From daee871ec22e8be10298119bc49c3f3dd6defcc4 Mon Sep 17 00:00:00 2001
From: Hiroshi Hatake <hiroshi@chronosphere.io>
Date: Mon, 6 Jan 2025 19:48:22 +0900
Subject: [PATCH] unescape: Try to fill out with \uFFFD for invalid sequence

Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
---
 src/flb_unescape.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

diff --git a/src/flb_unescape.c b/src/flb_unescape.c
index 71bd9ff4292..18df30e6d10 100644
--- a/src/flb_unescape.c
+++ b/src/flb_unescape.c
@@ -121,11 +121,13 @@ static int u8_read_escape_sequence(const char *str, int size, uint32_t *dest)
         }
         if (dno != 4) {
             /* Incomplete \u escape sequence */
+            ch = L'\uFFFD';
             goto invalid_sequence;
         }
         ch = strtol(digs, NULL, 16);
         if (u8_low_surrogate(ch)) {
             /* Invalid: low surrogate without preceding high surrogate */
+            ch = L'\uFFFD';
             goto invalid_sequence;
         }
         else if (u8_high_surrogate(ch)) {
@@ -139,6 +141,7 @@ static int u8_read_escape_sequence(const char *str, int size, uint32_t *dest)
                 }
                 if (dno != 4) {
                     /* Incomplete low surrogate */
+                    ch = L'\uFFFD';
                     goto invalid_sequence;
                 }
                 uint32_t low = strtol(digs, NULL, 16);
@@ -147,11 +150,13 @@ static int u8_read_escape_sequence(const char *str, int size, uint32_t *dest)
                 }
                 else {
                     /* Invalid: high surrogate not followed by low surrogate */
+                    ch = L'\uFFFD';
                     goto invalid_sequence;
                 }
             }
             else {
                 /* Invalid: high surrogate not followed by \u */
+                ch = L'\uFFFD';
                 goto invalid_sequence;
             }
         }
@@ -166,10 +171,9 @@ static int u8_read_escape_sequence(const char *str, int size, uint32_t *dest)
     }
     *dest = ch;
 
-    return i;
-
 invalid_sequence:
-    return -1;
+
+    return i;
 }
 
 int flb_unescape_string_utf8(const char *in_buf, int sz, char *out_buf)
@@ -223,6 +227,10 @@ int flb_unescape_string_utf8(const char *in_buf, int sz, char *out_buf)
                 size = end - next;
                 if (size > 0) {
                     esc_in = u8_read_escape_sequence(next, size, &ch) + 1;
+                    if (esc_in == -1) {
+                        flb_error("invalid sequence detected");
+                        break;
+                    }
                 }
                 else {
                     /* because char is unsigned char by default on arm, so we need to do a explicit conversion */