The presence of a trailing comma in the single and only input/output
declaration confused the parser and made zig fmt discard any element
placed after the comma.
Instead of having one function per precedence level, mirroring the BNF grammar,
this uses the "precedence climbing" algorithm with a table of operator
precedences (and a few special cases that don't fit into the table as it is).
This is a first pass -- it's probably possible to put more of the parser
into this form, e.g. to support prefix/suffix operators with precedence, if
necessary, or just to simplify the code more.
It may also be possible to speed this up by putting more useful information
into the tokens during tokenization, to avoid the extra branch on the token in
operInfo.
rowSize used to return null if all the elements were placed on the same
line as the right brace, making the rendering logic skip the whole set
of elements.
Given the usage of rowSize let's just drop the null and always return
the number of elements.
Fixes#8423
Bring this in line with how variable declarations are handled.
Open a new indentation level for the initialization expression to handle
nested expressions like blocks.
Closes#7618
The main realization here was that getting rid of the early returns
in renderWhile() and rewriting the logic into a mostly unified execution
path took things from ~200 lines to ~100 lines and improved consistency
by deduplicating code.
Also add several test cases and fix a few issues along the way:
Fixes https://github.com/ziglang/zig/issues/6114
Fixes https://github.com/ziglang/zig/issues/8022
Add failing testcase to reproduce issue 8088
Tidy up renderWhile(), factoring out renderWhilePayload()
Ensure correct newline is used before 'then' token in while/for/if
Handle indents for 'if' inside 'for' or 'while'
Stop special-casing 'if' compared to 'for' and 'while'
The modification to the grammar in the comment is in line with the
grammar in the zig-spec repo.
Note: checking if the previous token is a colon is insufficent to tell
if a block has a label, the identifier must be checked for as well. This
can be seen in sentinel terminated slicing: `foo[0..1:{}]`
Currently `// zig fmt: off` does not work as there are two spaces
after the `//` instead of one. This can cause confusion, so allow
arbitrary whitespace before the `zig fmt: (off|on)` in the comment but
trim this whitespace to the canonical single space in the output.
This regresses the test case of `zig fmt` deleting empty line comments.
Two open questions here:
* What should the rules be about deleting empty line comments?
It makes sense usually, but for array initization, empty line
comments cause a line break, affecting the row/column alignment.
Perhaps we should therefore respect all empty line comments?
Or should we special case array initializations?
* If we decide to special case some kinds of line comments to respect
them (which is status quo!), how should that be implemented?
I modified this test case to expect different results.
Now, the trailing comma on a list of struct fields is the only deciding
factor, not whether or not the field init expressions contain a newline.
