@minded-ai/mindedjs 1.0.0-ec2-beta-6 → 1.0.0-ec2-beta-7

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (3) hide show
  1. package/dist/browserTask/executeBrowserTask.py +9 -34
  2. package/package.json +1 -1
  3. package/src/browserTask/executeBrowserTask.py +9 -34
package/dist/browserTask/executeBrowserTask.py CHANGED
@@ -1,15 +1,8 @@
1
1
  """
2
2
  Browser Task Executor with Screenshot Capture Support
3
3
 
4
- This script runs browser automation tasks using browser-use and automatically captures
5
- screenshots at the end of each step, uploading them to S3.
6
-
7
- Environment Variables for Screenshots:
8
- - SCREENSHOT_S3_BUCKET: S3 bucket name (default: 'global-development-agentsforce')
9
- - SCREENSHOT_S3_PREFIX: S3 key prefix (default: 'browser-use-runs-screenshots/')
10
- - AWS_REGION: AWS region for S3 (default: 'us-east-1')
11
-
12
- AWS credentials should be configured via standard AWS SDK methods.
4
+ This script runs browser automation tasks using browser-use and can capture
5
+ screenshots at the end of each step, uploading them to S3 when configured.
13
6
  """
14
7
 
15
8
  import asyncio
@@ -90,15 +83,13 @@ class ScreenshotCapture:
90
83
  # Get current page
91
84
  page = await agent.browser_session.get_current_page()
92
85
 
93
- # Get current URL for context - use page.url() method or fallback
86
+ # Get current URL for logging (browser-use might use method instead of property)
94
87
  try:
95
- # Try as property first
96
88
  current_url = page.url if hasattr(page, 'url') else page.url()
97
89
  except:
98
- # Fallback if URL is not accessible
99
90
  current_url = "unknown"
100
91
 
101
- logger.info(f"📸 Taking step_end screenshot at: {current_url}")
92
+ logger.info(f"📸 Taking screenshot at: {current_url}")
102
93
 
103
94
  # Update step counter
104
95
  step_number = self.step_counter
@@ -107,45 +98,30 @@ class ScreenshotCapture:
107
98
  # Take screenshot - browser-use takes viewport by default
108
99
  screenshot_data = await page.screenshot()
109
100
 
110
- # Ensure we have bytes - browser-use might return different format
101
+ # Ensure we have bytes - browser-use might return base64 string or bytes
111
102
  if isinstance(screenshot_data, bytes):
112
103
  screenshot_bytes = screenshot_data
113
104
  elif isinstance(screenshot_data, str):
114
- # If it's base64 encoded string
105
+ # If it's base64 encoded string, decode it
115
106
  screenshot_bytes = base64.b64decode(screenshot_data)
116
107
  else:
117
108
  # Try to get bytes from whatever format it is
118
109
  screenshot_bytes = bytes(screenshot_data)
119
110
 
120
- logger.debug(f"Screenshot data type: {type(screenshot_data)}, size: {len(screenshot_bytes)} bytes")
121
-
122
111
  # Generate filename with metadata
123
112
  timestamp = datetime.now().strftime("%Y%m%d_%H%M%S_%f")[:-3]
124
113
  filename = f"{self.session_id}/step_{step_number:04d}_{timestamp}.png"
125
114
 
126
- # Optional: Save to temp file for debugging (like Lambda does)
127
- if os.getenv('DEBUG_SCREENSHOTS') == 'true':
128
- temp_dir = Path("/tmp/screenshots")
129
- temp_dir.mkdir(exist_ok=True, parents=True)
130
- temp_path = temp_dir / f"step_{step_number:04d}_{timestamp}.png"
131
- with open(temp_path, "wb") as f:
132
- f.write(screenshot_bytes)
133
- logger.debug(f"💾 Debug: Saved screenshot to: {temp_path}")
134
-
135
115
  # Upload to S3
136
116
  s3_key = f"{self.s3_prefix}{filename}"
137
117
 
138
118
  try:
139
- # Verify we have valid PNG data before uploading
119
+ # Verify we have valid data before uploading
140
120
  if not screenshot_bytes or len(screenshot_bytes) == 0:
141
121
  logger.error(f"❌ Screenshot bytes are empty!")
142
122
  return
143
123
 
144
- # Check for PNG header (89 50 4E 47 = PNG signature)
145
- if len(screenshot_bytes) > 4 and screenshot_bytes[:4] != b'\x89PNG':
146
- logger.warning(f"⚠️ Screenshot may not be valid PNG format. First 4 bytes: {screenshot_bytes[:4].hex()}")
147
-
148
- # Upload with KMS encryption if required by bucket (matching Lambda implementation)
124
+ # Upload to S3 with KMS encryption
149
125
  self.s3_client.put_object(
150
126
  Bucket=self.s3_bucket,
151
127
  Key=s3_key,
@@ -153,11 +129,10 @@ class ScreenshotCapture:
153
129
  ContentType='image/png',
154
130
  ServerSideEncryption='aws:kms',
155
131
  Tagging='retention=30d'
156
- # Using bucket's default KMS key
157
132
  )
158
133
 
159
134
  s3_url = f"s3://{self.s3_bucket}/{s3_key}"
160
- logger.info(f"✅ Screenshot uploaded: {s3_url} (size: {len(screenshot_bytes)} bytes)")
135
+ logger.info(f"✅ Screenshot uploaded: {s3_url}")
161
136
 
162
137
  except self.ClientError as e:
163
138
  logger.error(f"❌ Failed to upload screenshot to S3: {str(e)}")
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@minded-ai/mindedjs",
3
- "version": "1.0.0-ec2-beta-6",
3
+ "version": "1.0.0-ec2-beta-7",
4
4
  "description": "MindedJS is a TypeScript library for building agents.",
5
5
  "main": "dist/index.js",
6
6
  "types": "dist/index.d.ts",
package/src/browserTask/executeBrowserTask.py CHANGED
@@ -1,15 +1,8 @@
1
1
  """
2
2
  Browser Task Executor with Screenshot Capture Support
3
3
 
4
- This script runs browser automation tasks using browser-use and automatically captures
5
- screenshots at the end of each step, uploading them to S3.
6
-
7
- Environment Variables for Screenshots:
8
- - SCREENSHOT_S3_BUCKET: S3 bucket name (default: 'global-development-agentsforce')
9
- - SCREENSHOT_S3_PREFIX: S3 key prefix (default: 'browser-use-runs-screenshots/')
10
- - AWS_REGION: AWS region for S3 (default: 'us-east-1')
11
-
12
- AWS credentials should be configured via standard AWS SDK methods.
4
+ This script runs browser automation tasks using browser-use and can capture
5
+ screenshots at the end of each step, uploading them to S3 when configured.
13
6
  """
14
7
 
15
8
  import asyncio
@@ -90,15 +83,13 @@ class ScreenshotCapture:
90
83
  # Get current page
91
84
  page = await agent.browser_session.get_current_page()
92
85
 
93
- # Get current URL for context - use page.url() method or fallback
86
+ # Get current URL for logging (browser-use might use method instead of property)
94
87
  try:
95
- # Try as property first
96
88
  current_url = page.url if hasattr(page, 'url') else page.url()
97
89
  except:
98
- # Fallback if URL is not accessible
99
90
  current_url = "unknown"
100
91
 
101
- logger.info(f"📸 Taking step_end screenshot at: {current_url}")
92
+ logger.info(f"📸 Taking screenshot at: {current_url}")
102
93
 
103
94
  # Update step counter
104
95
  step_number = self.step_counter
@@ -107,45 +98,30 @@ class ScreenshotCapture:
107
98
  # Take screenshot - browser-use takes viewport by default
108
99
  screenshot_data = await page.screenshot()
109
100
 
110
- # Ensure we have bytes - browser-use might return different format
101
+ # Ensure we have bytes - browser-use might return base64 string or bytes
111
102
  if isinstance(screenshot_data, bytes):
112
103
  screenshot_bytes = screenshot_data
113
104
  elif isinstance(screenshot_data, str):
114
- # If it's base64 encoded string
105
+ # If it's base64 encoded string, decode it
115
106
  screenshot_bytes = base64.b64decode(screenshot_data)
116
107
  else:
117
108
  # Try to get bytes from whatever format it is
118
109
  screenshot_bytes = bytes(screenshot_data)
119
110
 
120
- logger.debug(f"Screenshot data type: {type(screenshot_data)}, size: {len(screenshot_bytes)} bytes")
121
-
122
111
  # Generate filename with metadata
123
112
  timestamp = datetime.now().strftime("%Y%m%d_%H%M%S_%f")[:-3]
124
113
  filename = f"{self.session_id}/step_{step_number:04d}_{timestamp}.png"
125
114
 
126
- # Optional: Save to temp file for debugging (like Lambda does)
127
- if os.getenv('DEBUG_SCREENSHOTS') == 'true':
128
- temp_dir = Path("/tmp/screenshots")
129
- temp_dir.mkdir(exist_ok=True, parents=True)
130
- temp_path = temp_dir / f"step_{step_number:04d}_{timestamp}.png"
131
- with open(temp_path, "wb") as f:
132
- f.write(screenshot_bytes)
133
- logger.debug(f"💾 Debug: Saved screenshot to: {temp_path}")
134
-
135
115
  # Upload to S3
136
116
  s3_key = f"{self.s3_prefix}{filename}"
137
117
 
138
118
  try:
139
- # Verify we have valid PNG data before uploading
119
+ # Verify we have valid data before uploading
140
120
  if not screenshot_bytes or len(screenshot_bytes) == 0:
141
121
  logger.error(f"❌ Screenshot bytes are empty!")
142
122
  return
143
123
 
144
- # Check for PNG header (89 50 4E 47 = PNG signature)
145
- if len(screenshot_bytes) > 4 and screenshot_bytes[:4] != b'\x89PNG':
146
- logger.warning(f"⚠️ Screenshot may not be valid PNG format. First 4 bytes: {screenshot_bytes[:4].hex()}")
147
-
148
- # Upload with KMS encryption if required by bucket (matching Lambda implementation)
124
+ # Upload to S3 with KMS encryption
149
125
  self.s3_client.put_object(
150
126
  Bucket=self.s3_bucket,
151
127
  Key=s3_key,
@@ -153,11 +129,10 @@ class ScreenshotCapture:
153
129
  ContentType='image/png',
154
130
  ServerSideEncryption='aws:kms',
155
131
  Tagging='retention=30d'
156
- # Using bucket's default KMS key
157
132
  )
158
133
 
159
134
  s3_url = f"s3://{self.s3_bucket}/{s3_key}"
160
- logger.info(f"✅ Screenshot uploaded: {s3_url} (size: {len(screenshot_bytes)} bytes)")
135
+ logger.info(f"✅ Screenshot uploaded: {s3_url}")
161
136
 
162
137
  except self.ClientError as e:
163
138
  logger.error(f"❌ Failed to upload screenshot to S3: {str(e)}")