1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
|
"""URL scraping tool that fetches a page and extracts readable text."""
from __future__ import annotations
from bs4 import BeautifulSoup
from ..config import settings
from .helper.pagination import format_paged_result, make_token, parse_token
from .helper.safe_http import UnsafeURL, fetch_public_bytes
_TOOL_NAME = "scrape_url"
def scrape_url(url: str, max_chars: int | None = None, continue_from: str | None = None) -> str:
"""Fetch a web page and return its main text content.
Scripts, styles and other non-content elements are removed before
collapsing whitespace. The returned text is a page of at most ``max_chars``
characters; when more content is available, a ``next_token`` is appended
that can be passed back via ``continue_from`` to fetch the next page.
Args:
url: The absolute URL of the page to scrape.
max_chars: Maximum number of characters to return per page. Defaults to
the value in settings.
continue_from: Opaque continuation token returned in a previous call's
``next_token``. When provided, the page is resumed from the encoded
offset instead of the beginning.
Returns:
The extracted plain text of the page (one page of ``max_chars``),
followed by a ``---page-info---`` footer when more content is available.
If the request fails a short error description is returned instead.
"""
if not url or not url.strip():
return "Invalid URL."
limit = max_chars if max_chars is not None and max_chars > 0 else settings.scrape_max_chars
# Resolve the resumption offset from the continuation token, if any.
offset = 0
if continue_from:
try:
payload = parse_token(continue_from, _TOOL_NAME)
except ValueError as exc:
return f"Invalid continuation token: {exc}"
# The token's URL takes precedence; it identifies the page being paged.
url = str(payload.get("url") or url)
offset = int(payload.get("offset", 0))
if offset < 0:
offset = 0
try:
data, final_url, content_type = fetch_public_bytes(url, timeout=settings.scrape_timeout)
encoding = "utf-8"
if "charset=" in content_type:
encoding = content_type.rsplit("charset=", 1)[-1].split(";", 1)[0].strip()
html = data.decode(encoding or "utf-8", errors="replace")
url = final_url
except UnsafeURL as exc:
return f"Blocked URL: {exc}"
except Exception as exc: # noqa: BLE001
return f"Failed to fetch {url}: {exc}"
soup = BeautifulSoup(html, "html.parser")
# Remove non-content tags.
for tag in soup(["script", "style", "noscript", "template", "head", "nav", "footer", "aside"]):
tag.decompose()
# Prefer article/main content when available.
container = soup.find("article") or soup.find("main") or soup
text = container.get_text(separator="\n")
# Collapse blank lines and trailing whitespace per line.
lines = [ln.strip() for ln in text.splitlines()]
cleaned = "\n".join(ln for ln in lines if ln)
if not cleaned:
return "No readable text content found."
# Slice the requested page from the full cleaned text.
page = cleaned[offset : offset + limit]
if not page:
# Offset is past the end; nothing left to return.
return format_paged_result("", has_more=False, next_token=None)
has_more = len(cleaned) > offset + limit
next_token = make_token(_TOOL_NAME, url, offset + limit) if has_more else None
return format_paged_result(page, has_more=has_more, next_token=next_token)
|
