From 784f2a2c2094ae30b7f07a80cf41b1b548929211 Mon Sep 17 00:00:00 2001
From: Michal Klempa <michal.klempa@gmail.com>
Date: Thu, 18 Feb 2016 09:42:58 +0100
Subject: [PATCH] NIFI-1518 InferAvroSchema note has an option to set CSV
 delimiter

Reviewed (and amended based on acknowledgement in PR review) by Tony Kurc (tkurc@apache.org). This closes #235.
---
 .../nifi/processors/kite/InferAvroSchema.java |  39 +-
 .../processors/kite/TestInferAvroSchema.java  |  43 +++
 .../resources/Shapes_Header_TabDelimited.csv  | 352 ++++++++++++++++++
 3 files changed, 431 insertions(+), 3 deletions(-)
 create mode 100644 nifi-nar-bundles/nifi-kite-bundle/nifi-kite-processors/src/test/resources/Shapes_Header_TabDelimited.csv

diff --git a/nifi-nar-bundles/nifi-kite-bundle/nifi-kite-processors/src/main/java/org/apache/nifi/processors/kite/InferAvroSchema.java b/nifi-nar-bundles/nifi-kite-bundle/nifi-kite-processors/src/main/java/org/apache/nifi/processors/kite/InferAvroSchema.java
index ad8b7e5c88..1923785293 100644
--- a/nifi-nar-bundles/nifi-kite-bundle/nifi-kite-processors/src/main/java/org/apache/nifi/processors/kite/InferAvroSchema.java
+++ b/nifi-nar-bundles/nifi-kite-bundle/nifi-kite-processors/src/main/java/org/apache/nifi/processors/kite/InferAvroSchema.java
@@ -20,6 +20,7 @@ package org.apache.nifi.processors.kite;
 
 import org.apache.avro.Schema;
 import org.apache.commons.lang.StringUtils;
+import org.apache.commons.lang3.StringEscapeUtils;
 import org.apache.nifi.annotation.behavior.InputRequirement;
 import org.apache.nifi.annotation.behavior.ReadsAttribute;
 import org.apache.nifi.annotation.behavior.ReadsAttributes;
@@ -28,6 +29,9 @@ import org.apache.nifi.annotation.behavior.WritesAttribute;
 import org.apache.nifi.annotation.documentation.CapabilityDescription;
 import org.apache.nifi.annotation.documentation.Tags;
 import org.apache.nifi.components.PropertyDescriptor;
+import org.apache.nifi.components.ValidationContext;
+import org.apache.nifi.components.ValidationResult;
+import org.apache.nifi.components.Validator;
 import org.apache.nifi.flowfile.FlowFile;
 import org.apache.nifi.flowfile.attributes.CoreAttributes;
 import org.apache.nifi.processor.ProcessContext;
@@ -42,7 +46,6 @@ import org.kitesdk.data.spi.JsonUtil;
 import org.kitesdk.data.spi.filesystem.CSVProperties;
 import org.kitesdk.data.spi.filesystem.CSVUtil;
 
-
 import java.io.InputStream;
 import java.io.IOException;
 import java.io.InputStreamReader;
@@ -81,7 +84,22 @@ import java.util.concurrent.atomic.AtomicReference;
 public class InferAvroSchema
         extends AbstractKiteProcessor {
 
-    public static final String CSV_DELIMITER = ",";
+    private static final Validator CHAR_VALIDATOR = new Validator() {
+        @Override
+        public ValidationResult validate(String subject, String input, ValidationContext context) {
+            // Allows special, escaped characters as input, which is then unescaped and converted to a single character.
+            // Examples for special characters: \t (or \u0009), \f.
+            input = unescapeString(input);
+
+            return new ValidationResult.Builder()
+                .subject(subject)
+                .input(input)
+                .explanation("Only non-null single characters are supported")
+                .valid(input.length() == 1 && input.charAt(0) != 0)
+                .build();
+        }
+    };
+
     public static final String USE_MIME_TYPE = "use mime.type value";
     public static final String JSON_CONTENT = "json";
     public static final String CSV_CONTENT = "csv";
@@ -154,6 +172,13 @@ public class InferAvroSchema
             .addValidator(StandardValidators.NON_NEGATIVE_INTEGER_VALIDATOR)
             .build();
 
+    public static final PropertyDescriptor DELIMITER = new PropertyDescriptor.Builder()
+            .name("CSV delimiter")
+            .description("Delimiter character for CSV records")
+            .addValidator(CHAR_VALIDATOR)
+            .defaultValue(",")
+            .build();
+
     public static final PropertyDescriptor ESCAPE_STRING = new PropertyDescriptor.Builder()
             .name("CSV Escape String")
             .description("This property only applies to CSV content type. String that represents an escape sequence" +
@@ -234,6 +259,7 @@ public class InferAvroSchema
         properties.add(CSV_HEADER_DEFINITION);
         properties.add(GET_CSV_HEADER_DEFINITION_FROM_INPUT);
         properties.add(HEADER_LINE_SKIP_COUNT);
+        properties.add(DELIMITER);
         properties.add(ESCAPE_STRING);
         properties.add(QUOTE_STRING);
         properties.add(PRETTY_AVRO_OUTPUT);
@@ -366,7 +392,7 @@ public class InferAvroSchema
 
         //Prepares the CSVProperties for kite
         final CSVProperties props = new CSVProperties.Builder()
-                .delimiter(CSV_DELIMITER)
+                .delimiter(context.getProperty(DELIMITER).getValue())
                 .escape(context.getProperty(ESCAPE_STRING).evaluateAttributeExpressions().getValue())
                 .quote(context.getProperty(QUOTE_STRING).evaluateAttributeExpressions().getValue())
                 .header(header.get())
@@ -457,4 +483,11 @@ public class InferAvroSchema
 
         return avroSchema;
     }
+
+    private static String unescapeString(String input) {
+        if (input.length() > 1) {
+            input = StringEscapeUtils.unescapeJava(input);
+        }
+        return input;
+    }
 }
diff --git a/nifi-nar-bundles/nifi-kite-bundle/nifi-kite-processors/src/test/java/org/apache/nifi/processors/kite/TestInferAvroSchema.java b/nifi-nar-bundles/nifi-kite-bundle/nifi-kite-processors/src/test/java/org/apache/nifi/processors/kite/TestInferAvroSchema.java
index 08185f64ad..510eaf97ad 100644
--- a/nifi-nar-bundles/nifi-kite-bundle/nifi-kite-processors/src/test/java/org/apache/nifi/processors/kite/TestInferAvroSchema.java
+++ b/nifi-nar-bundles/nifi-kite-bundle/nifi-kite-processors/src/test/java/org/apache/nifi/processors/kite/TestInferAvroSchema.java
@@ -173,4 +173,47 @@ public class TestInferAvroSchema {
         runner.assertTransferCount(InferAvroSchema.REL_ORIGINAL, 0);
         runner.assertTransferCount(InferAvroSchema.REL_SUCCESS, 0);
     }
+
+    @Test
+    public void inferAvroSchemaFromHeaderDefinitionOfCSVTabDelimitedFile() throws Exception {
+
+        runner.setProperty(InferAvroSchema.DELIMITER, "\\t");
+        runner.assertValid();
+
+        Map<String, String> attributes = new HashMap<>();
+        attributes.put(CoreAttributes.MIME_TYPE.key(), "text/csv");
+        runner.enqueue(new File("src/test/resources/Shapes_Header_TabDelimited.csv").toPath(), attributes);
+
+        runner.run();
+        runner.assertTransferCount(InferAvroSchema.REL_UNSUPPORTED_CONTENT, 0);
+        runner.assertTransferCount(InferAvroSchema.REL_FAILURE, 0);
+        runner.assertTransferCount(InferAvroSchema.REL_ORIGINAL, 1);
+        runner.assertTransferCount(InferAvroSchema.REL_SUCCESS, 1);
+
+        MockFlowFile flowFile = runner.getFlowFilesForRelationship(InferAvroSchema.REL_SUCCESS).get(0);
+        flowFile.assertContentEquals(new File("src/test/resources/Shapes_header.csv.avro").toPath());
+        flowFile.assertAttributeEquals(CoreAttributes.MIME_TYPE.key(), "application/avro-binary");
+    }
+
+    @Test
+    public void inferAvroSchemaFromHeaderDefinitionOfCSVTabDelimitedFileNegativeTest() throws Exception {
+
+        // Inproper InferAvroSchema.DELIMITER > original goes to InferAvroSchema.REL_FAILURE
+        runner.setProperty(InferAvroSchema.DELIMITER, ";");
+        runner.assertValid();
+
+        Map<String, String> attributes = new HashMap<>();
+        attributes.put(CoreAttributes.MIME_TYPE.key(), "text/csv");
+        runner.enqueue(new File("src/test/resources/Shapes_Header_TabDelimited.csv").toPath(), attributes);
+
+        runner.run();
+        runner.assertTransferCount(InferAvroSchema.REL_UNSUPPORTED_CONTENT, 0);
+        runner.assertTransferCount(InferAvroSchema.REL_FAILURE, 1);
+        runner.assertTransferCount(InferAvroSchema.REL_ORIGINAL, 0);
+        runner.assertTransferCount(InferAvroSchema.REL_SUCCESS, 0);
+
+        MockFlowFile flowFile = runner.getFlowFilesForRelationship(InferAvroSchema.REL_FAILURE).get(0);
+        flowFile.assertContentEquals(new File("src/test/resources/Shapes_Header_TabDelimited.csv").toPath());
+        flowFile.assertAttributeEquals(CoreAttributes.MIME_TYPE.key(), "text/csv");
+    }
 }
diff --git a/nifi-nar-bundles/nifi-kite-bundle/nifi-kite-processors/src/test/resources/Shapes_Header_TabDelimited.csv b/nifi-nar-bundles/nifi-kite-bundle/nifi-kite-processors/src/test/resources/Shapes_Header_TabDelimited.csv
new file mode 100644
index 0000000000..7ffa4f9bd2
--- /dev/null
+++ b/nifi-nar-bundles/nifi-kite-bundle/nifi-kite-processors/src/test/resources/Shapes_Header_TabDelimited.csv
@@ -0,0 +1,352 @@
+shape	color	width	height
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+square	red	100	100
+sphere	red	100	100
+triangle	red	100	100
+rectangle	red	100	100
+circle	red	100	100
+sphere	red	100	100
+circle	red	100	100
+circle	red	100	100
+triangle	red	100	100
+cone	red	100	100
+circle	red	100	100
+rectangle	red	100	100
\ No newline at end of file