|
| 1 | +// Copyright 2021 LinkedIn Corporation. All rights reserved. |
| 2 | +// Licensed under the BSD-2 Clause license. |
| 3 | +// See LICENSE in the project root for license information. |
| 4 | + |
| 5 | +package com.linkedin.cdi.configuration; |
| 6 | + |
| 7 | +import com.google.common.collect.Lists; |
| 8 | +import com.google.gson.JsonObject; |
| 9 | +import com.linkedin.cdi.util.CsvUtils; |
| 10 | +import java.util.List; |
| 11 | +import org.apache.commons.lang3.StringUtils; |
| 12 | +import org.apache.gobblin.configuration.State; |
| 13 | + |
| 14 | +import static com.linkedin.cdi.configuration.PropertyCollection.*; |
| 15 | +import static com.linkedin.cdi.configuration.StaticConstants.*; |
| 16 | + |
| 17 | + |
| 18 | +/** |
| 19 | + * CSV Parameters |
| 20 | + */ |
| 21 | +public class CsvProperties extends JsonObjectProperties{ |
| 22 | + final private static String DEFAULT_FIELD_TYPE = "defaultFieldType"; |
| 23 | + final private static String ESCAPE_CHARACTER = "escapeCharacter"; |
| 24 | + final private static String ESCAPE_CHARACTER_DEFAULT = "u005C"; |
| 25 | + final private static String QUOTE_CHARACTER = "quoteCharacter"; |
| 26 | + final private static String QUOTE_CHARACTER_DEFAULT = "\""; |
| 27 | + final private static String FIELD_SEPARATOR = "fieldSeparator"; |
| 28 | + final private static String FIELD_SEPARATOR_DEFAULT = KEY_WORD_COMMA; |
| 29 | + final private static String RECORD_SEPARATOR = "recordSeparator"; |
| 30 | + final private static String RECORD_SEPARATOR_DEFAULT = System.lineSeparator(); |
| 31 | + final private static String LINES_TO_SKIP = "linesToSkip"; |
| 32 | + final private static String COLUMN_HEADER_INDEX = "columnHeaderIndex"; |
| 33 | + final private static String COLUMN_PROJECTION = "columnProjection"; |
| 34 | + final private static String MAX_FAILURES = "maxFailures"; |
| 35 | + final private static String KEEP_NULL_STRING = "keepNullString"; |
| 36 | + |
| 37 | + final private static IntegerProperties linesToSkip = new IntegerProperties(LINES_TO_SKIP); |
| 38 | + final private static IntegerProperties columnHeaderIndex = new IntegerProperties(COLUMN_HEADER_INDEX, -1, Integer.MAX_VALUE, -1); |
| 39 | + final private static LongProperties maxFailures = new LongProperties(MAX_FAILURES); |
| 40 | + final private static BooleanProperties keepNullString = new BooleanProperties(KEEP_NULL_STRING, Boolean.FALSE); |
| 41 | + |
| 42 | + final private static List<String> csvAttributes = Lists.newArrayList( |
| 43 | + DEFAULT_FIELD_TYPE, |
| 44 | + ESCAPE_CHARACTER, QUOTE_CHARACTER, FIELD_SEPARATOR, RECORD_SEPARATOR, |
| 45 | + LINES_TO_SKIP, COLUMN_HEADER_INDEX, COLUMN_PROJECTION, |
| 46 | + MAX_FAILURES, KEEP_NULL_STRING |
| 47 | + ); |
| 48 | + |
| 49 | + @Override |
| 50 | + public boolean isValid(State state) { |
| 51 | + if (super.isValid(state) && !super.isBlank(state)) { |
| 52 | + JsonObject value = GSON.fromJson(state.getProp(getConfig()), JsonObject.class); |
| 53 | + if (!value.entrySet().stream().allMatch(p -> csvAttributes.contains(p.getKey()))) { |
| 54 | + return false; |
| 55 | + } |
| 56 | + |
| 57 | + if (value.has(COLUMN_PROJECTION)) { |
| 58 | + String columnProjections = value.get(COLUMN_PROJECTION).getAsString(); |
| 59 | + if (columnProjections.trim().isEmpty()) { |
| 60 | + return false; |
| 61 | + } |
| 62 | + } |
| 63 | + |
| 64 | + State tmpState = new State(); |
| 65 | + if (value.has(COLUMN_HEADER_INDEX)) { |
| 66 | + tmpState.setProp(COLUMN_HEADER_INDEX, value.get(COLUMN_HEADER_INDEX).getAsString()); |
| 67 | + if (!columnHeaderIndex.isValid(tmpState)) { |
| 68 | + return false; |
| 69 | + } |
| 70 | + } |
| 71 | + if (value.has(LINES_TO_SKIP)) { |
| 72 | + tmpState.setProp(LINES_TO_SKIP, value.get(LINES_TO_SKIP).getAsString()); |
| 73 | + if (!linesToSkip.isValid(tmpState)) { |
| 74 | + return false; |
| 75 | + } |
| 76 | + |
| 77 | + if (linesToSkip.get(tmpState) < columnHeaderIndex.get(tmpState) + 1) { |
| 78 | + return false; |
| 79 | + } |
| 80 | + } |
| 81 | + if (value.has(MAX_FAILURES)) { |
| 82 | + tmpState.setProp(MAX_FAILURES, value.get(MAX_FAILURES).getAsString()); |
| 83 | + if (!maxFailures.isValid(tmpState)) { |
| 84 | + return false; |
| 85 | + } |
| 86 | + } |
| 87 | + |
| 88 | + if (value.has(KEEP_NULL_STRING)) { |
| 89 | + tmpState.setProp(KEEP_NULL_STRING, value.get(KEEP_NULL_STRING).getAsString()); |
| 90 | + if (!keepNullString.isValid(tmpState)) { |
| 91 | + return false; |
| 92 | + } |
| 93 | + } |
| 94 | + } |
| 95 | + return super.isValid(state); |
| 96 | + } |
| 97 | + |
| 98 | + /** |
| 99 | + * Constructor with implicit default value |
| 100 | + * @param config property name |
| 101 | + */ |
| 102 | + CsvProperties(String config) { |
| 103 | + super(config); |
| 104 | + } |
| 105 | + |
| 106 | + public String getDefaultFieldType(State state) { |
| 107 | + JsonObject value = get(state); |
| 108 | + if (value.has(DEFAULT_FIELD_TYPE)) { |
| 109 | + return value.get(DEFAULT_FIELD_TYPE).getAsString(); |
| 110 | + } |
| 111 | + return StringUtils.EMPTY; |
| 112 | + } |
| 113 | + |
| 114 | + public String getEscapeCharacter(State state) { |
| 115 | + JsonObject value = get(state); |
| 116 | + if (value.has(ESCAPE_CHARACTER)) { |
| 117 | + return CsvUtils.unescape(value.get(ESCAPE_CHARACTER).getAsString().trim()); |
| 118 | + } |
| 119 | + return CsvUtils.unescape(ESCAPE_CHARACTER_DEFAULT); |
| 120 | + } |
| 121 | + |
| 122 | + public String getQuoteCharacter(State state) { |
| 123 | + JsonObject value = get(state); |
| 124 | + if (value.has(QUOTE_CHARACTER)) { |
| 125 | + return CsvUtils.unescape(value.get(QUOTE_CHARACTER).getAsString().trim()); |
| 126 | + } |
| 127 | + return QUOTE_CHARACTER_DEFAULT; |
| 128 | + } |
| 129 | + |
| 130 | + public String getFieldSeparator(State state) { |
| 131 | + JsonObject value = get(state); |
| 132 | + if (value.has(FIELD_SEPARATOR)) { |
| 133 | + return CsvUtils.unescape(value.get(FIELD_SEPARATOR).getAsString().trim()); |
| 134 | + } |
| 135 | + return FIELD_SEPARATOR_DEFAULT; |
| 136 | + } |
| 137 | + |
| 138 | + public String getRecordSeparator(State state) { |
| 139 | + JsonObject value = get(state); |
| 140 | + if (value.has(RECORD_SEPARATOR)) { |
| 141 | + return CsvUtils.unescape(value.get(RECORD_SEPARATOR).getAsString().trim()); |
| 142 | + } |
| 143 | + return RECORD_SEPARATOR_DEFAULT; |
| 144 | + } |
| 145 | + |
| 146 | + public Integer getLinesToSkip(State state) { |
| 147 | + JsonObject value = get(state); |
| 148 | + int skip = 0; |
| 149 | + if (value.has(LINES_TO_SKIP) && StringUtils.isNotBlank(value.get(LINES_TO_SKIP).getAsString())) { |
| 150 | + skip = value.get(LINES_TO_SKIP).getAsInt(); |
| 151 | + } |
| 152 | + return Math.max(skip, getColumnHeaderIndex(state) + 1); |
| 153 | + } |
| 154 | + |
| 155 | + public Integer getColumnHeaderIndex(State state) { |
| 156 | + JsonObject value = get(state); |
| 157 | + if (value.has(COLUMN_HEADER_INDEX) && StringUtils.isNotBlank(value.get(COLUMN_HEADER_INDEX).getAsString())) { |
| 158 | + return value.get(COLUMN_HEADER_INDEX).getAsInt(); |
| 159 | + } |
| 160 | + return -1; |
| 161 | + } |
| 162 | + |
| 163 | + public String getColumnProjection(State state) { |
| 164 | + JsonObject value = get(state); |
| 165 | + if (value.has(COLUMN_PROJECTION)) { |
| 166 | + return value.get(COLUMN_PROJECTION).getAsString(); |
| 167 | + } |
| 168 | + return StringUtils.EMPTY; |
| 169 | + } |
| 170 | + |
| 171 | + public Long getMaxFailures(State state) { |
| 172 | + JsonObject value = get(state); |
| 173 | + if (value.has(MAX_FAILURES) && StringUtils.isNotBlank(value.get(MAX_FAILURES).getAsString())) { |
| 174 | + return value.get(MAX_FAILURES).getAsLong(); |
| 175 | + } |
| 176 | + return 0L; |
| 177 | + } |
| 178 | + |
| 179 | + public Boolean getKeepNullString(State state) { |
| 180 | + JsonObject value = get(state); |
| 181 | + if (value.has(KEEP_NULL_STRING) && StringUtils.isNotBlank(value.get(KEEP_NULL_STRING).getAsString())) { |
| 182 | + return value.get(KEEP_NULL_STRING).getAsBoolean(); |
| 183 | + } |
| 184 | + return false; |
| 185 | + } |
| 186 | +} |
0 commit comments