Fixing the string_view issue.

lemire · lemire · commit 5e690c5d04b9 · 2020-06-25T10:02:10.000-04:00
diff --git a/doc/basics.md b/doc/basics.md
@@ -9,6 +9,7 @@ An overview of what you need to know to use simdjson, with examples.
 * [The Basics: Loading and Parsing JSON Documents](#the-basics-loading-and-parsing-json-documents)
 * [Using the Parsed JSON](#using-the-parsed-json)
 * [C++17 Support](#c++17-support)
+* [C++11 Support and string_view](#c++11-support-and-string_view)
 * [Minifying JSON strings without parsing](#minifying-json-strings-without-parsing)
 * [UTF-8 validation (alone)](#utf-8-validation-alone)
 * [JSON Pointer](#json-pointer)
@@ -192,6 +193,19 @@ And another one:
   cout << "number: " << v << endl;
 ```
 
+
+C++11 Support and string_view
+-------------
+
+The simdjson library builds on compilers supporting the C++11 standard.
+We represent parsed strings in simdjson using the `std::string_view` class. 
+This class has become standard as part of C++17 but it is not always available
+on compilers which only supports C++11. When we detect that it is unavailable,
+we use [string-view-lite](https://github.com/martinmoene/string-view-lite) as a
+substitute. In such cases, we use the type alias `using string_view = nonstd::string_view;` to  
+offer the same API, irrespective of the compiler and standard library.
+
+
 C++17 Support
 -------------
 
diff --git a/include/simdjson/dom/element.h b/include/simdjson/dom/element.h
@@ -62,21 +62,42 @@ class element {
    */
   inline simdjson_result<object> get_object() const noexcept;
   /**
-   * Cast this element to a string.
+   * Cast this element to a null-terminated C string. 
+   * 
+   * The string is guaranteed to be valid UTF-8.
    *
-   * Equivalent to get<const char *>().
+   * The get_c_str() function is equivalent to get<const char *>().
+   * 
+   * The length of the string is given by get_string_length(). Because JSON strings
+   * may contain null characters, it may be incorrect to use strlen to determine the 
+   * string length.
    *
-   * @returns An pointer to a null-terminated string. This string is stored in the parser and will
+   * It is possible to get a single string_view instance which represents both the string
+   * content and its length: see get_string().
+   *
+   * @returns A pointer to a null-terminated UTF-8 string. This string is stored in the parser and will
    *          be invalidated the next time it parses a document or when it is destroyed.
    *          Returns INCORRECT_TYPE if the JSON element is not a string.
    */
   inline simdjson_result<const char *> get_c_str() const noexcept;
   /**
-   * Cast this element to a string.
+   * Gives the length in bytes of the string.
+   * 
+   * It is possible to get a single string_view instance which represents both the string
+   * content and its length: see get_string().
+   *
+   * @returns A string length in bytes.
+   *          Returns INCORRECT_TYPE if the JSON element is not a string.
+   */
+  inline simdjson_result<size_t> get_string_length() const noexcept;
+  /**
+   * Cast this element to a string. 
+   * 
+   * The string is guaranteed to be valid UTF-8.
    *
    * Equivalent to get<std::string_view>().
    *
-   * @returns A string. The string is stored in the parser and will be invalidated the next time it
+   * @returns An UTF-8 string. The string is stored in the parser and will be invalidated the next time it
    *          parses a document or when it is destroyed.
    *          Returns INCORRECT_TYPE if the JSON element is not a string.
    */
@@ -253,7 +274,9 @@ class element {
   inline operator bool() const noexcept(false);
 
   /**
-   * Read this element as a null-terminated string.
+   * Read this element as a null-terminated UTF-8 string.
+   * 
+   * Be mindful that JSON allows strings to contain null characters.
    *
    * Does *not* convert other types to a string; requires that the JSON type of the element was
    * an actual string.
@@ -264,7 +287,7 @@ class element {
   inline explicit operator const char*() const noexcept(false);
 
   /**
-   * Read this element as a null-terminated string.
+   * Read this element as a null-terminated UTF-8 string.
    *
    * Does *not* convert other types to a string; requires that the JSON type of the element was
    * an actual string.
diff --git a/include/simdjson/inline/element.h b/include/simdjson/inline/element.h
@@ -50,6 +50,10 @@ really_inline simdjson_result<const char *> simdjson_result<dom::element>::get_c
   if (error()) { return error(); }
   return first.get_c_str();
 }
+really_inline simdjson_result<size_t> simdjson_result<dom::element>::get_string_length() const noexcept {
+  if (error()) { return error(); }
+  return first.get_string_length();
+}
 really_inline simdjson_result<std::string_view> simdjson_result<dom::element>::get_string() const noexcept {
   if (error()) { return error(); }
   return first.get_string();
@@ -190,6 +194,15 @@ inline simdjson_result<const char *> element::get_c_str() const noexcept {
       return INCORRECT_TYPE;
   }
 }
+inline simdjson_result<size_t> element::get_string_length() const noexcept {
+  switch (tape.tape_ref_type()) {
+    case internal::tape_type::STRING: {
+      return tape.get_string_length();
+    }
+    default:
+      return INCORRECT_TYPE;
+  }
+}
 inline simdjson_result<std::string_view> element::get_string() const noexcept {
   switch (tape.tape_ref_type()) {
     case internal::tape_type::STRING:
