hurlfmt producing repeated content

[CodeMan99]

What is the current bug behavior? 🪲

When using hurlfmt sometimes a variable placeholder is repeated in the output.

Applies only to --out hurl formatting. Using --out json or --out html do not produce the incorrect behavior.

Steps to reproduce 🔗

Source file, called get_hurl.hurl:

GET https://github.com/Orange-OpenSource/hurl/releases/latest
Accept: application/json
[Options]
location: true
HTTP 200
[Captures]
version: jsonpath "$.tag_name"


GET https://github.com/Orange-OpenSource/hurl/releases/download/{{version}}/hurl_{{version}}_amd64.deb
[Options]
location: true
output: ./hurl_{{version}}_amd64.deb

I then called cargo run --package hurlfmt get_hurl.hurl, which produced:

vscode ➜ /workspaces/hurl (master) $ cargo run --package hurlfmt get_hurl.hurl 
    Finished dev [unoptimized + debuginfo] target(s) in 0.06s
     Running `target/debug/hurlfmt get_hurl.hurl`
GET https://github.com/Orange-OpenSource/hurl/releases/latest
Accept: application/json
[Options]
location: true
HTTP 200
[Captures]
version: jsonpath "$.tag_name"


GET https://github.com/Orange-OpenSource/hurl/releases/download/{{version}}/hurl_{{version}}_amd64.deb
[Options]
location: true
output: ./hurl_{{version}}./hurl_{{version}}_amd64.deb

Noticed the last output line where ./hurl_{{version}} is stated twice.

What is the expected correct behavior? ✔️

The input file seems to be mostly formatted correctly. I would only expect colors to be applied. That said, I'm new here! 👋

Execution context 🖥️

• Hurl Version (hurlfmt --version): hurlfmt 4.3.0-SNAPSHOT @ 0931c32

vscode ➜ /workspaces/hurl (master) $ git log -1 --oneline
0931c3251095 (HEAD -> master, origin/master, origin/HEAD) Add experimental parallel flag.
vscode ➜ /workspaces/hurl (master) $ uname -a
Linux 1443c8565104 5.15.133.1-microsoft-standard-WSL2 #1 SMP Thu Oct 5 21:02:42 UTC 2023 x86_64 GNU/Linux
vscode ➜ /workspaces/hurl (master) $ devcontainer-info 

Development container image information

- Image version: 1.0.9
- Definition ID: rust
- Variant: bookworm
- Source code repository: https://github.com/devcontainers/images
- Source code release/branch: v0.3.32
- Timestamp: Tue, 05 Mar 2024 18:10:09 GMT

More info: https://github.com/devcontainers/images/tree/main/src/rust/history/1.0.9.md

Possible fixes

❔

[CodeMan99]

In case I did something strange while creating my input file, here's a binary dump:

$ xxd get_hurl.hurl
00000000: 4745 5420 6874 7470 733a 2f2f 6769 7468  GET https://gith
00000010: 7562 2e63 6f6d 2f4f 7261 6e67 652d 4f70  ub.com/Orange-Op
00000020: 656e 536f 7572 6365 2f68 7572 6c2f 7265  enSource/hurl/re
00000030: 6c65 6173 6573 2f6c 6174 6573 740a 4163  leases/latest.Ac
00000040: 6365 7074 3a20 6170 706c 6963 6174 696f  cept: applicatio
00000050: 6e2f 6a73 6f6e 0a5b 4f70 7469 6f6e 735d  n/json.[Options]
00000060: 0a6c 6f63 6174 696f 6e3a 2074 7275 650a  .location: true.
00000070: 4854 5450 2032 3030 0a5b 4361 7074 7572  HTTP 200.[Captur
00000080: 6573 5d0a 7665 7273 696f 6e3a 206a 736f  es].version: jso
00000090: 6e70 6174 6820 2224 2e74 6167 5f6e 616d  npath "$.tag_nam
000000a0: 6522 0a0a 0a47 4554 2068 7474 7073 3a2f  e"...GET https:/
000000b0: 2f67 6974 6875 622e 636f 6d2f 4f72 616e  /github.com/Oran
000000c0: 6765 2d4f 7065 6e53 6f75 7263 652f 6875  ge-OpenSource/hu
000000d0: 726c 2f72 656c 6561 7365 732f 646f 776e  rl/releases/down
000000e0: 6c6f 6164 2f7b 7b76 6572 7369 6f6e 7d7d  load/{{version}}
000000f0: 2f68 7572 6c5f 7b7b 7665 7273 696f 6e7d  /hurl_{{version}
00000100: 7d5f 616d 6436 342e 6465 620a 5b4f 7074  }_amd64.deb.[Opt
00000110: 696f 6e73 5d0a 6c6f 6361 7469 6f6e 3a20  ions].location:
00000120: 7472 7565 0a6f 7574 7075 743a 202e 2f68  true.output: ./h
00000130: 7572 6c5f 7b7b 7665 7273 696f 6e7d 7d5f  url_{{version}}_
00000140: 616d 6436 342e 6465 620a                 amd64.deb.

[jcamiel]

Hi @CodeMan99

Thanks for the detailed issue, it looks like a bug! Nice catch!

[CodeMan99]

@jcamiel

Best I tell it seems to be triggered by having any text after the variable placeholder.

[jcamiel]

Thanks again for the report @CodeMan99 , this was more than a format issue. The parsing logic for filename with variables was buggy and should be fixed by #2545.

[CodeMan99]

Very nice! Code changes look sane to me. That said I have zero context as I've really only glanced at the format code.

I'm alright with Rust, but it takes more than a once-over to grasp :)