Skip to content

Releases

TIE 0.6.0

Choose a tag to compare

@themarioga themarioga released this 18 Nov 22:30

A file whose name matches the patterns of several layouts no longer blocks every run when one of those layouts has already processed it. Without overlapping patterns, 0.6.0 behaves like 0.5.0. No API changes.

Changed

  • Ambiguous files that were already processed are ignored. A listed file matching the patterns of several root layouts used to fail the whole run with AMBIGUOUS_LAYOUT, even if it was old and already imported. Now it is ignored when any of those layouts already covers it:

    • its timestamp is not after that layout's watermark (with duplicateProtection.watermark), or
    • the alreadyProcessed predicate reports it.

    It is logged at DEBUG. A file none of them covers still fails the run with AMBIGUOUS_LAYOUT, and a file matching several layouts is never assigned to one of them. Old files left on a server no longer block a definition that gains an overlapping pattern.

Fixed

  • With a name matching several patterns, each layout's watermark is compared with the timestamp extracted by its own pattern, not by the last pattern that matched.

Artifacts

Artifact
io.github.tie-integration-engine:tie-core:0.6.0 The library
io.github.tie-integration-engine:tie-jexl:0.6.0 Optional programmable fields with Apache Commons JEXL
io.github.tie-integration-engine:tie-spring-boot-starter:0.6.0 Optional Spring Boot 4 auto-configuration

Available on Maven Central:

<dependency>
    <groupId>io.github.tie-integration-engine</groupId>
    <artifactId>tie-core</artifactId>
    <version>0.6.0</version>
</dependency>

The jars, with their sources and javadoc, are attached below. Requires Java 25.

Documentation

TIE 0.5.0

Choose a tag to compare

@themarioga themarioga released this 09 Nov 16:22

Definitions can now be written back as JSON or YAML, and fields can compute their value with an expression. Programmable fields come with a new optional module, tie-jexl. Without expressions, 0.5.0 behaves like 0.4.0.

Added

  • DefinitionWriter.toJson / toYaml: a definition built with the Java API, or loaded, written as the document DefinitionLoader reads back.
    • The definition is validated first.
    • Keys follow the schema order and defaults are left out, so equal definitions give the same text.
    • Literal credentials are written as ***.
  • Programmable fields: expression on a field computes its final value.
    • On import, expressions run after the other steps of every field, in declaration order. They see value, the other fields, the parent record (parent, parent.parent), today and the context variables. The result is converted to the field's type and validated.
    • On export, they run over the supplied values.
    • New SPI ExpressionEngine (Tie.Builder.expressionEngine) and line error code EXPRESSION_FAILED.
  • New module tie-jexl: JexlExpressionEngine, with Apache Commons JEXL in a restricted sandbox:
    • no class access, loops, new or side effects;
    • misspelt names are errors;
    • functions tie:age, tie:years, tie:blank, tie:date, tie:format and tie:coalesce.
  • Spring Boot starter: registers a JexlExpressionEngine when tie-jexl is on the classpath.
fields:
  - { name: birthDate, type: DATE }
  - name: idDocument
    expression: "value != null ? value : (tie:age(birthDate) < 16 ? parent.idDocument : null)"
Tie tie = Tie.builder().expressionEngine(new JexlExpressionEngine()).build();
String yaml = DefinitionWriter.toYaml(definition);

Changed

  • FieldDefinition has a new expression component, so its canonical constructor changed. The builder is unaffected.
  • A credential equal to ***, what DefinitionWriter writes for a literal one, is now a configuration problem.
  • copyFrom of a field with an expression is now a configuration problem.

Artifacts

Artifact
io.github.tie-integration-engine:tie-core:0.5.0 The library
io.github.tie-integration-engine:tie-jexl:0.5.0 Optional programmable fields with Apache Commons JEXL (new)
io.github.tie-integration-engine:tie-spring-boot-starter:0.5.0 Optional Spring Boot 4 auto-configuration

Available on Maven Central:

<dependency>
    <groupId>io.github.tie-integration-engine</groupId>
    <artifactId>tie-core</artifactId>
    <version>0.5.0</version>
</dependency>

The jars, with their sources and javadoc, are attached below. Requires Java 25.

Documentation

TIE 0.4.0

Choose a tag to compare

@themarioga themarioga released this 09 Nov 13:57

Records can now be found by position when their name is not known in advance, and layouts fed only by provided inputs no longer need a file name pattern. With the defaults, 0.4.0 behaves exactly like 0.3.0.

Added

  • XML imports without recordElement: every child of the root element is a record, whatever its name; sibling records may have different names. Records are read as before (attributes, dotted paths, nested layouts by nestedKey). Text, comments and processing instructions directly under the root are ignored. Exports still require recordElement.
  • json.firstArray: with no recordProperty, the records are the root array or, for a root object, its first top-level property whose value is an array. Earlier properties are skipped in streaming without being read; a root object without any array fails with STRUCTURAL. It cannot be combined with recordProperty on import and is ignored on export. JsonOptions gets a builder.
  • fileNamePattern is optional on a root import layout when every input channel is a ProvidedStreamChannel. A declared pattern is still validated.
inputChannels: [PROVIDED_STREAM]
layouts:
  - name: employees              # no fileNamePattern: only provided inputs
    allowedFormats: [XML, JSON]
    formatOptions:
      xml: {}                    # no recordElement: every child of the root is a record
      json: { firstArray: true }

Changed

  • JsonOptions has a new firstArray record component, so its canonical constructor changed; use JsonOptions.builder(). JsonOptions.DEFAULTS keeps the same values.
  • A blank (but not null) xml.recordElement on import is now reported as "must not be blank" instead of "is required".

Artifacts

Artifact
io.github.tie-integration-engine:tie-core:0.4.0 The library
io.github.tie-integration-engine:tie-spring-boot-starter:0.4.0 Optional Spring Boot 4 auto-configuration
<dependency>
    <groupId>io.github.tie-integration-engine</groupId>
    <artifactId>tie-core</artifactId>
    <version>0.4.0</version>
</dependency>

The jars, with their sources and javadoc, are attached below. Requires Java 25.

Documentation

TIE 0.3.0

Choose a tag to compare

@themarioga themarioga released this 09 Nov 12:25

Four additions for reading files as the V2 integration engine did: default values delivered as text, XLSX cells read as Excel displays them, blank XLSX rows skipped, and date patterns with a locale. With the defaults, 0.3.0 behaves exactly like 0.2.0.

Added

  • defaultMode on a field with defaultValue: CONVERT (default, the 0.2.0 behaviour) or TEXT. With TEXT the resolved default skips type conversion, still goes through the field's validations, and reaches the consumer as a String whatever the field's type, so an unresolved ${…} template in a DATE field no longer rejects the line.
  • XLSX reading options:
    • skipBlankRecords, as in CSV: a row whose cells are all blank is dropped and uses no line number.
    • cellValues: FORMATTED reads numbers and dates as Excel displays them per the cell's number format (08001, 25%, 1,234.50, 15/03/2024, 12345678A); CANONICAL remains the default.
    • locale (BCP 47, default ROOT) sets the separators and names for FORMATTED: 1.234,50 with es-ES.
    • XlsxOptions has a builder.
  • Date patterns with a locale: TypeFormat.date(pattern, locale) or typeFormat: { pattern, locale }, for month and day names and AM/PM when reading and writing (31-dic-2026 with es-ES).
formatOptions:
  xlsx: { cellValues: FORMATTED, locale: es-ES, skipBlankRecords: true }
fields:
  - { name: signed, type: DATE, typeFormat: { pattern: dd-MMM-yyyy, locale: es-ES } }   # 31-dic-2026
  - { name: loaded, inInput: false, type: DATE, defaultValue: "${loadDate}", defaultMode: TEXT }

Changed

  • RecordView.get(field) is no longer always of the field's Java type: a defaultMode: TEXT field that took its default holds a String, and get(field, LocalDate.class) throws ClassCastException for it. This only affects fields that declare TEXT.
  • Canonical constructors of XlsxOptions (use XlsxOptions.builder()), TypeFormat.DateFormat and FieldDefinition have new components; record patterns on DateFormat(String pattern) no longer compile. XlsxOptions.DEFAULTS, TypeFormat.date(pattern) and the builders are unchanged.

Artifacts

Artifact
io.github.tie-integration-engine:tie-core:0.3.0 The library
io.github.tie-integration-engine:tie-spring-boot-starter:0.3.0 Optional Spring Boot 4 auto-configuration

Available on Maven Central:

<dependency>
    <groupId>io.github.tie-integration-engine</groupId>
    <artifactId>tie-core</artifactId>
    <version>0.3.0</version>
</dependency>

The jars, with their sources and javadoc, are attached below. Requires Java 25.

Documentation

TIE 0.2.0

Choose a tag to compare

@themarioga themarioga released this 09 Nov 10:53

Two contract changes to the import engine: file hooks now run for provided inputs, and field copies can take the canonical value of their source.

Added

  • copyMode on a field with copyFrom: TEXT (default, the 0.1.0 behaviour) or CANONICAL, in the Java API, YAML/JSON and the JSON schema.
    • With CANONICAL, an imported copy receives the canonical text of the source's typed value (2026-12-31, 1234.50, true), whatever the source's typeFormat, so two files with the same date written in different formats give the same copy.
    • When the source is not STRING and the copy's own trimming, transforms, translations and transformer leave that text unchanged, the target tries the canonical form before its typeFormat.
    • Ignored on export, where a copy already takes the source's supplied value. copyMode without copyFrom is a configuration error.
fields:
  - { name: born, type: DATE, typeFormat: dd/MM/yyyy }
  - { name: bornIso, inInput: false, copyFrom: born, copyMode: CANONICAL }   # 2026-12-31

Changed

  • FileHook is called for provided inputs too. With a hook registered, a provided input is spooled once to the working directory and gets onFileReceived (null timestamp, size = bytes read) and onFileCompleted, like a channel file. Without a hook nothing changes. If you register a FileHook and use provided inputs, it now receives these calls.
  • FieldDefinition has a new copyMode record component after copyFrom, so its canonical constructor changed. Code using the builder is unaffected.

Artifacts

Artifact
io.github.tie-integration-engine:tie-core:0.2.0 The library
io.github.tie-integration-engine:tie-spring-boot-starter:0.2.0 Optional Spring Boot 4 auto-configuration

Available on Maven Central:

<dependency>
    <groupId>io.github.tie-integration-engine</groupId>
    <artifactId>tie-core</artifactId>
    <version>0.2.0</version>
</dependency>

The jars, with their sources and javadoc, are attached below. Requires Java 25.

Documentation

TIE 0.1.0

Choose a tag to compare

@themarioga themarioga released this 08 Nov 15:53

First release of TIE, a stateless Java library for file integrations.

Highlights

  • Imports from SFTP, local directories and provided streams; exports and deliverFile to the same channels.
  • CSV, XLSX, XML and JSON, read and written incrementally: memory is bounded by one record, not by the file.
  • Typed field engine: column lookup by header, position or letter; copies, trimming, text transforms, translation tables, custom transformers, type conversion, validations, empty-value rules and default values.
  • Nested child records in JSON and XML.
  • Delivery strategies STREAMING and VALIDATE_THEN_DELIVER, rejection thresholds, consumer error policies, cancellation, run listener and file hooks.
  • Stateless duplicate protection: watermarks, an already-processed predicate, and move or delete after success.
  • Safe deliveries: temporary .<name>.part file, final rename, onExistingFile (SUFFIX, FAIL, OVERWRITE).
  • Result reports in CSV, XLSX, JSON and XML.
  • Declarative definitions in YAML or JSON, with a published JSON schema.
  • Stable catalogs of line error codes, file error types and warning codes.
  • Spring Boot 4 starter.

Requirements

Java 25. tie-core has no Spring dependency.

Artifacts

Artifact
io.github.tie-integration-engine:tie-core:0.1.0 The library
io.github.tie-integration-engine:tie-spring-boot-starter:0.1.0 Optional Spring Boot 4 auto-configuration

Available on Maven Central:

<dependency>
    <groupId>io.github.tie-integration-engine</groupId>
    <artifactId>tie-core</artifactId>
    <version>0.1.0</version>
</dependency>

The jars, with their sources and javadoc, are also attached below.

Documentation

  • Wiki: getting started, configuration reference, how imports and exports work, error catalogs.
  • README and CHANGELOG.