lexbor: Merge upstream WHATWG URL and IDNA fixes

lexbor/lexbor@859f100
lexbor/lexbor@a36e09a
lexbor/lexbor@b0f7412
lexbor/lexbor@1b215a8
lexbor/lexbor@385afff
lexbor/lexbor@e6c068f
lexbor/lexbor@327a8b6
lexbor/lexbor@917742f
lexbor/lexbor@e89c258

Only the source/lexbor/url part of lexbor/lexbor@327a8b6 and
lexbor/lexbor@e89c258 is carried: the first squash also contains unrelated
HTML tree construction changes, and the second documents a percent-encoder
API that this branch does not bundle.

Author: alexandre-daubois

NEWS

@@ -87,6 +87,15 @@ PHP                                                                        NEWS
   . Merge patches lexbor/lexbor@8a14bc0 and lexbor/lexbor@f67ce4b, fixing a
     heap buffer overflow in :lexbor-contains() parsing and buffer overflows
     in malformed decode replay. (alexandre-daubois)
+  . Merge patches lexbor/lexbor@859f100, lexbor/lexbor@a36e09a,
+    lexbor/lexbor@b0f7412, lexbor/lexbor@1b215a8, lexbor/lexbor@385afff,
+    lexbor/lexbor@e6c068f, lexbor/lexbor@327a8b6, lexbor/lexbor@917742f and
+    lexbor/lexbor@e89c258, fixing dropped usernames containing an at sign,
+    uninitialized memory in IDNA buffer growth, the encoding of a space
+    before a query or fragment in an opaque path, a query or fragment lost
+    after a dot segment in a path, replacement file drive paths, the output
+    encoding used for percent-encoding, the URLSearchParams tail pointer and
+    fragment serialization without a query. (alexandre-daubois)
 
 - MBString:
   . Fixed bug GH-23106 (mb_strpos() reads past the end of a haystack ending in

ext/lexbor/lexbor/unicode/idna.c

@@ -117,12 +117,14 @@ lxb_unicode_idna_realloc(lxb_codepoint_t *buf, const lxb_codepoint_t *buffer,
     lxb_codepoint_t *tmp;
 
     nlen = ((*buf_end - buf) * 4) + len;
- 
+
     if (buf == buffer) {
         tmp = lexbor_malloc(nlen * sizeof(lxb_codepoint_t));
         if (tmp == NULL) {
             return NULL;
         }
+
+        memcpy(tmp, buf, (*buf_p - buf) * sizeof(lxb_codepoint_t));
     }
     else {
         tmp = lexbor_realloc(buf, nlen * sizeof(lxb_codepoint_t));
@@ -458,13 +460,17 @@ lxb_unicode_idna_ascii_puny_cb(const lxb_char_t *data, size_t length, void *ctx,
 
         if (asc->buf == asc->buffer) {
             tmp = lexbor_malloc(nlen);
+            if (tmp == NULL) {
+                return LXB_STATUS_ERROR_MEMORY_ALLOCATION;
+            }
+
+            memcpy(tmp, asc->buf, asc->p - asc->buf);
         }
         else {
             tmp = lexbor_realloc(asc->buf, nlen);
-        }
-
-        if (tmp == NULL) {
-            return LXB_STATUS_ERROR_MEMORY_ALLOCATION;
+            if (tmp == NULL) {
+                return LXB_STATUS_ERROR_MEMORY_ALLOCATION;
+            }
         }
 
         asc->p = tmp + (asc->p - asc->buf);
@@ -711,13 +717,17 @@ lxb_unicode_idna_to_unicode_cb(const lxb_codepoint_t *part, size_t len,
 
         if (asc->buf == asc->buffer) {
             tmp = lexbor_malloc(nlen);
+            if (tmp == NULL) {
+                return LXB_STATUS_ERROR_MEMORY_ALLOCATION;
+            }
+
+            memcpy(tmp, asc->buf, asc->p - asc->buf);
         }
         else {
             tmp = lexbor_realloc(asc->buf, nlen);
-        }
-
-        if (tmp == NULL) {
-            return LXB_STATUS_ERROR_MEMORY_ALLOCATION;
+            if (tmp == NULL) {
+                return LXB_STATUS_ERROR_MEMORY_ALLOCATION;
+            }
         }
 
         asc->p = tmp + (asc->p - asc->buf);