Introduce Config struct that holds parser configuration and implement #513
#677
Conversation
Old tests for that option that comes from xml-rs are removed
Old tests for that option that comes from xml-rs are removed
Old tests for that option that comes from xml-rs are removed
failures (ignored):
trim_text_end::true_
Failure is ignored for now because it is hard to fix it in current implementation
of a parser, but a new implementation coming soon, where that will be easy
failures:
check_end_names::true_::mismatched_tags
Fixed:
check_end_names::true_::mismatched_tags
failures:
dashes_in_comments
`dashes_in_comment` test repeat the `check_comments::true_` tests, so removed
… with access via `.config_mut()` accessor
text_trim_start and text_trim_end have retained their logical order
Codecov Report
❗ Your organization needs to install the Codecov GitHub app to enable full functionality. @@ Coverage Diff @@
## master #677 +/- ##
==========================================
+ Coverage 65.05% 65.16% +0.11%
==========================================
Files 38 38
Lines 17837 17851 +14
==========================================
+ Hits 11604 11633 +29
+ Misses 6233 6218 -15
Flags with carried forward coverage won't be shown. Click here to find out more.
|
|
|
||
| /// Self-closed elements should be reported as one `Empty` event | ||
| #[test] | ||
| fn false_() { |
There was a problem hiding this comment.
Perhaps "enabled" and "disabled" would be better names, to avoid keyword clashes?
There was a problem hiding this comment.
I chose those words because if we will have other options that are not booleans, tests would be named as a config value
| ); | ||
| assert_eq!( | ||
| reader.read_event().unwrap(), | ||
| Event::Comment(BytesText::new(" comment \t\r\n")) |
There was a problem hiding this comment.
Do we have any reason to support trimming the text values of "comments"? I cannot immediately think of a reason to do that, but perhaps one exists.
There was a problem hiding this comment.
I think not, at least nobody request such feature. If such a request appears, we can add a separate option.
Generally speaking, I would delete the current trim options as they simply do not work correctly for text alternating with CDATA / comments / processing instructions, but I suppose that would break many users. I was thinking about renaming current Event into RawEvent and DeEvent to Event and give users stream of Events. The RawEvent then would be a low-level event which usually not needed by most users. That is very raw thoughts currently, so I decided to not do revolutional changes for now.
tests/reader-config.rs
Outdated
| #[test] | ||
| fn check_end_names_false() { | ||
| let mut reader = Reader::from_str("<root></root \t\r\n>"); | ||
| reader.trim_markup_names_in_closing_tags(false); |
There was a problem hiding this comment.
What is it the reason that we only have this for closing tags, attributes?
There was a problem hiding this comment.
I think, that this is optimization option. Usually end tags does not contain spaces before >, so if we will assume that the name ends immediately before the >, we could save some time. Such optimisation make sense only for the closing tags -- for opening tags we in any case should check if it has attributes and find the actual end of tag name.
| @@ -11,153 +11,189 @@ use crate::reader::state::ReaderState; | |||
|
|
|||
| use memchr; | |||
|
|
|||
| macro_rules! configure_methods { | |||
There was a problem hiding this comment.
I appreciate being able to ditch these macros.
fuzz/fuzz_targets/fuzz_target_1.rs
Outdated
| let config = reader.config_mut(); | ||
| config.expand_empty_elements = true; | ||
| config.trim_text(true); | ||
| config.trim_text_end = true; |
There was a problem hiding this comment.
I slightly dislike the aesthetics here and inability to chain (because of converting methods to members), but ultimately it's an incredibly minor detail
This PR changes a way how reader is configured. Instead of having builder methods in
Reader/NsReadernow they provide aconfig()/config_mut()methods:config()method introduces the ability to read parser configuration, that was not be possible until now;config_mut()method allow to set parser options. You can even to set all options at once or store them (theConfigstruct implementsSerialize/Deserializewhenserde-typesfeature is active)To ensure, that the behavior was not changed, this PR also introduces a new integration test-suite which tests all possible parser options. That allowed me to find a bug in
trim_text_endoption, but it is hard to fix it today. Because I plan to rewrite parser (that task is actually mostly done) I justignorethat test for now.As a nice bonus, after implementing new tests it become obvious how to implement #513, which was done.
Closes #513