issue better error message on encountering old-style code chunk options

[rundel]

Chunks with old style options are mangled

An executable code block {r eval=FALSE} is re-written as {.{r} eval="FALSE"}.

$ printf -- '```{r eval=FALSE}\nx = 1\n```\n' | cargo run --bin pampa -- -t qmd
Warning: [Q-2-8] Code block options in header
. . .
```{.{r} eval="FALSE"}
x = 1
```

$ printf -- '```{r eval=FALSE}\nx = 1\n```\n' | cargo run --bin pampa --     
Warning: [Q-2-8] Code block options in header
. . .
[ CodeBlock ( "" , ["{r}"] , [("eval", "FALSE")] ) "x = 1" ]%    

---

Table caption attributes are not written

$ printf '%s\n' '| A | B |' '|---|---|' '| 1 | 2 |' '' ': ABCD {tbl-colwidths="[30,70]"}'
| A | B |
|---|---|
| 1 | 2 |

: ABCD {tbl-colwidths="[30,70]"}

$ printf '%s\n' '| A | B |' '|---|---|' '| 1 | 2 |' '' ': ABCD {tbl-colwidths="[30,70]"}' | cargo run --bin pampa -- -t qmd
| A   | B   |
| --- | --- |
| 1   | 2   |

: ABCD

$ printf '%s\n' '| A | B |' '|---|---|' '| 1 | 2 |' '' ': ABCD {tbl-colwidths="[30,70]"}' | cargo run --bin pampa --
[ Table ( "" , [] , [("tbl-colwidths", "[30,70]")] ) (Caption Nothing [ Plain [Str "ABCD"] ]) [(AlignDefault, ColWidthDefault), (AlignDefault, ColWidthDefault)] (TableHead ( "" , [] , [] ) [Row ( "" , [] , [] ) [Cell ( "" , [] , [] ) AlignDefault (RowSpan 1) (ColSpan 1) [Plain [Str "A"]] , Cell ( "" , [] , [] ) AlignDefault (RowSpan 1) (ColSpan 1) [Plain [Str "B"]] ] ]) [TableBody ( "" , [] , [] ) (RowHeadColumns 0) [] [Row ( "" , [] , [] ) [Cell ( "" , [] , [] ) AlignDefault (RowSpan 1) (ColSpan 1) [Plain [Str "1"]] , Cell ( "" , [] , [] ) AlignDefault (RowSpan 1) (ColSpan 1) [Plain [Str "2"]] ] ]] (TableFoot ( "" , [] , [] ) [] ) ]%   

[cscheid]

An executable code block {r eval=FALSE} is re-written as {.{r} eval="FALSE"}.

I'm sorry to report to you that this is no longer going to be valid syntax going forward. We plan to have helpers for people to convert, but I think we should actually be flagging that as a syntax error instead of a warning, because that round-tripping looks terrible.

table attributes

Hm. This looks like a bug. I'll ask Claude for a diagnosis and we'll take the fix design from there.

[cscheid]

Bugfix for table round-trip incoming on #154. I'm not sure what to do with the other one yet.

[cscheid]

Keeping this open for the first half of the issue, but the table roundtrip should work now on main. Thanks for the report!

[rundel]

Sounds good, removal of the chunk options makes sense in the longer term and I had already noticed that labels included in the {} resulted in an error (slightly more sad to see those going as I prefer the conciseness over #| label: but simplifying the parser is the greater good).

Maybe part of this fix would be to make the error thrown for those clearer? Not sure it needs to be label specific

printf '```{r test}\n1+1\n```\n' | cargo run --bin pampa -- -t qmd
Error: Parse error
   ╭─[ <stdin>:1:6 ]
   │
 1 │ ```{r test}
   │      ──┬──  
   │        ╰──── unexpected character or token here
───╯
```

[cscheid]

Yes, I think we might be able to do much better in that error using our tree-sitter LALR table error message system. Thanks for the idea, I'll come back to this in a few days or weeks.