bas-http 1.0.0__py3-none-any.whl

bas/_curl.py

@@ -0,0 +1,393 @@
+"""
+Low-level ctypes bindings for libcurl with curl-impersonate support.
+
+This module provides Python bindings to libcurl (or curl-impersonate if available)
+using ctypes. It avoids cffi compilation issues that plague curlcffi.
+"""
+
+import ctypes
+import ctypes.util
+import os
+import sys
+from ctypes import (
+    CFUNCTYPE,
+    POINTER,
+    c_char,
+    c_char_p,
+    c_int,
+    c_long,
+    c_off_t,
+    c_void_p,
+    c_slist,
+)
+from typing import Optional
+
+# ── libcurl loading ──────────────────────────────────────────────────────────
+# Priority: curl-impersonate first, then system libcurl.
+
+_CURL_LIB_NAMES = [
+    # curl-impersonate shared libraries
+    "curl-impersonate-chrome",
+    "curl-impersonate-firefox",
+    "libcurl-impersonate-chrome",
+    "libcurl-impersonate-firefox",
+    # Standard libcurl
+    "curl",
+    "libcurl",
+    "libcurl.so.4",
+    "libcurl.4.dylib",
+    "curl.dll",
+]
+
+
+def _find_curl_library() -> ctypes.CDLL:
+    """Locate and load the best available curl library."""
+    # 1. Check FREEBUFF_CURL_PATH env var
+    env_path = os.environ.get("FREEBUFF_CURL_PATH")
+    if env_path and os.path.isfile(env_path):
+        return ctypes.CDLL(env_path)
+
+    # 2. Check system paths
+    for name in _CURL_LIB_NAMES:
+        found = ctypes.util.find_library(name)
+        if found:
+            try:
+                return ctypes.CDLL(found)
+            except OSError:
+                continue
+
+    # 3. Try direct load on common paths
+    if sys.platform == "linux":
+        for path in [
+            "/usr/lib/x86_64-linux-gnu/libcurl.so.4",
+            "/usr/lib/libcurl.so.4",
+            "/usr/local/lib/libcurl.so.4",
+        ]:
+            if os.path.isfile(path):
+                return ctypes.CDLL(path)
+    elif sys.platform == "darwin":
+        for path in [
+            "/usr/local/lib/libcurl.4.dylib",
+            "/opt/homebrew/lib/libcurl.4.dylib",
+        ]:
+            if os.path.isfile(path):
+                return ctypes.CDLL(path)
+
+    raise OSError(
+        "Could not find libcurl. Install libcurl-dev (Linux) or "
+        "curl-impersonate (for browser impersonation). "
+        "Set FREEBUFF_CURL_PATH to a custom library path."
+    )
+
+
+_lib = _find_curl_library()
+
+
+# ── Constants ────────────────────────────────────────────────────────────────
+
+# CURLoptions
+CURL_GLOBAL_ALL = 3
+CURLOPT_URL = 10002
+CURLOPT_WRITEFUNCTION = 20011
+CURLOPT_WRITEDATA = 10001
+CURLOPT_HEADERFUNCTION = 20079
+CURLOPT_HEADERDATA = 10029
+CURLOPT_POSTFIELDS = 10015
+CURLOPT_POST = 47
+CURLOPT_CUSTOMREQUEST = 10036
+CURLOPT_FOLLOWLOCATION = 52
+CURLOPT_MAXREDIRS = 68
+CURLOPT_COOKIEFILE = 10031
+CURLOPT_COOKIEJAR = 10082
+CURLOPT_COOKIELIST = 10135
+CURLOPT_COOKIE = 10022
+CURLOPT_HTTPHEADER = 10023
+CURLOPT_HTTPGET = 80
+CURLOPT_SSL_VERIFYPEER = 64
+CURLOPT_SSL_VERIFYHOST = 81
+CURLOPT_CAINFO = 10065
+CURLOPT_TIMEOUT = 13
+CURLOPT_CONNECTTIMEOUT = 78
+CURLOPT_USERAGENT = 10018
+CURLOPT_REFERER = 10016
+CURLOPT_ENCODING = 10106
+CURLOPT_PROXY = 10004
+CURLOPT_PROXYPORT = 59
+CURLOPT_PROXYTYPE = 101
+CURLOPT_NOPROXY = 10177
+CURLOPT_RESOLVE = 10203
+CURLOPT_INTERFACE = 10062
+CURLOPT_LOCALPORT = 139
+CURLOPT_DNS_CACHE_TIMEOUT = 92
+CURLOPT_TCP_KEEPALIVE = 213
+CURLOPT_TCP_KEEPIDLE = 214
+CURLOPT_TCP_KEEPINTVL = 215
+CURLOPT_SSL_FALSESTART = 224
+CURLOPT_TCP_FASTOPEN = 244
+CURLOPT_HTTP_VERSION = 84
+CURLOPT_CERTINFO = 172
+CURLOPT_PINNEDPUBLICKEY = 10230
+CURLOPT_NO_SIGNAL = 99
+CURLOPT_TRANSFER_ENCODING = 20206
+CURLOPT_HTTP200ALIASES = 10260
+CURLOPT_PREREQFUNCTION = 20328
+CURLOPT_PREREQDATA = 10313
+CURLOPT_MAIL_FROM = 10186
+CURLOPT_MAIL_RCPT = 10187
+CURLOPT_UPLOAD = 46
+CURLOPT_INFILESIZE = 14
+CURLOPT_INFILESIZE_LARGE = 10311
+CURLOPT_UPLOAD = 46
+CURLOPT_PORT = 3
+
+# CURLinfo
+CURLINFO_RESPONSE_CODE = 0x200002
+CURLINFO_EFFECTIVE_URL = 0x100002
+CURLINFO_CONTENT_TYPE = 0x100004
+CURLINFO_REDIRECT_URL = 0x100007
+CURLINFO_HEADER_SIZE = 0x20000B
+CURLINFO_TOTAL_TIME = 0x300003
+CURLINFO_NAMELOOKUP_TIME = 0x300004
+CURLINFO_CONNECT_TIME = 0x300005
+CURLINFO_PRETRANSFER_TIME = 0x300006
+CURLINFO_STARTTRANSFER_TIME = 0x300009
+CURLINFO_PRIMARY_IP = 0x40000C
+CURLINFO_PRIMARY_PORT = 0x40000F
+CURLINFO_SSL_VERIFYRESULT = 0x400008
+CURLINFO_HTTP_VERSION = 0x400012
+
+# CURLcode
+CURL_OK = 0
+
+# CURLMcode
+CURLM_OK = 0
+
+# CURLHTTPVersion
+CURL_HTTP_VERSION_NONE = 0
+CURL_HTTP_VERSION_1_0 = 1
+CURL_HTTP_VERSION_1_1 = 2
+CURL_HTTP_VERSION_2_0 = 3
+CURL_HTTP_VERSION_2TLS = 4
+CURL_HTTP_VERSION_3 = 30
+
+# CURLProxyType
+CURLPROXY_HTTP = 0
+CURLPROXY_SOCKS4 = 4
+CURLPROXY_SOCKS5 = 5
+CURLPROXY_SOCKS4A = 6
+CURLPROXY_SOCKS5_HOSTNAME = 7
+
+
+# ── CURLINFO enum for curl_slist operations ──────────────────────────────────
+
+# CURLOPT_COOKIELIST action
+CURLCOOKIE_SET = 1
+CURLCOOKIE_UNSET = 3
+
+
+# ── Function signatures ──────────────────────────────────────────────────────
+
+_lib.curl_global_init.argtypes = [c_long]
+_lib.curl_global_init.restype = c_int
+
+_lib.curl_easy_init.argtypes = []
+_lib.curl_easy_init.restype = c_void_p
+
+_lib.curl_easy_cleanup.argtypes = [c_void_p]
+_lib.curl_easy_cleanup.restype = None
+
+_lib.curl_easy_setopt.argtypes = [c_void_p, c_int, ...]
+_lib.curl_easy_setopt.restype = c_int
+
+_lib.curl_easy_perform.argtypes = [c_void_p]
+_lib.curl_easy_perform.restype = c_int
+
+_lib.curl_easy_getinfo.argtypes = [c_void_p, c_int, c_void_p]
+_lib.curl_easy_getinfo.restype = c_int
+
+_lib.curl_easy_reset.argtypes = [c_void_p]
+_lib.curl_easy_reset.restype = None
+
+_lib.curl_easy_duphandle.argtypes = [c_void_p]
+_lib.curl_easy_duphandle.restype = c_void_p
+
+_lib.curl_slist_append.argtypes = [POINTER(c_slist), c_char_p]
+_lib.curl_slist_append.restype = POINTER(c_slist)
+
+_lib.curl_slist_free_all.argtypes = [POINTER(c_slist)]
+_lib.curl_slist_free_all.restype = None
+
+_lib.curl_free.argtypes = [c_void_p]
+_lib.curl_free.restype = None
+
+_lib.curl_easy_strerror.argtypes = [c_int]
+_lib.curl_easy_strerror.restype = c_char_p
+
+_lib.curl_version.argtypes = []
+_lib.curl_version.restype = c_char_p
+
+
+# ── Callback types ───────────────────────────────────────────────────────────
+
+WRITEFUNC = CFUNCTYPE(c_int, c_char_p, c_int, c_int, c_void_p)
+HEADERFUNC = CFUNCTYPE(c_int, c_char_p, c_int, c_int, c_void_p)
+
+
+# ── Global init ──────────────────────────────────────────────────────────────
+
+_lib.curl_global_init(CURL_GLOBAL_ALL)
+
+
+# ── Helper classes ───────────────────────────────────────────────────────────
+
+class CurlSList:
+    """Wrapper around curl_slist* for safe memory management."""
+
+    def __init__(self):
+        self._list = POINTER(c_slist)()
+        self._items: list[bytes] = []
+
+    def append(self, item: str) -> None:
+        encoded = item.encode("utf-8") if isinstance(item, str) else item
+        self._list = _lib.curl_slist_append(self._list, encoded)
+        self._items.append(encoded)
+
+    def __del__(self):
+        if self._list:
+            _lib.curl_slist_free_all(self._list)
+
+    def as_cpointer(self) -> POINTER(c_slist):
+        return self._list
+
+
+# ── Easy handle wrapper ─────────────────────────────────────────────────────
+
+class CurlEasy:
+    """
+    Wrapper around a CURL* easy handle.
+
+    Provides a Pythonic interface while keeping all the power of libcurl.
+    """
+
+    def __init__(self):
+        self._handle = _lib.curl_easy_init()
+        if not self._handle:
+            raise RuntimeError("Failed to initialize curl easy handle")
+
+        self._headers_list: Optional[CurlSList] = None
+        self._write_callback: Optional[WRITEFUNC] = None
+        self._header_callback: Optional[HEADERFUNC] = None
+        self._write_buffer = bytearray()
+        self._header_buffer = bytearray()
+
+        # Default settings
+        self.setopt(CURLOPT_FOLLOWLOCATION, 1)
+        self.setopt(CURLOPT_MAXREDIRS, 20)
+        self.setopt(CURLOPT_SSL_VERIFYPEER, 1)
+        self.setopt(CURLOPT_SSL_VERIFYHOST, 2)
+        self.setopt(CURLOPT_TIMEOUT, 30)
+        self.setopt(CURLOPT_CONNECTTIMEOUT, 10)
+        self.setopt(CURLOPT_DNS_CACHE_TIMEOUT, 300)
+        self.setopt(CURLOPT_TCP_KEEPALIVE, 1)
+        self.setopt(CURLOPT_NO_SIGNAL, 1)
+        self.setopt(CURLOPT_HTTP_VERSION, CURL_HTTP_VERSION_2TLS)
+
+    def setopt(self, option: int, value) -> None:
+        """Set a curl option."""
+        if isinstance(value, int):
+            _lib.curl_easy_setopt(self._handle, option, c_long(value))
+        elif isinstance(value, str):
+            _lib.curl_easy_setopt(self._handle, option, value.encode("utf-8"))
+        elif isinstance(value, bytes):
+            _lib.curl_easy_setopt(self._handle, option, value)
+        elif isinstance(value, CurlSList):
+            _lib.curl_easy_setopt(self._handle, option, value.as_cpointer())
+        else:
+            raise TypeError(f"Unsupported option type: {type(value)}")
+
+    def perform(self) -> int:
+        """Execute the curl request."""
+        return _lib.curl_easy_perform(self._handle)
+
+    def getinfo(self, info: int):
+        """Get information about the last performed request."""
+        if info in (CURLINFO_RESPONSE_CODE, CURLINFO_HEADER_SIZE, CURLINFO_HTTP_VERSION):
+            val = c_long()
+            _lib.curl_easy_getinfo(self._handle, info, ctypes.byref(val))
+            return val.value
+        elif info in (
+            CURLINFO_EFFECTIVE_URL,
+            CURLINFO_CONTENT_TYPE,
+            CURLINFO_REDIRECT_URL,
+            CURLINFO_PRIMARY_IP,
+        ):
+            val = c_char_p()
+            _lib.curl_easy_getinfo(self._handle, info, ctypes.byref(val))
+            return val.value.decode("utf-8") if val.value else None
+        elif info == CURLINFO_TOTAL_TIME:
+            val = c_double()
+            _lib.curl_easy_getinfo(self._handle, info, ctypes.byref(val))
+            return val.value
+        elif info == CURLINFO_SSL_VERIFYRESULT:
+            val = c_int()
+            _lib.curl_easy_getinfo(self._handle, info, ctypes.byref(val))
+            return val.value
+        raise ValueError(f"Unsupported info: {info}")
+
+    def set_headers(self, headers: Optional[dict[str, str]]) -> None:
+        """Set request headers."""
+        if self._headers_list:
+            _lib.curl_slist_free_all(self._headers_list)
+            self._headers_list = None
+
+        if headers:
+            self._headers_list = CurlSList()
+            for key, value in headers.items():
+                self._headers_list.append(f"{key}: {value}")
+            self.setopt(CURLOPT_HTTPHEADER, self._headers_list)
+
+    def reset(self) -> None:
+        """Reset the handle to default state."""
+        _lib.curl_easy_reset(self._handle)
+        # Re-apply defaults
+        self.setopt(CURLOPT_FOLLOWLOCATION, 1)
+        self.setopt(CURLOPT_MAXREDIRS, 20)
+        self.setopt(CURLOPT_SSL_VERIFYPEER, 1)
+        self.setopt(CURLOPT_SSL_VERIFYHOST, 2)
+        self.setopt(CURLOPT_TIMEOUT, 30)
+        self.setopt(CURLOPT_CONNECTTIMEOUT, 10)
+        self.setopt(CURLOPT_DNS_CACHE_TIMEOUT, 300)
+        self.setopt(CURLOPT_TCP_KEEPALIVE, 1)
+        self.setopt(CURLOPT_NO_SIGNAL, 1)
+        self.setopt(CURLOPT_HTTP_VERSION, CURL_HTTP_VERSION_2TLS)
+
+    def duphandle(self) -> "CurlEasy":
+        """Duplicate this handle (for session forking)."""
+        new = CurlEasy.__new__(CurlEasy)
+        new._handle = _lib.curl_easy_duphandle(self._handle)
+        if not new._handle:
+            raise RuntimeError("Failed to duplicate curl handle")
+        new._headers_list = None
+        new._write_callback = None
+        new._header_callback = None
+        new._write_buffer = bytearray()
+        new._header_buffer = bytearray()
+        return new
+
+    def strerror(self, code: int) -> str:
+        """Get human-readable error string."""
+        msg = _lib.curl_easy_strerror(code)
+        return msg.decode("utf-8") if msg else f"Unknown error {code}"
+
+    def __del__(self):
+        if hasattr(self, "_handle") and self._handle:
+            if self._headers_list:
+                _lib.curl_slist_free_all(self._headers_list)
+            _lib.curl_easy_cleanup(self._handle)
+            self._handle = None
+
+
+def curl_version() -> str:
+    """Return the libcurl version string."""
+    v = _lib.curl_version()
+    return v.decode("utf-8") if v else "unknown"