Add Jackson-based model generation for Trino 483 (prototyping)

.gitignore

@@ -2,3 +2,6 @@ Gemfile.lock
 tmp/*
 pkg/*
 coverage/
+/modelgen/java-extractor/.gradle/
+/modelgen/java-extractor/build/
+/modelgen/java-extractor/.idea/

modelgen/compare_model_shapes.rb

@@ -0,0 +1,128 @@
+# frozen_string_literal: true
+
+require "json"
+
+if ARGV.length != 3
+  warn <<~USAGE
+    Usage:
+      ruby modelgen/compare_model_shapes.rb \
+        <version> \
+        <old-model.rb> \
+        <new-model.rb>
+  USAGE
+
+  exit 1
+end
+
+version, old_path, new_path = ARGV
+
+module Trino
+  module Client
+    module ModelVersions
+    end
+  end
+end
+
+module_name = :"V#{version.tr('.', '_')}"
+
+def capture_shape(
+  module_name:,
+  path:
+)
+  model_versions =
+    Trino::Client::ModelVersions
+
+  if model_versions.const_defined?(
+    module_name,
+    false,
+    )
+    model_versions.send(
+      :remove_const,
+      module_name,
+      )
+  end
+
+  load File.expand_path(path)
+
+  version_module = model_versions.const_get(
+    module_name,
+    false,
+    )
+
+  shape = {}
+
+  version_module
+    .constants(false)
+    .sort
+    .each do |constant_name|
+    value = version_module.const_get(
+      constant_name,
+      false,
+      )
+
+    next unless value.respond_to?(:members)
+
+    shape[constant_name.to_s] =
+      value.members.map(&:to_s)
+  end
+
+  model_versions.send(
+    :remove_const,
+    module_name,
+    )
+
+  shape
+end
+
+old_shape = capture_shape(
+  module_name: module_name,
+  path: old_path,
+  )
+
+new_shape = capture_shape(
+  module_name: module_name,
+  path: new_path,
+  )
+
+missing_models =
+  old_shape.keys - new_shape.keys
+
+added_models =
+  new_shape.keys - old_shape.keys
+
+missing_fields = {}
+added_fields = {}
+
+(old_shape.keys & new_shape.keys).each do |model|
+  old_fields = old_shape.fetch(model)
+  new_fields = new_shape.fetch(model)
+
+  missing = old_fields - new_fields
+  added = new_fields - old_fields
+
+  missing_fields[model] = missing \
+    unless missing.empty?
+
+  added_fields[model] = added \
+    unless added.empty?
+end
+
+result = {
+  missingModels: missing_models.sort,
+  addedModels: added_models.sort,
+  missingFields: missing_fields,
+  addedFields: added_fields,
+}
+
+puts JSON.pretty_generate(result)
+
+has_regression =
+  !missing_models.empty? ||
+  !missing_fields.empty?
+
+if has_regression
+  warn "Compatibility check failed."
+  exit 1
+end
+
+puts "Compatibility check passed."

modelgen/generate_from_schema.rb

@@ -0,0 +1,62 @@
+# frozen_string_literal: true
+
+require "erb"
+require "fileutils"
+require "json"
+
+require_relative "schema_formatter"
+
+if ARGV.length != 3
+  warn <<~USAGE
+    Usage:
+      ruby modelgen/generate_from_schema.rb \
+        <version> \
+        <schema.json> \
+        <output.rb>
+  USAGE
+
+  exit 1
+end
+
+model_version, schema_path, output_path = ARGV
+
+schema = JSON.parse(
+  File.read(schema_path),
+  )
+
+schema_version = schema.fetch(
+  "trinoVersion",
+  )
+
+unless schema_version == model_version
+  raise ArgumentError,
+        "Version mismatch: argument=#{model_version.inspect}, " \
+          "schema=#{schema_version.inspect}"
+end
+
+formatter = TrinoModels::SchemaFormatter.new(
+  schema,
+  )
+
+@contents = formatter.format
+@model_version = model_version
+
+template_path = File.expand_path(
+  "model_versions.rb",
+  __dir__,
+  )
+
+template = ERB.new(
+  File.read(template_path),
+  )
+
+FileUtils.mkdir_p(
+  File.dirname(output_path),
+  )
+
+File.write(
+  output_path,
+  template.result(binding),
+  )
+
+puts "Generated #{output_path}."

modelgen/java-extractor/build.gradle

@@ -0,0 +1,83 @@
+plugins {
+    id 'application'
+    id 'java'
+}
+
+repositories {
+    mavenCentral()
+}
+
+def trinoVersion = providers
+    .gradleProperty('trinoVersion')
+    .getOrElse('483')
+
+def javaVersion = providers
+    .gradleProperty('javaVersion')
+    .getOrElse('25')
+    .toInteger()
+
+java {
+    toolchain {
+        languageVersion = JavaLanguageVersion.of(javaVersion)
+    }
+}
+
+dependencies {
+    implementation "io.trino:trino-client:${trinoVersion}"
+    implementation "io.trino:trino-main:${trinoVersion}"
+
+    implementation "com.fasterxml.jackson.core:jackson-databind:${jacksonVersion}"
+    implementation "com.fasterxml.jackson.datatype:jackson-datatype-jdk8:${jacksonVersion}"
+    implementation "com.fasterxml.jackson.module:jackson-module-parameter-names:${jacksonVersion}"
+
+    testImplementation platform('org.junit:junit-bom:5.11.4')
+    testImplementation 'org.junit.jupiter:junit-jupiter'
+    testRuntimeOnly 'org.junit.platform:junit-platform-launcher'
+}
+
+application {
+    mainClass = 'org.trinoruby.modelgen.ModelExtractor'
+}
+
+test {
+    useJUnitPlatform()
+}
+
+tasks.register('extractModels', JavaExec) {
+    group = 'model generation'
+    description = 'Extracts Trino models into an intermediate JSON schema'
+
+    classpath = sourceSets.main.runtimeClasspath
+    mainClass = application.mainClass
+
+    def version = trinoVersion
+
+    def rootsFile = layout.projectDirectory.file(
+            "config/roots-${version}.txt"
+    )
+
+    def outputFile = layout.projectDirectory.file(
+            "../schemas/${version}.json"
+    )
+
+    args(
+            '--trino-version', version,
+            '--roots', rootsFile.asFile.absolutePath,
+            '--output', outputFile.asFile.absolutePath
+    )
+
+    doFirst {
+        println "Extractor Java: ${javaLauncher.get().executablePath}"
+        println "Trino version: ${version}"
+        println "Roots: ${rootsFile.asFile}"
+        println "Output: ${outputFile.asFile}"
+    }
+}
+
+tasks.register('printRuntimeClasspath') {
+    group = 'help'
+
+    doLast {
+        println sourceSets.main.runtimeClasspath.asPath
+    }
+}

modelgen/java-extractor/config/roots-351.txt

@@ -0,0 +1,59 @@
+# Trino REST API root model
+# Fully qualified class name
+
+io.trino.client.QueryResults
+io.trino.execution.QueryInfo
+io.trino.server.BasicQueryInfo
+
+io.trino.sql.planner.plan.OutputNode
+io.trino.sql.planner.plan.ProjectNode
+io.trino.sql.planner.plan.TableScanNode
+io.trino.sql.planner.plan.ValuesNode
+io.trino.sql.planner.plan.AggregationNode
+io.trino.sql.planner.plan.MarkDistinctNode
+io.trino.sql.planner.plan.FilterNode
+io.trino.sql.planner.plan.WindowNode
+io.trino.sql.planner.plan.RowNumberNode
+io.trino.sql.planner.plan.TopNRowNumberNode
+io.trino.sql.planner.plan.LimitNode
+io.trino.sql.planner.plan.DistinctLimitNode
+io.trino.sql.planner.plan.TopNNode
+io.trino.sql.planner.plan.SampleNode
+io.trino.sql.planner.plan.SortNode
+io.trino.sql.planner.plan.RemoteSourceNode
+io.trino.sql.planner.plan.JoinNode
+io.trino.sql.planner.plan.SemiJoinNode
+io.trino.sql.planner.plan.SpatialJoinNode
+io.trino.sql.planner.plan.IndexJoinNode
+io.trino.sql.planner.plan.IndexSourceNode
+io.trino.sql.planner.plan.TableWriterNode
+io.trino.sql.planner.plan.DeleteNode
+io.trino.sql.planner.plan.TableFinishNode
+io.trino.sql.planner.plan.UnnestNode
+io.trino.sql.planner.plan.ExchangeNode
+io.trino.sql.planner.plan.UnionNode
+io.trino.sql.planner.plan.IntersectNode
+io.trino.sql.planner.plan.EnforceSingleRowNode
+io.trino.sql.planner.plan.GroupIdNode
+io.trino.sql.planner.plan.ExplainAnalyzeNode
+io.trino.sql.planner.plan.ApplyNode
+io.trino.sql.planner.plan.AssignUniqueId
+io.trino.sql.planner.plan.CorrelatedJoinNode
+io.trino.sql.planner.plan.StatisticsWriterNode
+io.trino.metadata.AnalyzeTableHandle
+io.trino.metadata.InsertTableHandle
+io.trino.metadata.OutputTableHandle
+io.trino.operator.ExchangeClientStatus
+io.trino.operator.JoinOperatorInfo
+io.trino.operator.PageBufferClientStatus
+io.trino.operator.PartitionedOutputOperator$PartitionedOutputInfo
+io.trino.operator.SplitOperatorInfo
+io.trino.operator.TableFinishInfo
+io.trino.operator.TableWriterOperator$TableWriterInfo
+io.trino.operator.WindowInfo
+io.trino.operator.WindowInfo$DriverWindowInfo
+io.trino.operator.exchange.LocalExchangeBufferInfo
+io.trino.sql.planner.plan.StatisticsWriterNode$WriteStatisticsHandle
+io.trino.sql.planner.plan.TableWriterNode$CreateTarget
+io.trino.sql.planner.plan.TableWriterNode$InsertTarget
+io.trino.sql.planner.plan.TableWriterNode$RefreshMaterializedViewTarget

modelgen/java-extractor/config/roots-483.txt

@@ -0,0 +1,53 @@
+# Trino REST API root model
+# Fully qualified class name
+
+io.trino.client.QueryResults
+io.trino.execution.QueryInfo
+io.trino.server.BasicQueryInfo
+
+io.trino.sql.planner.plan.OutputNode
+io.trino.sql.planner.plan.ProjectNode
+io.trino.sql.planner.plan.TableScanNode
+io.trino.sql.planner.plan.ValuesNode
+io.trino.sql.planner.plan.AggregationNode
+io.trino.sql.planner.plan.MarkDistinctNode
+io.trino.sql.planner.plan.FilterNode
+io.trino.sql.planner.plan.WindowNode
+io.trino.sql.planner.plan.RowNumberNode
+io.trino.sql.planner.plan.LimitNode
+io.trino.sql.planner.plan.DistinctLimitNode
+io.trino.sql.planner.plan.TopNNode
+io.trino.sql.planner.plan.SampleNode
+io.trino.sql.planner.plan.SortNode
+io.trino.sql.planner.plan.RemoteSourceNode
+io.trino.sql.planner.plan.JoinNode
+io.trino.sql.planner.plan.SemiJoinNode
+io.trino.sql.planner.plan.SpatialJoinNode
+io.trino.sql.planner.plan.IndexJoinNode
+io.trino.sql.planner.plan.IndexSourceNode
+io.trino.sql.planner.plan.TableWriterNode
+io.trino.sql.planner.plan.TableFinishNode
+io.trino.sql.planner.plan.UnnestNode
+io.trino.sql.planner.plan.ExchangeNode
+io.trino.sql.planner.plan.UnionNode
+io.trino.sql.planner.plan.IntersectNode
+io.trino.sql.planner.plan.EnforceSingleRowNode
+io.trino.sql.planner.plan.GroupIdNode
+io.trino.sql.planner.plan.ExplainAnalyzeNode
+io.trino.sql.planner.plan.ApplyNode
+io.trino.sql.planner.plan.AssignUniqueId
+io.trino.sql.planner.plan.CorrelatedJoinNode
+io.trino.sql.planner.plan.StatisticsWriterNode
+io.trino.metadata.AnalyzeTableHandle
+io.trino.metadata.InsertTableHandle
+io.trino.metadata.OutputTableHandle
+io.trino.operator.PageBufferClientStatus
+io.trino.operator.TableFinishInfo
+io.trino.operator.TableWriterOperator$TableWriterInfo
+io.trino.operator.WindowInfo
+io.trino.operator.WindowInfo$DriverWindowInfo
+io.trino.operator.exchange.LocalExchangeBufferInfo
+io.trino.sql.planner.plan.StatisticsWriterNode$WriteStatisticsHandle
+io.trino.sql.planner.plan.TableWriterNode$CreateTarget
+io.trino.sql.planner.plan.TableWriterNode$InsertTarget
+io.trino.sql.planner.plan.TableWriterNode$RefreshMaterializedViewTarget