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

CSVParser: Identify duplicates in null, empty and blank header names (WIP)

Author: garydgregory

src/main/java/org/apache/commons/csv/CSVFormat.java

@@ -289,7 +289,7 @@ public Builder setAllowDuplicateHeaderNames(final boolean allowDuplicateHeaderNa
         }
 
         /**
-         * 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.
          *
          * @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 Builder setIgnoreEmptyLines(final boolean ignoreEmptyLines) {
         }
 
         /**
-         * 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.
          * @return This instance.
@@ -1599,7 +1599,7 @@ public boolean getIgnoreEmptyLines() {
     }
 
     /**
-     * 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.
      * @since 1.3

src/main/java/org/apache/commons/csv/CSVParser.java

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