Apache
/
commons-csv
mirror of https://github.com/apache/commons-csv.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #279 from aherbert/csvparser-duplicate-missing-headers 

CSVParser: Identify duplicates in null, empty and blank header names (WIP)
This commit is contained in:
Gary Gregory 2022-10-25 21:23:51 -04:00 committed by GitHub
parent 8ad07df0d4 69aa686187
commit 5a7b030537
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 175 additions and 119 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
src/main/java/org/apache/commons/csv/CSVFormat.java
View File

@ -289,7 +289,7 @@ public final class CSVFormat implements Serializable {
} }
/** /**
* Sets the missing column names parser behavior, {@code true} to allow missing column names in the header line, {@code false} to cause an * Sets the parser missing column names behavior, {@code true} to allow missing column names in the header line, {@code false} to cause an
* {@link IllegalArgumentException} to be thrown. * {@link IllegalArgumentException} to be thrown.
* *
* @param allowMissingColumnNames the missing column names behavior, {@code true} to allow missing column names in the header line, {@code false} to * @param allowMissingColumnNames the missing column names behavior, {@code true} to allow missing column names in the header line, {@code false} to
@ -564,7 +564,7 @@ public final class CSVFormat implements Serializable {
} }
/** /**
* Sets the case mapping behavior, {@code true} to access name/values, {@code false} to leave the mapping as is. * Sets the parser case mapping behavior, {@code true} to access name/values, {@code false} to leave the mapping as is.
* *
* @param ignoreHeaderCase the case mapping behavior, {@code true} to access name/values, {@code false} to leave the mapping as is. * @param ignoreHeaderCase the case mapping behavior, {@code true} to access name/values, {@code false} to leave the mapping as is.
* @return This instance. * @return This instance.
@ -1599,7 +1599,7 @@ public final class CSVFormat implements Serializable {
} }
/** /**
* Gets whether header names will be accessed ignoring case. * Gets whether header names will be accessed ignoring case when parsing input.
* *
* @return {@code true} if header names cases are ignored, {@code false} if they are case sensitive. * @return {@code true} if header names cases are ignored, {@code false} if they are case sensitive.
* @since 1.3 * @since 1.3

5
src/main/java/org/apache/commons/csv/CSVParser.java
View File

@ -499,6 +499,8 @@ public final class CSVParser implements Iterable<CSVRecord>, Closeable {
// build the name to index mappings // build the name to index mappings
if (headerRecord != null) { if (headerRecord != null) {
// Track an occurrence of a null, empty or blank header.
boolean observedMissing = false;
for (int i = 0; i < headerRecord.length; i++) { for (int i = 0; i < headerRecord.length; i++) {
final String header = headerRecord[i]; final String header = headerRecord[i];
final boolean blankHeader = CSVFormat.isBlank(header); final boolean blankHeader = CSVFormat.isBlank(header);
@ -507,7 +509,7 @@ public final class CSVParser implements Iterable<CSVRecord>, Closeable {
"A header name is missing in " + Arrays.toString(headerRecord)); "A header name is missing in " + Arrays.toString(headerRecord));
} }
final boolean containsHeader = header != null && hdrMap.containsKey(header); final boolean containsHeader = blankHeader ? observedMissing : hdrMap.containsKey(header);
final DuplicateHeaderMode headerMode = this.format.getDuplicateHeaderMode(); final DuplicateHeaderMode headerMode = this.format.getDuplicateHeaderMode();
final boolean duplicatesAllowed = headerMode == DuplicateHeaderMode.ALLOW_ALL; final boolean duplicatesAllowed = headerMode == DuplicateHeaderMode.ALLOW_ALL;
final boolean emptyDuplicatesAllowed = headerMode == DuplicateHeaderMode.ALLOW_EMPTY; final boolean emptyDuplicatesAllowed = headerMode == DuplicateHeaderMode.ALLOW_EMPTY;
@ -518,6 +520,7 @@ public final class CSVParser implements Iterable<CSVRecord>, Closeable {
"The header contains a duplicate name: \"%s\" in %s. If this is valid then use CSVFormat.Builder.setDuplicateHeaderMode().", "The header contains a duplicate name: \"%s\" in %s. If this is valid then use CSVFormat.Builder.setDuplicateHeaderMode().",
header, Arrays.toString(headerRecord))); header, Arrays.toString(headerRecord)));
} }
observedMissing |= blankHeader;
if (header != null) { if (header != null) {
hdrMap.put(header, Integer.valueOf(i)); hdrMap.put(header, Integer.valueOf(i));
if (headerNames == null) { if (headerNames == null) {

283
src/test/java/org/apache/commons/csv/CSVDuplicateHeaderTest.java
View File

@ -46,144 +46,181 @@ public class CSVDuplicateHeaderTest {
*/ */
static Stream<Arguments> duplicateHeaderData() { static Stream<Arguments> duplicateHeaderData() {
return Stream.of( return Stream.of(
// TODO: Fix CSVParser which does not sanitise 'empty' names or
// equate null names with empty as duplications
// Any combination with a valid header // Any combination with a valid header
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {"A", "B"}, true), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {"A", "B"}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {"A", "B"}, true), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {"A", "B"}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {"A", "B"}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {"A", "B"}, true),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {"A", "B"}, true), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {"A", "B"}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {"A", "B"}, true), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {"A", "B"}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {"A", "B"}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {"A", "B"}, true),
// Any combination with a valid header including empty // Any combination with a valid header including empty
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {"A", ""}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {"A", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {"A", ""}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {"A", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {"A", ""}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {"A", ""}, false),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {"A", ""}, true), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {"A", ""}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {"A", ""}, true), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {"A", ""}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {"A", ""}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {"A", ""}, true),
// Any combination with a valid header including blank (1 space) // Any combination with a valid header including blank (1 space)
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {"A", " "}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {"A", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {"A", " "}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {"A", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {"A", " "}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {"A", " "}, false),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {"A", " "}, true), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {"A", " "}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {"A", " "}, true), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {"A", " "}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {"A", " "}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {"A", " "}, true),
// Any combination with a valid header including null // Any combination with a valid header including null
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {"A", null}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {"A", null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {"A", null}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {"A", null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {"A", null}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {"A", null}, false),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {"A", null}, true), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {"A", null}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {"A", null}, true), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {"A", null}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {"A", null}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {"A", null}, true),
// Duplicate non-empty names // Duplicate non-empty names
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {"A", "A"}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {"A", "A"}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {"A", "A"}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {"A", "A"}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {"A", "A"}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {"A", "A"}, true),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {"A", "A"}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {"A", "A"}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {"A", "A"}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {"A", "A"}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {"A", "A"}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {"A", "A"}, true),
// Duplicate empty names // Duplicate empty names
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {"", ""}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {"", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {"", ""}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {"", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {"", ""}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {"", ""}, false),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {"", ""}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {"", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {"", ""}, true), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {"", ""}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {"", ""}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {"", ""}, true),
// Duplicate blank names (1 space) // Duplicate blank names (1 space)
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {" ", " "}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {" ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {" ", " "}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {" ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {" ", " "}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {" ", " "}, false),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {" ", " "}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {" ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {" ", " "}, true), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {" ", " "}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {" ", " "}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {" ", " "}, true),
// Duplicate blank names (3 spaces) // Duplicate blank names (3 spaces)
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {" ", " "}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {" ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {" ", " "}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {" ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {" ", " "}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {" ", " "}, false),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {" ", " "}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {" ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {" ", " "}, true), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {" ", " "}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {" ", " "}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {" ", " "}, true),
// Duplicate null names // Duplicate null names
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {null, null}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {null, null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {null, null}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {null, null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {null, null}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {null, null}, false),
// Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {null, null}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {null, null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {null, null}, true), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {null, null}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {null, null}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {null, null}, true),
// Duplicate blank names (1+3 spaces) // Duplicate blank names (1+3 spaces)
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {" ", " "}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {" ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {" ", " "}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {" ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {" ", " "}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {" ", " "}, false),
// Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {" ", " "}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {" ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {" ", " "}, true), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {" ", " "}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {" ", " "}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {" ", " "}, true),
// Duplicate blank names and null names // Duplicate blank names and null names
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {" ", null}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {" ", null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {" ", null}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {" ", null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {" ", null}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {" ", null}, false),
// Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {" ", null}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {" ", null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {" ", null}, true), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {" ", null}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {" ", null}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {" ", null}, true),
// Duplicate non-empty and empty names // Duplicate non-empty and empty names
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {"A", "A", "", ""}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {"A", "A", "", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {"A", "A", "", ""}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {"A", "A", "", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {"A", "A", "", ""}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {"A", "A", "", ""}, false),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {"A", "A", "", ""}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {"A", "A", "", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {"A", "A", "", ""}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {"A", "A", "", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {"A", "A", "", ""}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {"A", "A", "", ""}, true),
// Non-duplicate non-empty and duplicate empty names
Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {"A", "B", "", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {"A", "B", "", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {"A", "B", "", ""}, false),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {"A", "B", "", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {"A", "B", "", ""}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {"A", "B", "", ""}, true),
// Duplicate non-empty and blank names // Duplicate non-empty and blank names
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {"A", "A", " ", " "}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {"A", "A", " ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {"A", "A", " ", " "}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {"A", "A", " ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {"A", "A", " ", " "}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {"A", "A", " ", " "}, false),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {"A", "A", " ", " "}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {"A", "A", " ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {"A", "A", " ", " "}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {"A", "A", " ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {"A", "A", " ", " "}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {"A", "A", " ", " "}, true),
// Duplicate non-empty and null names // Duplicate non-empty and null names
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {"A", "A", null, null}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {"A", "A", null, null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {"A", "A", null, null}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {"A", "A", null, null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {"A", "A", null, null}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {"A", "A", null, null}, false),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {"A", "A", null, null}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {"A", "A", null, null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {"A", "A", null, null}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {"A", "A", null, null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {"A", "A", null, null}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {"A", "A", null, null}, true),
// Duplicate blank names // Duplicate blank names
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {"A", "", ""}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {"A", "", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {"A", "", ""}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {"A", "", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {"A", "", ""}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {"A", "", ""}, false),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {"A", "", ""}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {"A", "", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {"A", "", ""}, true), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {"A", "", ""}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {"A", "", ""}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {"A", "", ""}, true),
// Duplicate null names // Duplicate null names
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {"A", null, null}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {"A", null, null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {"A", null, null}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {"A", null, null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {"A", null, null}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {"A", null, null}, false),
// Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {"A", null, null}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {"A", null, null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {"A", null, null}, true), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {"A", null, null}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {"A", null, null}, true), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {"A", null, null}, true),
// Duplicate blank names (1+3 spaces) // Duplicate blank names (1+3 spaces)
Arguments.of(DuplicateHeaderMode.DISALLOW, false, new String[] {"A", " ", " "}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, false, false, new String[] {"A", " ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, new String[] {"A", " ", " "}, false), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, false, new String[] {"A", " ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, new String[] {"A", " ", " "}, false), Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, false, new String[] {"A", " ", " "}, false),
// Arguments.of(DuplicateHeaderMode.DISALLOW, true, new String[] {"A", " ", " "}, false), Arguments.of(DuplicateHeaderMode.DISALLOW, true, false, new String[] {"A", " ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, new String[] {"A", " ", " "}, true), Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, false, new String[] {"A", " ", " "}, true),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, new String[] {"A", " ", " "}, true) Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, false, new String[] {"A", " ", " "}, true),
// Duplicate names (case insensitive)
Arguments.of(DuplicateHeaderMode.DISALLOW, false, true , new String[] {"A", "a"}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, true , new String[] {"A", "a"}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, true , new String[] {"A", "a"}, true),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, true , new String[] {"A", "a"}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, true , new String[] {"A", "a"}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, true , new String[] {"A", "a"}, true),
// Duplicate non-empty (case insensitive) and empty names
Arguments.of(DuplicateHeaderMode.DISALLOW, false, true, new String[] {"A", "a", "", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, true, new String[] {"A", "a", "", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, true, new String[] {"A", "a", "", ""}, false),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, true, new String[] {"A", "a", "", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, true, new String[] {"A", "a", "", ""}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, true, new String[] {"A", "a", "", ""}, true),
// Duplicate non-empty (case insensitive) and blank names
Arguments.of(DuplicateHeaderMode.DISALLOW, false, true, new String[] {"A", "a", " ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, true, new String[] {"A", "a", " ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, true, new String[] {"A", "a", " ", " "}, false),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, true, new String[] {"A", "a", " ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, true, new String[] {"A", "a", " ", " "}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, true, new String[] {"A", "a", " ", " "}, true),
// Duplicate non-empty (case insensitive) and null names
Arguments.of(DuplicateHeaderMode.DISALLOW, false, true, new String[] {"A", "a", null, null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, false, true, new String[] {"A", "a", null, null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, false, true, new String[] {"A", "a", null, null}, false),
Arguments.of(DuplicateHeaderMode.DISALLOW, true, true, new String[] {"A", "a", null, null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_EMPTY, true, true, new String[] {"A", "a", null, null}, false),
Arguments.of(DuplicateHeaderMode.ALLOW_ALL, true, true, new String[] {"A", "a", null, null}, true)
); );
} }
@ -191,23 +228,33 @@ public class CSVDuplicateHeaderTest {
* Return test cases for duplicate header data for use in CSVFormat. * Return test cases for duplicate header data for use in CSVFormat.
* <p> * <p>
* This filters the parsing test data to all cases where the allow missing column * This filters the parsing test data to all cases where the allow missing column
* names flag is true. The allow missing column names is exclusively for parsing. * names flag is true and ignore header case is false: these flags are exclusively for parsing.
* CSVFormat validation applies to both parsing and writing and thus validation * CSVFormat validation applies to both parsing and writing and thus validation
* is less strict and behaves as if the missing column names constraint is absent. * is less strict and behaves as if the allow missing column names constraint and
* The filtered data is then returned with the missing column names flag set to both * the ignore header case behaviour are absent.
* true and false for each test case. * The filtered data is then returned with the parser flags set to both true and false
* for each test case.
* </p> * </p>
* *
* @return the stream of arguments * @return the stream of arguments
*/ */
static Stream<Arguments> duplicateHeaderAllowsMissingColumnsNamesData() { static Stream<Arguments> duplicateHeaderAllowsMissingColumnsNamesData() {
return duplicateHeaderData() return duplicateHeaderData()
.filter(arg -> Boolean.TRUE.equals(arg.get()[1])) .filter(arg -> Boolean.TRUE.equals(arg.get()[1]) && Boolean.FALSE.equals(arg.get()[2]))
.flatMap(arg -> { .flatMap(arg -> {
// Return test case with flag as both true and false // Return test case with flags as all true/false combinations
final Object[] data = arg.get().clone(); final Object[][] data = new Object[4][];
data[1] = Boolean.FALSE; final Boolean[] flags = {Boolean.TRUE, Boolean.FALSE};
return Stream.of(arg, Arguments.of(data)); int i = 0;
for (final Boolean a : flags) {
for (final Boolean b : flags) {
data[i] = arg.get().clone();
data[i][1] = a;
data[i][2] = b;
i++;
}
}
return Arrays.stream(data).map(Arguments::of);
}); });
} }
@ -216,6 +263,7 @@ public class CSVDuplicateHeaderTest {
* *
* @param duplicateHeaderMode the duplicate header mode * @param duplicateHeaderMode the duplicate header mode
* @param allowMissingColumnNames the allow missing column names flag (only used for parsing) * @param allowMissingColumnNames the allow missing column names flag (only used for parsing)
* @param ignoreHeaderCase the ignore header case flag (only used for parsing)
* @param headers the headers * @param headers the headers
* @param valid true if the settings are expected to be valid, otherwise expect a IllegalArgumentException * @param valid true if the settings are expected to be valid, otherwise expect a IllegalArgumentException
*/ */
@ -223,12 +271,14 @@ public class CSVDuplicateHeaderTest {
@MethodSource(value = {"duplicateHeaderAllowsMissingColumnsNamesData"}) @MethodSource(value = {"duplicateHeaderAllowsMissingColumnsNamesData"})
public void testCSVFormat(final DuplicateHeaderMode duplicateHeaderMode, public void testCSVFormat(final DuplicateHeaderMode duplicateHeaderMode,
final boolean allowMissingColumnNames, final boolean allowMissingColumnNames,
final boolean ignoreHeaderCase,
final String[] headers, final String[] headers,
final boolean valid) { final boolean valid) {
final CSVFormat.Builder builder = final CSVFormat.Builder builder =
CSVFormat.DEFAULT.builder() CSVFormat.DEFAULT.builder()
.setDuplicateHeaderMode(duplicateHeaderMode) .setDuplicateHeaderMode(duplicateHeaderMode)
.setAllowMissingColumnNames(allowMissingColumnNames) .setAllowMissingColumnNames(allowMissingColumnNames)
.setIgnoreHeaderCase(ignoreHeaderCase)
.setHeader(headers); .setHeader(headers);
if (valid) { if (valid) {
final CSVFormat format = builder.build(); final CSVFormat format = builder.build();
@ -245,6 +295,7 @@ public class CSVDuplicateHeaderTest {
* *
* @param duplicateHeaderMode the duplicate header mode * @param duplicateHeaderMode the duplicate header mode
* @param allowMissingColumnNames the allow missing column names flag (only used for parsing) * @param allowMissingColumnNames the allow missing column names flag (only used for parsing)
* @param ignoreHeaderCase the ignore header case flag (only used for parsing)
* @param headers the headers (joined with the CSVFormat delimiter to create a string input) * @param headers the headers (joined with the CSVFormat delimiter to create a string input)
* @param valid true if the settings are expected to be valid, otherwise expect a IllegalArgumentException * @param valid true if the settings are expected to be valid, otherwise expect a IllegalArgumentException
* @throws IOException Signals that an I/O exception has occurred. * @throws IOException Signals that an I/O exception has occurred.
@ -253,17 +304,19 @@ public class CSVDuplicateHeaderTest {
@MethodSource(value = {"duplicateHeaderData"}) @MethodSource(value = {"duplicateHeaderData"})
public void testCSVParser(final DuplicateHeaderMode duplicateHeaderMode, public void testCSVParser(final DuplicateHeaderMode duplicateHeaderMode,
final boolean allowMissingColumnNames, final boolean allowMissingColumnNames,
final boolean ignoreHeaderCase,
final String[] headers, final String[] headers,
final boolean valid) throws IOException { final boolean valid) throws IOException {
final CSVFormat format = final CSVFormat format =
CSVFormat.DEFAULT.builder() CSVFormat.DEFAULT.builder()
.setDuplicateHeaderMode(duplicateHeaderMode) .setDuplicateHeaderMode(duplicateHeaderMode)
.setAllowMissingColumnNames(allowMissingColumnNames) .setAllowMissingColumnNames(allowMissingColumnNames)
.setIgnoreHeaderCase(ignoreHeaderCase)
.setNullString("NULL") .setNullString("NULL")
.setHeader() .setHeader()
.build(); .build();
final String input = Arrays.stream(headers) final String input = Arrays.stream(headers)
.map(s -> s == null ? "NULL" : s) .map(s -> s == null ? format.getNullString() : s)
.collect(Collectors.joining(format.getDelimiterString())); .collect(Collectors.joining(format.getDelimiterString()));
if (valid) { if (valid) {
try(CSVParser parser = CSVParser.parse(input, format)) { try(CSVParser parser = CSVParser.parse(input, format)) {