A `{% filter %}` block could have an unbound number of filters chained
one after another. This chain would wind up as a nested structure on the
heap. When dropped, this would cause a stackoverflow.
Resolves <https://issues.oss-fuzz.com/issues/386547457>.
Not for all operations the nesting level was incremented when needed
and/or the un-incremented nesting level was used in subfunction calls.
Binary operators such as `*` did not properly increment the nesting
level.
This PR changes `Level` in such a way that it can be used to keep track
of the nesting level when used in a loop. It is now impossible to
accidentally refer to an old nesting level value.
Resolves <https://issues.oss-fuzz.com/issues/385256115>.
* refactor(parser): Consistently use 'i' in skip_till
* refactor(parser): Clarify second input in skip_till
* refactor(parser): Consistently use 'i' in Comment::parse::content
* refactor(parser): Clarify second input in Comment::parse::content
* refactor(parser): Rename ParseResult to InputParseResult
This leaves room for a new `ParseResult` that doesn't have the input as
we transition to the Winnow 0.5 API.
* refactor(parser): Temporarily switch from parse_next to parse_peek
In 0.4, `parse_next` and `parse_peek` are the same.
In 0.5, `parse_next`s signature will change.
By switching to `parse_peek` now, we can make it so we gradually adopt
the new `parse_next` signature.
* refactor(parser): Temporarilly add unpeeks
In 0.4, `unpeek` is a no-op.
In 0.5, the `FnMut` signature impled for `Parser` changes and `unpeek`
will take the old signature and adapt it to work.
This allows us to more gradually migrate to the new `FnMut` signature.
Yes, the name `unpeek` isn't ideal but its only transitional.
* refactor(parser): Upgrade to Winnow 0.5
Note: `tests::fuzzed_target_recursion` fails on debug builds because of
the extra overhead from the deprecated APIs adapting to their
non-deprecated forms
* refactor(parser): Update signature for 'identifier'
* refactor(parser): Update signature for 'bool_lit'
* refactor(parser): Update signature for 'num_lit_suffix'
* refactor(parser): Update body for 'num_lit::int_with_base'
* refactor(parser): Update signature for 'num_lit::float'
* refactor(parser): Update signature for 'num_lit'
* refactor(parser): Update signature for 'str_lit_without_prefix'
* refactor(parser): Update signature for 'str_lit'
* refactor(parser): Update signature for 'char_lit'
* refactor(parser): Update signature for 'Char::parse'
* refactor(parser): Update signature for 'path_or_identifier'
* refactor(parser): Update signature for 'filter'
* refactor(parser): Update signature for 'State::tag_block_start'
* refactor(parser): Update signature for 'State::tag_block_end'
* refactor(parser): Update signature for 'State::tag_comment_start'
* refactor(parser): Update signature for 'State::tag_comment_end'
* refactor(parser): Update signature for 'State::tag_expr_start'
* refactor(parser): Update signature for 'State::tag_expr_end'
* refactor(parser): Update signature for 'State::nest'
* refactor(parser): Update signature for 'Level::nest'
I had considered making the input a `&mut &str` but didn't really see
the benefit, especially at this stage.
That can always be changed later.
* refactor(parser): Update signature for 'skip_ws0'
* refactor(parser): Update signature for 'skip_ws1'
* refactor(parser): Update signature for 'separated_digits' closure
* refactor(parser): Update signature for 'skip_till' closure
* refactor(parser): Update signature for 'Target::lit'
* refactor(parser): Update signature for 'Target::unnamed'
* refactor(parser): Update signature for 'Target::named'
* refactor(parser): Update signature for 'Target::rest'
* refactor(parser): Update signature for 'collect_targets'
* refactor(parser): Update signature for 'Target::parse'
* refactor(parser): Update signature for 'Target::parse_one'
* refactor(parser): Update signature for 'Suffix::try'
* refactor(parser): Update signature for 'Suffix::call'
* refactor(parser): Update signature for 'Suffix::index'
* refactor(parser): Update signature for 'Suffix::attr'
* refactor(parser): Update signature for 'Suffix::macro'
* refactor(parser): Update signature for 'token_bitand'
* refactor(parser): Update signature for 'token_xor'
* refactor(parser): Update signature for 'Expr::char'
* refactor(parser): Update signature for 'Expr::num'
* refactor(parser): Update signature for 'Expr::str'
* refactor(parser): Update signature for 'Expr::path_var_bool'
* refactor(parser): Update signature for 'Expr::array'
* refactor(parser): Update signature for 'Expr::group'
* refactor(parser): Update signature for 'Expr::single'
* refactor(parser): Update signature for 'Expr::prefix'
* refactor(parser): Update signature for 'Expr::filtered'
* refactor(parser): Update signature for 'Expr::is_as'
* refactor(parser): Update signature for 'Expr::concat'
* refactor(parser): Update signature for 'expr_prec_layer'
* refactor(parser): Update signature for 'Suffix::parse'
* refactor(parser): Update signature for 'Expr::parse'
* refactor(parser): Update signature for 'Expr::named_argument'
* refactor(parser): Update signature for 'Expr::arguments'
* refactor(parser): Update signature for 'Expr::arguments' closure
* refactor(parser): Update signature for 'Comment::parse::tag
* refactor(parser): Update signature for 'Comment::parse::content
* refactor(parser): Update signature for 'Comment::parse'
* refactor(parser): Update signature for 'Extends::parse'
* refactor(parser): Update signature for 'Include::parse'
* refactor(parser): Update signature for 'Import::parse'
* refactor(parser): Update signature for 'Whitespace::parse'
* refactor(parser): Update signature for 'Lit::parse'
* refactor(parser): Update signature for 'Node::expr'
* refactor(parser): Update signature for 'Node::many'
* refactor(parser): Update signature for 'Node::parse'
* refactor(parser): Update signature for 'Node::parse_template'
* refactor(parser): Update signature for 'cut_node' closure
* refactor(parser): Avoid 'parse_peek' in 'cut_node'
* refactor(parser): Update signature for 'unexpected_raw_tag'
* refactor(parser): Update signature for 'unexpected_tag'
* refactor(parser): Update signature for 'check_end_name'
* refactor(parser): Update signature for 'CondTest::parse_cond'
* refactor(parser): Update signature for 'check_block_start'
* refactor(parser): Update signature for 'Loop::parse::content'
* refactor(parser): Update signature for 'Macro::parse::parameters'
* refactor(parser): Update signature for 'Loop::parse::else_block'
* refactor(parser): Update signature for 'Loop::parse::body_and_end'
* refactor(parser): Reuse built-in empty parser
* refactor(parser): Update signature for remaining '*::parse'
* refactor(parser): Resolve remaining deprecations
In [#131] I added a check that ensures that the section in `Cargo.toml`
(and there values) are in a specific order. I invented the order quite
randomly, in the order which I thought might be useful.
This PR changes the order to be the same as in Cargo's reference about
[The Manifest Format]. Also some missing softlinks were added.
[#131]: <https://github.com/rinja-rs/rinja/pull/131>
[The Manifest Format]: <https://doc.rust-lang.org/cargo/reference/manifest.html>
* refactor(parser): In 'filter', name a checkpoint
* refactor(parser): In 'Target::named', name a checkpoint
* refactor(parser): In 'Target::named', remove a checkpoint
* refactor(parser): In 'Target::rest', name a checkpoint
* refactor(parser): In 'Target::parse_one', name a checkpoint
* refactor(parser): In 'Target::parse', align the input
* refactor(parser): In 'Expr::is_as', name a checkpoint
* refactor(parser): In 'Expr::concat', name a checkpoint
* refactor(parser): In 'Suffix::parse', name a checkpoint
* refactor(parser): In 'Node::break', name a checkpoint
* refactor(parser): In 'Node::continue', name a checkpoint
* refactor(parser): In 'Macro::parse', name a checkpoint
* refactor(parser): In 'When::else', name a checkpoint
The fuzzing we inherited from askama only tested the parser. In rinja we
parse multiple aspects of the project in a unified crate: `fuzzing`.
Everything the old fuzzer did is included in the new fuzzer, and more.
* refactor(parser): Rely on Parser trait rather than FnMut
* refactor(parser): Consistently use ParseResult
* refactor(parser): Accept Parser rather than FnMut
* refactor(parser): Chain the input through
* refactor(parser): Accept stateful Parser rather than stateless FnMut
* refactor(parser): Remove redundant closures
