proxy-scraper-cli 1.19.0__tar.gz → 1.20.0__tar.gz

{proxy_scraper_cli-1.19.0 → proxy_scraper_cli-1.20.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: proxy-scraper-cli
-Version: 1.19.0
+Version: 1.20.0
 Summary: Finds free HTTP, SOCKS4 and SOCKS5 proxies that actually work: 700+ sources, real checks, honeypot filtering and a rotating local proxy server.
 Author: Maximilian Feix
 License-Expression: MIT
@@ -456,6 +456,21 @@ with ProxyRotator(country="DE") as rotator:
 
 Usually a few seconds; when many proxies in a row are down it can take a minute (`timeout=` is per attempt). There's `aget()` with `async with` for asyncio code.
 
+Already using requests, httpx, Playwright or Scrapy? `proxy_url()` gives you one local proxy address that rotates behind the scenes – every connection goes out through another proxy from the list, best first, dead ones skipped:
+
+```python
+import requests
+from proxyscraper import ProxyRotator
+
+with ProxyRotator(country="DE") as rotator:
+    proxy = rotator.proxy_url()          # http://country-de:…@127.0.0.1:…
+    requests.get("https://api.ipify.org", proxies={"http": proxy, "https": proxy})
+    # httpx.Client(proxy=proxy) · scrapy: meta={"proxy": proxy} · curl -x "$proxy"
+    # Playwright wants the login apart: chromium.launch(proxy=rotator.playwright_proxy())
+```
+
+It runs on 127.0.0.1 with a random password, takes over each new hourly list by itself and stops with the `with` block.
+
 <a id="proxy-server"></a>
 
 ## Rotating proxy server

{proxy_scraper_cli-1.19.0 → proxy_scraper_cli-1.20.0}/README.md

@@ -415,6 +415,21 @@ with ProxyRotator(country="DE") as rotator:
 
 Usually a few seconds; when many proxies in a row are down it can take a minute (`timeout=` is per attempt). There's `aget()` with `async with` for asyncio code.
 
+Already using requests, httpx, Playwright or Scrapy? `proxy_url()` gives you one local proxy address that rotates behind the scenes – every connection goes out through another proxy from the list, best first, dead ones skipped:
+
+```python
+import requests
+from proxyscraper import ProxyRotator
+
+with ProxyRotator(country="DE") as rotator:
+    proxy = rotator.proxy_url()          # http://country-de:…@127.0.0.1:…
+    requests.get("https://api.ipify.org", proxies={"http": proxy, "https": proxy})
+    # httpx.Client(proxy=proxy) · scrapy: meta={"proxy": proxy} · curl -x "$proxy"
+    # Playwright wants the login apart: chromium.launch(proxy=rotator.playwright_proxy())
+```
+
+It runs on 127.0.0.1 with a random password, takes over each new hourly list by itself and stops with the `with` block.
+
 <a id="proxy-server"></a>
 
 ## Rotating proxy server

{proxy_scraper_cli-1.19.0 → proxy_scraper_cli-1.20.0}/proxy_scraper_cli.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: proxy-scraper-cli
-Version: 1.19.0
+Version: 1.20.0
 Summary: Finds free HTTP, SOCKS4 and SOCKS5 proxies that actually work: 700+ sources, real checks, honeypot filtering and a rotating local proxy server.
 Author: Maximilian Feix
 License-Expression: MIT
@@ -456,6 +456,21 @@ with ProxyRotator(country="DE") as rotator:
 
 Usually a few seconds; when many proxies in a row are down it can take a minute (`timeout=` is per attempt). There's `aget()` with `async with` for asyncio code.
 
+Already using requests, httpx, Playwright or Scrapy? `proxy_url()` gives you one local proxy address that rotates behind the scenes – every connection goes out through another proxy from the list, best first, dead ones skipped:
+
+```python
+import requests
+from proxyscraper import ProxyRotator
+
+with ProxyRotator(country="DE") as rotator:
+    proxy = rotator.proxy_url()          # http://country-de:…@127.0.0.1:…
+    requests.get("https://api.ipify.org", proxies={"http": proxy, "https": proxy})
+    # httpx.Client(proxy=proxy) · scrapy: meta={"proxy": proxy} · curl -x "$proxy"
+    # Playwright wants the login apart: chromium.launch(proxy=rotator.playwright_proxy())
+```
+
+It runs on 127.0.0.1 with a random password, takes over each new hourly list by itself and stops with the `with` block.
+
 <a id="proxy-server"></a>
 
 ## Rotating proxy server

{proxy_scraper_cli-1.19.0 → proxy_scraper_cli-1.20.0}/proxyscraper/__init__.py

@@ -1,7 +1,7 @@
 """proxy-scraper: collects public HTTP/SOCKS4/SOCKS5 proxies from hundreds of sources, checks them
 in parallel with its own protocol handshakes and learns which sources deliver good proxies."""
 
-__version__ = "1.19.0"
+__version__ = "1.20.0"
 
 
 def __getattr__(name):

{proxy_scraper_cli-1.19.0 → proxy_scraper_cli-1.20.0}/proxyscraper/agent.py

@@ -458,6 +458,22 @@ def _to_text(body: bytes, headers, content_type: str, raw_html: bool) -> str:
     return text if raw_html or "html" not in content_type.lower() else html_to_text(text)
 
 
+def _require_match(live: LiveList, protocol: str, countries: List[str], tls: bool) -> None:
+    if not any(True for _ in matching(live.rows, protocol, countries, https_only=tls, run_hours=live.run_hours)):
+        what = " ".join(x for x in (countries[0] if countries else "", "" if protocol == "any" else protocol,
+                                    "HTTPS-capable" if tls else "") if x)
+        raise AgentError(f"No {what} proxy in the live list right now."
+                         + (shortage_note(live.rows, 1, 0, countries) or ""))
+
+
+def _wishes(protocol: str, countries: List[str]) -> List[str]:
+    """What the rotating server reads from the proxy login: country-de, type-socks5."""
+    wishes = [f"country-{countries[0].lower()}"] if countries else []
+    if protocol != "any":
+        wishes.append(f"type-{protocol}")
+    return wishes
+
+
 class PageFetcher:
     """Loads pages through the rotating server of this package, running on 127.0.0.1 with a random password:
     the same failover as `--serve`, and HTTPS only through proxies that passed the verified-TLS test.
@@ -503,6 +519,19 @@ class PageFetcher:
                 self._list_id = list_id
         return live
 
+    async def local_proxy(self, protocol: str = "any", country: str = "") -> str:
+        """The URL of the rotating server behind this fetcher, with the login, for any HTTP client:
+        http://country-de:<password>@127.0.0.1:<port>. Every connection through it gets a proxy of its own."""
+        protocol = _check_protocol(protocol)
+        countries = normalize_countries([country] if country else [])
+        _require_match(await self._ready(), protocol, countries, tls=False)
+        user = "-".join(_wishes(protocol, countries)) or "any"
+        return f"http://{user}:{self._password}@127.0.0.1:{self.server.port}"
+
+    async def refresh(self) -> None:
+        """Take over a new run's list, if there is one (fetch does that by itself)."""
+        await self._ready()
+
     async def fetch(self, url: str, protocol: str = "any", country: str = "", max_chars: int = 20000,
                     raw_html: bool = False, progress: Optional[Progress] = None, tick: float = 5.0,
                     with_body: bool = False) -> dict:
@@ -511,15 +540,9 @@ class PageFetcher:
         countries = normalize_countries([country] if country else [])
         live = await self._ready()
         tls = urlsplit(url).scheme == "https"
-        if not any(True for _ in matching(live.rows, protocol, countries, https_only=tls, run_hours=live.run_hours)):
-            what = " ".join(x for x in (countries[0] if countries else "", "" if protocol == "any" else protocol,
-                                        "HTTPS-capable" if tls else "") if x)
-            raise AgentError(f"No {what} proxy in the live list right now."
-                             + (shortage_note(live.rows, 1, 0, countries) or ""))
+        _require_match(live, protocol, countries, tls)
         session = secrets.token_hex(6)
-        wishes = [f"country-{countries[0].lower()}"] if countries else []
-        if protocol != "any":
-            wishes.append(f"type-{protocol}")
+        wishes = _wishes(protocol, countries)
         pool, port = self.server.pool, self.server.port
         started = time.monotonic()
         result, faults, suspects = None, [], []

{proxy_scraper_cli-1.19.0 → proxy_scraper_cli-1.20.0}/proxyscraper/api.py

@@ -27,12 +27,14 @@ import asyncio
 import contextlib
 import email.message
 import io
+import logging
 import os
 import re
 import tempfile
 import threading
 from dataclasses import dataclass, field
 from typing import Dict, Iterable, List, Optional, Tuple
+from urllib.parse import urlsplit
 
 from rich.console import Console
 
@@ -42,6 +44,8 @@ from .parsing import PROXY_TYPES
 from .targets import parse_target
 from .ui import widgets
 
+log = logging.getLogger(__name__)
+
 __all__ = [
     "CheckResult",
     "LiveProxy",
@@ -309,6 +313,15 @@ class ProxyResponse:
             raise ConnectionError(f"the body from {self.final_url} was cut off at 2 MB")
 
 
+REFRESH_SECONDS = 300  # proxy_url(): how often the pool behind it looks for a new run's list
+
+
+def _playwright(url: str) -> Dict[str, str]:
+    parts = urlsplit(url)
+    return {"server": f"{parts.scheme}://{parts.hostname}:{parts.port}", "username": parts.username or "",
+            "password": parts.password or ""}
+
+
 class ProxyRotator:
     """Loads URLs through proxies from the hourly list and switches to the next one when a proxy fails –
     the loop every scraper writes by hand. HTTPS only through proxies with verified TLS.
@@ -330,6 +343,7 @@ class ProxyRotator:
         self._loop: Optional[asyncio.AbstractEventLoop] = None  # the sync API's own loop, in a thread
         self._thread: Optional[threading.Thread] = None
         self._start_lock = threading.Lock()
+        self._refresher: Optional[asyncio.Task] = None
 
     def _use(self, mode: str) -> None:
         if self._mode is None:
@@ -367,7 +381,43 @@ class ProxyRotator:
                              proxy_country=page["proxy_country"], elapsed_ms=page["elapsed_ms"],
                              truncated=page["body_truncated"])
 
+    async def aproxy_url(self) -> str:
+        self._use("async")
+        return await self._proxy_url()
+
+    async def _proxy_url(self) -> str:
+        from .agent import AgentError
+
+        if self._fetcher is None:
+            self._fetcher = self._make_fetcher()
+        try:
+            url = await self._fetcher.local_proxy(self.protocol, self.country)
+        except AgentError as e:
+            raise ConnectionError(str(e)) from None
+        if self._refresher is None:  # nobody calls get() on this path, so the pool refreshes by itself
+            self._refresher = asyncio.ensure_future(self._keep_fresh())
+        return url
+
+    async def aplaywright_proxy(self) -> Dict[str, str]:
+        self._use("async")
+        return _playwright(await self._proxy_url())
+
+    async def _keep_fresh(self) -> None:
+        from .agent import AgentError
+
+        while True:
+            await asyncio.sleep(REFRESH_SECONDS)
+            try:
+                await self._fetcher.refresh()
+            except AgentError:
+                pass  # GitHub not reachable right now: the pool keeps what it has, next try later
+            except Exception as e:  # a broken list must not end the refreshing for good
+                log.warning("proxy-scraper: could not refresh the proxy list: %s", e)
+
     async def aclose(self) -> None:
+        if self._refresher is not None:
+            self._refresher.cancel()
+            self._refresher = None
         if self._fetcher is not None:
             await self._fetcher.close()
             self._fetcher = None
@@ -380,13 +430,37 @@ class ProxyRotator:
 
     # sync: one event loop in a background thread keeps the proxy server alive between calls
     def get(self, url: str) -> ProxyResponse:
-        with self._start_lock:  # several threads calling get() at once still share one loop
-            self._use("sync")
+        return self._run_sync(self._aget(url))
+
+    def _run_sync(self, coro):
+        with self._start_lock:  # several threads calling at once still share one loop
+            try:
+                self._use("sync")
+            except RuntimeError:
+                coro.close()
+                raise
             if self._loop is None:
                 self._loop = asyncio.new_event_loop()
                 self._thread = threading.Thread(target=self._loop.run_forever, name="proxy-rotator", daemon=True)
                 self._thread.start()
-        return asyncio.run_coroutine_threadsafe(self._aget(url), self._loop).result()
+        return asyncio.run_coroutine_threadsafe(coro, self._loop).result()
+
+    def proxy_url(self) -> str:
+        """A local proxy URL for any library – requests, httpx, aiohttp, Playwright, Scrapy, curl:
+
+            with ProxyRotator(country="DE") as rotator:
+                proxy = rotator.proxy_url()
+                requests.get(url, proxies={"http": proxy, "https": proxy})
+
+        Every connection through it gets a proxy from the live list (best first, with failover), HTTPS only
+        through verified-TLS ones. It lives as long as the rotator and takes over each new run's list."""
+        return self._run_sync(self._proxy_url())
+
+    def playwright_proxy(self) -> Dict[str, str]:
+        """proxy_url() the way Playwright wants it – it ignores a login inside the server address:
+
+            browser = p.chromium.launch(proxy=rotator.playwright_proxy())"""
+        return _playwright(self.proxy_url())
 
     def close(self) -> None:
         with self._start_lock:

{proxy_scraper_cli-1.19.0 → proxy_scraper_cli-1.20.0}/tests/test_rotator.py

@@ -127,3 +127,95 @@ def test_mixing_sync_and_async_is_refused(monkeypatch):
     with pytest.raises(RuntimeError, match="async"):
         asyncio.run(rotator.aget("http://x/"))
     rotator.close()
+
+
+def get_through(proxy_url: str, url: str):
+    """A plain GET through an HTTP proxy URL with a login, the way requests or httpx send it."""
+    import base64
+    import http.client
+    from urllib.parse import urlsplit
+
+    p = urlsplit(proxy_url)
+    login = base64.b64encode(f"{p.username}:{p.password}".encode()).decode()
+    conn = http.client.HTTPConnection(p.hostname, p.port, timeout=10)
+    conn.request("GET", url, headers={"Proxy-Authorization": f"Basic {login}"})
+    response = conn.getresponse()
+    return response.status, response.read()
+
+
+def test_proxy_url_is_a_rotating_proxy_for_any_library():
+    loop = asyncio.new_event_loop()
+    target_srv, proxy_srv, port, rows = loop.run_until_complete(local_setup())
+    import threading
+    thread = threading.Thread(target=loop.run_forever, daemon=True)
+    thread.start()
+    try:
+        with ProxyRotator(_source=source_for(rows), _allow_private=True, timeout=5) as rotator:
+            url = rotator.proxy_url()
+            assert url.startswith("http://") and "@127.0.0.1:" in url
+            # the dead proxy is tried and skipped inside, the client just gets its page
+            assert get_through(url, f"http://127.0.0.1:{port}/ok") == (200, b"ok")
+            assert rotator.proxy_url() == url  # the same server for the rotator's lifetime
+    finally:
+        loop.call_soon_threadsafe(target_srv.close)
+        loop.call_soon_threadsafe(proxy_srv.close)
+        loop.call_soon_threadsafe(loop.stop)
+
+
+def test_proxy_url_carries_country_and_protocol_and_refuses_what_isnt_there():
+    async def go(**wishes):
+        rows = [row(1, country="DE", ptype="socks5")]
+        async with ProxyRotator(_source=source_for(rows), _allow_private=True, **wishes) as rotator:
+            return await rotator.aproxy_url()
+    assert "//country-de-type-socks5:" in asyncio.run(go(country="de", protocol="socks5"))
+
+    async def for_playwright():
+        async with ProxyRotator(_source=source_for([row(1)]), _allow_private=True) as rotator:
+            return await rotator.aproxy_url(), await rotator.aplaywright_proxy()
+    url, settings = asyncio.run(for_playwright())
+    # Playwright ignores a login inside "server": it wants the parts on their own
+    assert settings["server"] == "http://" + url.rsplit("@", 1)[1] and "@" not in settings["server"]
+    assert url == f"http://{settings['username']}:{settings['password']}@{settings['server'][7:]}"
+    with pytest.raises(ConnectionError, match="JP"):
+        asyncio.run(go(country="JP"))
+
+
+def test_proxy_url_keeps_the_pool_fresh(monkeypatch):
+    from proxyscraper import api
+
+    monkeypatch.setattr(api, "REFRESH_SECONDS", 0.05)
+    runs = iter([agent.LiveList(rows=[row(1)], stats={**STATS, "updated": "run 1"}, loaded_at=0.0),
+                 agent.LiveList(rows=[row(2)], stats={**STATS, "updated": "run 2"}, loaded_at=0.0)])
+
+    class Source:  # the first call sees run 1, every later one run 2
+        current = None
+
+        async def get(self):
+            self.current = next(runs, self.current)
+            return self.current
+
+    async def go():
+        async with ProxyRotator(_source=Source(), _allow_private=True) as rotator:
+            await rotator.aproxy_url()
+            await asyncio.sleep(0.3)
+            return {e.result.proxy for e in rotator._fetcher.server.pool.entries}
+    assert asyncio.run(go()) == {row(2)["proxy"]}
+
+
+def test_the_refresh_survives_a_surprise(monkeypatch):
+    from proxyscraper import api
+
+    monkeypatch.setattr(api, "REFRESH_SECONDS", 0.02)
+    calls = []
+
+    async def go():
+        async with ProxyRotator(_source=source_for([row(1)]), _allow_private=True) as rotator:
+            await rotator.aproxy_url()
+
+            async def broken():
+                calls.append(1)
+                raise KeyError("stats")
+            rotator._fetcher.refresh = broken
+            await asyncio.sleep(0.2)
+    asyncio.run(go())
+    assert len(calls) > 2  # kept trying after the first failure