Ocr and pdf operation

main.py

@@ -81,32 +81,58 @@ def main():
         task = example_tasks[0]
         print(f"Using example task: {task}")
 
-    # Execute task
+    # Execute first task
+    is_first_task = True
     try:
-        result = agent.execute(task)
-
-        print("\n" + "="*80)
-        print("📊 FINAL RESULT")
-        print("="*80)
-        print(result)
-        print()
-
-        # Show artifacts
-        artifacts = list(ARTIFACTS_DIR.glob("*"))
-        if artifacts:
-            print(f"\n📁 Generated {len(artifacts)} artifacts in {ARTIFACTS_DIR}/")
-            recent = sorted(artifacts, key=lambda p: p.stat().st_mtime, reverse=True)[:5]
-            for artifact in recent:
-                size = artifact.stat().st_size
-                print(f"   - {artifact.name} ({size} bytes)")
+        while True:
+            try:
+                result = agent.execute(task, is_followup=not is_first_task)
+
+                print("\n" + "="*80)
+                print("📊 RESULT")
+                print("="*80)
+                print(result)
+                print()
+
+                # Show artifacts
+                artifacts = list(ARTIFACTS_DIR.glob("*"))
+                if artifacts:
+                    print(f"\n📁 Generated {len(artifacts)} artifacts in {ARTIFACTS_DIR}/")
+                    recent = sorted(artifacts, key=lambda p: p.stat().st_mtime, reverse=True)[:5]
+                    for artifact in recent:
+                        size = artifact.stat().st_size
+                        print(f"   - {artifact.name} ({size} bytes)")
+
+                # Ask for follow-up task
+                print("\n" + "="*80)
+                print("💬 You can continue with a follow-up task in this session")
+                print("   (Type 'exit', 'quit', or 'q' to end the session)")
+                print("="*80)
+                followup = input("\nEnter follow-up task (or 'exit' to end): ").strip()
+
+                if not followup or followup.lower() in ['exit', 'quit', 'q']:
+                    print("\n👋 Ending session...")
+                    break
+
+                task = followup
+                is_first_task = False
+
+            except KeyboardInterrupt:
+                print("\n\n⚠️  Task interrupted by user")
+                break
 
-    except KeyboardInterrupt:
-        print("\n\n⚠️  Task interrupted by user")
     except Exception as e:
         print(f"\n❌ Error: {e}")
         import traceback
         traceback.print_exc()
 
+    finally:
+        # Always generate final interpretation and close session
+        try:
+            agent.close_session()
+        except Exception as e:
+            print(f"⚠️  Error during session close: {e}")
+
     print("\n" + "="*80)
     print("👋 Thank you for using Vision-LLM Web Agent!")
     print("="*80 + "\n")

tests/test_file_operations.py

@@ -11,9 +11,12 @@
     extract_pdf_text,
     extract_pdf_images,
     save_or_crop_image,
-    write_text_to_file
+    write_text_to_file,
+    ocr_image_to_text
 )
 from vision_llm_web_agent.tools.browser_control import browser_state
+from PIL import Image, ImageDraw,ImageFont
+import pytesseract
 
 
 @pytest.fixture
@@ -81,6 +84,34 @@ def sample_image(tmp_path):
     img.save(img_path)
     return img_path
 
+@pytest.fixture
+def sample_image_for_ocr(tmp_path):
+    """
+    创建一个包含可识别文本的图像，用于 OCR 测试。
+    注意：这需要系统安装了字体文件才能正确工作。
+    """
+    img_path = tmp_path / "ocr_test_image.png"
+
+    # 创建一个白色背景的图像
+    img = Image.new('RGB', (400, 100), color='white')
+    draw = ImageDraw.Draw(img)
+
+    test_text = "OCR Test: Hello World 123"
+
+    # 尝试使用一个通用字体，如果找不到可能会失败
+    try:
+        # 尝试加载一个相对可靠的默认字体（可能需要调整路径或使用系统字体名称）
+        font = ImageFont.truetype("arial.ttf", 24)
+    except IOError:
+        # Fallback to default font if the specified one is not found
+        font = ImageFont.load_default()
+
+    # 在图像上绘制文本
+    draw.text((10, 10), test_text, fill='black', font=font)
+
+    img.save(img_path)
+    return img_path, test_text # 返回路径和预期的文本内容
+
 
 class TestDownloadPdfFile:
     """Test the download_pdf_file function"""
@@ -425,3 +456,84 @@ def test_crop_and_save_chain(self, sample_image, test_artifacts_dir):
         img = Image.open(full_final)
         assert img.size == (100, 100)
 
+
+
+
+class TestOcrImageToText:
+    """Test the ocr_image_to_text function"""
+
+    def test_ocr_basic_success(self, sample_image_for_ocr, test_artifacts_dir):
+        """Test basic OCR and text saving"""
+        image_path, expected_text = sample_image_for_ocr
+        output_file_name = "test/ocr_result.txt"
+
+        # 运行 OCR 工具
+        result = ocr_image_to_text(str(image_path), output_file_name)
+
+        # 检查结果是否成功
+        assert "✅" in result, f"OCR failed or did not return success status: {result}"
+
+        # 检查输出文件是否存在于 artifacts 目录中
+        full_output_path = Path("artifacts") / output_file_name
+        assert full_output_path.exists()
+
+        # 读取内容并验证文本是否被识别（由于 OCR 准确性无法保证完美，我们检查内容长度和关键字）
+        extracted_content = full_output_path.read_text(encoding='utf-8').strip()
+
+        # OCR 引擎可能会在文本中添加换行符或其他空格
+        assert len(extracted_content) > 10, "Extracted text is too short or empty."
+        assert "Hello World" in extracted_content, "Expected text 'Hello World' not found in OCR output."
+
+
+    def test_ocr_creates_directory(self, sample_image_for_ocr, test_artifacts_dir):
+        """Test that OCR output creates parent directories"""
+        image_path, _ = sample_image_for_ocr
+        output_file_name = "test/nested/ocr/output.txt"
+
+        # 运行 OCR 工具
+        result = ocr_image_to_text(str(image_path), output_file_name)
+
+        assert "✅" in result
+
+        # 检查目录是否被创建
+        full_output_path = Path("artifacts") / output_file_name
+        assert full_output_path.parent.exists()
+        assert full_output_path.exists()
+
+
+    def test_ocr_nonexistent_image(self, test_artifacts_dir):
+        """Test OCR on a nonexistent image file"""
+        nonexistent_image = "nonexistent_ocr.png"
+        output_file_name = "test/ocr_fail.txt"
+
+        # 运行 OCR 工具
+        result = ocr_image_to_text(nonexistent_image, output_file_name)
+
+        # 检查结果是否为失败
+        assert "❌" in result
+        assert "file not found" in result
+
+        # 检查输出文件是否未被创建
+        full_output_path = Path("artifacts") / output_file_name
+        assert not full_output_path.exists()
+
+
+    def test_ocr_empty_output(self, sample_image, test_artifacts_dir):
+        """Test OCR on an image with little or no recognizable text (e.g., solid color)"""
+        image_path = sample_image # This is a 200x200 blue image with a white square, little text
+        output_file_name = "test/ocr_empty.txt"
+
+        # 运行 OCR 工具
+        result = ocr_image_to_text(str(image_path), output_file_name)
+
+        # 检查结果可能是成功但有警告，或者失败
+        # 如果 pytesseract 成功运行但识别不出文字，通常会返回成功状态和警告。
+        assert "✅" in result or "⚠️" in result
+
+        # 检查输出文件是否存在
+        full_output_path = Path("artifacts") / output_file_name
+        assert full_output_path.exists()
+
+        # 检查内容是否为空或接近空
+        extracted_content = full_output_path.read_text(encoding='utf-8').strip()
+        assert len(extracted_content) < 100 # 应该只有很少的字符或空格

tests/test_tesseract/tesseractTest.py

@@ -0,0 +1,4 @@
+from PIL import Image
+import pytesseract
+text=pytesseract.image_to_string(Image.open('test_image.jpg'))
+print(text)