pattex 0.4.0

Regex pattern matching and extraction library for Python.

pattex

A Python library for extracting common patterns from text using regex — emails, URLs, phone numbers, IP addresses, and more.

Installation

pip install pattex

Changelog

[0.4.0] - 2026-05-16

Added

url_extractor.py

• extract_urls_by_scheme(text, scheme) — filter URLs by a specific scheme
• extract_urls_by_domain(text, domain, include_subdomains) — filter URLs by domain, with optional subdomain matching
• extract_urls_by_tld(text, tld) — filter URLs by TLD, leading dot optional
• extract_secure_urls(text) — returns only https, ftps, sftp, wss URLs
• extract_insecure_urls(text) — returns only http, ftp, ws URLs
• extract_urls_with_auth(text) — returns only URLs containing userinfo (user:pass@host)
• extract_urls_with_port(text, port) — returns URLs with any port, or a specific port
• extract_unique_domains(text, root_only) — unique hostnames or root domains across all URLs
• is_url(text, strict) — validates whether a string is a valid URL
• allow_ipv4, allow_ipv6, allow_port, allow_tail, schemes parameters added to extract_urls
• _build_strict_regex() — dynamic regex builder with lru_cache(maxsize=16)
• _build_permissive_regex() — single combined permissive regex with named groups, lru_cache(maxsize=1)

ip_extractor.py ← new file

• extract_ip_urls(text, ipv4, ipv6) — extract URLs whose host is an IP address
• extract_localhost_urls(text, allow_port, allow_tail) — extract localhost URLs only

regexes.py

• _SCHEMES — raw scheme alternation string
• _BARE_DOMAIN — raw bare domain pattern without port/tail

_utils.py

• _get_root_domain(host) — extracts root domain from hostname
• has_auth(url) — checks whether a URL contains userinfo
• _is_ip_host(url) — checks whether a URL host is an IP address

extractor_constants.py

• URL_SCHEMES — Literal type for all supported schemes

Changed

• include_localhost renamed to allow_localhost — breaking change
• Permissive mode now runs a single finditer scan instead of 3–4 separate loops — protocol-relative, bare domain, bare IP, and localhost combined into one regex via named groups ((?P<name>...))
• match.lastgroup used to dispatch per-branch guard logic in permissive loop
• Localhost always included in permissive mode implicitly — allow_localhost only controls strict mode
• _build_permissive_regex takes no arguments — localhost always included in permissive
• Schemed http://localhost always extracted in strict mode regardless of allow_localhost
• extract_secure_urls and extract_insecure_urls have no mode parameter — scheme implies strictness

Internal

• All raw pattern constants kept as plain strings in regexes.py — compiled only at assembly time
• lru_cache on both builders avoids recompilation on repeated calls with same flags
• urlparse import consolidated at top of url_extractor.py
• Walrus operator used in extract_urls_with_port to avoid double urlparse call

---

v0.3.1 - 2026-05-13

Fixed

• Zoho extractor: removed hyphen (-) and plus (+) from allowed local part characters — Zoho usernames only permit letters, numbers, dots, and underscores per official documentation
• Zoho extractor: + is a subaddressing tag separator, not a valid account creation character; removed from base regex and validation logic
• Proton extractor: tightened leading/trailing character rule — local part must begin and end with an alphanumeric character (was checking against ._- set, now explicitly uses isalnum() to match Proton's documented rule)

Changed

• Renamed email_extraction.py → email_extractor.py
• Renamed urls_extraction.py → url_extractor.py
• Updated docstrings for extract_zoho_emails and extract_proton_emails to reflect corrected provider rules
• Added inline comments on each validation rule inside email_extractor.py for all six provider extractors
• Added documentation comments for all email regex patterns in constants/regexes.py and corresponding .pyi stubs

Internal

• General code cleanups across extractor modules

---

Status

Currently under active development. Full documentation and usage examples coming soon.

License

MIT