|
13 | 13 | import com.fasterxml.jackson.core.JsonParser; |
14 | 14 | import com.fasterxml.jackson.core.JsonToken; |
15 | 15 |
|
16 | | -import org.elasticsearch.common.Strings; |
17 | 16 | import org.elasticsearch.core.CheckedConsumer; |
18 | 17 | import org.elasticsearch.test.ESTestCase; |
19 | | -import org.elasticsearch.xcontent.XContentBuilder; |
20 | 18 | import org.elasticsearch.xcontent.XContentString; |
21 | | -import org.elasticsearch.xcontent.json.JsonXContent; |
22 | 19 | import org.hamcrest.Matchers; |
23 | 20 |
|
24 | 21 | import java.io.IOException; |
25 | 22 | import java.nio.charset.StandardCharsets; |
| 23 | +import java.util.Locale; |
26 | 24 |
|
27 | 25 | public class ESUTF8StreamJsonParserTests extends ESTestCase { |
28 | 26 |
|
@@ -62,8 +60,18 @@ public void testGetValueAsText() throws IOException { |
62 | 60 | assertThat(parser.nextFieldName(), Matchers.equalTo("foo")); |
63 | 61 | assertThat(parser.nextValue(), Matchers.equalTo(JsonToken.VALUE_STRING)); |
64 | 62 |
|
| 63 | + var textRef = parser.getValueAsText(); |
| 64 | + assertThat(textRef, Matchers.notNullValue()); |
| 65 | + assertTextRef(textRef.bytes(), "bar\"baz\""); |
| 66 | + }); |
| 67 | + |
| 68 | + testParseJson("{\"foo\": \"b\\u00e5r\"}", parser -> { |
| 69 | + assertThat(parser.nextToken(), Matchers.equalTo(JsonToken.START_OBJECT)); |
| 70 | + assertThat(parser.nextFieldName(), Matchers.equalTo("foo")); |
| 71 | + assertThat(parser.nextValue(), Matchers.equalTo(JsonToken.VALUE_STRING)); |
| 72 | + |
65 | 73 | assertThat(parser.getValueAsText(), Matchers.nullValue()); |
66 | | - assertThat(parser.getValueAsString(), Matchers.equalTo("bar\"baz\"")); |
| 74 | + assertThat(parser.getValueAsString(), Matchers.equalTo("bår")); |
67 | 75 | }); |
68 | 76 |
|
69 | 77 | testParseJson("{\"foo\": \"bår\"}", parser -> { |
@@ -112,43 +120,98 @@ public void testGetValueAsText() throws IOException { |
112 | 120 | }); |
113 | 121 | } |
114 | 122 |
|
115 | | - private boolean validForTextRef(String value) { |
116 | | - for (char c : value.toCharArray()) { |
117 | | - if (c == '"') { |
118 | | - return false; |
| 123 | + private record TestInput(String input, String result, boolean supportsOptimized) {} |
| 124 | + |
| 125 | + private static final TestInput[] ESCAPE_SEQUENCES = { |
| 126 | + new TestInput("\\b", "\b", false), |
| 127 | + new TestInput("\\t", "\t", false), |
| 128 | + new TestInput("\\n", "\n", false), |
| 129 | + new TestInput("\\f", "\f", false), |
| 130 | + new TestInput("\\r", "\r", false), |
| 131 | + new TestInput("\\\"", "\"", true), |
| 132 | + new TestInput("\\/", "/", true), |
| 133 | + new TestInput("\\\\", "\\", true) }; |
| 134 | + |
| 135 | + private int randomCodepoint(boolean includeAscii) { |
| 136 | + while (true) { |
| 137 | + char val = Character.toChars(randomInt(0xFFFF))[0]; |
| 138 | + if (val <= 0x7f && includeAscii == false) { |
| 139 | + continue; |
119 | 140 | } |
120 | | - if (c == '\\') { |
121 | | - return false; |
| 141 | + if (val >= Character.MIN_SURROGATE && val <= Character.MAX_SURROGATE) { |
| 142 | + continue; |
122 | 143 | } |
123 | | - if ((int) c < 32 || (int) c >= 128) { |
124 | | - return false; |
| 144 | + return val; |
| 145 | + } |
| 146 | + } |
| 147 | + |
| 148 | + private TestInput buildRandomInput(int length) { |
| 149 | + StringBuilder input = new StringBuilder(length); |
| 150 | + StringBuilder result = new StringBuilder(length); |
| 151 | + boolean forceSupportOptimized = randomBoolean(); |
| 152 | + boolean doesSupportOptimized = true; |
| 153 | + for (int i = 0; i < length; ++i) { |
| 154 | + if (forceSupportOptimized == false && randomBoolean()) { |
| 155 | + switch (randomInt(9)) { |
| 156 | + case 0 -> { |
| 157 | + var escape = randomFrom(ESCAPE_SEQUENCES); |
| 158 | + input.append(escape.input()); |
| 159 | + result.append(escape.result()); |
| 160 | + doesSupportOptimized = doesSupportOptimized && escape.supportsOptimized(); |
| 161 | + } |
| 162 | + case 1 -> { |
| 163 | + int value = randomCodepoint(true); |
| 164 | + input.append(String.format(Locale.ENGLISH, "\\u%04x", value)); |
| 165 | + result.append(Character.toChars(value)); |
| 166 | + doesSupportOptimized = false; |
| 167 | + } |
| 168 | + default -> { |
| 169 | + var value = Character.toChars(randomCodepoint(false)); |
| 170 | + input.append(value); |
| 171 | + result.append(value); |
| 172 | + doesSupportOptimized = false; |
| 173 | + } |
| 174 | + } |
| 175 | + } else { |
| 176 | + var value = randomAlphanumericOfLength(1); |
| 177 | + input.append(value); |
| 178 | + result.append(value); |
125 | 179 | } |
126 | 180 | } |
127 | | - return true; |
| 181 | + return new TestInput(input.toString(), result.toString(), doesSupportOptimized); |
128 | 182 | } |
129 | 183 |
|
130 | 184 | public void testGetValueRandomized() throws IOException { |
131 | | - XContentBuilder jsonBuilder = JsonXContent.contentBuilder().startObject(); |
| 185 | + StringBuilder inputBuilder = new StringBuilder(); |
| 186 | + inputBuilder.append('{'); |
| 187 | + |
132 | 188 | final int numKeys = 128; |
133 | 189 | String[] keys = new String[numKeys]; |
134 | | - String[] values = new String[numKeys]; |
| 190 | + TestInput[] inputs = new TestInput[numKeys]; |
135 | 191 | for (int i = 0; i < numKeys; i++) { |
136 | 192 | String currKey = randomAlphanumericOfLength(6); |
137 | | - String currVal = randomUnicodeOfLengthBetween(0, 512); |
138 | | - jsonBuilder.field(currKey, currVal); |
| 193 | + var currVal = buildRandomInput(randomInt(512)); |
| 194 | + inputBuilder.append('"'); |
| 195 | + inputBuilder.append(currKey); |
| 196 | + inputBuilder.append("\":\""); |
| 197 | + inputBuilder.append(currVal.input()); |
| 198 | + inputBuilder.append('"'); |
| 199 | + if (i < numKeys - 1) { |
| 200 | + inputBuilder.append(','); |
| 201 | + } |
139 | 202 | keys[i] = currKey; |
140 | | - values[i] = currVal; |
| 203 | + inputs[i] = currVal; |
141 | 204 | } |
142 | 205 |
|
143 | | - jsonBuilder.endObject(); |
144 | | - testParseJson(Strings.toString(jsonBuilder), parser -> { |
| 206 | + inputBuilder.append('}'); |
| 207 | + testParseJson(inputBuilder.toString(), parser -> { |
145 | 208 | assertThat(parser.nextToken(), Matchers.equalTo(JsonToken.START_OBJECT)); |
146 | 209 | for (int i = 0; i < numKeys; i++) { |
147 | 210 | assertThat(parser.nextFieldName(), Matchers.equalTo(keys[i])); |
148 | 211 | assertThat(parser.nextValue(), Matchers.equalTo(JsonToken.VALUE_STRING)); |
149 | 212 |
|
150 | | - String currVal = values[i]; |
151 | | - if (validForTextRef(currVal)) { |
| 213 | + String currVal = inputs[i].result(); |
| 214 | + if (inputs[i].supportsOptimized()) { |
152 | 215 | assertTextRef(parser.getValueAsText().bytes(), currVal); |
153 | 216 | } else { |
154 | 217 | assertThat(parser.getValueAsText(), Matchers.nullValue()); |
|
0 commit comments