From 15776790533dd2b1b5742c0349106a24dda2766a Mon Sep 17 00:00:00 2001
From: irl <iain@learmonth.me>
Date: Sat, 3 May 2025 15:19:47 +0100
Subject: [PATCH] feat: adds module to obtain registered domain name using PSL

---
 src/lua/psl.lua | 105 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 105 insertions(+)
 create mode 100644 src/lua/psl.lua

diff --git a/src/lua/psl.lua b/src/lua/psl.lua
new file mode 100644
index 0000000..6ce88ba
--- /dev/null
+++ b/src/lua/psl.lua
@@ -0,0 +1,105 @@
+local http = require "resty.http"
+
+local _M = {}
+
+local function fetch_psl()
+    local httpc = http.new()
+    local res, err = httpc:request_uri("https://publicsuffix.org/list/public_suffix_list.dat", {
+        method = "GET",
+        ssl_verify = true
+    })
+
+    if not res then
+        return nil, "Failed to fetch PSL: " .. err
+    end
+
+    local rules = {}
+    for line in res.body:gmatch("[^\r\n]+") do
+        line = line:match("^%s*(.-)%s*$")
+        if line ~= "" and not line:match("^//") then
+            table.insert(rules, line)
+        end
+    end
+    return rules
+end
+
+local function load_psl()
+    local cache = ngx.shared.jasima_cache
+    local cached = cache:get("psl")
+    if cached then return cached end
+    local rules, err = fetch_psl()
+    if rules then
+        cache:set("psl", rules, 86400 * 7)  -- 1 week
+        return rules
+    else
+        ngx.log(ngx.ERR, err or "Unknown error loading PSL")
+        return nil
+    end
+end
+
+local function split(domain)
+    local parts = {}
+    for part in domain:gmatch("[^%.]+") do table.insert(parts, part) end
+    return parts
+end
+
+local function domain_matches_rule(domain_parts, rule_parts)
+    for i = 1, #rule_parts do
+        local rule_part = rule_parts[#rule_parts - i + 1]
+        local domain_part = domain_parts[#domain_parts - i + 1]
+        if rule_part == "*" then
+            -- wildcard match
+        elseif not domain_part or rule_part ~= domain_part then
+            return false
+        end
+    end
+    return true
+end
+
+local function find_matching_rule(domain, rules)
+    local domain_parts = split(domain)
+    local match = nil
+    local max_len = 0
+
+    for _, rule in ipairs(rules) do
+        local is_exception = rule:sub(1, 1) == "!"
+        local clean_rule = is_exception and rule:sub(2) or rule
+        local rule_parts = split(clean_rule)
+
+        if domain_matches_rule(domain_parts, rule_parts) then
+            if #rule_parts > max_len then
+                match = { rule = rule, is_exception = is_exception }
+                max_len = #rule_parts
+            end
+        end
+    end
+
+    return match
+end
+
+function _M.get_registered_domain(domain)
+    if not domain or domain == "" then return nil end
+    local rules = load_psl()
+    if not rules then return nil end
+
+    local domain_parts = split(domain:lower())
+    local match = find_matching_rule(domain, rules)
+    if not match then return domain end
+
+    local rule = match.rule
+    local is_exception = match.is_exception
+    local rule_parts = split(rule:gsub("^!", ""))
+    local offset = is_exception and (#rule_parts + 1) or #rule_parts
+
+    local start_idx = #domain_parts - offset
+    if start_idx < 1 then return domain end
+
+    local reg_parts = {}
+    for i = start_idx, #domain_parts do
+        table.insert(reg_parts, domain_parts[i])
+    end
+
+    return table.concat(reg_parts, ".")
+end
+
+return _M