This commit takes into account also the case in which a prime follows
a superscript, a case not explicitly mentioned in bug 1633.
This case has to be accounted for in a different way. We cannot look
ahead when exporting, so we simply tell TeXMathStream to output an
empty group if we just output a superscript and a prime comes next.
Some lfuns are handled directly in GuiApplication while handling
of menu hiding in fullscreen was done on the level of GuiView.
We'll see if this code movement breaks in different scenarios,
but quick grep suggested that the only other way to guiview
dispatch is through drag&drop when menu isn't involved.
Add a new ForceAll update flag that forces to update metrics and
redraw all buffer views.
The situation in multi windows setting is very fragile and will need
to be adressed properly in the 2.5 cycle.
This patch takes into account the skewness of a character for properly
align a decoration. Unfortunately, the required amount of shift depends
on single chars, so that the recently introduced FontMetrics::italicSlope()
method cannot be used with all characters. A heuristics that produces
satisfactory results is using the italic slope only with characters whose
italic correction (kerning) is zero.
Part of #11491
Using unordered_map instead of map.
Reasons:
1.) The relevant maps contain 166(Keys) and 649(Accents) entries.
This mean that average access with 'map' needs 8 to 10 compares to find the value.
2.) Since we are using at least c++11, the unordered_map is available
2.) increasing the maps (in future) needs not to be considered anymore, because
the access-time will not increase.
When hitting backspace in the first position of a cell in mathed,
if the cell is part of an array, the whole array is dissolved
and the sole cell content pasted in the containing environment.
But if the cell is part of an eqnarray or align math hull type, the
cursor is simply (and unexpectedly) taken out of the whole equation,
which is left unchanged.
This patch makes backspace equivalent to the cursor left key in those
cases, while leaving the behavior unchanged in all other ones.
Fixes#11678
A reference can be directly inserted into mathed but cannot be pasted
because the pasted material is returned in plain text format. This patch
allows getting a string from the cut stack in a suitable format allowing
the math parser to actually create an InsetRef.
Fixes#11539
In Qt 5.15 many deprecation warnings are issued, presumably for
helping the port to Qt 6. They are really a load but can be
shuten up by using -DQT_NO_DEPRECATED_WARNINGS, except for the ones
about MidButton, AdjustToMinimumContentsLength, and DefaultLocaleLongDate.
This change should not affect compilation with Qt 4, as all changes are
backward compatible (only Qt versions prior 4.5 could be affected).
Following 4a4ded22, the enabling of some change-related functions is
handled in updateBuffer. However, this method is not ran at every
document change for performance reasons.
This patch adds code to every place that modifies
Paragraph::Private::changes_ that checks whether the `changedness' of
the paragraph, err... changes.
To this end, a new helper struct is introduced that remembers
paragraph state at contruction time, and compares it to new state in
the destructor.
New forceUpdate/needUpdate methods are added to Buffer class, since
the cursor is in general not available in the places where these
changes are made.
Fixes bug #12074.
This is a variant of Input for layout files that only searches build
and system directories and thus allows "InputGlobal name.inc" in a user
file name.inc that attempts to modify its global counterpart.
In case of search with format:
If the pattern contains language spec different to the document language
then do not ignore language setting.
Also renamed 'matchstart' in FindAndReplaceOptions to 'matchAtStart'
- Multicolumn now supports multiple paragraphs in non-fixed-width context.
- Multicolumn now supports valign in non-fixed-width context.
- varwidth columns now properly align horizontally and vertically.
Since commit c600906d92 all matches are match-results of examined strings starting
with a character of the same cursor depth, we can be sure to match the same string again if:
1.) the number of characters to the end of the examined strings match.
2.) the match-lengths are identical
The include inset has its own button_ object, that hides the button_
object of the InsetCommand parent. Remove that.
Remove the hand-made caching that uses the set_label_ boolean.
Let metrics() and draw() use the inherited methods. This requires to
use setBroken as needed and to implement inheritFont.
The use of InsetCommand::draw is what restores the hover effect.
Fixes bug #10415.
Set colors now apply to the active mode only. Matching reversed colors
are produced for the other mode if no color is assigned yet.
File format change.
1) Take care of different input if changed the search-mode (with/without format)
This amends ec387b6d
2) Make the braces used in text to be treated as single characters
e.g. transform '\braceleft' to some unicode value
3) Try to use '$' as 'end of sequence' in regex
We have to remove '}' and '\n' chars from the examined string
In format-search the chars '{' and '}' are understood as latex parentheses, which normally are not
part of text and are discarded.
Instead we fake them as if they were a char like \backslash or \guilemotright or such.
See https://marc.info/?l=lyx-devel&m=161058233016780&w=2.
Internal Buffers do not have TOCs associated with them, so we should
not try to update them. This fixes Kornel's crash, but there may be
other instances of the problem.
Makes the code with less exceptions
(no need to differentiate beteen use_regex and !use_regex)
Move the creation of regexes to own subroutine (Handles '#if QTSEARCH ... #endif')
Use cursor position differences instead of length of matched string. This is important for putSelectionAt()
Factor out the code that connects to a newly created dialog and then
use that also in GuiGraphics.
There must be many other dialogs that need this same treatment....
It turns out that the code is the same in texted ans mathed and that
whatever is done in InsetTabular is not useful.
This means that we do not need to deal Text::dispatch idiosyncrasies
(in particular forcing the cursor to be visible).
Fix bug #11225.
If a macro has only optionals and none is specified and a [ immediately
follows, we have to terminate the macro with {}, otherwise what follows
is taken to be an optional argument.
Fixes#11665.
The bookmarks are added as virtual elements in display Rows. Bookmarks
are shown with circled numbers. A new color "bookmarks" has been
added. Currently bookmark 0 (the return position) is not displayed
because it is very disturbing in practice.
To make this work, a new method BookmarksSection::bookmarksInPar
retuns the list of bookmarks in a paragraph along with their position.
Force redraw when using bookmark-save and bookmark-clear.
Fixes bug #2496.
This avoids generating % for HTML or DocBook: % indicates a comment in TeX, but not in XML, it's just a character. The snippet that is shown is therefore not valid.
As suggested there, we attach ourselves to the newly created inset,
if we in fact just created an inset. (This will only matter if Apply
was clicked, so the dialog is still open, but I did the calcuation
anyway, as it seems cheap enough.) This allows the user to modify
that inset (e.g., change the format) and also prevents another click
on Apply, or OK, from creating another dialog.
I've also removed the override of disconnectOnApply(). It does not
work, anyway, I think because it only works with GuiDialog, not with
DialogView-based dialogs.
Otherwise we are unable to distinguish text from latex commands.
For instance '\color{blue}' in text-part is normal text, while othervise
it defines following characters as being blue colored)
The ensuremath math inset derives from InsetMathNest, so that only the
first cell of a grid gets pasted. This patch makes it a fixed 1x1 grid
inset so that pasting works as in normal simple math hulls, i.e., all
cells are pasted, one after the other.
Fixes#11617.
At the moment, there is no support for plurals and capitalization.
A long comment explains why. Support could be added for that without
a format change.
The correct way of implementing this is at paragraph level. Once this
is clear, the implementation is straightforward.
Note that RtL is not handled correctly by the \lyxrightaddress macro.
Fixes bug #11918 and #8152.
When a paragraph is empty, it was not possible to query what the left
margin would be for an hypothetical second row.
Modify TextMetrics::leftMargin so that it does not test
whether position is after the last position of the paragraph.
Cosmetics: set the size of label to "small" for Abstract layout in
standard classes and reduce the spacing between label and text.
Fixes remainder of #11939.
Since the first row may be indented, it is necessary to find out what
the left margin would be on the next row (even if this next row does
not exst.
To this end, modify TextMetrics::leftMargin so that it does not test
whether position is after the last position of the paragraph.
Cosmetics: set the size of label to "small" for Abstract layout in
standard classes and reduce the spacing between label and text.
Fixes remainder of #11939.
References to environment variables embedded in a filename are expanded
and replaced by their value. However, if a variable does not exist, its
reference is simply erased from the filename, causing havoc (see #7801).
This has been like that since ever and cannot be changed, both for
backward compatibility and because this feature is currently used in
the Windows installer.
A possible backward compatible strategy is leaving as is the reference
to the environment variable (introduced by a $ sign) in the filename
if it does not exist. This is done in this patch, which also assumes
that an escape character is never used in a filename (inserting a $ in
the filename is easy, but I don't think one is able to easily insert
an escape character).
Use innermost nesting to start searches.
Some fine tuning to determine correct match.
(If the regex contains '(\S)\1' at the end, then this regex would match '}}',
but this is often the case at and of examined string. We have to disable this invalid match.
)
The previous code was mistakenly comparing addresses of the arrays.
This issue was caught by a warning when compiling with Clang in
C++20 mode, since comparing array addresses in this way is
deprecated.
This commit should prevent some unnecessary dispatches of
LFUN_SCREEN_FONT_UPDATE.
This regex handling is part of QT5. For lyx which uses QT4
findafv will still work, but is not good for caseinsensitive matchings
in handling non ASCII characters
I misinterpreted the unicode creation 'u8"\uF00xx"'.
The C++-compiler saw 'u8"\uF00x" "x"', but this was not intended.
The routine which mimicked is doing the right job now.
For search we used to lowercase for everything, but sonce the regex itself
should be left unchanged, this change was needed.
Works nice with ASCII, but fails miserably on on other UTF8 points (like Cyrillic chars)
Since we are not sure of where the cursor ends up after invoking
Cursor::handleNest it is safer to reset the cursor at each cell,
rather than reuse the same one over and over.
Related to bug #12035.
The master_ variable holds a FileName that points to the meta-data
for this file, e.g., CVS/Entries. There is no such thing in SVN or GIT.
So we remove that variable from those classes.
boost/signals2.hpp is a convenience header, it seems that
boost/signals2/signal.hpp is enough for our needs.
This allows to get rid of 6MiB of header files in 3rdparty/boost (and
probably to speed-up compilation).
- Add limits support to InsetMathScript
- Add limits support to InsetMathClass
- Fix bug where limits changing is disabled in inline math
Now the MathData objects remember whether they are in diaply mode.
Fixes bug #12045.
Mistakes are visible but word replacement is disabled. Disable relevant
widgets depending on the situation.
Move updateView code to SpellcheckerWidget::Private.
1/ remove boost regex code from 3rdparty/boost. Only the cmake stuff remains.
2/ remove traces of BOOST_LIB since there is no need anymore to
compile boost stuff.
3/ remove traces of boost regexp in config/lyxinclude.m4 and INSTALL
4/ require gcc 4.9 as minimal version in autoconf.
Unfortunately, the net gain is only 10% of included boost size (now 21M).
Includes a fix when wrappers are used with empty paragraphs. The logic is probably wrong if two legit wrapped paragraphs have one empty paragraph in the middle. This is only for slightly convoluted cases... Wrapping decisions should rather be taken for a series of paragraphs at once, but it would be more complex to implement and would only solve very peculiar cases like these ones.
CaretGeometry is defined as a struct, but was forward-declared as a
class. Now we forward-declare as a struct. This addresses the
following Clang warning (that does not appear with GCC):
src/frontends/CaretGeometry.h:27:1: warning: 'CaretGeometry' defined
as a struct here but previously declared as a class; this is valid,
but may result in linker errors under the Microsoft C++ ABI
[-Wmismatched-tags]
The current code did not give much information.
I am not sure why the <<FontInfo operator is not defined in
FontInfo.h, but I figured that it was not necessary to understand that
now.
The exception is raised when an included file fails to export
because e.g. it is from a newer version of LyX or because an
intermediate command such as knitr fails.
The exception was introduced at 1a374a93 so that the export of a
document exits with error if an included document exports with
error.
This commit catches the exception when the LaTeX is asked for in
order to generate a preview of the included file.
The caret geometry is now computed in BufferView as a list of shapes
(caret, horizontal l-shape if needed, completion triangle if needed)
kept in a variable of type CaretGeometry.
The code in WorkArea.cpp only has to draw these shapes. The
CaretWidget (which never was a widget) in GuiWorkArea.cpp is gone
now.
As a consequence, the bounding box for the cursor is known precisely
and therefore rows should be repainted correctly now. This avoids
caret droppings.
Fixes bug #12024.
Try to find the most inner row that is overwritten by the caret. This
allows to replace the hack in TextMetrics::draw, which did not really
work.
Note that there are still issues with emphasized caret at the
beginning of inset, which will require some code reorganisation.
Fixes current recipe of bug #12024.
This is particularly important for contents in RtL languages, where
the right margin may contain the paragraph indentation, or
itemize/enumeration symbols.
Fixes bug #12030.
In one paragraph, there was both a float and a list of floats in a paragraph: the list of floats has no DocBook output, so that there should be no surrounding <para>.
Because Compare uses threads, we need to make sure it is finished when a
compare is executed from the command line. This was a problem for command
sequences, because the next command would start running before the compare
was done, and the buffer with differences was available.
So this commit adds the "run-blocking" parameter when using
LFUN_DIALOG_SHOW to run a Compare. When calling Compare with run-sync, the
LFUN will wait for the compare worker thread to finish before returning and
possibly running the next command.
There was already an existing method in InsetCaptionable for captions, but not yet for labels. It was at least already useful at two places, that's why I moved it to InsetCaptionable.
These paragraphs will simply have no output (no enclosing DocBook tag, no content). This change is as backward-compatible as possible: without this commit, the paragraphs will be output, but in an <IGNORE> tag (not much more can be achieved without this). It is mostly made for metadata and formatting instructions. (As opposed to NONE, which corresponds to outputting the paragraph without enclosing DocBook tag.)
To do this, hard-coded test for xcb had to be replaced with a call to
platformName(). Since this method does not exist in Qt4, we emulate
it.
Note that Qt5 uses xcb bindings for X11 system, while Qt4 relies on
older X11 bindings. We return platorm == "qt4x11" in this case.
Fixes bug #11746.
The problem here is that the check against TEXT_CODE only
catches InsetText, and not its subclasses. Same for the check
against MATH_HULL_CODE. So there is reason not to check codes
here.
This reverts commit 5a54ccfa87.
When using InsetText::edit(), invoke Cursor::setCurrentFont (note that
editXY does it already). It is a good idea in general, especially
after commit 411333fa95, which sets languagecorrectly in ERT.
This allows to comment out (again!) the trick added to fix ticket #9042.
Fix issue when language is set to document language instead of
latex_language in ERT and pass thru layouts.
To trigger the issue
1/ select all in a ERT inset
2/ type "abc"
3/ look at code in code preview pane
==> result: the "a" is in document language, the "bc" in LaTeX language
(as should be)
And with it, 'master'. That is less problematic by itself (so I'm not
worried about 'master document'), but here it doesn't make a lot of
sense without 'slave'.
More generally, ensures that paragraphs in abstracts do not have something else configured.
A major problem in making the layout more useful is that article titles are not supposed to be in TOC.
The option here is to NOT indent the paragraph, so make that clear.
As it was, we would set "Indent" to false in the case of paragraphs
that do not permit indentation and then write that, uselessly, to the
LyX file if we wrote any other option (and then keep writing it, even
if other options were removed). Now, we will write it only if it's been
explicitly set.
For elements that should behave like sections (for now, mostly prefaces).
A second paragraph of PartBacktext can only wreak havok: it should start a section, and nothing else (otherwise, it's a real nightmare to implement).
The method horizontalAdvance() replaces width() starting with Qt 5.11.
To handle this, all direct calls to QFontMetrics::width() are replaced
by calls to GuiFontMetrics::width(), and the code for
GuiFontMetrics::width(QChar) uses horizontalAdvance on newer Qt
versions.
This will catch cases where there are multiple listings in a single paragraph, for instance.
The new code is also less brittle (checks on lyxCode instead of generated LaTeX or names).
It's invalid to have an <inlineequation> outside a paragraph. Another solution would have been to change InsetMathHull::docbook to generate an <informalequation>, but that function would have required more knowledge about its context than now.
Need more tests to determine if <inlineequation> should only be output for hullSimple or in more cases like:
const static std::set<HullType> inlineHulls = {
hullUnknown,
hullNone,
hullSimple,
hullEquation,
hullMultline,
hullGather,
hullRegexp
};
Without this patch, there would be errors like these (warnings turned into errors):
D:\LyX\lyx-unstable\src\frontends\qt\GuiView.cpp(2413): error C4101: 'e': unreferenced local variable
D:\LyX\lyx-unstable\src\frontends\qt\GuiView.cpp(4879): error C4101: 'ex': unreferenced local variable
The setting implies that the variable external_in_tmpdir is set to false.
If a lyx-file includes external material for spreadsheet,
the conversion results are in tmpdir.
The routine doSubstitution() (insets/ExternalSupport.cpp:188) therefore
does nothing for the entry $$Contents because it uses wrong path.
Setting the res value and returning from the function at the same time
does not make sense. After reflection, it is the "return" that shall
go.
No idea what effect it had.
Intended changes:
* code is shorter and cleaner
* caret scales better with zoom when cursor_width=0: completion
indicator, l-shaped cursor...
Details:
* Rename BufferView::getPosAndHeight to getPosAndDim because ascent is
needed too and width could in the future be set depending on font.
* Get rid of rect_ in CaretWidget and record a Dimension (and y value) instead.
Remove also caret_width_ and slant_, replace rtl_ with dir.
* Make CaretWidget members public and lose the trailing _.
* change CaretWidget::update to read its parameters from current bufferview.
When validating local layout, in particular, we create a dummy text TextClass
and so are not necessarily modifying previously declared material. Hence, we
get a spurious (but harmless) "Incomplete argument definition!" warning. This
suppresses it, but to do that we need to propogate the ReadType.
Without GUI an image is not available even if it is actually cached.
In this case we load the image for obtaining its dimensions in order
to correct the clipping bounding box.
This amends 380f34a1.
This can lead to mistakes. If no buffer is required at the beginning,
it is better to pass 'nullptr' explicitly.
Fix a case of missing parameter in GuiCompleter (introduced at
f3656600). This avoids some spurious "There is no group open (creating
one)" messages.
The code in TextMetrics::completionPosAndDim is needlessly complicated.
1/ It is easier to rely directly on Text::getWord() rather than on the
helper previousWord().
2/ Using a fake row to compute completion height is overkill. A
simpler solution is proposed here.
No intended effect. These changes prepare the fix to ticket #11818.
This is mostly important for subfigures, but debugging this issue showed that InsetCaption could be slightly simplified and InsetLabel should be made a tad more robust.
This ensures that all comparisons for DocBook are made on the same normalised version of the float type. This cased a strange bug where <table> was output within <informaltable> for Linguistics Tableaux.
This lead to a quite subtle bug. The major problem is that float tags
depend on whether the LyX float has a title (formal) or nor (informal).
This information cannot be encoded in the layout format as easily as a
simple DocBookTag.
This was due to Floating::docbookTag not returning anything with the floattype_ tableau. Another issue that happened with that document is that the standard library's isspace crashed for some characters. I therefore implemented a more efficient version of the part that required it, and inlined the definition of isspace (even though that part becomes irrespective of locale, but was that feature ever used?).
Includes: semantic markup (sorry about noun: Additional.lyx uses it to mark menus; there is something better in DocBook, but it looks like the LaTeX equivalent is really for person names), boxes, info layouts.
Actually output something when list item is empty. XMLStream discarded the sequence StartTag/EndTag (nothing in between).
New-line behaviour around term in description lists.
Before, it directly wrote to the XMLStream, but it made implementation of new lines tricky. Now, it returns the XML for each sub-paragraph (delimited by new lines) as a string, so that the caller can adopt a more precise behaviour (such as in lists).
Fixes for bug #11428. Emphasis is used regularly and it is helpful to
see whether the next text will be written emphasized directly from the
cursor.
This is also implemented in MS Word and Google docs but not in
LibreWriter.
Also fixes the caret for RTL languages whose width was previously
extended into the wrong direction.
The change includes a rewrite of the cursor movement when „mac style“ is enabled and don’t change anything otherwise.
The new code for mac uses the document iterator to go forward or backward. The traversal stops at word boundaries.
If going forward the position increments until a word is reached (if not already inside) and stops at the end of the word.
If going backward it does the same in opposite direction. The cursor jumps over non-editable insets and math.
Editable (open) insets are entered and the cursor move detects word boundaries inside them.
Instead of working around crashes in update of caret geometry, only
request it as needed. The actual computation will take place just
before painting the caret.
It might be that this is overkill and that caret geometry should be
updated unconditionally. One would have to to some timing while idle to
ascertain that.
Fixes bug #11912.
The macros defined in the symbols file are only used for on-screen
representation, so that their definition should not require the use
on any latex package. Their definition should be taken into account
only when they are overridden by user macros with same name.
The rules for typesetting math differ from the rules for typesetting
text. For example, two italic 'f' chars have to be typeset more closely
than two 'o' chars in text mode, but not in math mode. Qt provides a
method that returns the distance appropriate for drawing a subsequent
character in text mode, but nothing for math mode. Typically, the
distance appropriate for drawing the next character in math mode is
the actual width span by the character, corrected by the rules of
an appendix in the TeXbook. Recently, those rules are followed more
closely in LyX but not exactly, and we have to find a way to adapt to them.
Some symbols may need more spacing around them than the width they span.
So, we use the distance suggested by Qt, unless it is less than the
width of the rectangle bounding the symbol. Before Qt 5.11 the used method
was QFontMetrics::width(), but since then it has been declared obsolete
in favor of QFontMetrics::horizontalAdvance(), whose name conveys better
its meaning.
Instead of working around crashes in update of caret geometry, only
request it as needed. The actual computaiton will take place just
before painting the caret.
It might be that this is overkill and that caret geometry should be
updated unconditionally. One would have to to some timing while idle to
ascertain that.
Fixes bug #11912.
Move InsetMacro::limits_ to InsetMacro::Private, where it should have
been from the start. This means that limits_ is now copied when the
macro is cloned.
When the fonts distributed with texlive with same family name
as ours are also managed by fontconfig, Qt is not able to
discriminate ours by the style only. In order to be sure Qt
will load our fonts, we change the foundry name from 'PfEd'
to 'LyEd' and augment the font family name with this foundry
name. This only works on *nix, because adding a foundry name
seemingly breaks things on Windows. However, this is a *nix
only problem, seemingly.
This restores the debug output as it was intended before the
introduction of the LYXERR macro that was unconditionally outputing
an end of line. This is how this output was appearing until lyx 1.5:
Looking for font family cmr10 ... got: cmr10
got it normal!
instead of as it was since lyx 1.6 and until now:
Looking for font family cmr10 ...
got: cmr10
got it
normal!
Using US english desktop language LyX's Mac OS adds some items to the view menu:
1. Show/Hide Tab Bar and
2. Enter Full Screen
These items won't work properly and it's better to hide them.
Mac OS adds two extra menu entries to the edit menu.
These menu items are:
1. Start Dictation...
2. Emoji & Symbols
Both menu items don't work with LyX.
Therefore they shouldn't be present and the new code hides them.
Both QTextLine::naturalTextWidth() and QTextLine::horizontalAdvance()
return the same value for \fint. However, examining esint10.ttf with
fontforge does not reveal any issue with the metrics. The fact that
\fint seems to be the only affected symbol might be due to its code
point, which corresponds to a space, so that maybe Qt makes some
assumptions on the metrics.
As QTextLine::naturalTextWidth() returns the width of the line that is
occupied by text, in the case of a single symbol we can obtain the
same value by using the width of the rectangle bounding the symbol.
This extra spacing was missing and is important for detecting extra
par breaks before equations (which are most of the times not wanted).
To do that, add a new member vmode to MetricsInfo which is equivalent
to LaTeX's \ifvmode (start of paragraph).
If in vmode, add the equivalent of an empty line before a display math inset.
At the same time, tweak value of \(above|below)displayskip, which was
12pt, whereas LaTeX uses 10pt in 10pt size (our reference).
Fixes bug #11891.
Fix bug where selecting in first paragraph gave an end of selection on
wrong row.
Since pm.ascent() may contain the top margin, it makes sense in
setCursorFromCoordinates() to use the ascent of the front row instead,
like was none in907f0207 for getPitAndRowNearY().
Currently tested:
- url & doi fields for bibtex.
- all documented eprinttypes of biblatex
- absolute paths of first entry of 'file' field for jabref and kbibtex
- external script searching for author + year pdf
Additional polishing will follow.
https://www.mail-archive.com/lyx-devel@lists.lyx.org/msg212505.html
