From 6b97a57e8e7949b9b4872a132804261f5c4248e9 Mon Sep 17 00:00:00 2001
From: Zhaofeng Li <hello@zhaofeng.li>
Date: Wed, 14 May 2025 16:01:44 -0600
Subject: [PATCH] Support @-prefixed Unix epoch timestamps as date strings

Signed-off-by: Zhaofeng Li <hello@zhaofeng.li>
---
 libarchive/archive_parse_date.c           | 55 +++++++++++++++++++++--
 libarchive/test/test_archive_parse_date.c | 12 +++++
 2 files changed, 63 insertions(+), 4 deletions(-)

diff --git a/libarchive/archive_parse_date.c b/libarchive/archive_parse_date.c
index 5e2af2aa0..f907bf20a 100644
--- a/libarchive/archive_parse_date.c
+++ b/libarchive/archive_parse_date.c
@@ -819,6 +819,23 @@ RelativeMonth(time_t Start, time_t Timezone, time_t RelMonth)
 		Timezone, DSTmaybe));
 }
 
+/*
+ * Parses and consumes an unsigned number.
+ * Returns 1 if any number is parsed. Otherwise, *value is unchanged.
+ */
+static char
+consume_unsigned_number(const char **in, time_t *value)
+{
+	char c;
+	if (isdigit((unsigned char)(c = **in))) {
+		for (*value = 0; isdigit((unsigned char)(c = *(*in)++)); )
+			*value = 10 * *value + c - '0';
+		(*in)--;
+		return 1;
+	}
+	return 0;
+}
+
 /*
  * Tokenizer.
  */
@@ -895,10 +912,7 @@ nexttoken(const char **in, time_t *value)
 		 * Because '-' and '+' have other special meanings, I
 		 * don't deal with signed numbers here.
 		 */
-		if (isdigit((unsigned char)(c = **in))) {
-			for (*value = 0; isdigit((unsigned char)(c = *(*in)++)); )
-				*value = 10 * *value + c - '0';
-			(*in)--;
+		if (consume_unsigned_number(in, value)) {
 			return (tUNUMBER);
 		}
 
@@ -929,6 +943,32 @@ difftm (struct tm *a, struct tm *b)
 	    + (a->tm_sec - b->tm_sec));
 }
 
+/*
+ * Parses a Unix epoch timestamp (seconds).
+ * This supports a subset of what GNU tar accepts from black box testing,
+ * but covers common use cases.
+ */
+static time_t
+parse_unix_epoch(const char *p)
+{
+	time_t epoch;
+
+	/* may begin with + */
+	if (*p == '+') {
+		p++;
+	}
+
+	/* followed by some number */
+	if (!consume_unsigned_number(&p, &epoch))
+		return (time_t)-1;
+
+	/* ...and nothing else */
+	if (*p != '\0')
+		return (time_t)-1;
+
+	return epoch;
+}
+
 /*
  *
  * The public function.
@@ -948,6 +988,13 @@ archive_parse_date(time_t now, const char *p)
 	time_t		tod;
 	long		tzone;
 
+	/*
+	 * @-prefixed Unix epoch timestamps (seconds)
+	 * Skip the complex tokenizer - We do not want to accept strings like "@tenth"
+	 */
+	if (*p == '@')
+		return parse_unix_epoch(p + 1);
+
 	/* Clear out the parsed token array. */
 	memset(tokens, 0, sizeof(tokens));
 	/* Initialize the parser state. */
diff --git a/libarchive/test/test_archive_parse_date.c b/libarchive/test/test_archive_parse_date.c
index fffec9ea6..0a70971b1 100644
--- a/libarchive/test/test_archive_parse_date.c
+++ b/libarchive/test/test_archive_parse_date.c
@@ -81,5 +81,17 @@ DEFINE_TEST(test_archive_parse_date)
 	/* "last tuesday" is one week before "tuesday" */
 	assertEqualInt(get_date(now, "last tuesday UTC"),
 	    now - 6 * 24 * 60 * 60);
+
+	/* Unix epoch timestamps */
+	assertEqualInt(get_date(now, "@0"), 0);
+	assertEqualInt(get_date(now, "@100"), 100);
+	assertEqualInt(get_date(now, "@+100"), 100);
+
+	assertEqualInt(get_date(now, "@"), -1);
+	assertEqualInt(get_date(now, "@-"), -1);
+	assertEqualInt(get_date(now, "@+"), -1);
+	assertEqualInt(get_date(now, "@tenth"), -1);
+	assertEqualInt(get_date(now, "@100 tomorrow"), -1);
+
 	/* TODO: Lots more tests here. */
 }
-- 
2.47.2