disposable emails

.gitignore

@@ -1,3 +1,4 @@
 composer.phar
 /vendor/
 composer.lock
+.phpunit.result.cache

scripts/refresh_disposable_list.sh

@@ -0,0 +1,27 @@
+#!/bin/bash
+
+SRC=(
+	"https://raw.githubusercontent.com/disposable-email-domains/disposable-email-domains/refs/heads/main/disposable_email_blocklist.conf"
+	"https://raw.githubusercontent.com/FGRibreau/mailchecker/refs/heads/master/list.txt"
+)
+
+OUTPUT_FILE="src/disposable_email_blocklist.conf"
+if [ ! -f src/disposable_email_blocklist.conf ]; then
+	echo "run at repository top level";
+	exit 1;
+fi
+
+echo "" > $OUTPUT_FILE
+for src in ${SRC[@]}; do
+    curl -s $src -o $OUTPUT_FILE.tmp
+	echo "--- $src ---";
+	wc -l $OUTPUT_FILE.tmp;
+	cat $OUTPUT_FILE.tmp >> $OUTPUT_FILE
+	rm $OUTPUT_FILE.tmp
+done
+
+cat src/disposable_email_blocklist.conf | sort | uniq > src/disposable_email_blocklist.conf.tmp
+mv src/disposable_email_blocklist.conf.tmp src/disposable_email_blocklist.conf
+
+echo "--- Total in $OUTPUT_FILE ---";
+wc -l src/disposable_email_blocklist.conf

src/InvalidEmailException.php

@@ -5,4 +5,5 @@
 class InvalidEmailException extends \Exception {
     const SYNTAX = 10;
     const DNSRECORDS = 20;
+    const DISPOSABLE = 30;
 }

src/MailUtils.php

@@ -10,12 +10,16 @@ public static function normalize(string $address): string
     }
 
     /**
+     * Check if the address is valid, raise an exception if not.
+     * 
      * @param string $original_address_string
-     * @param bool $doDnsChecks
+     * @param bool $doDnsChecks If true, the address will be checked if it has DNS records to receive emails.
+     * @param bool $checkDisposable If true, the address will be checked if it is disposable.
+     * @param bool $useIndexCache If true, the index cache will be used (built on first use) to check if the address is disposable.
      * @return Address
      * @throws InvalidEmailException
      */
-    public static function address(string $original_address_string, bool $doDnsChecks = true): Address
+    public static function address(string $original_address_string, bool $doDnsChecks = true, bool $checkDisposable = true, bool $useIndexCache = true): Address
     {
         $address_string = self::normalize($original_address_string);
         if (filter_var($address_string, FILTER_VALIDATE_EMAIL) !== $address_string) {
@@ -43,6 +47,11 @@ public static function address(string $original_address_string, bool $doDnsCheck
                 throw new InvalidEmailException("no dns record for domain", InvalidEmailException::DNSRECORDS);
             }
         }
+        if ($checkDisposable) {
+            if (self::isDisposable($address, $useIndexCache)) {
+                throw new InvalidEmailException("disposable address", InvalidEmailException::DISPOSABLE);
+            }
+        }
         return $address;
     }
 
@@ -55,4 +64,80 @@ public static function hasA(Address $address): bool
     {
         return checkdnsrr($address->getDomain(), "A");
     }
+
+    /**
+     * Build the index for the disposable email block list
+     * @param string $filename
+     * @param bool $cache
+     * @return array<string,array<string,string>|null>
+     * @throws \Exception
+     */
+    private static function buildIndex(string $filename, bool $cache = true): ?array
+    {
+        $cacheFilename = sys_get_temp_dir() . "/mailcheck_index_" . md5($filename) . ".bin";
+        if (file_exists($cacheFilename) && $cache) {
+            return unserialize(file_get_contents($cacheFilename));
+        }
+        $tree = [];
+        $f = fopen($filename, "r");
+        if ($f === false) {
+            return null;
+        }
+        while (($line = fgets($f)) !== false) {
+            $treeItem = &$tree;
+            foreach (str_split($line) as $char) {
+                if ($char === "\n") {
+                    break;
+                }
+                if ($char === "0") {
+                    $char = "00";
+                }
+                /** @phpstan-ignore-next-line non falsy value */
+                if (!array_key_exists($char, $treeItem)) {
+                    $treeItem["$char"] = [];
+                    $treeItem = &$treeItem["$char"];
+                } else {
+                    $treeItem = &$treeItem["$char"];
+                }
+            }
+        }
+        if ($cache) {
+            file_put_contents($cacheFilename, serialize($tree));
+        }
+        return $tree;
+    }
+
+    private static function searchIndex(string $indexFilename, string $domain, bool $cache = true): ?bool
+    {
+        $wt = self::buildIndex($indexFilename, $cache);
+        if ($wt === null) {
+            return null;
+        }
+        foreach (str_split($domain) as $char) {
+            if ($char === "0") {
+                $char = "00";
+            }
+            if (!isset($wt["$char"])) {
+                return false;
+            }
+            $wt = $wt[$char];
+        }
+        return empty(array_keys($wt));
+    }
+
+    /**
+     * Check if address is disposable
+     * 
+     * Domains list comes from 
+     *  https://github.com/disposable-email-domains/disposable-email-domains
+     *  https://github.com/FGRibreau/mailchecker/blob/master/list.txt
+     * 
+     * @param Address $address
+     * @param bool $cache
+     * @return bool|null
+     */
+    public static function isDisposable(Address $address, bool $cache = true): ?bool
+    {
+        return self::searchIndex(__DIR__ . "/disposable_email_blocklist.conf", $address->getDomain(), $cache);
+    }
 }