#4317: SchemaExpService.java

tools/org.projectforge.tools.schemaexp/src/main/java/org/projectforge/tools/schemaexp/SchemaExpService.java

Path: tools/org.projectforge.tools.schemaexp/src/main/java/org/projectforge/tools/schemaexp/SchemaExpService.java

Type: Spring @Service · Lines: 95

Purpose: Core business logic for the SchemaExp developer tool — database XML export, XML import (with dual-engine support), and database clearing. The central service that all other SchemaExp components delegate to.

Source: GitHub

95 lines · 80 code · 0 comments · 15 blank

Architecture: Dual-engine import system

Commit	Date	Message
`a5bbdca6a`	2017-12-04	Change logger to slf4j
`9ebb88522`	2016-07-18	Initial commit

The class's most notable architectural feature is its dual-engine import system, controlled by useJpaXmlDumpService = true (line 47):

Engine	Flag	Library	Features
Genome JPA	`true` (default)	`de.micromata.genome.db.jpa.xmldump`	JPA-aware parsing, entity type resolution, relationship handling, RestoreMode support (InsertNew/OverWrite/InsertAll), primary key allocation via sequence generators
Legacy XStream	`false`	ThoughtWorks XStream via `XmlDump`	Raw XStream serialization, manual verification via `verifyDump()`, post-import integrity checks via `ConfigurationDao`

The Genome engine is preferred because it understands JPA entity metadata — it correctly handles @OneToMany cascades, @Inheritance strategies, and foreign key relationships during import. The legacy engine is retained for backward compatibility.

Key methods

clearDb() (lines 49-52): Calls emfac.getJpaSchemaService().clearDatabase() — drops all tables in reverse dependency order. This is a destructive operation with no undo. Used both standalone (-cleardb flag) and as a prerequisite for imports (-cleardb -imp).

doImport(boolean clearDb, String clientFileName, RestoreMode modus) (lines 54-87): The main import pipeline: (1) optionally clear the database, (2) validate file existence, (3) detect file format (.xml vs .xml.gz) and decompress if needed, (4) delegate to the selected engine, (5) handle cleanup in finally block. The RestoreMode enum controls conflict resolution: InsertNew skips existing records, OverWrite updates them, InsertAll inserts everything (fails on duplicates).

doExport(String fileName) (lines 89-94): Delegates to jpaXmlDumpService.dumpToXml(emfac, file). Unlike import, export always uses the Genome library — there's no legacy XStream export path. Queries all entity types from JPA metadata, loads all records (full table scans), and serializes to XML with JPA-aware marshalling.

Code quality notes

Unused field: TransactionTemplate txTemplate (line 32) is injected but never used — dead code.
Deprecated API: IOUtils.closeQuietly() (line 84) is deprecated in Apache Commons IO 2.7+. Should use try-with-resources.
Typo in parent class: SchemaExpMain.initApplicatinContext() (missing 'o' in "Application") is called by all tests.
Hardcoded flag: useJpaXmlDumpService = true should ideally be a Spring @Value property for runtime configuration.
No export compression: The export path doesn't gzip the output. For production dumps, manually gzip after export.