embulk-filter-timestamp_format 0.1.8 → 0.1.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 81e7bf1b428ed0d48164dbaaec74a2080be029be
-  data.tar.gz: 7c8e45fcc5cf72d8ae25eb973382e74bf3b16420
+  metadata.gz: 88c7a2edd52859e9304b268ecea6c766158857cd
+  data.tar.gz: 263540b3154fdd4c0a04892db3708b09dac23cc0
 SHA512:
-  metadata.gz: fbae3c05eb9a2808adab605035e5ad432f9fd61df48b5b077565fcc6372a7e3413ec4ea499ddd98b9ebfbbee3f1073f7fa4a73b4743b238c9532905ee60afcfb
-  data.tar.gz: b3760fc88ea943d92edc776ef400563affe983718e7d66854e7093ba7a25983d9395ac3a46642ed430c5ccde4dc3845437e10145d7e4d2385e6b722fb021c311
+  metadata.gz: 08c540df5bab3aeb3f03c3507042f623fead8488126f714a17dfa3ff09a8c536d03c94ed1773cb6fa37b6255d8cf55f86ed454cb4b6cb229c8bc68e5bcee8c27
+  data.tar.gz: 945fb0c472ee3d9cc42bcfb9dda0a4444919773c42a9b4cca13a274e61505475ce2733e8bfc430449103488cddae5617300d9d0f2195cf779fafc24a96716819

data/CHANGELOG.md

@@ -1,3 +1,9 @@
+# 0.1.9 (2016-05-10)
+
+Enhancements:
+
+* Support nano second resolution for Java formatter
+
 # 0.1.8 (2016-05-10)
 
 Enhancements:

data/README.md

@@ -111,10 +111,11 @@ yyyy-MM-dd HH:mm:ss.nnnnnnnnn z
 
 This plugin finds places of nano second from texts with regular expression `\.(\d+)`.
 
-For formatter, you can use jruby formatter as
+For formatter, you can use `nnnnnnnnn` for nano and `nnnnnn` for micro as
 
 ```
-%Y-%m-%d %H:%M:%S.%N %z
+yyyy-MM-dd HH:mm:ss.nnnnnnnnn z
+yyyy-MM-dd HH:mm:ss.nnnnnn z
 ```
 
 ## ToDo

data/bench/config_java.yml

@@ -7,7 +7,8 @@ in:
       - {name: timestamp, type: string}
 filters:
   - type: timestamp_format
+    stop_on_invalid_record: true
     columns:
-      - {name: timestamp, from_format: ["yyyy-MM-dd hh:mm:ss.SSSSSSSSS"], to_format: "yyyy-MM-dd"}
+      - {name: timestamp, from_format: ["yyyy-MM-dd HH:mm:ss.SSSSSSSSS"], to_format: "yyyy-MM-dd"}
 out:
   type: "null"

data/bench/config_jruby.yml

@@ -7,6 +7,7 @@ in:
       - {name: timestamp, type: string}
 filters:
   - type: timestamp_format
+    stop_on_invalid_record: true
     columns:
       - {name: timestamp, from_format: ["%Y-%m-%d %H:%M:%S.%N"], to_format: "%Y-%m-%d"}
 

data/bench/config_jruby_formatter.yml

@@ -7,7 +7,8 @@ in:
       - {name: timestamp, type: string}
 filters:
   - type: timestamp_format
+    stop_on_invalid_record: true
     columns:
-      - {name: timestamp, from_format: ["yyyy-MM-dd hh:mm:ss.SSSSSSSSS"], to_format: "%Y-%m-%d"}
+      - {name: timestamp, from_format: ["yyyy-MM-dd HH:mm:ss.SSSSSSSSS"], to_format: "%Y-%m-%d"}
 out:
   type: "null"

data/bench/config_jruby_parser.yml

@@ -7,6 +7,7 @@ in:
       - {name: timestamp, type: string}
 filters:
   - type: timestamp_format
+    stop_on_invalid_record: true
     columns:
       - {name: timestamp, from_format: ["%Y-%m-%d %H:%M:%S.%N"], to_format: "yyyy-MM-dd"}
 out:

data/bench/config_nano.yml

@@ -7,7 +7,8 @@ in:
       - {name: timestamp, type: string}
 filters:
   - type: timestamp_format
+    stop_on_invalid_record: true
     columns:
-      - {name: timestamp, from_format: ["yyyy-MM-dd hh:mm:ss.nnnnnnnnn"], to_format: "%Y-%m-%d"}
+      - {name: timestamp, from_format: ["yyyy-MM-dd HH:mm:ss.nnnnnnnnn"], to_format: "yyyy-MM-dd HH:mm:ss.nnnnnn"}
 out:
   type: "null"

data/build.gradle

@@ -13,7 +13,7 @@ configurations {
     provided
 }
 
-version = "0.1.8"
+version = "0.1.9"
 sourceCompatibility = 1.7
 targetCompatibility = 1.7
 

data/example/string_nano.yml

@@ -13,7 +13,7 @@ filters:
     default_from_timezone: "Asia/Taipei"
     default_from_timestamp_format: ["yyyy-MM-dd", "yyyy-MM-dd z", "yyyy-MM-dd HH:mm:ss.nnnnnnnnn z", "yyyy-MM-dd HH:mm:ss z", "yyyy-MM-dd HH:mm:ss"]
     default_to_timezone: "Asia/Taipei"
-    default_to_timestamp_format: "%Y-%m-%d %H:%M:%S.%N"
+    default_to_timestamp_format: "yyyy-MM-dd HH:mm:ss.nnnnnnnn"
     columns:
       - {name: string1}
       - {name: string2, type: timestamp}

data/src/main/java/org/embulk/filter/timestamp_format/TimestampFormatter.java

@@ -16,6 +16,9 @@ import org.jruby.embed.ScriptingContainer;
 import org.jruby.util.RubyDateFormat;
 
 import java.util.Locale;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
 import org.joda.time.format.DateTimeFormat;
 import org.joda.time.format.DateTimeFormatter;
 
@@ -45,6 +48,8 @@ public class TimestampFormatter
 
     private final RubyDateFormat jrubyFormatter;
     private final DateTimeFormatter javaFormatter;
+    private boolean handleNanoResolution = false;
+    private boolean handleMicroResolution = false;
     private final DateTimeZone toTimeZone;
 
     public TimestampFormatter(PluginTask task, Optional<? extends TimestampColumnOption> columnOption)
@@ -62,12 +67,24 @@ public class TimestampFormatter
     {
         this.toTimeZone = toTimeZone;
         if (format.contains("%")) {
-            this.jrubyFormatter = new RubyDateFormat(format, Locale.ENGLISH, true);
             this.javaFormatter = null;
+            this.jrubyFormatter = new RubyDateFormat(format, Locale.ENGLISH, true);
         }
         else {
             this.jrubyFormatter = null;
-            this.javaFormatter = DateTimeFormat.forPattern(format).withLocale(Locale.ENGLISH).withZone(toTimeZone);
+            if (format.contains("nnnnnnnnn")) {
+                this.handleNanoResolution = true;
+                String newFormat = format.replaceAll("nnnnnnnnn", "'%09d'");
+                this.javaFormatter = DateTimeFormat.forPattern(newFormat).withLocale(Locale.ENGLISH).withZone(toTimeZone);
+            }
+            else if (format.contains("nnnnnn")) {
+                this.handleMicroResolution = true;
+                String newFormat = format.replaceAll("nnnnnn", "'%06d'");
+                this.javaFormatter = DateTimeFormat.forPattern(newFormat).withLocale(Locale.ENGLISH).withZone(toTimeZone);
+            }
+            else {
+                this.javaFormatter = DateTimeFormat.forPattern(format).withLocale(Locale.ENGLISH).withZone(toTimeZone);
+            }
         }
     }
 
@@ -106,7 +123,17 @@ public class TimestampFormatter
 
     private String javaFormat(Timestamp value)
     {
-        long milliSecond = value.getEpochSecond() * 1000 + value.getNano() / 1000000;
-        return javaFormatter.print(milliSecond);
+        if (handleNanoResolution) {
+            String datetimeFormatted = javaFormatter.print(value.getEpochSecond() * 1000);
+            return String.format(datetimeFormatted, value.getNano());
+        }
+        else if (handleMicroResolution) {
+            String datetimeFormatted = javaFormatter.print(value.getEpochSecond() * 1000);
+            return String.format(datetimeFormatted, value.getNano() / 1000);
+        }
+        else {
+            long milliSecond = value.getEpochSecond() * 1000 + value.getNano() / 1000000;
+            return javaFormatter.print(milliSecond);
+        }
     }
 }

data/src/main/java/org/embulk/filter/timestamp_format/TimestampParser.java

@@ -74,7 +74,7 @@ public class TimestampParser {
                 this.jrubyParserList.add(helper);
             } else {
                 // special treatment for nano resolution. n is not originally supported by Joda-Time
-                if (format.contains("n")) {
+                if (format.contains("nnnnnnnnn")) {
                     this.handleNanoResolutionList.add(true);
                     String newFormat = format.replaceAll("n", "S");
                     DateTimeFormatter parser = DateTimeFormat.forPattern(newFormat).withLocale(Locale.ENGLISH).withZone(defaultFromTimeZone);

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: embulk-filter-timestamp_format
 version: !ruby/object:Gem::Version
-  version: 0.1.8
+  version: 0.1.9
 platform: ruby
 authors:
 - Naotoshi Seo
@@ -97,7 +97,7 @@ files:
 - src/main/java/org/embulk/filter/timestamp_format/cast/StringCast.java
 - src/main/java/org/embulk/filter/timestamp_format/cast/TimestampCast.java
 - src/test/java/org/embulk/filter/timestamp_format/TestTimestampUnit.java
-- classpath/embulk-filter-timestamp_format-0.1.8.jar
+- classpath/embulk-filter-timestamp_format-0.1.9.jar
 homepage: https://github.com/sonots/embulk-filter-timestamp_format
 licenses:
 - MIT