Fix #31: add CsvParser.Feature.TRIM_HEADER_SPACES (#521)

Author: cowtowncoder

csv/src/main/java/com/fasterxml/jackson/dataformat/csv/CsvParser.java

@@ -47,10 +47,29 @@ public enum Feature
          * values (white space outside of double-quotes is never included regardless
          * of trimming).
          *<p>
+         * NOTE: this setting has no effect on header rows: see {@link #TRIM_HEADER_SPACES}
+         * for corresponding setting.
+         *<p>
          * Default value is false, as per <a href="http://tools.ietf.org/html/rfc4180">RFC-4180</a>.
          */
         TRIM_SPACES(false),
 
+        /**
+         * Feature determines whether spaces around separator characters
+         * (commas) in header line entries (header names) are to be automatically
+         * trimmed before being reported or not.
+         * Note that this does NOT force trimming of possible white space from
+         * within double-quoted values, but only those surrounding unquoted
+         * values (white space outside of double-quotes is never included regardless
+         * of trimming).
+         *<p>
+         * Default value is {@code true} for backwards compatibility (before 2.19 trimming
+         * was always performed)
+         *
+         * @since 2.19
+         */
+        TRIM_HEADER_SPACES(true),
+
         /**
          * Feature that determines how stream of records (usually CSV lines, but sometimes
          * multiple lines when linefeeds are included in quoted values) is exposed:
@@ -898,10 +917,13 @@ protected void _readHeaderLine() throws IOException {
         CsvSchema.Builder builder = _schema.rebuild().clearColumns();
         int count = 0;
 
+        final boolean trimHeaderNames = Feature.TRIM_HEADER_SPACES.enabledIn(_formatFeatures);
         while ((name = _reader.nextString()) != null) {
             // one more thing: always trim names, regardless of config settings
-            // TODO!!! [dataformats-text#31]: Allow disabling of trimming
-            name = name.trim();
+            // [dataformats-text#31]: Allow disabling of trimming
+            if (trimHeaderNames) {
+                name = name.trim();
+            }
             // See if "old" schema defined type; if so, use that type...
             CsvSchema.Column prev = _schema.column(name);
             if (prev != null) {

csv/src/test/java/com/fasterxml/jackson/dataformat/csv/deser/ParserWithHeaderTest.java

@@ -49,7 +49,30 @@ public void testSimpleHeader() throws Exception
         }
 
         // But! Can change header name trimming:
-        // [dataformats-text#31]: Allow disabling header name trimming
+        // [dataformats-text#31]: Allow disabling header row trimming
+        try (CsvParser parser = (CsvParser) MAPPER.reader()
+                .without(CsvParser.Feature.TRIM_HEADER_SPACES)
+                .createParser(
+                "name, age,other  \nfoo,2,xyz\n")) {
+            // need to enable first-line-as-schema handling:
+            parser.setSchema(CsvSchema.emptySchema().withHeader());
+            assertToken(JsonToken.START_OBJECT, parser.nextToken());
+            CsvSchema schema = parser.getSchema();
+            assertEquals(3, schema.size());
+    
+            // Verify header names are NOT trimmed when disabled
+            assertEquals("name", schema.columnName(0));
+            assertEquals(" age", schema.columnName(1));
+            assertEquals("other  ", schema.columnName(2));
+
+            assertEquals("name", parser.nextFieldName());
+            assertEquals("foo", parser.nextTextValue());
+            assertEquals(" age", parser.nextFieldName());
+            assertEquals("2", parser.nextTextValue());
+            assertEquals("other  ", parser.nextFieldName());
+            assertEquals("xyz", parser.nextTextValue());
+            assertToken(JsonToken.END_OBJECT, parser.nextToken());
+        }
     }
 
     public void testSimpleQuotes() throws Exception

release-notes/CREDITS-2.x

@@ -285,3 +285,7 @@ Burdyug Pavel (@Pavel38l)
 * Reported #485: (csv) CSVDecoder: No Long and Int out of range exceptions
  (2.18.0)
 
+Robert DiFalco (@rdifalco)
+
+* Reported #31: Header names seem to be trimmed
+ (2.19.0)

release-notes/VERSION-2.x

@@ -16,6 +16,8 @@ Active Maintainers:
 
 2.19.0 (not yet released)
 
+#31: Header names seem to be trimmed
+ (reported by Robert D)
 #502: Add an optional extended parser subclass (`YAMLAnchorReplayingFactory`)
   able to inline anchors
  (contributed by Heiko B)