Correctly handle multiple chunks from read_in_chunks

read_in_chunks does not clear the intermediate string 'str'
between iterations, so every chunk has every other chunk prepended
to it.

A secondary issue is that it calls str2wcstring() on an intermediate
chunk, which may split multi-byte sequences. This needs to be deferred
to the end.

Test added. Fixes #3756

src/builtin.cpp

@@ -2024,20 +2024,19 @@ static int read_in_chunks(int fd, wcstring &buff, bool split_null) {
             break;
         }
 
-        long i;
-        for (i = 0; i < bytes_read && !finished; i++) {
-            if ((!split_null && inbuf[i] == L'\n') || (split_null && inbuf[i] == L'\0')) {
+        const char *end = std::find(inbuf, inbuf + bytes_read, split_null ? L'\0' : L'\n');
+        long bytes_consumed = end - inbuf; // note: must be signed for use in lseek
+        assert(bytes_consumed <= bytes_read);
+        str.append(inbuf, bytes_consumed);
+        if (bytes_consumed < bytes_read) {
+            // We found a splitter
+            // +1 because we need to treat the splitter as consumed, but not append it to the string
+            CHECK(lseek(fd, bytes_consumed - bytes_read + 1, SEEK_CUR) != -1, STATUS_BUILTIN_ERROR)
             finished = true;
-            } else {
-                str.push_back(inbuf[i]);
-            }
-        }
-        buff += str2wcstring(str);
-        if (i < bytes_read) {
-            CHECK(lseek(fd, i - bytes_read, SEEK_CUR) != -1, STATUS_BUILTIN_ERROR)
         }
     }
 
+    buff = str2wcstring(str);
     if (buff.empty() && eof) {
         exit_res = STATUS_BUILTIN_ERROR;
     }

tests/read.in

@@ -117,4 +117,14 @@ echo -ne 'foo\nbar\0baz\nquux' | while read -lza foo
     print_vars foo
 end
 
+echo
+echo '# chunked read tests'
+set -l path /tmp/fish_chunked_read_test.txt
+set -l longstr (seq 1024 | string join ',')
+echo -n $longstr > $path
+read -l longstr2 < $path
+test "$longstr" = "$longstr2"
+and echo "Chunked reads test pass"
+or echo "Chunked reads test failure: long strings don't match!"
+
 true

tests/read.out

@@ -55,3 +55,6 @@ newline
 1 'foo' 1 'bar'
 2 'foo' 'bar'
 2 'baz' 'quux'
+
+# chunked read tests
+Chunked reads test pass