Fix several unescaping issues

src/fuzz.cpp

@@ -37,9 +37,19 @@ std::pair<std::string, parser3::Status> runBatch(std::string copy) {
 }
 
 void testStreaming(std::string const &json) {
-  auto result1 = runStreaming(json);
-  auto result2 = runBatch(json);
-  if (result1 != result2) {
+  auto streaming = runStreaming(json);
+  auto batch = runBatch(json);
+  if (streaming != batch) {
+    if (streaming.second == batch.second && streaming.second != parser3::S_OK) {
+      // It's ok if the processed data doesn't match if parsing failed
+      return;
+    }
+    printf("streaming: %s, %s\n",
+           streaming.second == parser3::S_OK ? "accept" : "reject",
+           streaming.first.c_str());
+    printf("batch: %s, %s\n",
+           streaming.second == parser3::S_OK ? "accept" : "reject",
+           batch.first.c_str());
     abort();
   }
 }

src/parser3.h

@@ -353,7 +353,7 @@ inline Status n_string(Parser3 *self) {
   }
   self->callbacks->on_begin_string(self->data);
   ++self->buf;
-  self->dataBegin = self->buf;
+  self->dataBegin = self->writeBuf = self->buf;
   self->pop();
   if (auto s = self->push({N_STRING2})) {
     return s;
@@ -372,10 +372,10 @@ inline Status n_string2(Parser3 *self) {
     // one byte utf-8 encoding
     switch (*self->buf) {
     case '"':
-      ++self->buf;
-      self->pop();
       self->flushString();
       self->callbacks->on_end_string(self->data);
+      ++self->buf;
+      self->pop();
       MUSTTAIL return Parser3::keepGoing(self);
     case '\\':
       ++self->buf;
@@ -385,15 +385,14 @@ inline Status n_string2(Parser3 *self) {
       }
       MUSTTAIL return Parser3::keepGoing(self);
     default:
-      ++self->buf;
+      *self->writeBuf++ = *self->buf++;
       MUSTTAIL return Parser3::keepGoing(self);
     }
   } else if ((*self->buf & 0b11100000) == 0b11000000) {
     // two byte utf-8 encoding
     self->utf8Codepoint = *self->buf & 0b00011111;
     self->minCodepoint = 0x80;
-    ++self->buf;
-    ++self->writeBuf;
+    *self->writeBuf++ = *self->buf++;
     self->pop();
     if (auto s = self->push({T_UTF8_LAST_CONTINUATION_BYTE, N_STRING2})) {
       return s;
@@ -404,8 +403,7 @@ inline Status n_string2(Parser3 *self) {
     // three byte utf-8 encoding
     self->utf8Codepoint = *self->buf & 0b00001111;
     self->minCodepoint = 0x800;
-    ++self->buf;
-    ++self->writeBuf;
+    *self->writeBuf++ = *self->buf++;
     self->pop();
     if (auto s = self->push({T_UTF8_CONTINUATION_BYTE,
                              T_UTF8_LAST_CONTINUATION_BYTE, N_STRING2})) {
@@ -416,8 +414,7 @@ inline Status n_string2(Parser3 *self) {
     // four byte utf-8 encoding
     self->utf8Codepoint = *self->buf & 0b00000111;
     self->minCodepoint = 0x10000;
-    ++self->buf;
-    ++self->writeBuf;
+    *self->writeBuf++ = *self->buf++;
     self->pop();
     if (auto s = self->push({T_UTF8_CONTINUATION_BYTE, T_UTF8_CONTINUATION_BYTE,
                              T_UTF8_LAST_CONTINUATION_BYTE, N_STRING2})) {
@@ -470,8 +467,7 @@ inline Status t_utf8_continuation_byte(Parser3 *self) {
   if ((*self->buf & 0b11000000) == 0b10000000) {
     self->utf8Codepoint <<= 6;
     self->utf8Codepoint |= *self->buf & 0b00111111;
-    ++self->buf;
-    ++self->writeBuf;
+    *self->writeBuf++ = *self->buf++;
     self->pop();
     MUSTTAIL return Parser3::keepGoing(self);
   }
@@ -494,8 +490,7 @@ inline Status t_utf8_last_continuation_byte(Parser3 *self) {
       return S_REJECT;
     }
     // TODO tell valgrind utf8Codepoint and minCodepoint are uninitialized
-    ++self->buf;
-    ++self->writeBuf;
+    *self->writeBuf++ = *self->buf++;
     self->pop();
     MUSTTAIL return Parser3::keepGoing(self);
   }
@@ -533,8 +528,7 @@ inline Status t_hex(Parser3 *self) {
   if (('0' <= *self->buf && *self->buf <= '9') ||
       ('a' <= *self->buf && *self->buf <= 'f') ||
       ('A' <= *self->buf && *self->buf <= 'F')) {
-    ++self->buf;
-    ++self->writeBuf;
+    *self->writeBuf++ = *self->buf++;
     self->pop();
     MUSTTAIL return Parser3::keepGoing(self);
   }