpxESs4biHcejU3mVLUQY5sMepWk,45047
+camel_ai-0.2.69a2.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+camel_ai-0.2.69a2.dist-info/licenses/LICENSE,sha256=id0nB2my5kG0xXeimIu5zZrbHLS6EQvxvkKkzIHaT2k,11343
+camel_ai-0.2.69a2.dist-info/RECORD,,