Merge pull request #356 from yozachar/workshop

feat: adds `private` parameter to `ip_address`, `hostname` & `url`

Author: yozachar

CHANGES.md

@@ -9,6 +9,25 @@ Note to self: Breaking changes must increment either
 
 -->
 
+## 0.27.0 (2024-04-03)
+
+_**Breaking**_ ⚠️
+
+- patch: moves `base58` and `base64` into `encoding` by @yozachar in [#354](https://github.com/python-validators/validators/pull/354)
+
+_**Features**_
+
+- feat: lays foundation for URI validation by @yozachar in [#353](https://github.com/python-validators/validators/pull/353)
+- feat: adds `private` parameter to `ip_address`, `hostname` & `url` by @yozachar in [#356](https://github.com/python-validators/validators/pull/356)
+
+_**Maintenance**_
+
+- patch: adds `encoding` tests and docs by @yozachar in [#355](https://github.com/python-validators/validators/pull/355)
+
+**Full Changelog**: [`0.26.0...0.27.0`](https://github.com/python-validators/validators/compare/0.26.0...0.27.0)
+
+---
+
 ## 0.26.0 (2024-04-02)
 
 _**Breaking**_

SECURITY.md

@@ -4,7 +4,7 @@
 
 | Version    | Supported          |
 | ---------- | ------------------ |
-| `>=0.26.0` | :white_check_mark: |
+| `>=0.27.0` | :white_check_mark: |
 
 ## Reporting a Vulnerability
 

src/validators/__init__.py

@@ -85,4 +85,4 @@
     "validator",
 )
 
-__version__ = "0.26.0"
+__version__ = "0.27.0"

src/validators/hostname.py

@@ -3,6 +3,7 @@
 # standard
 from functools import lru_cache
 import re
+from typing import Optional
 
 from .domain import domain
 
@@ -54,6 +55,7 @@ def hostname(
     skip_ipv4_addr: bool = False,
     may_have_port: bool = True,
     maybe_simple: bool = True,
+    private: Optional[bool] = None,  # only for ip-addresses
     rfc_1034: bool = False,
     rfc_2782: bool = False,
 ):
@@ -92,6 +94,8 @@ def hostname(
             Hostname string may contain port number.
         maybe_simple:
             Hostname string maybe only hyphens and alpha-numerals.
+        private:
+            Embedded IP address is public if `False`, private/local if `True`.
         rfc_1034:
             Allow trailing dot in domain/host name.
             Ref: [RFC 1034](https://www.rfc-editor.org/rfc/rfc1034).
@@ -110,13 +114,13 @@ def hostname(
         return (
             (_simple_hostname_regex().match(host_seg) if maybe_simple else False)
             or domain(host_seg, rfc_1034=rfc_1034, rfc_2782=rfc_2782)
-            or (False if skip_ipv4_addr else ipv4(host_seg, cidr=False))
+            or (False if skip_ipv4_addr else ipv4(host_seg, cidr=False, private=private))
             or (False if skip_ipv6_addr else ipv6(host_seg, cidr=False))
         )
 
     return (
         (_simple_hostname_regex().match(value) if maybe_simple else False)
         or domain(value, rfc_1034=rfc_1034, rfc_2782=rfc_2782)
-        or (False if skip_ipv4_addr else ipv4(value, cidr=False))
+        or (False if skip_ipv4_addr else ipv4(value, cidr=False, private=private))
         or (False if skip_ipv6_addr else ipv6(value, cidr=False))
     )

src/validators/ip_address.py

@@ -9,13 +9,44 @@
     IPv6Network,
     NetmaskValueError,
 )
+import re
+from typing import Optional
 
 # local
 from .utils import validator
 
 
+def _check_private_ip(value: str, is_private: Optional[bool]):
+    if is_private is None:
+        return True
+    if is_private and (
+        any(
+            value.startswith(l_bit)
+            for l_bit in {
+                "10.",  # private
+                "192.168.",  # private
+                "169.254.",  # link-local
+                "127.",  # localhost
+                "0.0.0.0",  # loopback #nosec
+            }
+        )
+        or re.match(r"^172\.(?:1[6-9]|2\d|3[0-1])\.", value)  # private
+        or re.match(r"^(?:22[4-9]|23[0-9]|24[0-9]|25[0-5])\.", value)  # broadcast
+    ):
+        return True
+    return False
+
+
 @validator
-def ipv4(value: str, /, *, cidr: bool = True, strict: bool = False, host_bit: bool = True):
+def ipv4(
+    value: str,
+    /,
+    *,
+    cidr: bool = True,
+    strict: bool = False,
+    private: Optional[bool] = None,
+    host_bit: bool = True,
+):
     """Returns whether a given value is a valid IPv4 address.
 
     From Python version 3.9.5 leading zeros are no longer tolerated
@@ -36,9 +67,11 @@ def ipv4(value: str, /, *, cidr: bool = True, strict: bool = False, host_bit: bo
         value:
             IP address string to validate.
         cidr:
-            IP address string may contain CIDR notation
+            IP address string may contain CIDR notation.
         strict:
-            IP address string is strictly in CIDR notation
+            IP address string is strictly in CIDR notation.
+        private:
+            IP address is public if `False`, private/local/loopback/broadcast if `True`.
         host_bit:
             If `False` and host bits (along with network bits) _are_ set in the supplied
             address, this function raises a validation error. ref [IPv4Network][2].
@@ -54,8 +87,8 @@ def ipv4(value: str, /, *, cidr: bool = True, strict: bool = False, host_bit: bo
         if cidr:
             if strict and value.count("/") != 1:
                 raise ValueError("IPv4 address was expected in CIDR notation")
-            return IPv4Network(value, strict=not host_bit)
-        return IPv4Address(value)
+            return IPv4Network(value, strict=not host_bit) and _check_private_ip(value, private)
+        return IPv4Address(value) and _check_private_ip(value, private)
     except (ValueError, AddressValueError, NetmaskValueError):
         return False
 
@@ -81,9 +114,9 @@ def ipv6(value: str, /, *, cidr: bool = True, strict: bool = False, host_bit: bo
         value:
             IP address string to validate.
         cidr:
-            IP address string may contain CIDR annotation
+            IP address string may contain CIDR annotation.
         strict:
-            IP address string is strictly in CIDR notation
+            IP address string is strictly in CIDR notation.
         host_bit:
             If `False` and host bits (along with network bits) _are_ set in the supplied
             address, this function raises a validation error. ref [IPv6Network][2].

src/validators/url.py

@@ -3,6 +3,7 @@
 # standard
 from functools import lru_cache
 import re
+from typing import Optional
 from urllib.parse import parse_qs, unquote, urlsplit
 
 # local
@@ -80,6 +81,7 @@ def _validate_netloc(
     skip_ipv4_addr: bool,
     may_have_port: bool,
     simple_host: bool,
+    private: Optional[bool],
     rfc_1034: bool,
     rfc_2782: bool,
 ):
@@ -97,6 +99,7 @@ def _validate_netloc(
             skip_ipv4_addr=skip_ipv4_addr,
             may_have_port=may_have_port,
             maybe_simple=simple_host,
+            private=private,
             rfc_1034=rfc_1034,
             rfc_2782=rfc_2782,
         )
@@ -111,6 +114,7 @@ def _validate_netloc(
         skip_ipv4_addr=skip_ipv4_addr,
         may_have_port=may_have_port,
         maybe_simple=simple_host,
+        private=private,
         rfc_1034=rfc_1034,
         rfc_2782=rfc_2782,
     ) and _validate_auth_segment(basic_auth)
@@ -151,6 +155,7 @@ def url(
     may_have_port: bool = True,
     simple_host: bool = False,
     strict_query: bool = True,
+    private: Optional[bool] = None,  # only for ip-addresses
     rfc_1034: bool = False,
     rfc_2782: bool = False,
 ):
@@ -191,6 +196,8 @@ def url(
             URL string maybe only hyphens and alpha-numerals.
         strict_query:
             Fail validation on query string parsing error.
+        private:
+            Embedded IP address is public if `False`, private/local if `True`.
         rfc_1034:
             Allow trailing dot in domain/host name.
             Ref: [RFC 1034](https://www.rfc-editor.org/rfc/rfc1034).
@@ -220,6 +227,7 @@ def url(
             skip_ipv4_addr,
             may_have_port,
             simple_host,
+            private,
             rfc_1034,
             rfc_2782,
         )

tests/test_url.py

@@ -1,5 +1,8 @@
 """Test URL."""
 
+# standard
+from typing import Optional
+
 # external
 import pytest
 
@@ -106,6 +109,19 @@ def test_returns_true_on_valid_url(value: str):
     assert url(value)
 
 
+@pytest.mark.parametrize(
+    "value, private",
+    [
+        ("http://username:password@10.0.10.1/", True),
+        ("http://username:password@192.168.10.10:4010/", True),
+        ("http://127.0.0.1", True),
+    ],
+)
+def test_returns_true_on_valid_private_url(value: str, private: Optional[bool]):
+    """Test returns true on valid private url."""
+    assert url(value, private=private)
+
+
 @pytest.mark.parametrize(
     "value",
     [
@@ -188,3 +204,17 @@ def test_returns_true_on_valid_url(value: str):
 def test_returns_failed_validation_on_invalid_url(value: str):
     """Test returns failed validation on invalid url."""
     assert isinstance(url(value), ValidationError)
+
+
+@pytest.mark.parametrize(
+    "value, private",
+    [
+        ("http://username:password@192.168.10.10:4010", False),
+        ("http://username:password@127.0.0.1:8080", False),
+        ("http://10.0.10.1", False),
+        ("http://255.255.255.255", False),
+    ],
+)
+def test_returns_failed_validation_on_invalid_private_url(value: str, private: Optional[bool]):
+    """Test returns failed validation on invalid private url."""
+    assert isinstance(url(value, private=private), ValidationError)