Safety: check index of scalar values before use

jkeiser · jkeiser · commit 73c7b7db3da1 · 2020-12-11T11:57:09.000-08:00
diff --git a/include/simdjson/generic/ondemand/document-inl.h b/include/simdjson/generic/ondemand/document-inl.h
@@ -8,8 +8,8 @@ simdjson_really_inline document::document(ondemand::json_iterator &&_iter) noexc
   logger::log_start_value(iter, "document");
 }
 
-simdjson_really_inline void document::assert_at_start() const noexcept {
-  iter.assert_at_start();
+simdjson_really_inline void document::assert_at_root() const noexcept {
+  iter.assert_at_root();
 }
 simdjson_really_inline document document::start(json_iterator &&iter) noexcept {
   return document(std::forward<json_iterator>(iter));
@@ -19,8 +19,8 @@ simdjson_really_inline value document::as_value() noexcept {
   return as_value_iterator();
 }
 simdjson_really_inline value_iterator document::as_value_iterator() noexcept {
-  assert_at_start();
-  return value_iterator(&iter, 1);
+  assert_at_root();
+  return value_iterator(&iter, 1, iter.token.checkpoint());
 }
 
 simdjson_really_inline simdjson_result<array> document::get_array() & noexcept {
@@ -30,15 +30,12 @@ simdjson_really_inline simdjson_result<object> document::get_object() & noexcept
   return as_value().get_object();
 }
 simdjson_really_inline simdjson_result<uint64_t> document::get_uint64() noexcept {
-  assert_at_start();
   return as_value_iterator().require_root_uint64();
 }
 simdjson_really_inline simdjson_result<int64_t> document::get_int64() noexcept {
-  assert_at_start();
   return as_value_iterator().require_root_int64();
 }
 simdjson_really_inline simdjson_result<double> document::get_double() noexcept {
-  assert_at_start();
   return as_value_iterator().require_root_double();
 }
 simdjson_really_inline simdjson_result<std::string_view> document::get_string() & noexcept {
@@ -48,11 +45,9 @@ simdjson_really_inline simdjson_result<raw_json_string> document::get_raw_json_s
   return as_value().get_raw_json_string();
 }
 simdjson_really_inline simdjson_result<bool> document::get_bool() noexcept {
-  assert_at_start();
   return as_value_iterator().require_root_bool();
 }
 simdjson_really_inline bool document::is_null() noexcept {
-  assert_at_start();
   return as_value_iterator().is_root_null();
 }
 
diff --git a/include/simdjson/generic/ondemand/document.h b/include/simdjson/generic/ondemand/document.h
@@ -232,7 +232,7 @@ class document {
   simdjson_really_inline value_iterator as_value_iterator() noexcept;
   static simdjson_really_inline document start(ondemand::json_iterator &&iter) noexcept;
 
-  simdjson_really_inline void assert_at_start() const noexcept;
+  simdjson_really_inline void assert_at_root() const noexcept;
 
   //
   // Fields
diff --git a/include/simdjson/generic/ondemand/json_iterator-inl.h b/include/simdjson/generic/ondemand/json_iterator-inl.h
@@ -90,11 +90,11 @@ simdjson_warn_unused simdjson_really_inline error_code json_iterator::skip_child
   return report_error(TAPE_ERROR, "not enough close braces");
 }
 
-simdjson_really_inline bool json_iterator::at_start() const noexcept {
-  return token.index == &parser->dom_parser.structural_indexes[0];
+simdjson_really_inline bool json_iterator::at_root() const noexcept {
+  return token.index == parser->dom_parser.structural_indexes.get();
 }
 
-simdjson_really_inline void json_iterator::assert_at_start() const noexcept {
+simdjson_really_inline void json_iterator::assert_at_root() const noexcept {
   SIMDJSON_ASSUME( _depth == 1 );
   // Visual Studio Clang treats unique_ptr.get() as "side effecting."
 #ifndef SIMDJSON_CLANG_VISUAL_STUDIO
diff --git a/include/simdjson/generic/ondemand/json_iterator.h b/include/simdjson/generic/ondemand/json_iterator.h
@@ -60,12 +60,12 @@ class json_iterator {
   /**
    * Tell whether the iterator is still at the start
    */
-  simdjson_really_inline bool at_start() const noexcept;
+  simdjson_really_inline bool at_root() const noexcept;
 
   /**
    * Assert if the iterator is not at the start
    */
-  simdjson_really_inline void assert_at_start() const noexcept;
+  simdjson_really_inline void assert_at_root() const noexcept;
 
   /**
    * Tell whether the iterator is at the EOF mark
diff --git a/include/simdjson/generic/ondemand/token_iterator-inl.h b/include/simdjson/generic/ondemand/token_iterator-inl.h
@@ -39,6 +39,10 @@ simdjson_really_inline bool token_iterator::operator<=(const token_iterator &oth
   return index <= other.index;
 }
 
+simdjson_really_inline const uint32_t *token_iterator::checkpoint() const noexcept {
+  return index;
+}
+
 } // namespace ondemand
 } // namespace SIMDJSON_IMPLEMENTATION
 } // namespace simdjson
diff --git a/include/simdjson/generic/ondemand/token_iterator.h b/include/simdjson/generic/ondemand/token_iterator.h
@@ -49,6 +49,11 @@ class token_iterator {
    */
   simdjson_really_inline const uint8_t *advance() noexcept;
 
+  /**
+   * Save the current index to be restored later.
+   */
+  simdjson_really_inline const uint32_t *checkpoint() const noexcept;
+
   // NOTE: we don't support a full C++ iterator interface, because we expect people to make
   // different calls to advance the iterator based on *their own* state.
 
@@ -77,6 +82,7 @@ class token_iterator {
   const uint32_t *index{};
 
   friend class json_iterator;
+  friend class value_iterator;
   friend simdjson_really_inline void logger::log_line(const json_iterator &iter, const char *title_prefix, const char *title, std::string_view detail, int delta, int depth_delta) noexcept;
 };
 
diff --git a/include/simdjson/generic/ondemand/value_iterator-inl.h b/include/simdjson/generic/ondemand/value_iterator-inl.h
diff --git a/include/simdjson/generic/ondemand/value_iterator.h b/include/simdjson/generic/ondemand/value_iterator.h

Original file line number	Diff line number	Diff line change
`@@ -8,8 +8,8 @@ simdjson_really_inline document::document(ondemand::json_iterator &&_iter) noexc`
`8`	`8`	`logger::log_start_value(iter, "document");`
`9`	`9`	`}`
`10`	`10`
`11`		`-simdjson_really_inline void document::assert_at_start() const noexcept {`
`12`		`- iter.assert_at_start();`
	`11`	`+simdjson_really_inline void document::assert_at_root() const noexcept {`
	`12`	`+ iter.assert_at_root();`
`13`	`13`	`}`
`14`	`14`	`simdjson_really_inline document document::start(json_iterator &&iter) noexcept {`
`15`	`15`	`return document(std::forward<json_iterator>(iter));`
`@@ -19,8 +19,8 @@ simdjson_really_inline value document::as_value() noexcept {`
`19`	`19`	`return as_value_iterator();`
`20`	`20`	`}`
`21`	`21`	`simdjson_really_inline value_iterator document::as_value_iterator() noexcept {`
`22`		`- assert_at_start();`
`23`		`- return value_iterator(&iter, 1);`
	`22`	`+ assert_at_root();`
	`23`	`+ return value_iterator(&iter, 1, iter.token.checkpoint());`
`24`	`24`	`}`
`25`	`25`
`26`	`26`	`simdjson_really_inline simdjson_result<array> document::get_array() & noexcept {`
`@@ -30,15 +30,12 @@ simdjson_really_inline simdjson_result<object> document::get_object() & noexcept`
`30`	`30`	`return as_value().get_object();`
`31`	`31`	`}`
`32`	`32`	`simdjson_really_inline simdjson_result<uint64_t> document::get_uint64() noexcept {`
`33`		`- assert_at_start();`
`34`	`33`	`return as_value_iterator().require_root_uint64();`
`35`	`34`	`}`
`36`	`35`	`simdjson_really_inline simdjson_result<int64_t> document::get_int64() noexcept {`
`37`		`- assert_at_start();`
`38`	`36`	`return as_value_iterator().require_root_int64();`
`39`	`37`	`}`
`40`	`38`	`simdjson_really_inline simdjson_result<double> document::get_double() noexcept {`
`41`		`- assert_at_start();`
`42`	`39`	`return as_value_iterator().require_root_double();`
`43`	`40`	`}`
`44`	`41`	`simdjson_really_inline simdjson_result<std::string_view> document::get_string() & noexcept {`
`@@ -48,11 +45,9 @@ simdjson_really_inline simdjson_result<raw_json_string> document::get_raw_json_s`
`48`	`45`	`return as_value().get_raw_json_string();`
`49`	`46`	`}`
`50`	`47`	`simdjson_really_inline simdjson_result<bool> document::get_bool() noexcept {`
`51`		`- assert_at_start();`
`52`	`48`	`return as_value_iterator().require_root_bool();`
`53`	`49`	`}`
`54`	`50`	`simdjson_really_inline bool document::is_null() noexcept {`
`55`		`- assert_at_start();`
`56`	`51`	`return as_value_iterator().is_root_null();`
`57`	`52`	`}`
`58`	`53`
Original file line number	Diff line number	Diff line change
`@@ -90,11 +90,11 @@ simdjson_warn_unused simdjson_really_inline error_code json_iterator::skip_child`
`90`	`90`	`return report_error(TAPE_ERROR, "not enough close braces");`
`91`	`91`	`}`
`92`	`92`
`93`		`-simdjson_really_inline bool json_iterator::at_start() const noexcept {`
`94`		`- return token.index == &parser->dom_parser.structural_indexes[0];`
	`93`	`+simdjson_really_inline bool json_iterator::at_root() const noexcept {`
	`94`	`+ return token.index == parser->dom_parser.structural_indexes.get();`
`95`	`95`	`}`
`96`	`96`
`97`		`-simdjson_really_inline void json_iterator::assert_at_start() const noexcept {`
	`97`	`+simdjson_really_inline void json_iterator::assert_at_root() const noexcept {`
`98`	`98`	`SIMDJSON_ASSUME( _depth == 1 );`
`99`	`99`	`// Visual Studio Clang treats unique_ptr.get() as "side effecting."`
`100`	`100`	`#ifndef SIMDJSON_CLANG_VISUAL_STUDIO`