Backtrack one char if the global/char-literal var matcher hits
inside a string. The next char could be the beginning of an
expression expansion.
* test/automated/ruby-mode-tests.el
(ruby-interpolation-after-dollar-sign): New test.
* lisp/progmodes/ruby-mode.el (ruby-align-chained-calls): New option.
(ruby-smie-grammar): Make "." right-associative. Make its priority
lower than the ternary and all binary operators.
(ruby-smie-rules): Indent "(" relative to the first non-"."
parent, or the first "." parent at indentation. Use
`ruby-align-chained-calls' for indentation of "." tokens.
* test/automated/ruby-mode-tests.el (ruby-align-chained-calls):
New test.
Fixes: debbugs:16593
option.
(ruby-smie--indent-to-stmt-p): Use it.
(ruby-smie-rules): Revert the logic in the handling of `when'.
Expand the `begin' clause to handle `ruby-align-to-stmt-keywords'.
(ruby-deep-arglist, ruby-deep-indent-paren)
(ruby-deep-indent-paren-style): Update docstrings to note that the
vars don't have any effect with SMIE.
* test/automated/ruby-mode-tests.el: Add tests for
`ruby-align-to-stmt-keywords'.
* test/indent/ruby.rb: Update examples to reflect the lack of change in
default indentation of `begin' blocks.
Fixes: debbugs:16182
* lisp/progmodes/ruby-mode.el (ruby-mode-syntax-table): Don't modify
syntax for `?'.
(ruby-expr-beg): Expect that `!' will have syntax class "symbol"
where appropriate already.
(ruby-syntax-propertize-function): Propertize `?' and `!' at the
end of method names.
(ruby-smie--backward-token): Tokenize heredocs as semicolons.
* test/automated/ruby-mode-tests.el: Remove outdated comment.
* test/indent/ruby.rb: Add a statement on the line after heredoc.
specific in what the first arg can be: a non-keyword word,
string/regexp/percent literal opener, opening paren, or unary
operator followed directly by word.
* test/automated/ruby-mode-tests.el (ruby-toggle-block-to-brace): Fix
the test, in respect to adding the space after the curly.
iuwu-mod token.
(ruby-smie--implicit-semi-p): Prohibit implicit semicolon after
hanging iuwu-mod token.
(ruby-smie--forward-token): Do not include a dot after a token in
that token.
(ruby-smie--backward-token): Likewise.
middle of block statement initially, lower the depth. Remove
FIXME comment, not longer valid. Remove middle of block statement
detection, no need to do that anymore since we've been using
`ruby-parse-region' here.
function, checks if expression expansion is allowed in given parse
state.
(ruby-syntax-propertize-expansion): Use it.
(ruby-syntax-propertize-function): Bind `case-fold-search' to nil
around the body.
* test/automated/ruby-mode-tests.el: New tests, for percent literals
and expression expansion.
start at point, so that expansion starting right after opening
slash in a regexp is recognized.
(ruby-syntax-before-regexp-re): New defvar, extracted from
ruby-syntax-propertize-function. Since the value of this regexp
is looked up at runtime now, we should be able to turn
`ruby-syntax-methods-before-regexp' into a defcustom later.
(ruby-syntax-propertize-function): Split regexp matching into two
parts, for opening and closing slashes. That allows us to skip
over string interpolations and support multiline regexps.
Don't call `ruby-syntax-propertize-expansions', instead use another rule
for them, which calls `ruby-syntax-propertize-expansion'.
(ruby-syntax-propertize-expansions): Move `remove-text-properties'
call to `ruby-syntax-propertize-function'.
(ruby-syntax-propertize-expansion): Extracted from
`ruby-syntax-propertize-expansions'. Handles one expansion.
(ruby-syntax-propertize-heredoc): Explicitly call
`ruby-syntax-propertize-expansions'.
(ruby-syntax-propertize-percent-literal): Leave point right after
the percent symbol, so that the expression expansion rule can
propertize the contents.
* test/automated/ruby-mode-tests.el (ruby-heredoc-highlights-interpolations)
(ruby-regexp-skips-over-interpolation)
(ruby-regexp-continues-till-end-when-unclosed)
(ruby-regexp-can-be-multiline)
(ruby-interpolation-inside-percent-literal): New tests.
* test/indent/ruby.rb: Add multiline regexp example.
(ruby-end-of-defun, ruby-move-to-block): Bind `case-fold-search'
to nil.
(ruby-end-of-defun): Remove the unused arg, change the docstring
to reflect that this function is only used as the value of
`end-of-defun-function'.
(ruby-beginning-of-defun): Remove "top-level" from the docstring,
to reflect an earlier change that beginning/end-of-defun functions
jump between methods in a class definition, as well as top-level
functions.
of accuracy. `ruby-block-contains-point' is relatively slow, so
only use it for method and singleton class blocks.
* test/automated/ruby-mode-tests.el
(ruby-add-log-current-method-after-inner-class): Lower
expectations: move point inside a method, initially.
depth for unfinished percent literal. Not using it in the caller.
(ruby-move-to-block): Jump over multiline literals of all types,
ignoring code-looking contents inside them.
* test/automated/ruby-mode-tests.el
(ruby-move-to-block-skips-percent-literal): Add depth-affecting
bits inside the examples.
(ruby-move-to-block-skips-heredoc): New test.
performance. Instead of recalculating indentation fully for each
line, sum up indentation depth based only on visited lines.
(ruby-parse-partial): Increase the depth after "do" even when END
is right after it.
(ruby-parse-partial): When END is in the middle of a percent
literal, increase the depth if the delimiter chars belong to the
paren syntax class.
* test/automated/ruby-mode-tests.el
(ruby-move-to-block-skips-percent-literal): New test.
temporarily) broken indentation.
* automated/ruby-mode-tests.el (ruby-block-test-example):
Break indentation of the do block opener and add a line inside it.
* automated/ruby-mode-tests.el (works-on-do, ok-with-three):
Adjust line numbers.
"module" and "def" to have indentation before them. Regression
from 2012-09-07T04:15:56Z!dgutov@yandex.ru (see the new test).
* test/automated/ruby-mode-tests.el
(ruby-indent-after-block-in-continued-expression): New test.
Extract `ruby-syntax-propertize-expansions'.
(ruby-syntax-propertize-expansions): Only change syntax on
certain string delimiters, to punctuation. This way the common
functions like forward-word and thing-at-point still work.
(ruby-match-expression-expansion): Improve readability.
(ruby-block-contains-point): New function.
(ruby-add-log-current-method): Handle several edge cases.
* test/automated/ruby-mode-tests.el
Rename one interpolation test; add three more.
(ruby-with-temp-buffer): New macro, use it where appropriate.
(ruby-add-log-current-method-examples): Use "_" for target point.
Add four tests for ruby-add-log-current-method.
Extract `ruby-syntax-propertize-expansions'.
(ruby-syntax-propertize-expansions): Only change syntax on
certain string delimiters, to punctuation. This way the common
functions like forward-word and thing-at-point still work.
* test/automated/ruby-mode-tests.el
Rename one interpolation test; add three more.
(ruby-syntax-propertize-function): After everything else, search
for expansions in string literals, mark their insides as
whitespace syntax and save match data for font-lock.
(ruby-font-lock-keywords): Highlight just the 2nd group from
expression expansion matches.
(ruby-match-expression-expansion): Use the match data saved to the
text property in ruby-syntax-propertize-function.
* test/automated/ruby-mode-tests.el
Change direct font-lock face references to var references.
(ruby-interpolation-suppresses-syntax-inside): New test.
(ruby-interpolation-inside-percent-literal-with-paren): New
failing test.
start/end keyword a bit harder. Works with different values of N.
Add more comments.
(ruby-end-of-block): Update accordingly.
* test/automated/ruby-mode-tests.el (ruby-heredoc-font-lock)
(ruby-singleton-class-no-heredoc-font-lock)
(ruby-add-log-current-method-examples): New tests.
(ruby-test-string): Extract from ruby-should-indent-buffer.
(ruby-deftest-move-to-block): New macro.
Add several move-to-block tests.
period before class method names, not after. Remove handling of
one impossible case. Add comments.
* test/automated/ruby-mode-tests.el
(ruby-add-log-current-method-examples): New test.
(ruby-test-string): Extract from ruby-should-indent-buffer.
font-lock the beginning of singleton class as heredoc.
* test/automated/ruby-mode-tests.el (ruby-heredoc-font-lock)
(ruby-singleton-class-no-heredoc-font-lock): New tests.
not just expect to be at its beginning. Adjust callees.
Succeed when do-end block has no space before the pipe character.
(ruby-brace-to-do-end): When the original block is one-liner,
convert to multiline. Reindent the result.
* test/automated/ruby-mode-tests.el:
(ruby-toggle-block-to-multiline): New test.
(ruby-should-indent-buffer, ruby-toggle-block-to-do-end)
(ruby-toggle-block-to-brace): Use buffer-string.
ruby-beginning-of-indent, simplify, allow all keywords to have
indentation before them.
(ruby-beginning-of-indent): Adjust for above. Search until the
found point is not inside a string or comment.
(ruby-font-lock-keywords): Allow symbols to start with "@"
character, give them higher priority than variables.
(ruby-syntax-propertize-function)
(ruby-font-lock-syntactic-keywords): Remove the "not comments"
matchers. Expression expansions are not comments when inside a
string, and there comment syntax status is irrelevant.
(ruby-match-expression-expansion): New function. Check that
expression expansion is inside a string, and it's not escaped.
(ruby-font-lock-keywords): Use it.
* test/automated/ruby-mode-tests.el: New tests (Bug#11613).
