When switching screen rotation, have the re-rendering
properly done only via :onUpdatePos().
When changing alt status bar font size, also have it
done via :onUpdatePos() instead of the next painting
(this also update any dogear position after the resize).
When in scroll mode (no alt status bar), be sure to
draw crengine progress bar at top (and not below the
not shown alt status bar).
Wrap document opening and re-renderings (which can block
the app for some time) with setIgnoreInput() to avoid ANR
on Android.
Any setting update that could possibly cause a re-rendering
should send the UpdatePos event, to ensure the re-rendering
happens in ReaderRolling:onUpdatePos() where precautions
are taken to avoid ANR.
In response to the comment by @ilyats on Weblate:
> Text appears in code twice, as menu item and as dialog button label. Ideally, translations should be different.
When trying to open/delete a supported document with empty name (eg .txt) got an error
./luajit: frontend/docsettings.lua:118: attempt to concatenate a nil value
Everything in string.gsub() is replaced, not only the matching group. By using two groups, we can correctly return the full matched string when none of the special conditions apply.
Fixes <https://github.com/koreader/koreader/issues/8486>.
Move Default and extra buttons above Cancel/OK.
Default values shown in the default button.
Precisions can be set for both values separately.
Minor geometry fix for consistence with SpinWidget.
Allow setting a folder as a book shortcut, with 2 options:
- open file browser: opens the FM in that folder
- last book: opens the most recently read book (via
ReadHistory) in that folder
Bookmark/highlight formats in crengine and mupdf are incompatible.
This backups and restores bookmarks and highlights when opening the
document with an incompatible engine, instead of deleting them.
Fuzzy searching doesn't work with CJK text: with Japanese,
we get large numbers of useless results because sdcv
decides to strip off the wrong part of the word.
It seems unlikely that sdcv correctly handles Korean
or Chinese, so just disable fuzzy searching on all
CJK-containing word lookups.
Add a "Select" button in the highlight dialog to initiate
text selection; on the next text selection, the text between
these 2 points will be selected.
Limited to a single page with non-CRE documents.
Also move "Search" button at end, so it's the one that
will be wide in case of an odd number of buttons.
Previously getTextFromBoxes would just pass the first and last three
bytes of the current and previous words when trying to detect CJK
characters (which shouldn't have spaces inserted).
However, this handling was not correct because CJK characters can be
longer than 3 bytes, and internally BaseUtil.utf8charcode doesn't ensure
that it was only given a single utf8 character (it blindly does the bit
operations on whatever length code you give it).
As a result, before this patch selections in PDF documents would have
lots of spaces stripped because getTextFromBoxes would think that almost
all characters were CJK characters.
Fixes: 6f1b70e5eb ("util.utf8: improve CJK character detection")
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
This layout is far more commonly used on mobile devices, and allows for
much easier typing. The keyboard primarily functions through gestures in
the four cardinal directions to select which vowel kana to select. In
addition, users can cycle through each kana row by tapping the key
within a 2-second window (this is the equivalent to T9 input for
Japanese phone keyboards).
This also resolves the long-standing issue that the old keyboard did not
correctly handle dakuten (there was a standalone dakuten key which added
a stray dakuten mark, and the umlat mode which added dakuten to all of
the keys it could) and could not input handakuten characters at all.
In order to allow adding dakuten and cycling through the various
modifiers for the previous kana, we need to wrap the input-box (similar
to korean) but luckily we don't need any state machine magic since we
just need to modify the last character in the character buffer. However
because the tap timeout for T9-like-cycling needs to be reset after any
non-tap key we need to add some basic wrappers around a few other
input-box methods.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
A layout might want to make some specific feature configurable, so
create an addToMainMenu-like system for allowing layouts to add their
own configuration sub-menu to the keyboard configuration menu.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
This allows for InputText wrappers (namely the Japanese keyboard which
needs to be able to apply modifiers to the character before the cursor)
to nicely access the character list.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
In some cases, it's useful to be able to wrap a function and either
replace its contents entirely or have some callback be run before
calling the underlying function.
The most obvious users for this feature are the Japanese and Korean
keyboards (both of which need to wrap the inputbox methods with either
their own versions or have basic callbacks be run before the method is
executed).
This is loosely based on how busted/luassert spies work.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
It's possible for the user to have selected nothing, and trying to
operate on the nil highlight can cause confusion or crashes. This
restores the behaviour before commit 7a0e3d5e68 ("readerhighlight:
remove selected_word and use selected_text everywhere"), which missed
this case.
In addition, add some debug guards to ReaderHighlight methods which
cannot handle selected_text being nil (or at least, shouldn't be called
with selected_text being nil).
Fixes: 7a0e3d5e68 ("readerhighlight: remove selected_word and use selected_text everywhere")
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
The previous version of JMdict comes from 2009 and doesn't appear to
work at all when trying to do basic lookups (likely due to some kind of
encoding problem). In addition, the license information and sourcing was
not really in line with the requirements specified by the JMdict
license. This version is far more up-to-date and also includes synonym-based
deinflection (though because KOReader has a Japanese plugin now, this is
technically not necessary).
Since there didn't exist a nicely-maintained place to download these
dictionaries (because StarDict is not widely used for Japanese
dictionaries), I've set up a personal GitHub repository where I've
hosted them. Note that we're intentionally not pinning the commit hash
because GitHub only recommends we use gh-pages for CDN purposes, and one
of the requirements of the JMdict license is that you need to be able to
update to later versions.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
If we had a local prefered AP with a higher RSSI,
we attempted to associate with it over wpa_supplicant
being already attempting to associate with its own preferred AP.
That... failed horribly.
Also adapt to the new lj-wpaclient API, fixing a few other edge-cases,
and making the whole thing slightly faster because we no longer
uselessly sleep.
And more reliable because we now actually wait for replies to our
requests.
Bump base
https://github.com/koreader/koreader-base/pull/1424
screen.
Otherwise, on the Sage, weird flash glitches may happen, depending on
what was on screen...
(e.g., there's some weird update merging shenanigans going on
despite those updates being flagged NO_MERGE...).
The second argument is a ddjvu_render_mode_t
Try to actually honor the user settings instead of enforcing COLOR
while we're there.
Fix#8376
Regression since #8250
Now that FileManager registers its UI modules in the same way as Reader,
this shouldn't be necessary but this protects us against some other app
creating a ReaderDictionary instance without having ui.languagesupport
registered properly.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
With the addition of the language support module, ReaderDictionary
tries to use modules registered with the UI instance, but the
FileManager doesn't provide key-based registration of its UI modules.
In order to allow the same code to be used by both FileManager and
Reader seamlessly, copy the :registerPlugin() method from Reader and use
it with FileManager. This will ensure any other hidden assumptions about
UI module registration are handled properly.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
- Add an icon to distinguish between page bookmarks, plain
highlights, and highlights with an added note
- Bookmark details: show both highlighted text and added note
- Bookmark list: allow filtering by type and/or by keyword
- New bookmark selection mode, to allow multiple removals
- New option: show separator line
This creates a new plugin system which hooks into a handful of reader
operations in order to allow plugins to add language-specific support
where the default reader falls short. The two hooks added are:
* During hold-without-pan taps, language plugins can modify the
selection in order to better match what users expect koreader to
highlight when selecting a single word.
The vast majority of CJK language words are more than one character,
but KOReader treats all CJK characters as a single word by default,
so adding this hook means that readers no longer need to manually
select the whole word every time they need to look something.
* During dictionary lookup, language plugins can propose alternative
candidate words to look up if the selected word could not be found in
the dictionary.
This is pretty necessary for Japanese and Korean, both of which are
highly agglutinative languages and the fuzzy searching system of
StarDict is simply not usable because often the inflection of the
word is so much longer than the dictionary form that sdcv decides to
chop off the actual word and search for the inflection (which yields
useless results).
This system is of particular interest for readers of CJK languages
(without this, looking up words using KOReader was fairly painful) but
this system is designed to be minimal and language-agnostic enough that
other languages could make use of it by creating their own plugins if
the default "whole word" highlight and fuzzy-search system doesn't match
their needs.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
This is necessary in order to allow the language support module to be
added to the menu outside of the common_settings menu table definition.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
In order to make startSdcv usable for plugins that might need to do
dictionary lookups in order to work, it is necessary to split out the
core of startSdcv into another method which returns the raw data from
sdcv.
In addition, in order to make it possible to amortise the cost of each
lookup (which could be fairly expensive) make it possible to pass
multiple words to rawSdcv at the same time. Sdcv supports passing
multiple words as arguments (which it then looks up in order and returns
a separate JSON array per line for each word) so we just need to tweak
the return style accordingly.
All of the deduplication and dummy results handling remains in startSdcv
because plugins might strongly depend on whether sdcv returned actual
results.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
There were two ways of specifing selected text for a highlight depending
on whether it was a "single word" or text selected using hold-and-pan.
In addition to being a bit more complicated than is necessary, with the
addition of the language support plugin system (where the "single word"
selected might be expanded), it makes more sense to simply use the same
logic and table structure for both cases.
The dictionary lookup special case (hold-without-pan triggering a
dictionary lookup by default) still works as before.
In addition, this patch fixes a minor inefficiency during dictionary
quick lookup -- before this patch, the highlight would be re-selected
because the quick lookup window is run concurrently and tries to fetch
ReaderHighlight.selected_text but this is set to nil immediately after
triggering the lookup. This is unnecessary because :clear() will be
called anyway when the quick pop-up closes, and so clearing this can be
left until then.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
There were a handful of cases where if there was no cached kctx there
was no fallback and several KoptInterface methods would return nil,
causing issues in various parts of KOReader (this happened with the
migration to selected_text everywhere but it's unclear how that change
caused this regression).
In any case, from a correctness perspective it makes sense to have the
corresponding fallback paths to create a new kctx if we couldn't find a
cached one.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
With the latest koreader-base update, we can now create native
selections using getTextFromXPointers. In order to make the wrapper less
annoying to use, always enable segmented selection if selections are
enabled (to match getTextFromPositions).
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
It is a bit cleaner to do all of the necessary looping over lists of
Geoms within a straight-forward Geom.boundingBox function rather than
looping over :combine every time (or reimplementing :combine in some
cases). Geom:combine can be trivially reimplemented in terms of
Geom.boundingBox as well.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
Previously the CJK character detection defined only characters in the
range U+4000..U+AFFF as "CJK characters". This excludes an incredibly
large number of CJK characters within the BMP, let alone the whole two
planes dedicated to rarer CJK characters (the SIP and TIP). As a result,
a very large number of Chinese, Japanese, and Korean characters were not
detected as being CJK characters.
While slightly less elegant-looking, it is far more accurate to compute
the codepoint from the utf8 character and then see if it falls within
one of the defined CJK blocks. This is not future-proof against future
CJK ideograph extensions in future Unicode versions, but there is no
real way to accurately predict such changes so this is the best we can
do without accidentally treating characters explicitily defined as being
non-CJK in Unicode as CJK.
While we're at it, copy Lua 5.3's utf8.charpattern constant definition
so that we can more easily write utf8 iterators with string.gmatch (at
least in the interim until there is a rework of utf8 handling in
KOReader and everything is rebuilt on top of utf8proc).
Some unit tests are added for Korean and Japanese text, and the existing
unit tests needed a minor adjustment to handle the fact that
isSplittable now correctly detects CJK punctuation as a character to
compare against the forbidden split rules.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
* Use paintRect and plain colors instead of lightenRect and a weird
dimming factor.
* Update call sites to the new API
* Handle FP maths properly (i.e., floor coordinates & ceil dimensions at
the latest possible time).
* Fix border handling in the fill bar (make sure we actually honor it
when computin the x position, and that we won't overflow into it when
computing the width).
* Update docs
Reword "Purge .sdr" to "Reset settings".
When purging, remove only the known document metadata
files, and not those for a document with the same name but
a different suffix.
Bookmarks list:
- page numbers are displayed
- page bookmarks are marked with a star
- new setting: Sort by largest page number (default: checked)
New bookmark setting: Add page number / timestamp to bookmark
- If enabled (default), bookmark name is 'Page # notes @ time'.
- If disabled, bookmark name is equal to the notes field.
Rename bookmark dialog:
- page number and timestamp are displayed in the input
dialog description
- blank input renames bookmark to the default name in
accordance with the new setting
Also fix: changing boundaries of the highlight: the name of the
highlight is not changed if it was previously edited by the user.
- ButtonTable, ButtonDialog, ButtonDialogTitle
- ConfirmBox, MultiConfirmBox, SkimToWidget
- KeyboardLayoutDialog (and initially move the dialog
down to show the title in landscape mode)
- InputText's Clipboard dialog
Also: Notification: truncate long text
cores
* Only keep a single core online most of the time.
* Device: Add an enableCPUCores method to allow controlling the amount of
online CPU cores.
* Move the initial core onlining setup to Kobo:init, instead of the startup script.
* Enable two CPU cores while hinting new (e.g., cache miss) pages in PDF land.
* Enable two CPU cores while processing book metadata.
* Drive-by fix to isolate the DocCache pressure check to KoptInterface
and actually apply it when it matters most (e.g., k2pdfopt stuff).
* Make sure we have a BB to measure in getSize, in case the instance is
recycled. (fix#8241)
* nil `line_num_to_image` early in `:free`
* Hide the _renderText calls that are used across the whole module to
simply update the text layout & instantiate the inner bb behind a
wrapper function with a slightly less obscure name.
Discussion in #6409.
Highlight action renamed to Long-press on text and moved from Gear - Document to Gear - Taps and gestures.
Added action Do nothing.
Removed menu item Typeset - Highlighting - Allow highlighting.
Fixed untranslated strings in the Cycle highlight action notification.
Long-press on images always opens ImageViewer. Closes#6409.
* Gestures: standardize hold to long-press
* Common settings menu: standardize hold to long-press
* Readerstyletweak: standardize hold to long-press
* Readersearch: standardize hold to long-press
* Geom:transformByScale:
* Apply the right scaling factor to the y axis
* Round in a more sensible fashion (àla fz_round_rect, since we pretty much exclusively use it in a similar fashion).
* Bump base (https://github.com/koreader/koreader-base/pull/1407)
Also: Text editor now closes its keyboard when calling Find
and Go to line (which open their own keyboards) to avoid
conflicts between multiple keyboards.
Includes among others:
- (Upstream) Various CHM handling fixes, and others
- HTML documents: rebuild TOC from headings after load
- Font: use metrics for underline offset and thickness
- epub.css, html5.css: tweak ruby styling
- CSS: fix EPUB's head>style content encoding
- CSS: add support for 'box-sizing: content-box/border-box'
- CSS: support for styling the <html> element
Also bump KoboUSBMS to v1.2.2 and FBInk to v1.24.0.
ReaderFont's "Generate font test document": update the
generated HTML so its ToC is build from proper HTML headings.
Store list of layouts in settings file as array of enabled
layouts only (up to 4 elements). Optimize code.
Allows sorting of the abbreviations in the globe popup:
just check layouts in the desired order (the first checked
will be northeast).
Requires onetime migration to clean up the settings.
If this is not done, the URL when the file is downloaded will be
something like hostdir/path, rather than host/dir/path.
Also add a debug log to make it more clear when a bogus URL
is being fetched,
* Cleanup util.secondsFrom*Clock stuff (simpler maths, tail calls, meaningful printf tokens).
* Use util.secondsToClockDuration in ReadTimer instead of reinventing the wheel three different ways.
* Reschedule unexpired timers properly on resume (as best as we can, given the unreliable nature of REALTIME).
* Make clock timers tick on the dot, instead of at the same second as when being set.
* Speaking of clock timers, leave the math to os.date & os.time, don't reinvent the wheel yet again.
We should always unschedule suspend before scheduling it again (i.e.,
use rescheduleSuspend ;)).
Fix#8097 (many thanks to @Mel-kior for the detailed repro!)
When enabled, if the book has some supported language tag
in its metadata, use it as the source language. Otherwise,
fallback to the current settings (auto-detect or selected
source language).
Before: when holding the input box in input dialogs
for calling the Clipboard, hold release was passed to
MovableContainer and input dialog moved a little bit.
If the button text would be truncated, try to avoid
it by reducing the font size, and even switching to
a 2-lines TextBoxWidget.
TextBoxWidget: fix possible glyphs truncations when
a small line_height is used. Also avoid some bad
result from getFontSizeToFitHeight(), possible due
to some rounding errors.
* keep_dialog_open, default to false.
Set to true to keep dialog open upon pressing any button, except Cancel and dismissable tap.
* other_buttons_first, default to false.
Set to true to put other buttons above Cancel - OK row
Go to the directory of the deleted file, instead of the folder you happend to switch into the reader from as this may have changed (via changing books from history etc)
It turns out that the kernel needs a little push now that the dedicated
wifi power control module is gone ;).
Issue was only exposed if you booted KOReader while the Wi-Fi was down.
* Decode EV_KEY:KEY_BATTERY
* Input: Only drop hovering *pen* events.
There are currently too many broken 0-pressure *finger* events being
reported on the Elipsa, making a dumb rejection highly annoying.
* Bump base
https://github.com/koreader/koreader-base/pull/1393
* Rely on actual events to detect loss of contact for the "snow"
protocol.
Allows simplifying the whole thing.
* Use `ipairs` over `pairs` for pure arrays.
Make "Taps and gestures - Page turns" available only in reader.
Move there other page turn related menu items from Navigation.
Remove duplicated code. Added standard "star" for default RTL.
TileCacheItem: add created_ts property.
Document: manage a tile_cache_validity_ts and ignore
older cached tiles.
This timestamps is updated when highlights are written
as annotations in, or deleted from, the PDF, so we can
get the most current rendered bitmap from MuPDF and
avoid highlight ghosts on old tiles.
Save this timestamp in doc settings so older cached to
disk tiles will also be ignored across re-openings.
Bump base for: mupdf.lua: update frontend pboxes with
MuPDF adjusted ones.
We may get multiple boxes when selecting texts, one for each
word, and we have to add spaces between the extracted words
ourselves. Previously, we were only adding a space if the
last char of previous word was ASCII, so missing spaces
after accents or greek words.
Try to do better by measuring the distances between boxes
and comparing to box heights, with a few heuristics.
Bump base for cre.cpp cleanup and utf8proc FFI.
Add a checkbutton for case sensitive search in FileBrowser,
and use Utf8Proc.lowercase() for case insensitive search.
Also use it in ReaderUserHyph as a replacement for
crengine getLowercasedWord().
- bump crengine: findText(): add support for regular
expression search.
- bump base: add thirdparty/srell/srell.hpp, a C++ library
that provides Unicode regex support, used by crengine.
- ReaderSearch: with credocuments, add checkboxes for case
sensitive and regular expression search.
I've encountered an issue when Calibre Content Server's OPDS feed produced ``text/fb2-xml`` mimetype. Don't know if it is actually Calibre to blame, but thought this simple fix will save some poor souls' time.
- New way to hide the VirtualKeyboard: to hide the keyboard
tap any point of the screen outside the inputbox and above
the keyboard; to show the keyboard tap the inputbox.
(Removed hacky "holding the arrow-down key" which is no
longer needed).
- InputDialog windows are movable/translucent by default
- Redesign of the Clipboard dialog
get any boxes
Exposed by #7624, but we were arguably putting garbage in the Cache
before that anyway, so, it w<asn't all that great either ;p.
Fix#7850
Instead of baking in a hyphen, so as to avoid signed zeroes, or
confusion with page progress if page maps are enabled.
Also, made compact itemps RTL friendly.
Menu: display the right item ("mandatory") with a
smaller font size related to the left item font size,
instead of a function of perpage.
ToC popups: use same font size as in ToC.
Migrate *global* zoom_mode settings to genus/type, too.
Nothing can actually set this as a global anymore, but we still honored
it nonetheless.
Fix#7778
The hyphenation of a word can be changed from its default
by long pressing for 3 seconds and selecting 'Hyphenate'.
These overrides are stored in a per-language file, i.e:
koreader/settings/user-German.hyph.
Don't close the popup on the first hold_release, ensuring it won't
immediately close on a *different* key if the finger hasn't moved.
If the finger *did* move, that's a quick swipe, and that's unaffected by
this tweak.
That seems like a rather terrible idea to beign with, and that that may actually have fatal consequences.
Re #7738
Co-authored-by: Frans de Jonge <fransdejonge@gmail.com>
By showing a warning, instead of not passing any -u flag to sdcv and letting it query *all* dictionaries if FS order...
Co-authored-by: Frans de Jonge <fransdejonge@gmail.com>
This is handled in an Event handler, but we have zero guarantee that
we're actually the *final* Event sent to the Document, and other Events
usually kinda need the Document instance to still be alive ;).
Delay action until the next tick to avoid potential crashes.
* Namely, ensure zoom_mode is consistent with genus & type *both ways*. (I only dealt with the "no zoom_mode" case in my original fixup).
Because documents with settings dating back from before the new zoom modes had "old" zoom_mode settings mixed with "new" genus/type defaults that didn't agree with each other.
It lead to super-confusing ConfigDialog behavior, because ConfigDialog was in fact not reflecting the reality.
(As the source of truth is actually `zoom_mode`).
* There was a snafu in manual mode, because of the extremely weird way prefixes are handled by Configurable/ReaderConfig/DocSettings/ConfigDialog.
So, make sure we only have a *single* zoom_factor, and that it's updated and saved properly under the right name everywhere.
Fixes inconsistencies between first swapping to manual mode, and what the ConfigDialog said/did (because again: possibly a lie), vs., re-opening the same document, which would magically use *different* settings, closer to what was expected (but still broken because of the prefix mismatch and a disagreement on defaults between the two variants).
Fallout from #6885
(i.e., when History is spawned from ReaderUI, delay the flash until we
*show* the next widget, instead of when we close the current RD
instance).
Prevents flashing the InfoMessage.
It can happen in perfectly sane contexts.
CReDocument: Don't destroy internal engine data when Document just
decreased the refcount (as opposed to actually tore down the document
userdata if it were the last ref).
PdfDocument: Only write edited documents if the Doc instance was torn
down.
PicDocument: Silence some DocumentRegistry related warnings
Notification: adds some functions so it can be used as
a notification manager.
Have various bits of code emitting events that may generate
notifications advertize themselves as the source for following
notifications.
Add a menu to allow selecting some subsets of sources
to show or hide.
Add a new reader module: ReaderScrolling, that exposes
some Scrolling options to the menu (which are to be used
by and implemented in ReaderPaging and ReaderRolling
themselves) and implement some inertial scrolling logic
used by both of them.
Default to "Classic scrolling" which is the expected
behaviour on phones and tablets.
The old CreDocument buggy behaviour is available as
"Turbo scrolling" for both Paging and Rolling documents.
Added a "On release scrolling" option that might be
useful on eInk to avoid dynamic pan/scrolling.
Try to avoid bad interactions between pan and swipe,
cancelling unwanted panning if we ended up doing a
swipe or multiswipe.
- TextWidget: avoid crash with small max_width (could happen
when opening menu on a small emulator window)
- TextWidget: avoid crash if re-used after :free()
- ReaderHighlight:clear(): fix possible crash when
scheduled and run after document closed
- DictQuickLookup: minor canSearch() tweak ('selected_text'
is normally available only on multiple words selection,
but is currently available on single word selection thanks
to some unrelated side effect)
Whith the new font size step of 0.5 (46a2d9c), the gesture
for increasing or decreasing font size would change the
size by only half of the given value.
* Tear down FM instances properly
* Don't manhandle ReaderUI too much, and document when the tests do
actively broken shit, like bypassing safeties to open two // ReaderUI
instances.
We're now more careful about this, so, I suppose weird timings may mean
we might be trying to access a nil here.
Fix#7706
Guard a few other similar constructs
FFI finalizers can fire in unspecified orders, but for MuPDF,
we need to ensure that the context is the *very* last thing that get
destroyed.
As such, we need to make sure we close open documents properly on our
end, first.
This prevents potential crashes while switchign to USBMS inside an open
document handled by MuPDF.
Fix#7428
Namely, instead of making it lower priority than readerhighlight and
hoping for the best, make it higher priority, as it should (ReaderFooter
itself is *above* Reader, after all), with a sane set of fallthroughs:
* No footer
* No hold-to-skim
* Held outside the footer (but inside the footer tap zone)
This made the crap workaround from #7466 unnecessary, and actually fixes
the behavior in PDFs (because readerhighlight will match the *physical*
page, which may include off-screen content) and ePubs w/r eclaim bar
height (in which case, there's actual content behind the footer that
readerhighlight could have matched on).
Fix#7697
Searches for a string in the edited text.
Available in the Text editor and other input dialogs with the navigation bar enabled.
Find first searches from the beginning of the text.
Find next searches from the next to cursor position, used for continious search.
By now, the Search input window is closed after the search. You need to press the Find button again for continious search, the search string is kept in the input.
Is it better to keep the dialog open for the comfortable continious search? And close it with the Cancel only?
Case insensitive. Cursor jumps to the beginning of the found string.
Notifications are shown for better results visibility.
Unfortunately, violated our standartization to "search", couldn't invent better short wordings.
To avoid a few log lines on all platforms:
```
ffi.load: SDL2
ffi.load (warning): libSDL2.so: cannot open shared object file: No such file or directory
ffi.load: libSDL2-2.0.so
ffi.load (warning): libSDL2-2.0.so: cannot open shared object file: No such file or directory
ffi.load: libSDL2-2.0.so.0
ffi.load (warning): libSDL2-2.0.so.0: cannot open shared object file: No such file or directory
```
We've managed to trip a few of those on dimen fields post-init but
pre-paintTo in a few weird coner-cases, a point at which dimen is often
nil.
ConfigDialog: Deal with that very thing in update()
Fix#7656
A dotted line joining the ToC entry text to the
page number may make it easier to use.
(This replaces the menu item separator from d879062e.)
Also fix baselines aligment, which could be a bit off.
* FileManager/ReaderUI: Clarify the current instance accessor
Make it clearer that we actually store it in a *module/class* member, not an *instance* member.
Also, warn if there's a close/open mismatch.
Use a table & table.concat instead of individual concats.
And then use that same table for every hash-related operation.
(Nothing else uses the configurable hash function, otherwise I'd have
limited the table shenanigans to the function itself).
* Allow doing away with CacheItem
Now that we have working FFI finalizers on BBs, it's mostly useless overhead.
We only keep it for DocCache, because it's slightly larger, and memory pressure might put us in a do or die situation where waiting for the GC might mean an OOM kill.
* Expose's LRU slot-only mode
And use it for CatalogCache, which doesn't care about storage space
* Make GlyphCache slots only (storage space is insignificant here, it was
always going to be evicted by running out of slots).
* More informative warning when we chop the cache in half
* Neuter timekeeping when statistics are disabled
Saves a few syscalls ;).
* Port to ffi/lru
Only a tiny bit of it actually requires any sort of LRU logic, so it's fairly painless.
* Release the cache on close
* Use string.buffer to serialize function arguments
Ought to be faster than the custom approach ;).
(Still requires wrapping them in a table, though).
It's much less human-readable, but then again, this doesn't need to be :).
Ought to be faster than our naive array-based approach.
Especially for the glyph cache, which has a solid amount of elements,
and is mostly cache hits.
(There are few things worse for performance in Lua than
table.remove @ !tail and table.insert @ !tail, which this was full of :/).
DocCache: New module that's now an actual Cache instance instead of a
weird hack. Replaces "Cache" (the instance) as used across Document &
co.
Only Cache instance with on-disk persistence.
ImageCache: Update to new Cache.
GlyphCache: Update to new Cache.
Also, actually free glyph bbs on eviction.
* Minor updates to the min & max cache sizes (16 & 64MB). Mostly to satisfy my power-of-two OCD.
* Purge broken on-disk cache files
* Optimize free RAM computations
* Start dropping LRU items when running low on memory before pre-rendring (hinting) pages in non-reflowable documents.
* Make serialize dump the most recently *displayed* page, as the actual MRU item is the most recently *hinted* page, not the current one.
* Use more accurate item size estimations across the whole codebase.
TileCacheItem:
* Drop lua-serialize in favor of Persist.
KoptInterface:
* Drop lua-serialize in favor of Persist.
* Make KOPTContext caching actually work by ensuring its hash is stable.
* Ensure that going from one to the other tears down the former and
its plugins before instantiating the latter and its plugins.
UIManager: Unify Event sending & broadcasting
* Make the two behave the same way (walk the widget stack from top to
bottom), and properly handle the window stack shrinking shrinking
*and* growing.
Previously, broadcasting happened bottom-to-top and didn't really
handle the list shrinking/growing, while sending only handled the list
shrinking by a single element, and hopefully that element being the one
the event was just sent to.
These two items combined allowed us to optimize suboptimal
refresh behavior with Menu and other Menu classes when
opening/closing a document.
e.g., the "opening document" Notification is now properly regional,
and the "open last doc" option no longer flashes like a crazy person
anymore.
Plugins: Allow optimizing Menu refresh with custom menus, too.
Requires moving Menu's close_callback *after* onMenuSelect, which, eh,
probably makes sense, and is probably harmless in the grand scheme of
things.
* PluginLoader: Allow querying whether a specific plugin is loaded (instantiated, actually).
Re #7598
* PluginLoader: Add a method to access a specific Plugin's instance
(Besides `self.ui[plugin_name]`, that is).
Requires some nastiness to avoid pinning stale references in memory :s.
* PluginLoader: Tweak genPluginManagerSubItem to keep from rebuilding `self.all_plugins`, it's using `loadPlugins`, which is cached, so it made no sense to re-compute this one.
Includes:
- MathML: a few minor fixes
- (Upstream) lvtext: fix possible index out of range
- Fonts: RegisterExternalFont() should take a documentId
- Fonts: fix: letter-spacing should not be applied on diacritic
- (Upstream) Fonts: more granular synthetic weights
- Fonts: synthesized weights: tweak some comments
- Fonts: keep hinting with synthetic weight
- Fonts: fix synthesized weight inconsitencies
- Fonts: fix getFontFileNameAndFaceIndex()
- Fonts: adds LVFontMan::RegularizeRegisteredFontsWeights()
- Fonts: handle synth_weight tweaks in glyph/glyphinfo slots
- (Upstream) Fonts: fix some compiler warnings
- Fix hyphenation on Armenian and Georgian text
Update the bottom menu widget "Font Weight" to allow more
granular weights than the previous "regular | bold".
Also bump thirdparty/luasec to v1.0.1.
successful authentication.
Fully tearing down Wi-Fi was a bit optimistic, as the AP list can
technically still be up, so the user might want to try again and/or
connect to another AP.
Fix#5912, regression since #4616.
The reasoning behind #4616 doesn't really apply anymore anyway, as the
Wi-Fi prompt now handles this inconsistent state properly.
The whole codepath should be *extremely* rare anyway (and/or require
super-broken network conditions).
Similar in spirit to the previous commit, although since we're drawn as
part of the CRe page, this is just the refresh per-se that's optimized,
it had no chance to actually be visible.
Make sure we don't repaint the footer over anything else than ReaderUI,
as some events are fired without direct UI interaction, so we can't be
assured of the actual state of the UI.
Fix#7556
Exposed since #7379, but could arguably happen with other existing
handlers before.
* Support cancelling individual moves.
Coopt the cancel button to actually do just that, cancel, instead of close.
* Don't close when hitting the accept button, allowing to chain multiple moves.
Changes are still propagated to the caller on each individual move, though.
* Update the new icons to match our usual stroke width.
* ReaderToc: Tweak padding around items to be a tad more balanced for multi-level ToCs, and perfectly balanced for flat ToCs. (Re #7548)
* ReaderToc: Make the onHold popup's width match said padding. (Re #7548)
* InfoMessage: Allow passing the alignment flag to TextBoxWidget
* ReaderToc: Center text inside the InfoMessage onHold popup
The object was never re-assigned, so closing a smaller menu (e.g.,
Calibre metadata search) made the underlying one (e.g., CoverBrowser's
ListMenu) inherit the smaller dimensions...
Instead of creating the object in the Class constructor, create it in the
instance constructor (i.e., :init).
Similar cleanups in other Menu* related classes.
* Input: Don't create a new TimeVal object for input frame timestamps, just promote our existing table by assigning it the `TimeVal` metatable.
* TimeVal: Export (const) `zero` & `huge` TimeVal objects, because they're common enough in our codebase. (NOTE: not actually const, that's a Lua 5.4 feature ;p).
* GestureDetector: Explain the behavior of the `last_tevs` & `first_tevs` tables, and why one needs a new object and not the other.
* Speaking of, simplify the copy method for `first_tevs`, because it doesn't need to create a new TimeVal object, we can just reference the original, it's unique and re-assigned for each frame.
* Wireless: Optimize memory usage in StreamMessageQueue (use an array of string ropes, that we only concatenate once). Allowed to relax the throttling, making transfers that much faster.
* Persist: Add a "zstd" codec, that uses the "luajit" codec, but compressed via zstd. Since both of those are very fast, it pretty much trounces everything in terms of speed and size ;).
* Persist: Implemented a "writes_to_file" framework, much like the existing "reads_from_file" one. And use it in the zstd codec to avoid useless temporary string interning.
* Metadata: Switch to the zstd codec.
There have been a couple of these this month, and keeping stuff that should only ever run once piling up in their respective module was getting ugly, especially when it's usually simple stuff (settings, files).
So, move everything to a dedicated module, run by reader.lua on startup, and that will actually only do things once, when necessary.
* TimeVal: Log the results of the COARSE probes in debug logs
* GestureDetector: Print details of failed clock probes in debug logs
* GestureDetector: Skip the BOOTTIME probe when CLOCK_BOOTTIME is unsupported.
* Input: Decode ABS_DISTANCE events in debug logs
* Get rid of duplicated <linux/input.h> constants, use the FFI module everywhere (re #7536)
* Kobo: Get rid of the `touch_alyssum_protocol` quirk. Replace it by setting `main_finger_slot` to `1`, like on the H2O.
The Wacom pen reports sane CLOCK_REALTIME timestamps, but the Touchscreen reports timestamps frozen at the boot time ts (in REALTIME) :s.
So, make everyone use synthetic CLOCK_MONOTONIC timestamps in order to make gesture detection behave.
* CalibreMetadata: Get rid of the now useless NULL-hunt: here, this was basically looking for `rapidjson.null` to replace them with... `rapidjson.null` :?. IIRC, that's a remnant of a quirk of the previous JSON parser (possibly even the previous, *previous* JSON parser ^^).
* CalibreSearch: Update the actually relevant NULL-hunt to make it explicit: replace JSON NULLs with Lua nils, instead of relying on an implementation detail of Lua-RapidJSON, because that detail just changed data type ;).
* UIManager: Make sure tasks scheduled during the final ZMQ callback are honored. e.g., the Calibre "Disconnect" handler. This happened to mostly work purely by chance before the event loop rework.
* Calibre: Restore a proper receiveCallback handler after receiving a book, in order not to break the "Disconnect" handler's state (and, well, get a working Disconnect handler, period ^^).
* Calibre: Unbreak metadata cache when it's initialized by a search (regression since #7159).
* Calibre: Handle UTC <-> local time conversions when checking the cache's timestamp against the Calibre metadata timestamp.
* Bump base (Unbreak CRe on Android, update RapidJSON)
Attempting to update a highlight when matching on a bookmark doesn't
make sense, and no longer works since #7411
Also, switch to ipairs(), and stop piggybacking on a new local that
shadows or uses the function's item argument.
It works in Lua, but it's super weird and error-prone.
Fix#7520 (regression since #7411).
Cache's state tracking
Re #7510
Legacy cache has to be deleted in Cache,
*before* the first getDiskCache call.
Reword the "Clear caches" Dev menu entry & callback:
We now have an actual "Restart" action (on most, if not all, platforms).
And we clear the whole cache folder, which may host other things (e.g.,
fontlist)
