Python: ipextract

A Python package for fast IP address extraction from text, powered by the ip-extract Rust crate via PyO3.

Installation

pip install ipextract

Requires Python 3.10+. Pre-built wheels are available for Linux (x86_64, aarch64), macOS (x86_64, arm64), and Windows (x64). No Rust toolchain needed.

Quick Start

import ipextract

# Extract all IPs from a string
ips = ipextract.extract("Connection from 192.168.1.1 and 8.8.8.8")
# ["192.168.1.1", "8.8.8.8"]

# Deduplicate, preserving first-seen order
ips = ipextract.extract_unique("8.8.8.8 1.1.1.1 8.8.8.8")
# ["8.8.8.8", "1.1.1.1"]

# bytes input works too (useful when reading log files directly)
ips = ipextract.extract(b"host 10.0.0.1 connected")
# ["10.0.0.1"]

Reusable Extractor

For processing many lines (log analysis, batch jobs), create an Extractor once and reuse it. This avoids redundant initialization on each call.

extractor = ipextract.Extractor().only_public()

for line in log_file:
    ips = extractor.extract(line)
    if ips:
        process(ips)

Filtering

By default, all IP addresses are extracted — IPv4, IPv6, private ranges, loopback, and broadcast. Use the builder methods to filter:

# Only publicly routable IPs (excludes RFC 1918, loopback, broadcast)
e = ipextract.Extractor().only_public()

# Specific exclusions
e = ipextract.Extractor().ignore_private().ignore_loopback()

# Constructor kwargs for one-shot config
e = ipextract.Extractor(private=False, loopback=False, ipv6=False)

Fluent methods return a new Extractor — the original is not modified, making partial configs safe to reuse:

base = ipextract.Extractor()
public = base.only_public()   # new object
ipv4_only = base.ipv6(False)  # new object, base unchanged

Filter reference

Extraction Methods

extract(text) → list[str]

Returns all IP addresses found, in order of appearance.

ipextract.Extractor().extract("a 1.1.1.1 b 2.2.2.2")
# ["1.1.1.1", "2.2.2.2"]

extract_unique(text) → list[str]

Returns unique IP addresses, preserving first-seen order.

ipextract.Extractor().extract_unique("1.1.1.1 2.2.2.2 1.1.1.1")
# ["1.1.1.1", "2.2.2.2"]

extract_with_offsets(text) → list[tuple[str, int, int]]

Returns (ip, start, end) tuples. The byte offsets index directly into the original input — useful for annotation, highlighting, or structured log parsing.

text = "host 1.2.3.4 port 80"
for ip, start, end in ipextract.Extractor().extract_with_offsets(text):
    print(f"{ip} at [{start}:{end}]")
    assert text[start:end] == ip
# 1.2.3.4 at [5:12]

Performance

ipextract is designed for high-throughput applications. It uses a compile-time DFA (Deterministic Finite Automaton) from the Rust ip-extract crate, which scans at O(n) without backtracking.

Typical benchmark results comparing ipextract to Python re + ipaddress.ip_address() validation (both sides extract and validate):

The larger the input and the more non-IP text it contains, the greater the performance advantage. ipextract excels at high-speed log scanning because it can reject non-IP text much faster than a backtracking regex engine.

Module-Level Convenience Functions

ipextract.extract() and ipextract.extract_unique() are shorthand for Extractor().extract() with default settings (all IPs included). For repeated calls, prefer creating an Extractor instance explicitly.

Source

The ipextract Python package lives in crates/ipextract-py/ and wraps the ip-extract Rust crate.