From 483dfdcd7cc0ca1234b1d6af83126215938aa75f Mon Sep 17 00:00:00 2001
From: Michael Schmitz <michael@schmitztech.com>
Date: Wed, 16 Oct 2013 08:29:10 -0700
Subject: [PATCH 1/2] Add Format class and a brat implementation.

---
 core/src/main/scala/edu/knowitall/tool/Format.scala | 12 ++++++++++++
 .../scala/edu/knowitall/tool/postag/Postagger.scala | 13 +++++++++++++
 2 files changed, 25 insertions(+)
 create mode 100644 core/src/main/scala/edu/knowitall/tool/Format.scala

diff --git a/core/src/main/scala/edu/knowitall/tool/Format.scala b/core/src/main/scala/edu/knowitall/tool/Format.scala
new file mode 100644
index 0000000..07de1aa
--- /dev/null
+++ b/core/src/main/scala/edu/knowitall/tool/Format.scala
@@ -0,0 +1,12 @@
+package edu.knowitall.tool
+
+trait Writer[F, T] {
+  def write(from: F): T
+}
+
+trait Reader[F, T] {
+  def read(from: F): T
+}
+
+trait Format[F, T]
+extends Writer[F, T] with Reader[T, F]
diff --git a/core/src/main/scala/edu/knowitall/tool/postag/Postagger.scala b/core/src/main/scala/edu/knowitall/tool/postag/Postagger.scala
index 833d781..e843e49 100644
--- a/core/src/main/scala/edu/knowitall/tool/postag/Postagger.scala
+++ b/core/src/main/scala/edu/knowitall/tool/postag/Postagger.scala
@@ -64,6 +64,19 @@ object Postagger {
   def tokensFrom(postags: Seq[String], tokens: Seq[Token]): Seq[PostaggedToken] = {
     (postags zip tokens).map { case (postag, token) => PostaggedToken(token, postag) }
   }
+
+  object bratFormat extends Format[PostaggedToken, String] {
+    def write(token: PostaggedToken): String = {
+      Iterator(token.postag, token.offset, token.offsets.end, token.string).mkString("\t")
+    }
+
+    def read(string: String): PostaggedToken = {
+      string.split("\t") match {
+        case Array(postag, offset, _, token) => PostaggedToken(postag, token, offset.toInt)
+        case _ => throw new MatchError("Could not match BRAT PostaggedToken: " + string)
+      }
+    }
+  }
 }
 
 abstract class PostaggerMain extends LineProcessor("postagger") {

From 34c33b256bccef449fe8ce93170aeea327df6345 Mon Sep 17 00:00:00 2001
From: Michael Schmitz <michael@schmitztech.com>
Date: Wed, 16 Oct 2013 10:17:11 -0700
Subject: [PATCH 2/2] Move serialization to PostaggedToken.

---
 .../tool/postag/PostaggedToken.scala          | 21 +++++++++++++++++--
 .../edu/knowitall/tool/postag/Postagger.scala | 13 ------------
 2 files changed, 19 insertions(+), 15 deletions(-)

diff --git a/core/src/main/scala/edu/knowitall/tool/postag/PostaggedToken.scala b/core/src/main/scala/edu/knowitall/tool/postag/PostaggedToken.scala
index 59a4b85..617ff39 100644
--- a/core/src/main/scala/edu/knowitall/tool/postag/PostaggedToken.scala
+++ b/core/src/main/scala/edu/knowitall/tool/postag/PostaggedToken.scala
@@ -1,5 +1,5 @@
-package edu.knowitall
-package tool.postag
+package edu.knowitall.tool
+package postag
 
 import edu.knowitall.common.HashCodeHelper
 import edu.knowitall.tool.tokenize.Token
@@ -78,4 +78,21 @@ object PostaggedToken {
   def apply(token: Token, postag: String): PostaggedToken = PostaggedToken(postag, token.string, token.offset)
 
   def unapply(token: PostaggedToken): Option[(String, String, Int)] = Some((token.postag, token.string, token.offset))
+
+  object bratFormat extends Format[PostaggedToken, String] {
+    def write(token: PostaggedToken): String = {
+      Iterator(token.postag + " " + token.offset + " " + token.offsets.end, token.string).mkString("\t")
+    }
+
+    def read(string: String): PostaggedToken = {
+      string.split("\t") match {
+        case Array(meat, token) =>
+          meat.split("\\s+") match {
+            case Array(postag, token, offset) => PostaggedToken(postag, token, offset.toInt)
+            case _ => throw new MatchError("Could not match BRAT PostaggedToken: " + string)
+          }
+        case _ => throw new MatchError("Could not match BRAT PostaggedToken: " + string)
+      }
+    }
+  }
 }
diff --git a/core/src/main/scala/edu/knowitall/tool/postag/Postagger.scala b/core/src/main/scala/edu/knowitall/tool/postag/Postagger.scala
index e843e49..833d781 100644
--- a/core/src/main/scala/edu/knowitall/tool/postag/Postagger.scala
+++ b/core/src/main/scala/edu/knowitall/tool/postag/Postagger.scala
@@ -64,19 +64,6 @@ object Postagger {
   def tokensFrom(postags: Seq[String], tokens: Seq[Token]): Seq[PostaggedToken] = {
     (postags zip tokens).map { case (postag, token) => PostaggedToken(token, postag) }
   }
-
-  object bratFormat extends Format[PostaggedToken, String] {
-    def write(token: PostaggedToken): String = {
-      Iterator(token.postag, token.offset, token.offsets.end, token.string).mkString("\t")
-    }
-
-    def read(string: String): PostaggedToken = {
-      string.split("\t") match {
-        case Array(postag, offset, _, token) => PostaggedToken(postag, token, offset.toInt)
-        case _ => throw new MatchError("Could not match BRAT PostaggedToken: " + string)
-      }
-    }
-  }
 }
 
 abstract class PostaggerMain extends LineProcessor("postagger") {