Adjust lastIndex to leading surrogate when inside a surrogate pair in unicode RegExp...

diff --git a/libregexp.c b/libregexp.c
index 1c519af51688e60bbd996236acebda941bc2b25f..8ae4bac9a7a908c669ee0524ec326836101efc83 100644 (file)
--- a/libregexp.c
+++ b/libregexp.c
@@ -3163,6 +3163,7 @@ int lre_exec(uint8_t **capture,
     REExecContext s_s, *s = &s_s;
     int re_flags, i, alloca_size, ret;
     StackInt *stack_buf;
+    const uint8_t *cptr;
 
     re_flags = lre_get_flags(bc_buf);
     s->is_unicode = (re_flags & (LRE_FLAG_UNICODE | LRE_FLAG_UNICODE_SETS)) != 0;
@@ -3187,8 +3188,17 @@ int lre_exec(uint8_t **capture,
         capture[i] = NULL;
     alloca_size = s->stack_size_max * sizeof(stack_buf[0]);
     stack_buf = alloca(alloca_size);
+
+    cptr = cbuf + (cindex << cbuf_type);
+    if (0 < cindex && cindex < clen && s->is_unicode) {
+        const uint16_t *p = (const uint16_t *)cptr;
+        if (is_lo_surrogate(*p) && is_hi_surrogate(p[-1])) {
+            cptr = (const uint8_t *)(p - 1);
+        }
+    }
+
     ret = lre_exec_backtrack(s, capture, stack_buf, 0, bc_buf + RE_HEADER_LEN,
-                             cbuf + (cindex << cbuf_type), FALSE);
+                             cptr, FALSE);
     lre_realloc(s->opaque, s->state_stack, 0);
     return ret;
 }

diff --git a/tests/test_builtin.js b/tests/test_builtin.js
index a541c198157289f8bc92dd7698ba49d58c7c2193..14a883cd89ed6dc96bfb711d558ef8da8c5d150e 100644 (file)
--- a/tests/test_builtin.js
+++ b/tests/test_builtin.js
@@ -779,6 +779,23 @@ function test_regexp()
     /* Note: SpiderMonkey and v8 may not be correct */
     assert("abcAbC".replace(/[\q{BC|A}]/gvi,"X"), "XXXX");
     assert("abcAbC".replace(/[\q{BC|A}--a]/gvi,"X"), "aXAX");
+
+    /* case where lastIndex points to the second element of a
+       surrogate pair */
+    a = /(?:)/gu;
+    a.lastIndex = 1;
+    a.exec("🐱");
+    assert(a.lastIndex, 0);
+
+    a.lastIndex = 1;
+    a.exec("a\udc00");
+    assert(a.lastIndex, 1);
+
+    a = /\u{10000}/vgd;
+    a.lastIndex = 1;
+    a = a.exec("\u{10000}_\u{10000}");
+    assert(a.indices[0][0], 0);
+    assert(a.indices[0][1], 2);
 }
 
 function test_symbol()