super-code-assistant 3.3.6__py3-none-any.whl

tools/web_search.py

@@ -0,0 +1,220 @@
+"""WebSearch tool: search the web using Bing and return structured results.
+
+Uses Bing.com search — zero dependencies, pure stdlib. Works in China.
+"""
+from __future__ import annotations
+
+import html
+import os
+import re
+import urllib.request
+import urllib.error
+import urllib.parse
+from html.parser import HTMLParser
+
+from core.tool import Tool, ToolResult
+
+_MAX_RESULTS = 10
+_DEFAULT_RESULTS = 5
+_MAX_CHARS = 3_000
+
+
+class _BingParser(HTMLParser):
+    """Parse Bing search result page HTML.
+
+    Bing results are in <li class="b_algo"> blocks:
+      <h2><a href="URL">title</a></h2>
+      <div class="b_caption"><p>snippet</p></div>
+
+    Uses depth counters to handle nested elements inside b_caption
+    (e.g. <div class="b_caption"><div class="f">...</div></div>).
+    """
+
+    def __init__(self):
+        super().__init__()
+        self.results: list[dict[str, str]] = []
+        self._in_algo = False
+        self._in_title = False           # inside <h2><a>
+        self._in_caption = False          # inside <div class="b_caption">
+        self._caption_depth = 0           # 嵌套 div 深度计数
+        self._text: list[str] = []
+        self._current_href = ""
+        self._current_title = ""
+        self._current_snippet = ""
+
+    def handle_starttag(self, tag: str, attrs: list[tuple[str, str | None]]) -> None:
+        d = dict(attrs)
+        cls = d.get("class", "")
+
+        if tag == "li" and "b_algo" in cls.split():
+            self._in_algo = True
+            self._in_title = False
+            self._in_caption = False
+            self._caption_depth = 0
+            self._current_href = ""
+            self._current_title = ""
+            self._current_snippet = ""
+
+        elif self._in_algo and not self._current_href:
+            # 第一个 <h2><a href="..."> 就是结果标题链接
+            if tag == "h2":
+                self._in_title = True
+            elif self._in_title and tag == "a":
+                self._current_href = d.get("href", "")
+                self._text = []
+
+        elif self._in_algo:
+            if tag == "div" and "b_caption" in cls.split():
+                self._in_caption = True
+                self._caption_depth = 1
+            elif self._in_caption and tag == "div":
+                self._caption_depth += 1
+
+    def handle_endtag(self, tag: str) -> None:
+        if self._in_title and tag == "a":
+            self._current_title = "".join(self._text).strip()
+        elif self._in_title and tag == "h2":
+            self._in_title = False
+        elif self._in_caption and tag == "div":
+            self._caption_depth -= 1
+            if self._caption_depth > 0:
+                return  # 内层 </div>，跳过
+            self._in_caption = False
+            # 外层 b_caption 的 </div>，收尾
+            if self._current_href:
+                self.results.append({
+                    "title": self._current_title,
+                    "href": self._current_href,
+                    "snippet": self._current_snippet,
+                })
+            self._in_algo = False
+        elif self._in_algo and tag == "li":
+            # 有些结果可能没有 b_caption（如视频/图片结果），在 </li> 时兜底
+            if self._current_href and not self._current_snippet:
+                self.results.append({
+                    "title": self._current_title,
+                    "href": self._current_href,
+                    "snippet": self._current_snippet,
+                })
+            self._in_algo = False
+
+    def handle_data(self, data: str) -> None:
+        if self._in_title and self._current_href:
+            self._text.append(data)
+        elif self._in_caption:
+            self._current_snippet += data
+
+    def finalize(self) -> None:
+        if self._in_algo and self._current_href:
+            self.results.append({
+                "title": self._current_title,
+                "href": self._current_href,
+                "snippet": self._current_snippet.strip(),
+            })
+
+
+class WebSearchTool(Tool):
+    name = "WebSearch"
+    description = (
+        "Searches the web using Bing and returns results as structured text. "
+        "Use this to find documentation, code examples, or any information on the web. "
+        "Each result includes a title, URL, and text snippet."
+    )
+    input_schema = {
+        "type": "object",
+        "properties": {
+            "query": {
+                "type": "string",
+                "description": "Search query string",
+            },
+            "max_results": {
+                "type": "integer",
+                "description": f"Maximum number of results (1-{_MAX_RESULTS}, default {_DEFAULT_RESULTS})",
+                "default": _DEFAULT_RESULTS,
+            },
+        },
+        "required": ["query"],
+    }
+
+    def is_read_only(self) -> bool:
+        return True
+
+    def get_activity_description(self, **kwargs) -> str | None:
+        query = kwargs.get("query", "")
+        return f"Searching: {query}" if query else None
+
+    def execute(self, query: str, max_results: int = _DEFAULT_RESULTS, **kwargs) -> ToolResult:
+        if not query.strip():
+            return ToolResult(content="Error: query must not be empty.", is_error=True)
+
+        max_results = min(max(max_results, 1), _MAX_RESULTS)
+
+        encoded_query = urllib.parse.quote_plus(query)
+        url = f"https://www.bing.com/search?q={encoded_query}"
+
+        # 代理支持：读取环境变量 HTTPS_PROXY / HTTP_PROXY
+        proxy_url = os.environ.get("HTTPS_PROXY") or os.environ.get("HTTP_PROXY")
+        opener = (
+            urllib.request.build_opener(
+                urllib.request.ProxyHandler({"https": proxy_url, "http": proxy_url})
+            )
+            if proxy_url
+            else urllib.request.build_opener()
+        )
+
+        try:
+            req = urllib.request.Request(
+                url,
+                headers={
+                    "User-Agent": (
+                        "Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
+                        "AppleWebKit/537.36 (KHTML, like Gecko) "
+                        "Chrome/120.0.0.0 Safari/537.36"
+                    ),
+                },
+            )
+            with opener.open(req, timeout=15) as resp:
+                raw_bytes = resp.read(200_000)
+        except urllib.error.HTTPError as e:
+            return ToolResult(content=f"Search error: HTTP {e.code} {e.reason}", is_error=True)
+        except urllib.error.URLError as e:
+            return ToolResult(content=f"Search error: {e.reason}", is_error=True)
+        except TimeoutError:
+            return ToolResult(content="Search error: request timed out after 15s", is_error=True)
+        except Exception as e:
+            return ToolResult(content=f"Search error: {e}", is_error=True)
+
+        charset = resp.headers.get_content_charset() or "utf-8"
+        try:
+            raw_html = raw_bytes.decode(charset, errors="replace")
+        except LookupError:
+            raw_html = raw_bytes.decode("utf-8", errors="replace")
+
+        parser = _BingParser()
+        try:
+            parser.feed(raw_html)
+            parser.finalize()
+        except Exception:
+            return ToolResult(content="Search error: failed to parse results.", is_error=True)
+
+        results = parser.results[:max_results]
+
+        if not results:
+            return ToolResult(content="No results found.")
+
+        lines = []
+        total = 0
+        for i, r in enumerate(results, 1):
+            title = r["title"] or "(no title)"
+            href = r["href"]
+            # 清理 HTML 标签和实体
+            snippet = html.unescape(re.sub(r"<[^>]+>", "", r["snippet"].strip()))
+            line = f"{i}. **{title}**\n   {href}\n   {snippet}"
+            total += len(line)
+            if total > _MAX_CHARS:
+                if i == 1:
+                    lines.append(line[:_MAX_CHARS] + "...")
+                break
+            lines.append(line)
+
+        return ToolResult(content="\n\n".join(lines))