logpile/internal/privacy/allowlist.go

package privacy

import (
	"path"
	"strings"
)

// Public infrastructure, documentation, and vendor-default values that are not
// customer leaks. These are reference data (RFC 2606 / 5737 names, well-known
// NTP pools, standards-body domains, factory defaults), not vendor-detection
// logic.
var (
	allowlistedZones = []string{
		"example.com", "example.net", "example.org", "example.local", "example.edu",
		"foobar.edu", "issue.net",
		"localhost", "localdomain", "local.lan", "localhost.localdomain",
		"host.name", "other.host.name", "some.host.name", "other.domain", "some.domain",
		"pool.ntp.org", "ntp.org", "nist.gov", "windows.com", "microsoft.com",
		"dmtf.org", "iana.org", "openssl.org", "openssh.com", "openssh.org",
		"libssh.org", "rsyslog.com", "adiscon.com", "redhat.com", "kernel.org",
		"megarac.com", "ami.com", "commond.com",
		"oasis-open.org", "w3.org", "xmlsoap.org", "purl.org", "ietf.org",
		"nvidia.com", "mellanox.com", "gnu.org", "debian.org", "ubuntu.com",
		"smartmontools.org", "openib.org", "apache.org", "freebsd.org", "freedesktop.org",
		"golang.org", "go.dev", "x.org", "sourceforge.net", "xyz.com", "yandex.net",
		"haxx.se", "curl.se", "python.org", "perl.org", "isc.org", "sourceware.org",
		"ipxe.org", "gnupg.org", "gnutls.org", "openldap.org", "netfilter.org",
		"github.com", "githubusercontent.com", "typoland.com", "schemas.dell.com",
		"googleapis.com", "gstatic.com", "jquery.com", "jsdelivr.net", "unpkg.com",
		"cloudflare.com", "cloudflare.net", "certificate.fi",
		"inspur.com", "inspurcloud.com", "inservice-iq.com", "ieisystem.com", "kaytus.com",
		"jd.com", "jd.local", "jdcloud.com", "in-addr.arpa", "ip6.arpa", "arpa",
	}

	allowlistedValues = map[string]struct{}{
		"asia/shanghai": {},
		"etc/utc":       {},
		"etc/universal": {},
		"utc":           {},
		// neutral decoy zones the sanitizer writes in place of a real timezone
		"pacific/tarawa":         {},
		"atlantic/azores":        {},
		"antarctica/troll":       {},
		"antarctica/vostok":      {},
		"antarctica/mcmurdo":     {},
		"pacific/guadalcanal":    {},
		"pacific/bougainville":   {},
		"to be filled by o.e.m.": {},
		"default string":         {},
		"unknown":                {},
		"n/a":                    {},
		"none":                   {},
		"null":                   {},
		"0":                      {},
		"0.0.0.0":                {},
		"not specified":          {},
		"not available":          {},
		"not present":            {},
		"unspecified":            {},
		"no asset tag":           {},
		"no asset information":   {},
		"empty":                  {},
		"[empty]":                {},
	}

	// Archive members that are vendor factory templates or LOGPile's own derived
	// export artifacts (operator-entered target host, Redfish annotation keys) -
	// not customer data from the source.
	allowlistedFilenameParts = []string{
		"_tencent", "_jingdong", "_pdd", "_baidu", "_kuaishou", "_tianyiyun",
		"_jd.", "syslog_jd", "snmptrapcfg", ".json_bak", ".conf_bak", "_bak",
		"ntp_auto", "raw_export.json", "parser_fields.json", "collect.log",
	}
)

// IsAllowlistedFile reports whether a member path is a vendor factory template
// or a LOGPile-derived artifact that should not be scanned or redacted.
func IsAllowlistedFile(name string) bool { return isAllowlistedFilename(name) }

func isAllowlistedDomain(domain string) bool {
	d := strings.ToLower(strings.TrimSuffix(strings.TrimSpace(domain), "."))
	for _, z := range allowlistedZones {
		if d == z || strings.HasSuffix(d, "."+z) {
			return true
		}
	}
	return false
}

func isAllowlistedValue(v string) bool {
	lv := strings.ToLower(strings.TrimSpace(v))
	if _, ok := allowlistedValues[lv]; ok {
		return true
	}
	return isRedactionFiller(lv) || isAllowlistedDomain(v)
}

// isRedactionFiller recognises the internal/sanitize output so a re-scan or a
// second sanitize pass of an already-cleaned file finds nothing: an all-'x'
// hostname filler ("xxxxx.xxxx.xx", "x@xxxx.xxxx.xx", "xxxxxx/xxxxxx") or an
// all-zero IP / offset filler ("00.000.000.00", "000", "-000").
func isRedactionFiller(s string) bool {
	if len(s) < 2 {
		return false
	}
	hasFill := false
	for _, r := range s {
		switch {
		case r == 'x' || r == '0':
			hasFill = true
		case r == '.' || r == '-' || r == '@' || r == ':' || r == '/' || r == '+':
		default:
			return false
		}
	}
	return hasFill
}

func isAllowlistedFilename(p string) bool {
	lp := strings.ToLower(p)
	for _, part := range allowlistedFilenameParts {
		if strings.Contains(lp, part) {
			return true
		}
	}
	return false
}

func isCertFilename(p string) bool {
	switch strings.ToLower(path.Ext(p)) {
	case ".pem", ".csr", ".crt", ".cer":
		return true
	}
	return false
}