eclipse-aascw / aas-core-codegen / 39383890623

90%

Build Type

push

github

Committed by web-flow

Commit Message

Preserve encoded characters in Go regexes (#855)

We fix the Go renderer to preserve the literal meaning of explicitly
encoded characters in constant regex strings.

Previously, the renderer wrote these characters directly into the regex
pattern, potentially causing encoded metacharacters to be interpreted as
regex syntax.

This problem affects V3.2's `matches_reference_identifier`, which
contains the following ranges:

```regex
\x23-\x5B\x5D-\uD7FF
```

These ranges represent characters from `#` through `[` and from `]`
through U+D7FF.

The old renderer produces `#-[]-` within the pattern. Go interprets
`]` as the end of the character class rather than as a literal range
endpoint. As a result, the regex compiles successfully but rejects
valid reference identifiers.

The corrected renderer preserves explicitly encoded characters using
Go's supported `\x{...}` notation:

```regex
\x{23}-\x{5b}\x{5d}-\x{d7ff}
```

The changes also include focused renderer tests covering encoded
punctuation, bracket range endpoints, Unicode ranges, and ordinary
escaped characters. The affected Go verification output for the
existing AAS test metamodel has been regenerated.

This PR addresses the regex-escaping problem tracked in #840.

## Runtime demonstration

The metamodel explicitly documents two valid examples in
[`matches_reference_identifier`]:

[`matches_reference_identifier`]: https://github.com/eclipse-aascw/aas-core-meta/blob/560851dc9/aas_core_meta/v3_2.py#L11083

```text
$sm("https://example.com/sm")#idShort
$sme("https://example.com/sm").someProperty#value
```

The following demonstration uses the complete regexes emitted by the
old and corrected generators, without adding anchors or quantifiers.

Save the program as `main.go` and execute it with:

```shell
go run main.go
```

Both examples produce:

```text
old=false, corrected=true, metamodel=true
```

The `metamodel` value is a hardcoded expected result, independently
verifi... (continued)

Coverage Stats

1 of 1 new or added line in 1 file covered. (100.0%)

2 existing lines in 1 file now uncovered.

45908 of 50905 relevant lines covered (90.18%)

2.71 hits per line