drydock-cli 3.0.49__tar.gz → 3.0.50__tar.gz

{drydock_cli-3.0.49/drydock_cli.egg-info → drydock_cli-3.0.50}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: drydock-cli
-Version: 3.0.49
+Version: 3.0.50
 Summary: Drydock — a local, provider-agnostic terminal coding agent for local LLMs
 Author: Frank Bobe III
 License-Expression: Apache-2.0

{drydock_cli-3.0.49 → drydock_cli-3.0.50}/drydock/tools/__init__.py

@@ -161,6 +161,40 @@ SCHEMAS = [
             "required": ["prompt"],
         },
     },
+    {
+        "name": "WebSearch",
+        "description": (
+            "Search the internet and get back the top results (title, URL, "
+            "snippet). Use it for current events, docs, library/API details, "
+            "error messages, or anything you're unsure about or that may have "
+            "changed since training. Follow up with WebFetch to read a result in "
+            "full. Returns a clean message if the machine is offline."
+        ),
+        "input_schema": {
+            "type": "object",
+            "properties": {
+                "query": {"type": "string", "description": "The search query."},
+                "k": {"type": "integer", "description": "Max results (default 5)."},
+            },
+            "required": ["query"],
+        },
+    },
+    {
+        "name": "WebFetch",
+        "description": (
+            "Fetch a web page (or any URL) and return its readable text content, "
+            "HTML stripped. Use it to read a page found via WebSearch or a URL the "
+            "user gave you. Returns a clean message if the machine is offline."
+        ),
+        "input_schema": {
+            "type": "object",
+            "properties": {
+                "url": {"type": "string", "description": "The URL to fetch."},
+                "max_chars": {"type": "integer", "description": "Max chars to return (default 6000)."},
+            },
+            "required": ["url"],
+        },
+    },
     {
         "name": "Knowledge",
         "description": (
@@ -735,6 +769,42 @@ def tool_task(params: dict, config: dict) -> str:
     return f"[sub-agent finished {steps} step(s) with no summary]"
 
 
+def tool_websearch(params: dict, config: dict) -> str:
+    """Search the internet (DuckDuckGo). Read-only; clean message when offline."""
+    from drydock import web
+
+    query = (params.get("query") or "").strip()
+    if not query:
+        return "Error: `WebSearch` needs a `query`."
+    try:
+        k = int(params.get("k") or 5)
+    except (TypeError, ValueError):
+        k = 5
+    try:
+        results = web.search(query, k=max(1, min(k, 10)))
+    except web.WebError as e:
+        return f"Web search unavailable: {e}. You appear to be offline — answer from your own knowledge."
+    return web.format_search(query, results)
+
+
+def tool_webfetch(params: dict, config: dict) -> str:
+    """Fetch a URL and return readable text. Read-only; clean message offline."""
+    from drydock import web
+
+    url = (params.get("url") or "").strip()
+    if not url:
+        return "Error: `WebFetch` needs a `url`."
+    try:
+        mc = int(params.get("max_chars") or 6000)
+    except (TypeError, ValueError):
+        mc = 6000
+    try:
+        text = web.fetch(url, max_chars=max(500, min(mc, 30000)))
+    except web.WebError as e:
+        return f"Could not fetch the page: {e}."
+    return text or f"(no readable text extracted from {url})"
+
+
 def tool_knowledge(params: dict, config: dict) -> str:
     """Query the project's GraphRAG knowledge base (built with /graphrag build).
     Read-only; returns the most relevant passages plus related graph entities.
@@ -773,6 +843,8 @@ _TOOLS = [
     ("todo", tool_todo, False),
     ("task", tool_task, True),
     ("Knowledge", tool_knowledge, True),
+    ("WebSearch", tool_websearch, True),
+    ("WebFetch", tool_webfetch, True),
 ]
 
 def register_all():
@@ -782,9 +854,12 @@ def register_all():
             "Read": tool_read, "Write": tool_write, "Edit": tool_edit,
             "Bash": tool_bash, "Glob": tool_glob, "Grep": tool_grep,
             "todo": tool_todo, "task": tool_task, "Knowledge": tool_knowledge,
+            "WebSearch": tool_websearch, "WebFetch": tool_webfetch,
         }[name]
-        # task + Knowledge are read-only w.r.t. the parent's files.
-        read_only = name in ("Read", "Glob", "Grep", "task", "Knowledge")
+        # These are read-only w.r.t. the parent's files.
+        read_only = name in (
+            "Read", "Glob", "Grep", "task", "Knowledge", "WebSearch", "WebFetch",
+        )
         register(ToolDef(name=name, schema=schema, func=func, read_only=read_only))
 
 register_all()

drydock_cli-3.0.50/drydock/web.py

@@ -0,0 +1,98 @@
+"""Internet access — web search + page fetch, stdlib only (no deps, no API key).
+
+Search goes through DuckDuckGo's HTML endpoint (a form POST; the GET form is
+bot-challenged). Fetch pulls a page and reduces it to readable text. Both are
+best-effort and degrade cleanly when offline — drydock targets local LLMs that
+may run air-gapped, so a network failure returns a plain message, never raises.
+
+All logic original to Drydock.
+"""
+from __future__ import annotations
+
+import html
+import re
+import urllib.error
+import urllib.parse
+import urllib.request
+
+_UA = "Mozilla/5.0 (X11; Linux x86_64; rv:124.0) Gecko/20100101 Firefox/124.0"
+_DDG = "https://html.duckduckgo.com/html/"
+
+_RE_RESULT = re.compile(r'result__a"[^>]*href="([^"]+)".*?>(.*?)</a>', re.S)
+_RE_SNIPPET = re.compile(r'result__snippet"[^>]*>(.*?)</a>', re.S)
+_RE_TAG = re.compile(r"<[^>]+>")
+_RE_SCRIPT = re.compile(r"<(script|style|noscript|svg)\b.*?</\1>", re.S | re.I)
+_RE_WS = re.compile(r"[ \t]+")
+_RE_BLANKS = re.compile(r"\n\s*\n\s*\n+")
+
+
+def _strip_html(s: str) -> str:
+    return html.unescape(_RE_TAG.sub("", s)).strip()
+
+
+class WebError(RuntimeError):
+    """Network/HTTP failure reaching the internet (carries a user-facing msg)."""
+
+
+def search(query: str, k: int = 5, *, timeout: float = 12.0) -> list[dict]:
+    """Return up to k web results: [{title, url, snippet}]. Raises WebError on a
+    network failure so the caller can show a clean offline message."""
+    data = urllib.parse.urlencode({"q": query}).encode()
+    req = urllib.request.Request(_DDG, data=data, headers={"User-Agent": _UA})
+    try:
+        with urllib.request.urlopen(req, timeout=timeout) as r:  # noqa: S310 (trusted)
+            body = r.read().decode("utf-8", "ignore")
+    except (urllib.error.URLError, OSError) as e:
+        raise WebError(f"could not reach the internet ({e})") from e
+
+    links = _RE_RESULT.findall(body)
+    snippets = _RE_SNIPPET.findall(body)
+    out: list[dict] = []
+    for (url, title), snip in zip(links, snippets + [""] * len(links)):
+        url = html.unescape(url)
+        if url.startswith("//"):
+            url = "https:" + url
+        # DDG sometimes wraps the target in a redirect with ?uddg=<real-url>.
+        m = re.search(r"[?&]uddg=([^&]+)", url)
+        if m:
+            url = urllib.parse.unquote(m.group(1))
+        out.append({"title": _strip_html(title), "url": url, "snippet": _strip_html(snip)})
+        if len(out) >= k:
+            break
+    return out
+
+
+def fetch(url: str, *, max_chars: int = 6000, timeout: float = 12.0) -> str:
+    """Fetch a URL and return its readable text (HTML stripped). Raises WebError
+    on a network/HTTP failure."""
+    if not url.startswith(("http://", "https://")):
+        url = "https://" + url
+    req = urllib.request.Request(url, headers={"User-Agent": _UA})
+    try:
+        with urllib.request.urlopen(req, timeout=timeout) as r:  # noqa: S310 (user-supplied URL)
+            ctype = r.headers.get("Content-Type", "")
+            raw = r.read(4_000_000)  # cap the download
+    except (urllib.error.URLError, OSError) as e:
+        raise WebError(f"could not fetch {url} ({e})") from e
+
+    text = raw.decode("utf-8", "ignore")
+    if "html" in ctype.lower() or "<html" in text[:2000].lower():
+        text = _RE_SCRIPT.sub(" ", text)
+        text = _RE_TAG.sub(" ", text)
+        text = html.unescape(text)
+    text = _RE_WS.sub(" ", text)
+    text = _RE_BLANKS.sub("\n\n", "\n".join(ln.strip() for ln in text.splitlines()))
+    text = text.strip()
+    if len(text) > max_chars:
+        text = text[:max_chars] + f"\n[... truncated, {len(text) - max_chars} more chars ...]"
+    return text
+
+
+def format_search(query: str, results: list[dict]) -> str:
+    if not results:
+        return f"No web results for {query!r}."
+    parts = [f"Web results for {query!r}:"]
+    for i, r in enumerate(results, 1):
+        parts.append(f"\n[{i}] {r['title']}\n    {r['url']}\n    {r['snippet']}")
+    parts.append("\nTo read one in full, use WebFetch with its URL.")
+    return "\n".join(parts)

{drydock_cli-3.0.49 → drydock_cli-3.0.50/drydock_cli.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: drydock-cli
-Version: 3.0.49
+Version: 3.0.50
 Summary: Drydock — a local, provider-agnostic terminal coding agent for local LLMs
 Author: Frank Bobe III
 License-Expression: Apache-2.0

{drydock_cli-3.0.49 → drydock_cli-3.0.50}/drydock_cli.egg-info/SOURCES.txt

@@ -16,6 +16,7 @@ drydock/loop_detect.py
 drydock/providers.py
 drydock/tool_registry.py
 drydock/tuning.py
+drydock/web.py
 drydock/tools/__init__.py
 drydock/tui/__init__.py
 drydock/tui/app.py
@@ -60,4 +61,5 @@ tests/test_tool_arg_parsing.py
 tests/test_tools_undo.py
 tests/test_tui.py
 tests/test_tuning.py
-tests/test_vision_input.py
+tests/test_vision_input.py
+tests/test_web_tools.py

{drydock_cli-3.0.49 → drydock_cli-3.0.50}/pyproject.toml

@@ -7,7 +7,7 @@ build-backend = "setuptools.build_meta"
 # PyPI distribution name is drydock-cli (the established install name, continued
 # from the retired v2 fork); the import package + CLI command stay `drydock`.
 name = "drydock-cli"
-version = "3.0.49"
+version = "3.0.50"
 description = "Drydock — a local, provider-agnostic terminal coding agent for local LLMs"
 readme = "README.md"
 requires-python = ">=3.11"

drydock_cli-3.0.50/tests/test_web_tools.py

@@ -0,0 +1,72 @@
+"""Internet access: WebSearch + WebFetch tools (drydock/web.py). Network is
+mocked so tests are deterministic and run offline; the offline path is tested
+explicitly (drydock targets local LLMs that may be air-gapped — must degrade,
+never raise)."""
+from __future__ import annotations
+
+
+from drydock import web
+from drydock.tools import tool_websearch, tool_webfetch
+
+
+def test_websearch_formats_results(monkeypatch):
+    monkeypatch.setattr(web, "search", lambda q, k=5, **kw: [
+        {"title": "Python 3.13", "url": "https://docs.python.org/3/whatsnew/3.13.html",
+         "snippet": "new features"},
+    ])
+    out = tool_websearch({"query": "python 3.13 features"}, {})
+    assert "Python 3.13" in out and "docs.python.org" in out and "WebFetch" in out
+
+
+def test_websearch_offline_is_graceful(monkeypatch):
+    def boom(*a, **k):
+        raise web.WebError("could not reach the internet (offline)")
+    monkeypatch.setattr(web, "search", boom)
+    out = tool_websearch({"query": "anything"}, {})
+    assert "unavailable" in out.lower() and "offline" in out.lower()
+
+
+def test_websearch_needs_query():
+    assert "needs a `query`" in tool_websearch({}, {})
+
+
+def test_webfetch_returns_text(monkeypatch):
+    monkeypatch.setattr(web, "fetch", lambda url, **kw: "Readable page text here.")
+    out = tool_webfetch({"url": "example.com"}, {})
+    assert out == "Readable page text here."
+
+
+def test_webfetch_offline_is_graceful(monkeypatch):
+    def boom(*a, **k):
+        raise web.WebError("could not fetch (offline)")
+    monkeypatch.setattr(web, "fetch", boom)
+    out = tool_webfetch({"url": "https://x.com"}, {})
+    assert "could not fetch" in out.lower()
+
+
+def test_webfetch_needs_url():
+    assert "needs a `url`" in tool_webfetch({}, {})
+
+
+def test_strip_html_and_parsing_helpers():
+    # The HTML→text reducer drops tags + unescapes entities.
+    assert web._strip_html("<b>Hello &amp; bye</b>") == "Hello & bye"
+
+
+def test_search_parses_ddg_html(monkeypatch):
+    sample = (
+        '<a class="result__a" href="//duckduckgo.com/l/?uddg=https%3A%2F%2Fexample.com%2Fa">'
+        'Example A</a>'
+        '<a class="result__snippet" href="x">snippet A text</a>'
+    )
+
+    class _Resp:
+        def __enter__(self): return self
+        def __exit__(self, *a): return False
+        def read(self): return sample.encode()
+
+    monkeypatch.setattr(web.urllib.request, "urlopen", lambda *a, **k: _Resp())
+    res = web.search("q", k=5)
+    assert res and res[0]["title"] == "Example A"
+    assert res[0]["url"] == "https://example.com/a"   # uddg redirect unwrapped
+    assert res[0]["snippet"] == "snippet A text"