| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
default-input-method is only the default choice when the user is
activating an input method, while evil-input-method is what evil
uses in insert-state.
|
|
We need to be able to set input methods locally (see evil-search.el) and
globally (see following commit), so the macro should only deactivate the
evil's input method hooks and leave the setting to be done in the body.
|
|
|
|
|
|
|
|
If point is on a match but not on its first character, then backward
search in Vim finds this match. However, standard Emacs search
functions (in particular `re-search-backward`) only match if the end
of the match is *before* the starting point. This change modifies
Evil's search function so that it behaves like vim using an additional
search step.
|
|
The new custom variable `evil-symbol-word-search` determines whether
the word search function family (bound to `*` and `#`) looks for the
symbol (if non-nil) or word (if nil) under point. The variable is
buffer-local so it can be set appropriately for each major-mode. The
default value is `nil` to mimic Vim's default behaviour.
|
|
Word search now uses the buffer's definition of a "word" to determine
the boundaries of the object at point to be searched for. This is
consistent with the generated search pattern that uses word boundaries
\< ... \> and not symbol boundaries \_< ... \_> due to a recent
change 6d3d4df57e81.
This addresses #354.
|
|
|
|
|
|
|
|
The new functions `evil-ex-make-search-pattern` and
`evil-ex-make-substitute-pattern` are responsible to create patterns
w.r.t. to flags and global options. Ensure that all replacement
functions respect `case-replace` and `case-fold-search`, both
variables are set according to the pattern's case flags.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
(fix #299).
`isearch-filter-predicate` became a list `isearch-filter-predicates`.
|
|
|
|
|
|
|
|
This function is called from change-hooks and these hooks are supposed
to preserve the match-data according to the Elisp manual.
|
|
|
|
Setting this option to non-nil reverses the meaning of the 'g' flag in
:substitute.
|
|
|
|
If the "r" flag is present and the pattern is empty, then either the
previous search pattern or the previous substitute pattern should be
used, whichever is more recent.
Furthermore this flag is implicitly set for :substitute commands with
argument (i.e. with pattern/replacement part).
This fixes #252.
|
|
|
|
|
|
|
|
|
|
The ex-range can be an arbitrary range. The using code must not make
any assumption on the type of the range or its expansion state.
|
|
The headers of all source files are updated to be more standard
conform and include the license information. Furthermore the license
is put in the COPYING file.
|
|
Zero size matches are invisible but their substitute may not be.
Therefore empty overlays are created in this case.
|
|
|
|
This ensures that the search module is initialized correctly even if
the variable had been set before evil is loaded.
|
|
This addresses issue #148.
|
|
The pattern is added to the search history if and only if it does not
equal the first pattern in the history.
This addresses issue #148.
|
|
Formerly we used `evil-ex-lazy-highlight' for both, search and
substitute. Now it is only used for search and substitute uses face
`evil-ex-substitute-matches' so both can be configured separately.
This addresses issue #147.
|
|
More descriptive name.
|
|
Even if the search failed the search pattern should be stored for
subsequent search commands like 'n', e.g. in another buffer. The
pattern is not stored if the search is cancel with C-g.
This addresses issue #147.
|
|
If the search pattern for some highlight is empty then all old
overlays should be removed.
|
|
If the search pattern is empty and there was no previous search then
an error has been raised because `evil-ex-search-pattern' is nil in
this case. Now the condition is checked appropriately.
|
|
Ensure that point remains unchanged when the search fails, even if a
prefix of the search string matched. Cf. commits 568480b and 6302c05.
|
|
If the search pattern may match the empty string and the search during
the update of the highlighting reaches the end of the buffer then the
update went to an infinite loop. This change ensure highlighting is
stopped when reaching the end of the buffer.
This addresses issue #140.
|
|
The new behavior is to always use the global value of `case-replace'
and do not set it depending on the case setting of the search pattern.
This addresses #140.
|
