Avoid some checks for len == 0

src/parser3.h

@@ -160,9 +160,6 @@ inline Status n_json(Parser3 *self) {
 }
 
 inline Status n_value(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   switch (*self->buf) {
   case '{':
     self->pop();
@@ -226,9 +223,6 @@ inline Status n_value(Parser3 *self) {
 }
 
 inline Status n_object(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   if (*self->buf != '{') {
     return S_REJECT;
   }
@@ -242,9 +236,6 @@ inline Status n_object(Parser3 *self) {
 }
 
 inline Status n_object2(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   switch (*self->buf) {
   case '}':
     ++self->buf;
@@ -264,9 +255,6 @@ inline Status n_object2(Parser3 *self) {
 }
 
 inline Status n_object3(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   switch (*self->buf) {
   case '}':
     ++self->buf;
@@ -287,9 +275,6 @@ inline Status n_object3(Parser3 *self) {
 }
 
 inline Status n_array(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   if (*self->buf != '[') {
     return S_REJECT;
   }
@@ -303,9 +288,6 @@ inline Status n_array(Parser3 *self) {
 }
 
 inline Status n_array2(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   switch (*self->buf) {
   case ']':
     ++self->buf;
@@ -322,9 +304,6 @@ inline Status n_array2(Parser3 *self) {
 }
 
 inline Status n_array3(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   switch (*self->buf) {
   case ']':
     ++self->buf;
@@ -352,9 +331,6 @@ inline Status n_element(Parser3 *self) {
 }
 
 inline Status n_string(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   if (*self->buf != '"') {
     return S_REJECT;
   }
@@ -369,9 +345,6 @@ inline Status n_string(Parser3 *self) {
 }
 
 inline Status n_string2(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   if (tables.invalidStringByte[uint8_t(*self->buf)]) {
     return S_REJECT;
   }
@@ -433,9 +406,6 @@ inline Status n_string2(Parser3 *self) {
 }
 
 inline Status n_string_following_escape(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   switch (*self->buf) {
   case '"':
   case '\\':
@@ -465,9 +435,6 @@ inline Status n_string_following_escape(Parser3 *self) {
 }
 
 inline Status t_utf8_continuation_byte(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   if (tables.invalidStringByte[uint8_t(*self->buf)]) {
     return S_REJECT;
   }
@@ -482,9 +449,6 @@ inline Status t_utf8_continuation_byte(Parser3 *self) {
 }
 
 inline Status t_utf8_last_continuation_byte(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   if (tables.invalidStringByte[uint8_t(*self->buf)]) {
     return S_REJECT;
   }
@@ -505,9 +469,6 @@ inline Status t_utf8_last_continuation_byte(Parser3 *self) {
 }
 
 inline Status t_digit(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   if ('0' <= *self->buf && *self->buf <= '9') {
     ++self->buf;
     self->pop();
@@ -517,9 +478,6 @@ inline Status t_digit(Parser3 *self) {
 }
 
 inline Status t_onenine(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   if ('1' <= *self->buf && *self->buf <= '9') {
     ++self->buf;
     self->pop();
@@ -529,9 +487,6 @@ inline Status t_onenine(Parser3 *self) {
 }
 
 inline Status t_hex(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   self->utf8Codepoint <<= 4;
   if (('0' <= *self->buf && *self->buf <= '9')) {
     self->utf8Codepoint |= *self->buf - '0';
@@ -548,9 +503,6 @@ inline Status t_hex(Parser3 *self) {
 }
 
 inline Status t_hex2(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   self->utf8Codepoint <<= 4;
   if (('0' <= *self->buf && *self->buf <= '9')) {
     self->utf8Codepoint |= *self->buf - '0';
@@ -618,9 +570,6 @@ inline Status t_hex2(Parser3 *self) {
 }
 
 inline Status t_hex3(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   self->utf8Codepoint <<= 4;
   if (('0' <= *self->buf && *self->buf <= '9')) {
     self->utf8Codepoint |= *self->buf - '0';
@@ -679,9 +628,6 @@ inline Status n_number(Parser3 *self) {
 }
 
 inline Status n_integer(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   self->callbacks->on_begin_number(self->data);
   self->dataBegin = self->buf;
   switch (*self->buf) {
@@ -717,9 +663,6 @@ inline Status n_integer(Parser3 *self) {
 }
 
 inline Status n_integer2(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   switch (*self->buf) {
   case '0':
     ++self->buf;
@@ -746,9 +689,6 @@ inline Status n_integer2(Parser3 *self) {
 }
 
 inline Status n_digits(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   switch (*self->buf) {
   case '0':
   case '1':
@@ -867,9 +807,6 @@ inline Status n_whitespace(Parser3 *self) {
 }
 
 inline Status n_true(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   if (*self->buf == 'e') {
     ++self->buf;
     self->pop();
@@ -880,9 +817,6 @@ inline Status n_true(Parser3 *self) {
 }
 
 inline Status n_false(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   if (*self->buf == 'e') {
     ++self->buf;
     self->pop();
@@ -893,9 +827,6 @@ inline Status n_false(Parser3 *self) {
 }
 
 inline Status n_null(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   if (*self->buf == 'l') {
     ++self->buf;
     self->pop();
@@ -906,9 +837,6 @@ inline Status n_null(Parser3 *self) {
 }
 
 template <char kChar> inline Status singleChar(Parser3 *self) {
-  if (self->len() == 0) {
-    return S_REJECT;
-  }
   if (*self->buf == kChar) {
     ++self->buf;
     self->pop();
@@ -982,6 +910,7 @@ constexpr inline struct ContinuationTable {
     continuations[T_EOF] = t_eof;
     continuations[T_END_NUMBER] = t_end_number;
     continuations[T_BACKSLASH] = singleChar<'\\'>;
+
     symbolNames[N_JSON] = "n_json";
     symbolNames[N_VALUE] = "n_value";
     symbolNames[N_OBJECT] = "n_object";
@@ -1022,13 +951,25 @@ constexpr inline struct ContinuationTable {
     symbolNames[T_EOF] = "t_eof";
     symbolNames[T_BACKSLASH] = "singleChar<'\\'>";
     symbolNames[T_END_NUMBER] = "t_end_number";
+
+    // All others can assume that there's at least one byte when they're called
+    acceptsEmptyString[N_DIGITS2] = true;
+    acceptsEmptyString[N_FRACTION] = true;
+    acceptsEmptyString[N_EXPONENT] = true;
+    acceptsEmptyString[N_SIGN] = true;
+    acceptsEmptyString[N_WHITESPACE] = true;
+    acceptsEmptyString[T_EOF] = true;
+    acceptsEmptyString[T_END_NUMBER] = true;
   }
   Continuation continuations[N_SYMBOL_COUNT]{};
   const char *symbolNames[N_SYMBOL_COUNT]{};
+  bool acceptsEmptyString[N_SYMBOL_COUNT]{};
 } symbolTables;
 
 inline Status Parser3::keepGoing(Parser3 *self) {
-  if (self->len() == 0 && !self->complete) {
+  // self->debugPrint();
+  if (self->len() == 0) {
+    if (!self->complete) {
       switch (self->top()) {
       case N_NUMBER:
       case N_INTEGER:
@@ -1080,7 +1021,10 @@ inline Status Parser3::keepGoing(Parser3 *self) {
       }
       return S_AGAIN;
     }
-  // self->debugPrint();
+    if (!symbolTables.acceptsEmptyString[self->top()]) {
+      return S_REJECT;
+    }
+  }
   MUSTTAIL return symbolTables.continuations[self->top()](self);
 }