Fixed bug #80022: Support ISO 8601 years outside 0000-9999 range better

Author: derickr

ext/date/php_date.c

@@ -134,6 +134,7 @@ PHPAPI time_t php_time(void)
  *  zone            =       (( "+" / "-" ) 4DIGIT)
  */
 #define DATE_FORMAT_RFC2822  "D, d M Y H:i:s O"
+
 /*
  * RFC3339, Section 5.6: http://www.ietf.org/rfc/rfc3339.txt
  *  date-fullyear   = 4DIGIT
@@ -156,8 +157,42 @@ PHPAPI time_t php_time(void)
  */
 #define DATE_FORMAT_RFC3339  "Y-m-d\\TH:i:sP"
 
+/*
+ * This format does not technically match the ISO 8601 standard, as it does not
+ * use : in the UTC offset format specifier. This is kept for BC reasons. The
+ * DATE_FORMAT_ISO8601_EXPANDED format does correct this, as well as adding
+ * support for years out side of the traditional 0000-9999 range.
+ */
 #define DATE_FORMAT_ISO8601  "Y-m-d\\TH:i:sO"
 
+/* ISO 8601:2004(E)
+ *
+ * Section 3.5 Expansion:
+ * By mutual agreement of the partners in information interchange, it is
+ * permitted to expand the component identifying the calendar year, which is
+ * otherwise limited to four digits. This enables reference to dates and times
+ * in calendar years outside the range supported by complete representations,
+ * i.e. before the start of the year [0000] or after the end of the year
+ * [9999]."
+ *
+ * Section 4.1.2.4 Expanded representations:
+ * If, by agreement, expanded representations are used, the formats shall be as
+ * specified below. The interchange parties shall agree the additional number of
+ * digits in the time element year. In the examples below it has been agreed to
+ * expand the time element year with two digits.
+ * Extended format: ±YYYYY-MM-DD
+ * Example: +001985-04-12
+ *
+ * PHP's year expansion digits are variable.
+ */
+#define DATE_FORMAT_ISO8601_EXPANDED    "X-m-d\\TH:i:sP"
+
+/* Internal Only
+ * This format only extends the year when needed, keeping the 'P' format with
+ * colon for UTC offsets
+ */
+#define DATE_FORMAT_ISO8601_LARGE_YEAR  "x-m-d\\TH:i:sP"
+
 /*
  * RFC3339, Appendix A: http://www.ietf.org/rfc/rfc3339.txt
  *  ISO 8601 also requires (in section 5.3.1.3) that a decimal fraction
@@ -659,6 +694,8 @@ static zend_string *date_format(const char *format, size_t format_len, timelib_t
 			case 'L': length = slprintf(buffer, sizeof(buffer), "%d", timelib_is_leap((int) t->y)); break;
 			case 'y': length = slprintf(buffer, sizeof(buffer), "%02d", (int) (t->y % 100)); break;
 			case 'Y': length = slprintf(buffer, sizeof(buffer), "%s%04lld", t->y < 0 ? "-" : "", php_date_llabs((timelib_sll) t->y)); break;
+			case 'x': length = slprintf(buffer, sizeof(buffer), "%s%04lld", t->y < 0 ? "-" : (t->y >= 10000 ? "+" : ""), php_date_llabs((timelib_sll) t->y)); break;
+			case 'X': length = slprintf(buffer, sizeof(buffer), "%s%04lld", t->y < 0 ? "-" : "+", php_date_llabs((timelib_sll) t->y)); break;
 
 			/* time */
 			case 'a': length = slprintf(buffer, sizeof(buffer), "%s", t->h >= 12 ? "pm" : "am"); break;
@@ -1810,7 +1847,7 @@ static void date_object_to_hash(php_date_obj *dateobj, HashTable *props)
 	zval zv;
 
 	/* first we add the date and time in ISO format */
-	ZVAL_STR(&zv, date_format("Y-m-d H:i:s.u", sizeof("Y-m-d H:i:s.u")-1, dateobj->time, 1));
+	ZVAL_STR(&zv, date_format("x-m-d H:i:s.u", sizeof("x-m-d H:i:s.u")-1, dateobj->time, 1));
 	zend_hash_str_update(props, "date", sizeof("date")-1, &zv);
 
 	/* then we add the timezone name (or similar) */
@@ -3800,7 +3837,7 @@ PHP_FUNCTION(timezone_transitions_get)
 #define add_nominal() \
 		array_init(&element); \
 		add_assoc_long(&element, "ts",     timestamp_begin); \
-		add_assoc_str(&element, "time", php_format_date(DATE_FORMAT_ISO8601, 13, timestamp_begin, 0)); \
+		add_assoc_str(&element, "time", php_format_date(DATE_FORMAT_ISO8601_LARGE_YEAR, 13, timestamp_begin, 0)); \
 		add_assoc_long(&element, "offset", tzobj->tzi.tz->type[0].offset); \
 		add_assoc_bool(&element, "isdst",  tzobj->tzi.tz->type[0].isdst); \
 		add_assoc_string(&element, "abbr", &tzobj->tzi.tz->timezone_abbr[tzobj->tzi.tz->type[0].abbr_idx]); \
@@ -3809,7 +3846,7 @@ PHP_FUNCTION(timezone_transitions_get)
 #define add(i,ts) \
 		array_init(&element); \
 		add_assoc_long(&element, "ts",     ts); \
-		add_assoc_str(&element, "time", php_format_date(DATE_FORMAT_ISO8601, 13, ts, 0)); \
+		add_assoc_str(&element, "time", php_format_date(DATE_FORMAT_ISO8601_LARGE_YEAR, 13, ts, 0)); \
 		add_assoc_long(&element, "offset", tzobj->tzi.tz->type[tzobj->tzi.tz->trans_idx[i]].offset); \
 		add_assoc_bool(&element, "isdst",  tzobj->tzi.tz->type[tzobj->tzi.tz->trans_idx[i]].isdst); \
 		add_assoc_string(&element, "abbr", &tzobj->tzi.tz->timezone_abbr[tzobj->tzi.tz->type[tzobj->tzi.tz->trans_idx[i]].abbr_idx]); \
@@ -3818,7 +3855,7 @@ PHP_FUNCTION(timezone_transitions_get)
 #define add_by_index(i,ts) \
 		array_init(&element); \
 		add_assoc_long(&element, "ts",     ts); \
-		add_assoc_str(&element, "time", php_format_date(DATE_FORMAT_ISO8601, 13, ts, 0)); \
+		add_assoc_str(&element, "time", php_format_date(DATE_FORMAT_ISO8601_LARGE_YEAR, 13, ts, 0)); \
 		add_assoc_long(&element, "offset", tzobj->tzi.tz->type[i].offset); \
 		add_assoc_bool(&element, "isdst",  tzobj->tzi.tz->type[i].isdst); \
 		add_assoc_string(&element, "abbr", &tzobj->tzi.tz->timezone_abbr[tzobj->tzi.tz->type[i].abbr_idx]); \
@@ -3827,7 +3864,7 @@ PHP_FUNCTION(timezone_transitions_get)
 #define add_from_tto(to,ts) \
 		array_init(&element); \
 		add_assoc_long(&element, "ts",     ts); \
-		add_assoc_str(&element, "time", php_format_date(DATE_FORMAT_ISO8601, 13, ts, 0)); \
+		add_assoc_str(&element, "time", php_format_date(DATE_FORMAT_ISO8601_LARGE_YEAR, 13, ts, 0)); \
 		add_assoc_long(&element, "offset", (to)->offset); \
 		add_assoc_bool(&element, "isdst",  (to)->is_dst); \
 		add_assoc_string(&element, "abbr", (to)->abbr); \

ext/date/php_date.stub.php

@@ -20,6 +20,12 @@
  */
 const DATE_ISO8601 = "Y-m-d\\TH:i:sO";
 
+/**
+ * @var string
+ * @cvalue DATE_FORMAT_ISO8601_EXPANDED
+ */
+const DATE_ISO8601_EXPANDED = "X-m-d\\TH:i:sP";
+
 /**
  * @var string
  * @cvalue DATE_FORMAT_RFC822
@@ -285,6 +291,8 @@ interface DateTimeInterface
     /** @var string */
     public const ISO8601 = DATE_ISO8601;
     /** @var string */
+    public const ISO8601_EXPANDED = DATE_ISO8601_EXPANDED;
+    /** @var string */
     public const RFC822 = DATE_RFC822;
     /** @var string */
     public const RFC850 = DATE_RFC850;

ext/date/php_date_arginfo.h

@@ -32,7 +32,7 @@ array(5) {
   ["ts"]=>
   int(-213228000)
   ["time"]=>
-  string(24) "1963-03-31T02:00:00+0000"
+  string(25) "1963-03-31T02:00:00+00:00"
   ["offset"]=>
   int(3600)
   ["isdst"]=>

ext/date/tests/DateTimeZone_getTransitions_basic1.phpt

@@ -24,35 +24,35 @@ showTransitions('Europe/Paris', 1645095600);        // GH Issue 8108
 --EXPECT--
 Europe/London from @1648342200-@1727398200:
 
-  1648342200 2022-03-27T00:50:00+0000      0  x  GMT
-  1648342800 2022-03-27T01:00:00+0000   3600 DST BST
-  1667091600 2022-10-30T01:00:00+0000      0  x  GMT
-  1679792400 2023-03-26T01:00:00+0000   3600 DST BST
-  1698541200 2023-10-29T01:00:00+0000      0  x  GMT
-  1711846800 2024-03-31T01:00:00+0000   3600 DST BST
+  1648342200 2022-03-27T00:50:00+00:00      0  x  GMT
+  1648342800 2022-03-27T01:00:00+00:00   3600 DST BST
+  1667091600 2022-10-30T01:00:00+00:00      0  x  GMT
+  1679792400 2023-03-26T01:00:00+00:00   3600 DST BST
+  1698541200 2023-10-29T01:00:00+00:00      0  x  GMT
+  1711846800 2024-03-31T01:00:00+00:00   3600 DST BST
 
 America/Los_Angeles from @1648557596-@1727613596:
 
-  1648557596 2022-03-29T12:39:56+0000 -25200 DST PDT
-  1667725200 2022-11-06T09:00:00+0000 -28800  x  PST
-  1678615200 2023-03-12T10:00:00+0000 -25200 DST PDT
-  1699174800 2023-11-05T09:00:00+0000 -28800  x  PST
-  1710064800 2024-03-10T10:00:00+0000 -25200 DST PDT
+  1648557596 2022-03-29T12:39:56+00:00 -25200 DST PDT
+  1667725200 2022-11-06T09:00:00+00:00 -28800  x  PST
+  1678615200 2023-03-12T10:00:00+00:00 -25200 DST PDT
+  1699174800 2023-11-05T09:00:00+00:00 -28800  x  PST
+  1710064800 2024-03-10T10:00:00+00:00 -25200 DST PDT
 
 America/Chicago from @1293861600-@1372917600:
 
-  1293861600 2011-01-01T06:00:00+0000 -21600  x  CST
-  1300003200 2011-03-13T08:00:00+0000 -18000 DST CDT
-  1320562800 2011-11-06T07:00:00+0000 -21600  x  CST
-  1331452800 2012-03-11T08:00:00+0000 -18000 DST CDT
-  1352012400 2012-11-04T07:00:00+0000 -21600  x  CST
-  1362902400 2013-03-10T08:00:00+0000 -18000 DST CDT
+  1293861600 2011-01-01T06:00:00+00:00 -21600  x  CST
+  1300003200 2011-03-13T08:00:00+00:00 -18000 DST CDT
+  1320562800 2011-11-06T07:00:00+00:00 -21600  x  CST
+  1331452800 2012-03-11T08:00:00+00:00 -18000 DST CDT
+  1352012400 2012-11-04T07:00:00+00:00 -21600  x  CST
+  1362902400 2013-03-10T08:00:00+00:00 -18000 DST CDT
 
 Europe/Paris from @1645095600-@1724151600:
 
-  1645095600 2022-02-17T11:00:00+0000   3600  x  CET
-  1648342800 2022-03-27T01:00:00+0000   7200 DST CEST
-  1667091600 2022-10-30T01:00:00+0000   3600  x  CET
-  1679792400 2023-03-26T01:00:00+0000   7200 DST CEST
-  1698541200 2023-10-29T01:00:00+0000   3600  x  CET
-  1711846800 2024-03-31T01:00:00+0000   7200 DST CEST
+  1645095600 2022-02-17T11:00:00+00:00   3600  x  CET
+  1648342800 2022-03-27T01:00:00+00:00   7200 DST CEST
+  1667091600 2022-10-30T01:00:00+00:00   3600  x  CET
+  1679792400 2023-03-26T01:00:00+00:00   7200 DST CEST
+  1698541200 2023-10-29T01:00:00+00:00   3600  x  CET
+  1711846800 2024-03-31T01:00:00+00:00   7200 DST CEST

ext/date/tests/DateTimeZone_getTransitions_bug1.phpt

@@ -0,0 +1,29 @@
+--TEST--
+Bug #75035 (Datetime fails to unserialize "extreme" dates)
+--INI--
+date.timezone=UTC
+--FILE--
+<?php
+var_dump('PHP version', PHP_VERSION);
+
+foreach ([PHP_INT_MIN, PHP_INT_MAX] as $extreme) {
+    $i = 64;
+    while ($i --> 0) {
+        $d = new DateTime('@' . ($extreme >> $i));
+        $s = serialize($d);
+        try {
+            $u = unserialize($s);
+        } catch (Error $e) {
+            $u = "failed unserialization: " . $e->getMessage() . ' : ' . $s;
+        }
+        $original = $d->format('Y-m-d H:i:s');
+        $serializedUnserialized = is_string($u) ? $u : $u->format('Y-m-d H:i:s');
+        if ($original !== $serializedUnserialized) {
+            var_dump('[' . ($extreme >> $i) . '] ' . $original . ' => ' . $serializedUnserialized);
+        }
+    }
+}
+?>
+--EXPECTF--
+string(11) "PHP version"
+string(%d) "%s"

ext/date/tests/bug75035.phpt

@@ -19,7 +19,7 @@ array(5) {
   ["ts"]=>
   int(2140045200)
   ["time"]=>
-  string(24) "2037-10-25T01:00:00+0000"
+  string(25) "2037-10-25T01:00:00+00:00"
   ["offset"]=>
   int(0)
   ["isdst"]=>
@@ -32,7 +32,7 @@ array(5) {
   ["ts"]=>
   int(2140668000)
   ["time"]=>
-  string(24) "2037-11-01T06:00:00+0000"
+  string(25) "2037-11-01T06:00:00+00:00"
   ["offset"]=>
   int(-18000)
   ["isdst"]=>
@@ -45,7 +45,7 @@ array(5) {
   ["ts"]=>
   int(2140045200)
   ["time"]=>
-  string(24) "2037-10-25T01:00:00+0000"
+  string(25) "2037-10-25T01:00:00+00:00"
   ["offset"]=>
   int(3600)
   ["isdst"]=>

ext/date/tests/bug80963.phpt

@@ -11,8 +11,8 @@ foreach ($tz->getTransitions(strtotime("1996-01-01"), strtotime("1997-12-31")) a
 }
 ?>
 --EXPECT--
-1996-01-01T00:00:00+0000 3600 CET
-1996-03-31T01:00:00+0000 7200 CEST
-1996-10-27T01:00:00+0000 3600 CET
-1997-03-30T01:00:00+0000 7200 CEST
-1997-10-26T01:00:00+0000 3600 CET
+1996-01-01T00:00:00+00:00 3600 CET
+1996-03-31T01:00:00+00:00 7200 CEST
+1996-10-27T01:00:00+00:00 3600 CET
+1997-03-30T01:00:00+00:00 7200 CEST
+1997-10-26T01:00:00+00:00 3600 CET

ext/date/tests/bug81504.phpt

@@ -35,7 +35,7 @@ array(5) {
   ["ts"]=>
   int(-213228000)
   ["time"]=>
-  string(24) "1963-03-31T02:00:00+0000"
+  string(25) "1963-03-31T02:00:00+00:00"
   ["offset"]=>
   int(3600)
   ["isdst"]=>