feat: add parsing of surrogate pairs in string literals

mend: related tests
leviongit · May 23, 2024 · 3fdd234 · 3fdd234 · github-actions · May 23, 2024
1 parent 3af474d
commit 3fdd234
Show file tree

Hide file tree

Showing 6 changed files with 26 additions and 5 deletions.
diff --git a/json.rb b/json.rb
@@ -278,6 +278,27 @@ def __read_escape(str)
           cp = cp * 0x10 + __readexpect_hexdigit
           cp = cp * 0x10 + __readexpect_hexdigit
 
+          if (cp >= 0xd800 && cp <=0xdfff)
+            __failed("unexpected unpaired high surrogate") if cp >= 0xdc00
+
+            __expectb!(0x5c) && # \
+              __expectb!(0x75) || # u
+              __failed("expected second unicode escape in surrogate pair")
+
+            cp2 = __readexpect_hexdigit
+            cp2 = cp2 * 0x10 + __readexpect_hexdigit
+            cp2 = cp2 * 0x10 + __readexpect_hexdigit
+            cp2 = cp2 * 0x10 + __readexpect_hexdigit
+
+            __failed("low surrogate not in low surrogate range") unless cp2 >= 0xdc00 && cp2 <= 0xdfff
+
+            cp  &= 0x3ff
+            cp2 &= 0x3ff
+            cp  *= 0x400
+            cp  += cp2
+            cp  += 0x10000
+          end
+
           str << [cp].pack(STRING_U) # is this faster than just doing the ~~mental~~ bit arithmetic? idk probably?
           return true
         when nil

diff --git a/test/data/n_string_1_surrogate_then_escape.err b/test/data/n_string_1_surrogate_then_escape.err
@@ -1 +1 @@
-
+%q(Expected "u", but got "\"" at [1:10])
diff --git a/test/data/n_string_1_surrogate_then_escape_u.err b/test/data/n_string_1_surrogate_then_escape_u.err
@@ -1 +1 @@
-
+%q(expected hex digit [0-9a-fA-F] got "\"" at [1:11])
diff --git a/test/data/n_string_1_surrogate_then_escape_u1.err b/test/data/n_string_1_surrogate_then_escape_u1.err
@@ -1 +1 @@
-
+%q(expected hex digit [0-9a-fA-F] got "\"" at [1:12])
diff --git a/test/data/n_string_1_surrogate_then_escape_u1x.err b/test/data/n_string_1_surrogate_then_escape_u1x.err
@@ -1 +1 @@
-
+%q(expected hex digit [0-9a-fA-F] got "x" at [1:12])
diff --git a/test/data/n_string_incomplete_surrogate_escape_invalid.err b/test/data/n_string_incomplete_surrogate_escape_invalid.err
@@ -1 +1 @@
-%q(unexpected escape "x" at [1:16])
+%q(low surrogate not in low surrogate range at [1:15])