Handle string framing (i.e. \")

src/test.cpp

@@ -151,6 +151,7 @@ enum Symbol : int8_t {
   T_DUBQUOTE,
   // Nonterminals
   N_STRING, // Not including leading double quote, but including trailing quote
+  N_STRING_FROM_ESCAPE, // Immediately after a backslach
   N_NUMBER,
   N_VALUE,
   N_ARRAY_VALUE_OR_END,
@@ -173,6 +174,7 @@ static const char *symbolNames[] = {
     "T_S",
     "T_DUBQUOTE",
     "N_STRING",
+    "N_STRING_FROM_ESCAPE",
     "N_NUMBER",
     "N_VALUE",
     "N_ARRAY_VALUE_OR_END",
@@ -366,16 +368,23 @@ private:
     if (!parseLiteral("\"")) {
       return false;
     }
-    auto *result = (char *)memchr(buf, '"', len());
-    if (result == nullptr) {
-      return false;
+    char *result = buf;
+    for (;;) {
+      result = (char *)memchr(result, '"', bufEnd - result);
+      if (result == nullptr) {
+        return false;
+      }
+      if (result != buf && result[-1] == '\\') {
+        ++result;
+        continue;
+      }
+      break;
     }
     int stringLen = result - buf;
-    callbacks->on_string_data(data, buf, stringLen);
-    buf += stringLen;
-    if (!parseLiteral("\"")) {
-      return false;
+    if (stringLen > 0) {
+      callbacks->on_string_data(data, buf, stringLen);
     }
+    buf += stringLen + 1;
     callbacks->on_end_string(data);
     return true;
   }
@@ -494,12 +503,38 @@ private:
     callbacks->on_end_number(data);
     return S_OK;
   }
-  Status parse_string() {
-    auto *result = (char *)memchr(buf, '"', len());
-    if (result == nullptr) {
-      callbacks->on_string_data(data, buf, len());
-      buf += len();
-      return S_AGAIN;
+  Status parse_string(bool fromEscape) {
+    auto *result = buf;
+    if (fromEscape) {
+      if (*result == '\"') {
+        ++result;
+      }
+      pop();
+      if (Status s = push({N_STRING})) {
+        return s;
+      }
+    }
+    for (;;) {
+      result = (char *)memchr(result, '"', bufEnd - result);
+      if (result == nullptr) {
+        callbacks->on_string_data(data, buf, len());
+        if (bufEnd[-1] == '\\') {
+          pop();
+          if (Status s = push({N_STRING_FROM_ESCAPE})) {
+            return s;
+          }
+        }
+        return S_AGAIN;
+      }
+      if (result != buf && result[-1] == '\\') {
+        ++result;
+        if (result == bufEnd) {
+          callbacks->on_string_data(data, buf, len());
+          return S_AGAIN;
+        }
+        continue;
+      }
+      break;
     }
     int stringLen = result - buf;
     if (stringLen > 0) {
@@ -531,7 +566,14 @@ private:
   }
 
   static Status string(Parser2 *self) {
-    if (Status s = self->parse_string()) {
+    if (Status s = self->parse_string(false)) {
+      return s;
+    }
+    self->pop();
+    MUSTTAIL return keepGoing(self);
+  }
+  static Status stringFromEscape(Parser2 *self) {
+    if (Status s = self->parse_string(true)) {
       return s;
     }
     self->pop();
@@ -730,6 +772,7 @@ private:
       /*T_S*/ singleChar<'s'>,
       /*T_DUBQUOTE*/ dubquote,
       /*N_STRING*/ string,
+      /*N_STRING_FROM_ESCAPE*/ stringFromEscape,
       /*N_NUMBER*/ number,
       /*N_VALUE*/ value,
       /*N_ARRAY_VALUE_OR_END*/ arrayOrEnd,
@@ -813,10 +856,30 @@ Callbacks printCallbacks() {
   return result;
 }
 
+Callbacks minifyCallbacks() {
+  Callbacks result;
+  result.on_begin_object = +[](void *) { printf("{"); };
+  result.on_end_object = +[](void *) { printf("}\n"); };
+  result.on_begin_string = +[](void *) { printf("\""); };
+  result.on_string_data =
+      +[](void *, const char *buf, int len) { printf("%.*s", len, buf); };
+  result.on_end_string = +[](void *) { printf("\""); };
+  result.on_begin_array = +[](void *) { printf("["); };
+  result.on_end_array = +[](void *) { printf("]\n"); };
+  result.on_begin_number = +[](void *) {};
+  result.on_number_data =
+      +[](void *, const char *buf, int len) { printf("%.*s", len, buf); };
+  result.on_end_number = +[](void *) {};
+  result.on_true_literal = +[](void *) { printf("true"); };
+  result.on_false_literal = +[](void *) { printf("false"); };
+  result.on_null_literal = +[](void *) { printf("null"); };
+  return result;
+}
+
 } // namespace
 
 TEST_CASE("parser1") {
-  Callbacks c = printCallbacks();
+  Callbacks c = minifyCallbacks();
   {
     auto copy = json;
     Parser1 parser(copy.data(), copy.length(), &c, nullptr);
@@ -830,7 +893,7 @@ TEST_CASE("parser1") {
 }
 
 TEST_CASE("parser2") {
-  Callbacks c = printCallbacks();
+  Callbacks c = minifyCallbacks();
   {
     auto copy = json;
     Parser2 parser(&c, nullptr);