simdjson  3.11.0
Ridiculously Fast JSON
array-inl.h
1 #ifndef SIMDJSON_GENERIC_ONDEMAND_ARRAY_INL_H
2 
3 #ifndef SIMDJSON_CONDITIONAL_INCLUDE
4 #define SIMDJSON_GENERIC_ONDEMAND_ARRAY_INL_H
5 #include "simdjson/jsonpathutil.h"
6 #include "simdjson/generic/ondemand/base.h"
7 #include "simdjson/generic/ondemand/array.h"
8 #include "simdjson/generic/ondemand/array_iterator-inl.h"
9 #include "simdjson/generic/ondemand/json_iterator.h"
10 #include "simdjson/generic/ondemand/value.h"
11 #include "simdjson/generic/ondemand/value_iterator-inl.h"
12 #endif // SIMDJSON_CONDITIONAL_INCLUDE
13 
14 namespace simdjson {
15 namespace SIMDJSON_IMPLEMENTATION {
16 namespace ondemand {
17 
18 //
19 // ### Live States
20 //
21 // While iterating or looking up values, depth >= iter->depth. at_start may vary. Error is
22 // always SUCCESS:
23 //
24 // - Start: This is the state when the array is first found and the iterator is just past the `{`.
25 // In this state, at_start == true.
26 // - Next: After we hand a scalar value to the user, or an array/object which they then fully
27 // iterate over, the iterator is at the `,` before the next value (or `]`). In this state,
28 // depth == iter->depth, at_start == false, and error == SUCCESS.
29 // - Unfinished Business: When we hand an array/object to the user which they do not fully
30 // iterate over, we need to finish that iteration by skipping child values until we reach the
31 // Next state. In this state, depth > iter->depth, at_start == false, and error == SUCCESS.
32 //
33 // ## Error States
34 //
35 // In error states, we will yield exactly one more value before stopping. iter->depth == depth
36 // and at_start is always false. We decrement after yielding the error, moving to the Finished
37 // state.
38 //
39 // - Chained Error: When the array iterator is part of an error chain--for example, in
40 // `for (auto tweet : doc["tweets"])`, where the tweet element may be missing or not be an
41 // array--we yield that error in the loop, exactly once. In this state, error != SUCCESS and
42 // iter->depth == depth, and at_start == false. We decrement depth when we yield the error.
43 // - Missing Comma Error: When the iterator ++ method discovers there is no comma between elements,
44 // we flag that as an error and treat it exactly the same as a Chained Error. In this state,
45 // error == TAPE_ERROR, iter->depth == depth, and at_start == false.
46 //
47 // ## Terminal State
48 //
49 // The terminal state has iter->depth < depth. at_start is always false.
50 //
51 // - Finished: When we have reached a `]` or have reported an error, we are finished. We signal this
52 // by decrementing depth. In this state, iter->depth < depth, at_start == false, and
53 // error == SUCCESS.
54 //
55 
56 simdjson_inline array::array(const value_iterator &_iter) noexcept
57  : iter{_iter}
58 {
59 }
60 
61 simdjson_inline simdjson_result<array> array::start(value_iterator &iter) noexcept {
62  // We don't need to know if the array is empty to start iteration, but we do want to know if there
63  // is an error--thus `simdjson_unused`.
64  simdjson_unused bool has_value;
65  SIMDJSON_TRY( iter.start_array().get(has_value) );
66  return array(iter);
67 }
68 simdjson_inline simdjson_result<array> array::start_root(value_iterator &iter) noexcept {
69  simdjson_unused bool has_value;
70  SIMDJSON_TRY( iter.start_root_array().get(has_value) );
71  return array(iter);
72 }
73 simdjson_inline simdjson_result<array> array::started(value_iterator &iter) noexcept {
74  bool has_value;
75  SIMDJSON_TRY(iter.started_array().get(has_value));
76  return array(iter);
77 }
78 
79 simdjson_inline simdjson_result<array_iterator> array::begin() noexcept {
80 #if SIMDJSON_DEVELOPMENT_CHECKS
81  if (!iter.is_at_iterator_start()) { return OUT_OF_ORDER_ITERATION; }
82 #endif
83  return array_iterator(iter);
84 }
85 simdjson_inline simdjson_result<array_iterator> array::end() noexcept {
86  return array_iterator(iter);
87 }
88 simdjson_inline error_code array::consume() noexcept {
89  auto error = iter.json_iter().skip_child(iter.depth()-1);
90  if(error) { iter.abandon(); }
91  return error;
92 }
93 
95  const uint8_t * starting_point{iter.peek_start()};
96  auto error = consume();
97  if(error) { return error; }
98  // After 'consume()', we could be left pointing just beyond the document, but that
99  // is ok because we are not going to dereference the final pointer position, we just
100  // use it to compute the length in bytes.
101  const uint8_t * final_point{iter._json_iter->unsafe_pointer()};
102  return std::string_view(reinterpret_cast<const char*>(starting_point), size_t(final_point - starting_point));
103 }
104 
105 SIMDJSON_PUSH_DISABLE_WARNINGS
106 SIMDJSON_DISABLE_STRICT_OVERFLOW_WARNING
107 simdjson_inline simdjson_result<size_t> array::count_elements() & noexcept {
108  size_t count{0};
109  // Important: we do not consume any of the values.
110  for(simdjson_unused auto v : *this) { count++; }
111  // The above loop will always succeed, but we want to report errors.
112  if(iter.error()) { return iter.error(); }
113  // We need to move back at the start because we expect users to iterate through
114  // the array after counting the number of elements.
115  iter.reset_array();
116  return count;
117 }
118 SIMDJSON_POP_DISABLE_WARNINGS
119 
120 simdjson_inline simdjson_result<bool> array::is_empty() & noexcept {
121  bool is_not_empty;
122  auto error = iter.reset_array().get(is_not_empty);
123  if(error) { return error; }
124  return !is_not_empty;
125 }
126 
127 inline simdjson_result<bool> array::reset() & noexcept {
128  return iter.reset_array();
129 }
130 
131 inline simdjson_result<value> array::at_pointer(std::string_view json_pointer) noexcept {
132  if (json_pointer[0] != '/') { return INVALID_JSON_POINTER; }
133  json_pointer = json_pointer.substr(1);
134  // - means "the append position" or "the element after the end of the array"
135  // We don't support this, because we're returning a real element, not a position.
136  if (json_pointer == "-") { return INDEX_OUT_OF_BOUNDS; }
137 
138  // Read the array index
139  size_t array_index = 0;
140  size_t i;
141  for (i = 0; i < json_pointer.length() && json_pointer[i] != '/'; i++) {
142  uint8_t digit = uint8_t(json_pointer[i] - '0');
143  // Check for non-digit in array index. If it's there, we're trying to get a field in an object
144  if (digit > 9) { return INCORRECT_TYPE; }
145  array_index = array_index*10 + digit;
146  }
147 
148  // 0 followed by other digits is invalid
149  if (i > 1 && json_pointer[0] == '0') { return INVALID_JSON_POINTER; } // "JSON pointer array index has other characters after 0"
150 
151  // Empty string is invalid; so is a "/" with no digits before it
152  if (i == 0) { return INVALID_JSON_POINTER; } // "Empty string in JSON pointer array index"
153  // Get the child
154  auto child = at(array_index);
155  // If there is an error, it ends here
156  if(child.error()) {
157  return child;
158  }
159 
160  // If there is a /, we're not done yet, call recursively.
161  if (i < json_pointer.length()) {
162  child = child.at_pointer(json_pointer.substr(i));
163  }
164  return child;
165 }
166 
167 inline simdjson_result<value> array::at_path(std::string_view json_path) noexcept {
168  auto json_pointer = json_path_to_pointer_conversion(json_path);
169  if (json_pointer == "-1") { return INVALID_JSON_POINTER; }
170  return at_pointer(json_pointer);
171 }
172 
173 simdjson_inline simdjson_result<value> array::at(size_t index) noexcept {
174  size_t i = 0;
175  for (auto value : *this) {
176  if (i == index) { return value; }
177  i++;
178  }
179  return INDEX_OUT_OF_BOUNDS;
180 }
181 
182 } // namespace ondemand
183 } // namespace SIMDJSON_IMPLEMENTATION
184 } // namespace simdjson
185 
186 namespace simdjson {
187 
188 simdjson_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::array>::simdjson_result(
189  SIMDJSON_IMPLEMENTATION::ondemand::array &&value
190 ) noexcept
191  : implementation_simdjson_result_base<SIMDJSON_IMPLEMENTATION::ondemand::array>(
192  std::forward<SIMDJSON_IMPLEMENTATION::ondemand::array>(value)
193  )
194 {
195 }
196 simdjson_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::array>::simdjson_result(
197  error_code error
198 ) noexcept
199  : implementation_simdjson_result_base<SIMDJSON_IMPLEMENTATION::ondemand::array>(error)
200 {
201 }
202 
203 simdjson_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::array_iterator> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::array>::begin() noexcept {
204  if (error()) { return error(); }
205  return first.begin();
206 }
207 simdjson_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::array_iterator> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::array>::end() noexcept {
208  if (error()) { return error(); }
209  return first.end();
210 }
211 simdjson_inline simdjson_result<size_t> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::array>::count_elements() & noexcept {
212  if (error()) { return error(); }
213  return first.count_elements();
214 }
215 simdjson_inline simdjson_result<bool> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::array>::is_empty() & noexcept {
216  if (error()) { return error(); }
217  return first.is_empty();
218 }
219 simdjson_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::array>::at(size_t index) noexcept {
220  if (error()) { return error(); }
221  return first.at(index);
222 }
223 simdjson_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::array>::at_pointer(std::string_view json_pointer) noexcept {
224  if (error()) { return error(); }
225  return first.at_pointer(json_pointer);
226 }
227 simdjson_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::array>::at_path(std::string_view json_path) noexcept {
228  if (error()) { return error(); }
229  return first.at_path(json_path);
230 }
231 simdjson_inline simdjson_result<std::string_view> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::array>::raw_json() noexcept {
232  if (error()) { return error(); }
233  return first.raw_json();
234 }
235 } // namespace simdjson
236 
237 #endif // SIMDJSON_GENERIC_ONDEMAND_ARRAY_INL_H
simdjson_inline simdjson_result< value > at(size_t index) noexcept
Get the value at the given index.
Definition: array-inl.h:173
value_iterator iter
Iterator marking current position.
Definition: array.h:181
simdjson_inline simdjson_result< bool > is_empty() &noexcept
This method scans the beginning of the array and checks whether the array is empty.
Definition: array-inl.h:120
simdjson_result< value > at_pointer(std::string_view json_pointer) noexcept
Get the value associated with the given JSON pointer.
Definition: array-inl.h:131
static simdjson_inline simdjson_result< array > start(value_iterator &iter) noexcept
Begin array iteration.
Definition: array-inl.h:61
simdjson_inline simdjson_result< array_iterator > begin() noexcept
Begin array iteration.
Definition: array-inl.h:79
simdjson_inline array() noexcept=default
Create a new invalid array.
simdjson_inline error_code consume() noexcept
Go to the end of the array, no matter where you are right now.
Definition: array-inl.h:88
simdjson_inline simdjson_result< size_t > count_elements() &noexcept
This method scans the array and counts the number of elements.
Definition: array-inl.h:107
simdjson_result< value > at_path(std::string_view json_path) noexcept
Get the value associated with the given JSONPath expression.
Definition: array-inl.h:167
static simdjson_inline simdjson_result< array > start_root(value_iterator &iter) noexcept
Begin array iteration from the root.
Definition: array-inl.h:68
static simdjson_inline simdjson_result< array > started(value_iterator &iter) noexcept
Begin array iteration.
Definition: array-inl.h:73
simdjson_inline simdjson_result< std::string_view > raw_json() noexcept
Consumes the array and returns a string_view instance corresponding to the array as represented in JS...
Definition: array-inl.h:94
simdjson_inline simdjson_result< array_iterator > end() noexcept
Sentinel representing the end of the array.
Definition: array-inl.h:85
simdjson_result< bool > reset() &noexcept
Reset the iterator so that we are pointing back at the beginning of the array.
Definition: array-inl.h:127
An ephemeral JSON value returned during iteration.
Definition: value.h:21
The top level simdjson namespace, containing everything the library provides.
Definition: base.h:8
error_code
All possible errors returned by simdjson.
Definition: error.h:19
@ INCORRECT_TYPE
JSON element has a different type than user expected.
Definition: error.h:37
@ OUT_OF_ORDER_ITERATION
tried to iterate an array or object out of order (checked when SIMDJSON_DEVELOPMENT_CHECKS=1)
Definition: error.h:46
@ INDEX_OUT_OF_BOUNDS
JSON array index too large.
Definition: error.h:39
@ INVALID_JSON_POINTER
Invalid JSON pointer syntax.
Definition: error.h:42
std::string json_path_to_pointer_conversion(std::string_view json_path)
Converts JSONPath to JSON Pointer.
Definition: jsonpathutil.h:13
The result of a simdjson operation that could fail.
Definition: error.h:215
simdjson_inline error_code error() const noexcept
The error.
Definition: error-inl.h:131