From a6d7e9d1a8e2b8b8fb84e3d79409657c2fa694b6 Mon Sep 17 00:00:00 2001
From: Ryan Ernst <ryan@iernst.net>
Date: Fri, 24 Mar 2017 00:27:08 -0700
Subject: [PATCH] Build: Rewrite antlr regeneration in gradle

This change ports the regeneration of antlr parser/lexer into gradle
(but does still take advantage of ant calls where appropriate).
---
 modules/lang-painless/ant.xml      | 157 -----------------------------
 modules/lang-painless/build.gradle | 109 ++++++++++++++++----
 2 files changed, 92 insertions(+), 174 deletions(-)
 delete mode 100644 modules/lang-painless/ant.xml

diff --git a/modules/lang-painless/ant.xml b/modules/lang-painless/ant.xml
deleted file mode 100644
index 90e66b7b1a9b9..0000000000000
--- a/modules/lang-painless/ant.xml
+++ /dev/null
@@ -1,157 +0,0 @@
-<?xml version="1.0"?>
-<project name="ant-stuff">
-
-<!--
- grammar regeneration logic
- we do this with ant for several reasons:
- * remove generated tabs for forbidden-apis
- * remove generated timestamps/filenames for reproducible build
- * fix CRLF line endings for windows consistency
- * ability to make classes package-private
- * keeping in source code control is easier on IDEs
- * regeneration should be rare, no reason to be religious about generated files
- * all logic already written and battle tested in lucene build
--->
-  <target name="regenerate" description="Regenerate antlr lexer and parser" depends="run-antlr"/>
-
-  <target name="run-antlr">
-    <regen-delete grammar="Painless"/>
-    <regen-lexer grammar="Painless"/>
-    <regen-parser grammar="Painless"/>
-    <regen-fix grammar="Painless"/>
-  </target>
-
-  <macrodef name="replace-value">
-    <attribute name="value" />
-    <attribute name="property" />
-    <attribute name="from" />
-    <attribute name="to" />
-    <sequential>
-      <loadresource property="@{property}">
-        <string value="@{value}"/>
-        <filterchain>
-          <tokenfilter>
-            <filetokenizer/>
-            <replacestring from="@{from}" to="@{to}"/>
-          </tokenfilter>
-        </filterchain>
-      </loadresource>
-    </sequential>
-  </macrodef>
-
-  <macrodef name="regen-delete">
-    <attribute name="grammar" />
-    <sequential>
-      <local name="grammar.path"/>
-      <property name="grammar.path" location="src/main/antlr"/>
-      <!-- delete token files so files will be generated -->
-      <delete dir="${grammar.path}" includes="@{grammar}*.tokens"/>
-      <local name="output.path"/>
-      <patternset id="grammar.@{grammar}.patternset">
-        <include name="@{grammar}Lexer.java" />
-        <include name="@{grammar}Parser.java" />
-        <include name="@{grammar}Lexer.tokens" />
-        <include name="@{grammar}Parser.tokens" />
-        <include name="@{grammar}ParserVisitor.java" />
-        <include name="@{grammar}ParserBaseVisitor.java" />
-      </patternset>
-      <property name="output.path" location="src/main/java/org/elasticsearch/painless/antlr"/>
-      <!-- delete parser and lexer so files will be generated -->
-      <delete dir="${output.path}">
-        <patternset refid="grammar.@{grammar}.patternset"/>
-      </delete>
-    </sequential>
-  </macrodef>
-
-  <macrodef name="regen-lexer">
-    <attribute name="grammar" />
-    <sequential>
-      <local name="grammar.path"/>
-      <local name="output.path"/>
-      <property name="grammar.path" location="src/main/antlr"/>
-      <property name="output.path" location="src/main/java/org/elasticsearch/painless/antlr"/>
-      <!-- invoke ANTLR4 -->
-      <java classname="org.antlr.v4.Tool" fork="true" failonerror="true" classpathref="regenerate.classpath" taskname="antlr">
-        <sysproperty key="file.encoding" value="UTF-8"/>
-        <sysproperty key="user.language" value="en"/>
-        <sysproperty key="user.country" value="US"/>
-        <sysproperty key="user.variant" value=""/>
-        <arg value="-Werror"/>
-        <arg value="-package"/>
-        <arg value="org.elasticsearch.painless.antlr"/>
-        <arg value="-o"/>
-        <arg path="${output.path}"/>
-        <arg path="${grammar.path}/@{grammar}Lexer.g4"/>
-      </java>
-    </sequential>
-  </macrodef>
-
-  <macrodef name="regen-parser">
-    <attribute name="grammar" />
-    <sequential>
-      <local name="grammar.path"/>
-      <local name="output.path"/>
-      <property name="grammar.path" location="src/main/antlr"/>
-      <property name="output.path" location="src/main/java/org/elasticsearch/painless/antlr"/>
-      <!-- invoke ANTLR4 -->
-      <java classname="org.antlr.v4.Tool" fork="true" failonerror="true" classpathref="regenerate.classpath" taskname="antlr">
-        <sysproperty key="file.encoding" value="UTF-8"/>
-        <sysproperty key="user.language" value="en"/>
-        <sysproperty key="user.country" value="US"/>
-        <sysproperty key="user.variant" value=""/>
-        <arg value="-Werror"/>
-        <arg value="-package"/>
-        <arg value="org.elasticsearch.painless.antlr"/>
-        <arg value="-no-listener"/>
-        <arg value="-visitor"/>
-        <!-- <arg value="-Xlog"/> -->
-        <arg value="-o"/>
-        <arg path="${output.path}"/>
-        <arg path="${grammar.path}/@{grammar}Parser.g4"/>
-      </java>
-    </sequential>
-  </macrodef>
-
-  <macrodef name="regen-fix">
-    <attribute name="grammar" />
-    <sequential>
-      <local name="grammar.path"/>
-      <local name="output.path"/>
-      <property name="grammar.path" location="src/main/antlr"/>
-      <property name="output.path" location="src/main/java/org/elasticsearch/painless/antlr"/>
-      <patternset id="grammar.@{grammar}.patternset">
-        <include name="@{grammar}Lexer.java" />
-        <include name="@{grammar}Parser.java" />
-        <include name="@{grammar}ParserVisitor.java" />
-        <include name="@{grammar}ParserBaseVisitor.java" />
-      </patternset>
-      <!-- fileset with files to edit -->
-      <fileset id="grammar.fileset" dir="${output.path}">
-        <patternset refid="grammar.@{grammar}.patternset"/>
-      </fileset>
-      <!-- moves token files to grammar directory for use with IDE's -->
-      <move file="${output.path}/@{grammar}Lexer.tokens" todir="${grammar.path}"/>
-      <move file="${output.path}/@{grammar}Parser.tokens" todir="${grammar.path}"/>
-      <!-- make the generated classes package private -->
-      <replaceregexp match="public ((interface|class) \Q@{grammar}\E\w+)" replace="\1" encoding="UTF-8">
-        <fileset refid="grammar.fileset"/>
-      </replaceregexp>
-      <!-- make the lexer abstract -->
-      <replaceregexp match="(class \Q@{grammar}\ELexer)" replace="abstract \1" encoding="UTF-8">
-        <fileset refid="grammar.fileset"/>
-      </replaceregexp>
-      <!-- nuke timestamps/filenames in generated files -->
-      <replaceregexp match="\Q// Generated from \E.*" replace="\/\/ ANTLR GENERATED CODE: DO NOT EDIT" encoding="UTF-8">
-        <fileset refid="grammar.fileset"/>
-      </replaceregexp>
-      <!-- remove tabs in antlr generated files -->
-      <replaceregexp match="\t" flags="g" replace="  " encoding="UTF-8">
-        <fileset refid="grammar.fileset"/>
-      </replaceregexp>
-      <!-- fix line endings -->
-      <fixcrlf srcdir="${output.path}">
-        <patternset refid="grammar.@{grammar}.patternset"/>
-      </fixcrlf>
-    </sequential>
-  </macrodef>
-</project>
diff --git a/modules/lang-painless/build.gradle b/modules/lang-painless/build.gradle
index 943d953ea484e..31b41261b3a77 100644
--- a/modules/lang-painless/build.gradle
+++ b/modules/lang-painless/build.gradle
@@ -33,23 +33,6 @@ dependencyLicenses {
   mapping from: /asm-.*/, to: 'asm'
 }
 
-// regeneration logic, comes in via ant right now
-// don't port it to gradle, it works fine.
-
-/*
-configurations {
-  regenerate
-}
-
-dependencies {
-  regenerate 'org.antlr:antlr4:4.5.1-1'
-}
-
-// TODO: this is causing a resolve at configurationt time. we really should not be using ant anymore
-ant.references['regenerate.classpath'] = new Path(ant.project, configurations.regenerate.asPath)
-ant.importBuild 'ant.xml'
-*/
-
 integTestCluster {
   setting 'script.max_compilations_per_minute', '1000'
 }
@@ -73,3 +56,95 @@ task generatePainlessApi(type: JavaExec) {
   classpath = sourceSets.test.runtimeClasspath
   args file('../../docs/reference/painless-api-reference')
 }
+
+/**********************************************
+ *            Parser regeneration             *
+ **********************************************/
+
+configurations {
+  regenerate
+}
+
+dependencies {
+  regenerate 'org.antlr:antlr4:4.5.1-1'
+}
+
+String grammarPath = 'src/main/antlr'
+String outputPath = 'src/main/java/org/elasticsearch/painless/antlr'
+
+task cleanGenerated(type: Delete) {
+  delete fileTree(grammarPath) {
+    include '*.tokens'
+  }
+  delete fileTree(outputPath) {
+    include 'Painless*.java'
+  }
+}
+
+task regenLexer(type: JavaExec) {
+  dependsOn cleanGenerated
+  main = 'org.antlr.v4.Tool'
+  classpath = configurations.regenerate
+  systemProperty 'file.encoding', 'UTF-8'
+  systemProperty 'user.language', 'en'
+  systemProperty 'user.country', 'US'
+  systemProperty 'user.variant', ''
+  args '-Werror',
+       '-package', 'org.elasticsearch.painless.antlr',
+       '-o', outputPath,
+       "${file(grammarPath)}/PainlessLexer.g4"
+}
+
+task regenParser(type: JavaExec) {
+  dependsOn regenLexer
+  main = 'org.antlr.v4.Tool'
+  classpath = configurations.regenerate
+  systemProperty 'file.encoding', 'UTF-8'
+  systemProperty 'user.language', 'en'
+  systemProperty 'user.country', 'US'
+  systemProperty 'user.variant', ''
+  args '-Werror',
+       '-package', 'org.elasticsearch.painless.antlr',
+       '-no-listener',
+       '-visitor',
+       // '-Xlog',
+       '-o', outputPath,
+       "${file(grammarPath)}/PainlessParser.g4"
+}
+
+task regen {
+  dependsOn regenParser
+  doLast {
+    // moves token files to grammar directory for use with IDE's
+    ant.move(file: "${outputPath}/PainlessLexer.tokens", toDir: grammarPath)
+    ant.move(file: "${outputPath}/PainlessParser.tokens", toDir: grammarPath)
+    // make the generated classes package private
+    ant.replaceregexp(match: 'public ((interface|class) \\QPainless\\E\\w+)',
+                      replace: '\\1',
+                      encoding: 'UTF-8') {
+      fileset(dir: outputPath, includes: 'Painless*.java')
+    }
+    // make the lexer abstract
+    ant.replaceregexp(match: '(class \\QPainless\\ELexer)',
+                      replace: 'abstract \\1',
+                      encoding: 'UTF-8') {
+      fileset(dir: outputPath, includes: 'PainlessLexer.java')
+    }
+    // nuke timestamps/filenames in generated files
+    ant.replaceregexp(match: '\\Q// Generated from \\E.*',
+                      replace: '\\/\\/ ANTLR GENERATED CODE: DO NOT EDIT',
+                      encoding: 'UTF-8') {
+      fileset(dir: outputPath, includes: 'Painless*.java')
+    }
+    // remove tabs in antlr generated files
+    ant.replaceregexp(match: '\t', flags: 'g', replace: '  ', encoding: 'UTF-8') {
+      fileset(dir: outputPath, includes: 'Painless*.java')
+    }
+    // fix line endings
+    ant.fixcrlf(srcdir: outputPath) {
+      patternset(includes: 'Painless*.java')
+    }
+  }
+}
+
+