|
1 | 1 | from pyspark.sql.functions import to_timestamp |
2 | 2 | from pyspark.sql.types import ( |
3 | 3 | BooleanType, |
| 4 | + DoubleType, |
| 5 | + LongType, |
4 | 6 | StringType, |
5 | 7 | StructField, |
6 | 8 | StructType, |
7 | | - TimestampType, |
8 | 9 | ) |
9 | 10 |
|
10 | 11 | logSchema = StructType( |
11 | 12 | [ |
12 | | - StructField("time", TimestampType(), True), |
| 13 | + StructField("time", DoubleType(), True), |
13 | 14 | StructField("index", StringType(), True), |
14 | 15 | StructField("host", StringType(), True), |
15 | 16 | StructField("source", StringType(), True), |
|
38 | 39 | StructField("accept-encoding", StringType(), True), |
39 | 40 | StructField("Authorization", StringType(), True), |
40 | 41 | StructField("Host", StringType(), True), |
| 42 | + StructField( |
| 43 | + "NHSD-Client-RP-Details", StringType(), True |
| 44 | + ), |
41 | 45 | StructField( |
42 | 46 | "NHSD-Connection-Metadata", StringType(), True |
43 | 47 | ), |
44 | 48 | StructField("NHSD-Correlation-Id", StringType(), True), |
45 | 49 | StructField("User-Agent", StringType(), True), |
46 | 50 | StructField("X-Forwarded-For", StringType(), True), |
47 | 51 | StructField("X-Request-Id", StringType(), True), |
| 52 | + StructField( |
| 53 | + "NHSD-End-User-Organisation-ODS", StringType(), True |
| 54 | + ), |
| 55 | + StructField("NHSD-Request-ID", StringType(), True), |
| 56 | + StructField("X-Forwarded-Port", StringType(), True), |
| 57 | + StructField("X-Forwarded-Proto", StringType(), True), |
48 | 58 | ] |
49 | 59 | ), |
50 | 60 | True, |
51 | 61 | ), |
52 | 62 | StructField("log_reference", StringType(), True), |
53 | 63 | StructField("xray_trace_id", StringType(), True), |
| 64 | + StructField( |
| 65 | + "config", |
| 66 | + StructType( |
| 67 | + [ |
| 68 | + StructField("AWS_REGION", StringType(), True), |
| 69 | + StructField("PREFIX", StringType(), True), |
| 70 | + StructField("ENVIRONMENT", StringType(), True), |
| 71 | + StructField("SPLUNK_INDEX", StringType(), True), |
| 72 | + StructField("SOURCE", StringType(), True), |
| 73 | + StructField("AUTH_STORE", StringType(), True), |
| 74 | + StructField("TABLE_NAME", StringType(), True), |
| 75 | + ] |
| 76 | + ), |
| 77 | + True, |
| 78 | + ), |
| 79 | + StructField( |
| 80 | + "metadata", |
| 81 | + StructType( |
| 82 | + [ |
| 83 | + StructField("ods_code", StringType(), True), |
| 84 | + StructField("ods_code_extension", StringType(), True), |
| 85 | + StructField("nrl_app_id", StringType(), True), |
| 86 | + StructField("is_test_event", BooleanType(), True), |
| 87 | + StructField( |
| 88 | + "client_rp_details", |
| 89 | + StructType( |
| 90 | + [ |
| 91 | + StructField( |
| 92 | + "developer_app_name", StringType(), True |
| 93 | + ), |
| 94 | + StructField( |
| 95 | + "developer_app_id", StringType(), True |
| 96 | + ), |
| 97 | + ] |
| 98 | + ), |
| 99 | + True, |
| 100 | + ), |
| 101 | + ] |
| 102 | + ), |
| 103 | + True, |
| 104 | + ), |
| 105 | + StructField( |
| 106 | + "params", |
| 107 | + StructType( |
| 108 | + [ |
| 109 | + StructField("subject:identifier", StringType(), True), |
| 110 | + ] |
| 111 | + ), |
| 112 | + True, |
| 113 | + ), |
| 114 | + StructField("model", StringType(), True), |
| 115 | + StructField( |
| 116 | + "parsed_params", |
| 117 | + StructType( |
| 118 | + [ |
| 119 | + StructField("subject_identifier", StringType(), True), |
| 120 | + ] |
| 121 | + ), |
| 122 | + True, |
| 123 | + ), |
| 124 | + StructField("table_name", StringType(), True), |
| 125 | + StructField("item_type", StringType(), True), |
| 126 | + StructField("original_kwargs_keys", StringType(), True), |
| 127 | + StructField("filtered_kwargs_keys", StringType(), True), |
| 128 | + StructField("nhs_number", StringType(), True), |
| 129 | + StructField( |
| 130 | + "query", |
| 131 | + StructType( |
| 132 | + [ |
| 133 | + StructField("IndexName", StringType(), True), |
| 134 | + StructField( |
| 135 | + "KeyConditionExpression", StringType(), True |
| 136 | + ), |
| 137 | + StructField( |
| 138 | + "ExpressionAttributeValues", |
| 139 | + StructType( |
| 140 | + [ |
| 141 | + StructField( |
| 142 | + ":patient_key", StringType(), True |
| 143 | + ), |
| 144 | + StructField( |
| 145 | + ":patient_sort", StringType(), True |
| 146 | + ), |
| 147 | + ] |
| 148 | + ), |
| 149 | + True, |
| 150 | + ), |
| 151 | + StructField("Select", StringType(), True), |
| 152 | + StructField( |
| 153 | + "ReturnConsumedCapacity", StringType(), True |
| 154 | + ), |
| 155 | + ] |
| 156 | + ), |
| 157 | + True, |
| 158 | + ), |
| 159 | + StructField("count", LongType(), True), |
| 160 | + StructField( |
| 161 | + "result", |
| 162 | + StructType( |
| 163 | + [ |
| 164 | + StructField("Count", LongType(), True), |
| 165 | + StructField("ScannedCount", LongType(), True), |
| 166 | + StructField( |
| 167 | + "ConsumedCapacity", |
| 168 | + StructType( |
| 169 | + [ |
| 170 | + StructField( |
| 171 | + "TableName", StringType(), True |
| 172 | + ), |
| 173 | + StructField( |
| 174 | + "CapacityUnits", DoubleType(), True |
| 175 | + ), |
| 176 | + StructField( |
| 177 | + "Table", |
| 178 | + StructType( |
| 179 | + [ |
| 180 | + StructField( |
| 181 | + "CapacityUnits", |
| 182 | + DoubleType(), |
| 183 | + True, |
| 184 | + ), |
| 185 | + ] |
| 186 | + ), |
| 187 | + True, |
| 188 | + ), |
| 189 | + StructField( |
| 190 | + "GlobalSecondaryIndexes", |
| 191 | + StructType( |
| 192 | + [ |
| 193 | + StructField( |
| 194 | + "patient_gsi", |
| 195 | + StructType( |
| 196 | + [ |
| 197 | + StructField( |
| 198 | + "CapacityUnits", |
| 199 | + DoubleType(), |
| 200 | + True, |
| 201 | + ), |
| 202 | + ] |
| 203 | + ), |
| 204 | + True, |
| 205 | + ), |
| 206 | + ] |
| 207 | + ), |
| 208 | + True, |
| 209 | + ), |
| 210 | + ] |
| 211 | + ), |
| 212 | + True, |
| 213 | + ), |
| 214 | + StructField( |
| 215 | + "ResponseMetadata", |
| 216 | + StructType( |
| 217 | + [ |
| 218 | + StructField( |
| 219 | + "RequestId", StringType(), True |
| 220 | + ), |
| 221 | + StructField( |
| 222 | + "HTTPStatusCode", LongType(), True |
| 223 | + ), |
| 224 | + StructField( |
| 225 | + "HTTPHeaders", |
| 226 | + StructType( |
| 227 | + [ |
| 228 | + StructField( |
| 229 | + "server", StringType(), True |
| 230 | + ), |
| 231 | + StructField( |
| 232 | + "date", StringType(), True |
| 233 | + ), |
| 234 | + StructField( |
| 235 | + "content-type", |
| 236 | + StringType(), |
| 237 | + True, |
| 238 | + ), |
| 239 | + StructField( |
| 240 | + "content-length", |
| 241 | + StringType(), |
| 242 | + True, |
| 243 | + ), |
| 244 | + StructField( |
| 245 | + "connection", |
| 246 | + StringType(), |
| 247 | + True, |
| 248 | + ), |
| 249 | + StructField( |
| 250 | + "x-amzn-requestid", |
| 251 | + StringType(), |
| 252 | + True, |
| 253 | + ), |
| 254 | + StructField( |
| 255 | + "x-amz-crc32", |
| 256 | + StringType(), |
| 257 | + True, |
| 258 | + ), |
| 259 | + ] |
| 260 | + ), |
| 261 | + True, |
| 262 | + ), |
| 263 | + StructField( |
| 264 | + "RetryAttempts", LongType(), True |
| 265 | + ), |
| 266 | + ] |
| 267 | + ), |
| 268 | + True, |
| 269 | + ), |
| 270 | + ] |
| 271 | + ), |
| 272 | + True, |
| 273 | + ), |
| 274 | + StructField("status_code", StringType(), True), |
| 275 | + StructField( |
| 276 | + "response", |
| 277 | + StructType( |
| 278 | + [ |
| 279 | + StructField("statusCode", StringType(), True), |
| 280 | + StructField("body", StringType(), True), |
| 281 | + StructField("isBase64Encoded", BooleanType(), True), |
| 282 | + ] |
| 283 | + ), |
| 284 | + True, |
| 285 | + ), |
54 | 286 | ] |
55 | 287 | ), |
56 | 288 | True, |
|
0 commit comments