Fix panic when reconstructing Dockerfile from malformed image history

[0xVijay]

What

The code that rebuilds a Dockerfile from image history (reverse.go) indexes the result of strings.SplitN for HEALTHCHECK, ENTRYPOINT and CMD without checking the length. A malformed value in a layer's CreatedBy history makes it panic, for example:

HEALTHCHECK &{[                 -> index out of range [1] with length 1
HEALTHCHECK &{[CMD] 5s 10s 0s   -> index out of range [3] with length 3
ENTRYPOINT [   /   CMD [        -> slice bounds out of range [1:0]

There is no recover() around this, so one bad instruction aborts the whole xray, build or profile run.

Why

mint is often pointed at images it did not build (xray/build on pulled images, CI pipelines). A single image with malformed history metadata should not crash the tool. xray does not even run the target app, so the history alone is enough to trigger it.

How Tested

Added TestHealthCheckMalformed and TestDockerfileFromHistoryDataMalformed which feed the malformed values above and assert the parser returns an error instead of panicking. Existing TestHealthCheck still passes.

go test ./pkg/docker/dockerfile/reverse/
ok  github.com/mintoolkit/mint/pkg/docker/dockerfile/reverse

[kcq]

auggie review

[augmentcode]

🤖 Augment PR Summary

Summary: Prevents crashes when reconstructing a Dockerfile from images with malformed history entries.

Changes:

• Hardened JSON-array reconstruction for CMD/ENTRYPOINT by guarding against short/empty strings.
• Added defensive length checks in deserialiseHealtheckInstruction to avoid out-of-range panics when parsing malformed HEALTHCHECK encodings.
• Adjusted retries parsing to avoid indexing into an empty retries value.
• Added regression tests asserting malformed history strings do not panic during HEALTHCHECK parsing or full Dockerfile reconstruction.

Technical Notes: Malformed inputs now return errors instead of panicking, allowing callers to continue processing other history layers.

_{🤖 Was this summary useful? React with 👍 or 👎}

[augmentcode]

Review completed. 1 suggestion posted.

Comment augment review to trigger a new review at any time.

[augmentcode]

deserialiseHealtheckInstruction now returns an error here but still returns strTest (often empty) as the reconstructed instruction; upstream code currently assigns the returned string even when err != nil, which can silently drop the original HEALTHCHECK line and emit a blank/invalid Dockerfile line instead of “skipping”.

Severity: medium

Other Locations

• pkg/docker/dockerfile/reverse/reverse.go:819

_{🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.}