Integration Test Guide

Integration tests run the abapgit-agent CLI end-to-end against a real ABAP system. The test suite is self-organizing: on the first run it automatically clones all required test repos and activates the necessary ABAP objects. Subsequent runs skip any setup that is already in place.

Prerequisites

Requirement	Notes
Node.js ≥ 16	`node --version`
Configured ABAP system	BAS/on-premise with abapgit-agent ABAP backend installed
`.abapGitAgent` in project root	Created by `abapgit-agent init`; must contain valid credentials
Write access to test repos	Your own forks of the `github.com/SylvosCai/abgagt-*` reference repos — the public repos are read-only for external users. See Test Repo Access Requirements.

.abapGitAgent minimum fields needed for integration tests:

{
  "host": "your-sap-system.com",
  "sapport": 443,
  "protocol": "https",
  "client": "100",
  "user": "TECH_USER",
  "password": "your-password",
  "language": "EN",
  "gitUsername": "git-user",
  "gitPassword": "git-token"
}

Quick Start

# First run — setup + all tests (takes ~5 min):
npm run test:all

# Setup only (verify prerequisites are in place, then stop):
npm run test:setup

# Unit tests only (no ABAP system needed):
npm test

On first run npm run test:all prints a setup phase that:

Clones each test repo to /tmp/abgagt-*-test/
Registers it in the ABAP system (create)
Activates all test objects (pull --conflict-mode ignore)

On subsequent runs setup detects that the ABAP objects are already active and skips all registration/activation steps — adding only a few seconds overhead.

Test Suites

Suite	Command	What it tests	ABAP package
Unit (Jest)	`npm test`	JavaScript code, no ABAP needed	—
AUnit	`npm run test:aunit`	ABAP test classes	project package
Command	`npm run test:cmd`	All CLI commands end-to-end	project package (see note)
Pull workflow	`npm run test:pull`	Git ref switching (tags/branches)	`$ABGAGT_PULL_TEST`
Full pull	`npm run test:full-pull`	Pull without `--files`	`$ABGAGT_PULL_TEST`
Conflict detection	`npm run test:conflict`	Conflict abort / ignore modes	`$ABGAGT_PULL_TEST`
Sync-XML	`npm run test:sync-xml`	`--sync-xml` flag end-to-end	`$ABGAGT_PULL_TEST`
XML-only	`npm run test:xml-only`	TABL/DTEL/TTYP via `.xml` files	`$ABGAGT_PULL_TEST`
Drop	`npm run test:drop`	`drop` command per object type	`$ABGAGT_DROP_TEST`
Customize	`npm run test:customize`	`customize` command (C-class tables)	`$ABGAGT_CUS_TST`
Lifecycle	`npm run test:lifecycle`	`init → create → import → delete`	`$ABGAGT_LIFECYCLE_TEST`
Run	`npm run test:cmd -- --command=run`	`run --program` and `run --class`	`$ABGAGT_RUN_TEST`
Exclude	`npm run test:exclude`	`exclude` command — pattern + unsupportedTypes filtering	`$ABGAGT_EXCL_TEST`
Syntax (FUGR)	`npm run test:cmd -- --command=syntax`	FUGR function group syntax checks	`$ABGAGT_SYNTAX_TEST`
JUnit output	`npm run test:junit`	`--junit-output` flag	`$ABGAGT_UNIT_TEST`
Debug	`npm run test:debug`	Breakpoint management (ADT)	`$ABGAGT_DEBUG_TEST`
HANA	`npm run test:hana`	`hana view/schema/table` commands against a real HANA tenant	— (HANA-only repo; no ABAP package)
DTEL SCRLEN	`npm run test:dtel-scrlen`	`pull` correctly fails (exit 1, ❌) and surfaces the cause when a DTEL’s SCRTEXT exceeds its SCRLEN	`$ABGABT_PULL_TEST` (uses fixture branch `test/dtel-scrlen` of `abgagt-pull-test`)

Note: Most command tests use objects in the project package. Exceptions: run requires $ABGAGT_RUN_TEST, syntax FUGR tests require $ABGAGT_SYNTAX_TEST, junit requires $ABGAGT_UNIT_TEST, debug requires $ABGAGT_DEBUG_TEST, and exclude requires $ABGAGT_EXCL_TEST — all are set up automatically by the setup phase. The hana suite is different: its test repo (abgagt-hta-test) holds only HANA artifacts (.calculationview etc.), so setup just clones it — the HANA tenant connection comes from your .abapGitAgent and the test deploys/drops via the hana command itself.

Running a single command suite

npm run test:cmd -- --command=syntax   # syntax command only
npm run test:cmd -- --command=pull     # pull command only
npm run test:cmd -- --command=view
npm run test:cmd -- --command=debug
# etc.

Test Repo Access Requirements

The integration tests write to the test repos during execution (branch creation, pushes, abapGit registrations). This means:

Read access is not enough — gitUsername/gitPassword must have push access to the configured test repo URLs.
The github.com/SylvosCai/abgagt-* repos are public reference implementations. They are not writable by external contributors, so you cannot run the full integration suite against them directly.
You must fork the repos (or host your own copies) and configure your forks via testRepos in .abapGitAgent.

Forking the reference repos

Fork each repo on GitHub into your own account or organisation:
- github.com/SylvosCai/abgagt-pull-test
- github.com/SylvosCai/abgagt-drop-test
- github.com/SylvosCai/abgagt-customize-test
- github.com/SylvosCai/abgagt-lifecycle-test
- github.com/SylvosCai/abgagt-debug-test
- github.com/SylvosCai/abgagt-run-test
- github.com/SylvosCai/abgagt-syntax-test
- github.com/SylvosCai/abgagt-unit-test
- github.com/SylvosCai/abgagt-pull-type-test
- github.com/SylvosCai/abgagt-exclude-test
- github.com/SylvosCai/abgagt-hta-test
Add a testRepos section to .abapGitAgent pointing to your forks (this file is gitignored and never committed):

{
  "testRepos": {
    "pull":      "https://github.com/your-org/abgagt-pull-test.git",
    "drop":      "https://github.com/your-org/abgagt-drop-test.git",
    "customize": "https://github.com/your-org/abgagt-customize-test.git",
    "lifecycle": "https://github.com/your-org/abgagt-lifecycle-test.git",
    "debug":     "https://github.com/your-org/abgagt-debug-test.git",
    "run":       "https://github.com/your-org/abgagt-run-test.git",
    "syntax":    "https://github.com/your-org/abgagt-syntax-test.git",
    "unit":      "https://github.com/your-org/abgagt-unit-test.git",
    "pullType":  "https://github.com/your-org/abgagt-pull-type-test.git",
    "exclude":   "https://github.com/your-org/abgagt-exclude-test.git",
    "hana":      "https://github.com/your-org/abgagt-hta-test.git"
  }
}

Any key that is absent falls back to the public default URL, so you only need to specify repos you are overriding.

Ensure gitUsername/gitPassword in .abapGitAgent have push access to all configured repos.

Skipping Auto-Setup

If you manage test prerequisites manually (e.g. in CI where setup already ran in a prior stage), pass --no-setup to skip the setup phase:

npm run test:all -- --no-setup

CI Environment Variables

In CI pipelines you can override git credentials for test repos that live under a different account than the main project:

Variable	Purpose
`TEST_GIT_USR`	Git username injected into test repo `.abapGitAgent`
`TEST_GIT_PSW`	Git password/token injected into test repo `.abapGitAgent`

Troubleshooting

Setup fails with “cannot clone repo”

Check that the repo URL is reachable and your gitUsername/gitPassword have read access.
If using testRepos overrides, verify the URL is correct.
Stale clones at /tmp/abgagt-*-test/ with a broken .git/ are auto-detected and re-cloned.

Setup fails with “create failed”

The ABAP system may not have the required package (e.g. $ABGAGT_DROP_TEST). The $ prefix means it is a local (non-transportable) package — create it manually in SE80 or via abapgit-agent tree --package '$'.
Check that the ABAP backend version supports all commands (abapgit-agent health).

Pull / drop / customize tests fail after setup succeeded

Run setup standalone to verify the key object is active:

npm run test:setup

If it reports already active but tests still fail, the object may exist but be in an error state. Use abapgit-agent inspect or SE80 to investigate.

Lifecycle test: “create failed”

The lifecycle test creates and deletes an abapGit repo registration on each run. If a previous run crashed mid-test the registration may already exist. The lifecycle-runner.js detects this and calls delete automatically before re-running create.

Debug tests: breakpoint line numbers wrong

Debug tests use hardcoded ADT-verified line numbers. If the abgagt-debug-test objects were modified after the test was written, Group A tests will fail. Re-run setup (npm run test:setup) to ensure the latest object versions are active, then check the failing test’s expected line against view --full --lines output.