Spaces:
Sleeping
Sleeping
Update tools/getDeviceInfo.py
Browse files- tools/getDeviceInfo.py +13 -26
tools/getDeviceInfo.py
CHANGED
@@ -8,68 +8,55 @@ import re
|
|
8 |
import yaml
|
9 |
from smolagents import CodeAgent, HfApiModel
|
10 |
from tools.final_answer import FinalAnswerTool
|
|
|
11 |
|
12 |
class getDeviceInfo(Tool):
|
13 |
name = "visit_webpage"
|
14 |
description = "Visits a webpage at the given URL, reads its content as a markdown string, and extracts device information."
|
15 |
inputs = {'url': {'type': 'string', 'description': 'The URL of the webpage to visit.'}}
|
16 |
-
output_type = "
|
17 |
|
18 |
def forward(self, url: str) -> str: # ✅ Return a JSON string
|
19 |
-
import json
|
20 |
-
|
21 |
try:
|
22 |
-
import requests
|
23 |
-
from markdownify import markdownify
|
24 |
-
from requests.exceptions import RequestException
|
25 |
from smolagents.utils import truncate_content
|
26 |
except ImportError as e:
|
27 |
raise ImportError(
|
28 |
-
"You must install
|
29 |
) from e
|
30 |
|
31 |
try:
|
32 |
-
#
|
33 |
response = requests.get(url, timeout=20)
|
34 |
-
response.raise_for_status()
|
35 |
-
|
36 |
-
# Convert the HTML content to Markdown
|
37 |
-
markdown_content = markdownify(response.text).strip()
|
38 |
-
|
39 |
-
# Remove multiple line breaks
|
40 |
markdown_content = re.sub(r"\n{3,}", "\n\n", markdown_content)
|
41 |
-
|
42 |
webpage_content = truncate_content(markdown_content, 10000)
|
43 |
|
44 |
except requests.exceptions.Timeout:
|
45 |
-
webpage_content = "The request timed out.
|
46 |
-
except RequestException as e:
|
47 |
webpage_content = f"Error fetching the webpage: {str(e)}"
|
48 |
except Exception as e:
|
49 |
-
webpage_content = f"
|
50 |
|
51 |
# Extract device information using Playwright
|
52 |
with playwright.sync_api.sync_playwright() as p:
|
53 |
browser = p.chromium.launch(headless=True)
|
54 |
page = browser.new_page()
|
55 |
-
|
56 |
-
script = """
|
57 |
-
() => ({
|
58 |
userAgent: navigator.userAgent,
|
59 |
platform: navigator.platform,
|
60 |
language: navigator.language,
|
61 |
screenWidth: window.screen.width,
|
62 |
screenHeight: window.screen.height
|
63 |
-
})
|
64 |
-
"""
|
65 |
-
|
66 |
page.goto("about:blank")
|
67 |
device_info = page.evaluate(script)
|
68 |
browser.close()
|
69 |
|
70 |
-
# ✅ Convert output to JSON string
|
71 |
result = {
|
72 |
"webpage_content": webpage_content,
|
73 |
"device_info": device_info
|
74 |
}
|
75 |
-
return json.dumps(result) # ✅ Ensure return type matches "
|
|
|
8 |
import yaml
|
9 |
from smolagents import CodeAgent, HfApiModel
|
10 |
from tools.final_answer import FinalAnswerTool
|
11 |
+
import json
|
12 |
|
13 |
class getDeviceInfo(Tool):
|
14 |
name = "visit_webpage"
|
15 |
description = "Visits a webpage at the given URL, reads its content as a markdown string, and extracts device information."
|
16 |
inputs = {'url': {'type': 'string', 'description': 'The URL of the webpage to visit.'}}
|
17 |
+
output_type = "string" # ✅ Fix: Use "string" instead of "json"
|
18 |
|
19 |
def forward(self, url: str) -> str: # ✅ Return a JSON string
|
|
|
|
|
20 |
try:
|
|
|
|
|
|
|
21 |
from smolagents.utils import truncate_content
|
22 |
except ImportError as e:
|
23 |
raise ImportError(
|
24 |
+
"You must install the `smolagents` package to run this tool."
|
25 |
) from e
|
26 |
|
27 |
try:
|
28 |
+
# Fetch the webpage
|
29 |
response = requests.get(url, timeout=20)
|
30 |
+
response.raise_for_status()
|
31 |
+
markdown_content = markdownify.markdownify(response.text).strip()
|
|
|
|
|
|
|
|
|
32 |
markdown_content = re.sub(r"\n{3,}", "\n\n", markdown_content)
|
|
|
33 |
webpage_content = truncate_content(markdown_content, 10000)
|
34 |
|
35 |
except requests.exceptions.Timeout:
|
36 |
+
webpage_content = "The request timed out."
|
37 |
+
except requests.exceptions.RequestException as e:
|
38 |
webpage_content = f"Error fetching the webpage: {str(e)}"
|
39 |
except Exception as e:
|
40 |
+
webpage_content = f"Unexpected error: {str(e)}"
|
41 |
|
42 |
# Extract device information using Playwright
|
43 |
with playwright.sync_api.sync_playwright() as p:
|
44 |
browser = p.chromium.launch(headless=True)
|
45 |
page = browser.new_page()
|
46 |
+
script = """() => ({
|
|
|
|
|
47 |
userAgent: navigator.userAgent,
|
48 |
platform: navigator.platform,
|
49 |
language: navigator.language,
|
50 |
screenWidth: window.screen.width,
|
51 |
screenHeight: window.screen.height
|
52 |
+
})"""
|
|
|
|
|
53 |
page.goto("about:blank")
|
54 |
device_info = page.evaluate(script)
|
55 |
browser.close()
|
56 |
|
57 |
+
# ✅ Convert output to JSON string for compatibility
|
58 |
result = {
|
59 |
"webpage_content": webpage_content,
|
60 |
"device_info": device_info
|
61 |
}
|
62 |
+
return json.dumps(result) # ✅ Ensure return type matches "string"
|