Make use of signals to connect the selection in plot to the associated
table widget. Every time that the plot is updated from the Main Window
the table widget associated to the table or query is connected to the plot
and the previous widget is disconnected. This allows the selection of the
correct table widget.
Line selection methods moved to the Extended Table Widget to be used as
slots for this connection.
The destroyed signal is also connected for resetting the plot. This fixes
a crash that already existed before this PR, when closing a SQL tab while
the plot is still associated to the table results model.
Revert the savepoint in the Execute SQL whenever possible. We need to
do this after each statement because there are some rare cases where
the next statement might be affected by what is only a temporary and
unnecessary savepoint. For example in this case:
ATTACH 'xxx' AS 'db2'
SELECT * FROM db2.xy; -- Savepoint created here
DETACH db2; -- Savepoint makes this statement fail
See issue #1249.
When dragging and dropping a table from one instance of the application
to the other, the tree structure representing the database was broken.
We would show the 'Browsables' and 'All' nodes at the top level instead
of the child nodes of the 'All' node. This happened because after
dropping a table, we would reload the database structure and rebuild the
tree structure but didn't notify the tree view in the main window about
the update. This is fixed by this commit, so the main window's widgets
are always notified about the new tree structure.
See issue #1288.
The tables/queries sorted by X are drawn using QCPGraph as before.
Since QCPGraph does automatically sort by X, we change to QCPCurve that
requires a third data vector to reflect the order. We get that from the
current row order.
In the case of curves, only None and Line is supported as line style.
Since the order is now important for the plot, it is automatically updated
whenever the user sorts by another column in the browsed table.
This addresses issue #821 and indirectly fixes the problem of incorrect
point->row selection link when the table is not sorted by X, reported in
issue #838.
Before bbac655499 it was possible to delete
a set of selected rows. This makes it possible again by only selecting the
row if it is not already inside the selected rows. See issue #1283.
Additionally and for coherence, the "Duplicate record" from the context
menu is also made to apply to the list of selected rows. See issue #1090
This makes sense from the user point of view, since it remarks that the
options from the menu applies to that row. This change will also fix the
different behaviour of the "Delete record" and "Duplicate record" in the
context menu. See issue #1283
Don't set the modified flag of the main database if we're attaching or
detaching another database since these actions don't alter the original
database.
See issue #1249.
The current table encoding is used in the binary check, so text encoded in
8-bit encodings is properly recognised as text in the table widget.
Make it possible to reset the encoding used in the way suggested to the
user: "Leave the field empty for using the database encoding". Currently it
was rejecting it with the message: "This encoding is either not valid or
not supported."
See issue #1279
The shortcut is constrained to the Widget context so it does not conflict
with other SqlTextEdit widgets in the application. Now all SqlTextEdit
widgets are able of opening the dialog with the same shortcut and behave
in the same way (do not block the parent window).
The find/replace dialog can now be discovered by the user in the context
menu at the SqlTextEdit widgets.
This was mistakenly introduced in 7ed1b1db05
The unwanted effect was two new duplicate records inserted.
Additionally the shortcut is made self-documented for the user in the
corresponding context menu.
That avoids overwriting existing record as reported in issue #1255.
This doesn't improve the underlying situation, that is inserting empty rows
before duplicating the content. But it is safer to not ignore the error in
the initial row insertion.
Disable 'Attach Database' when no database is opened.
Added to the function that enables and disable actions depending on
database state. See issue #1254
Display more specific 'Choose a file...' messages in two places. See issue
#331.
Two new application toolbars: Extra DB (attach, close database) and Project
(Open, Save). Can be disabled and moved as usual.
Icons for Close Database and Attach Database actions, which lacked ones.
Different icon for the dock configuration, so they are visually
distinguishable from the toolbar configurations.
New preference for the toolbar style, with default value: text besides icon
as before. But in order for the style to change, the value in UI must be
ToolButtonIconOnly.
Needed new icons from famfamfam.com's Silk collection.
See related issue #331
The query used in sqlitetablemodel does not rename the columns, instead
it applies the display formats in the WHERE part. In this way the saved
view has the original column names.
Changes in Button "Save filter as view": more information in tooltip and it
is disabled when no database is open.
See discussion in PR #1246
New view name validated with isNull().
Text in information box reworded.
Column names are preserved in the new view, except for the ones with
display formats, for which an underline is appended in order to fix the
reported filtering defect.
Make sure the keyboard shortcut for the Find & Replace dialog is always
set to Ctrl+H instead of using Qt's QKeySequence::Replace. At least on
KDE desktops but probably on other systems too it break the Ctrl+R
shortcut which is used for refresh and SQL execution.
A new button is added to the Browse Data tab for saving the current display
of the table (current filter, sort column and display formats) as a new
view. This allows (specially for non advanced users) the creation of simple
views. It can be seen, either as a way of storing the current
filtering or as an easy way of creating views.
This reuses the query set in sqlitetablemodel, but the column aliases when
a display format is used has been changed from col%1 to the original column
names, i.e. format(`orig`) AS `orig`.
Find/Replace dialog does not block the main window.
Dialog reused, so checks changed by user are preserved.
The selection of the last occurrence in Find All and Replace All
is cleared, so all the occurrences have the same appearance.
A classic Find/Replace dialog including all the options provided
by the QScintilla editor. Additionally a "Replace All" and a
Find All" button are implemented. This complements issue #191
New shortcut in main window for opening the find/replace dialog
connected to the focused widget. Standard for Replace is Ctrl+H.
Consequently the shortcut for Copy with Headers is changed to
Ctrl+Shift+C (see issue #1058).
For SQL text widgets not embedded in the main window, the standard
shortcut cannot be enabled, otherwise it conflicts with the main
window. Therefore the shortcut Ctrl+Shift+H is enabled for all the
SQL widgets as fallback. A better solution would be desirable.
* Add Signal if File dropped
* Check if Index is valid
Send Signal and return if Index is invalid and File was dropped
* Connect Dropsignal to fileopen
results pane
Copy for multi-line text data surrounds the text with the
<pre> tag in the HTML version for avoiding line white-space
adjustments in the pasted text. See #1058
Do not use binary data in the "Use as Filter" option.
Indicate the existence of an error in the background of the
SQL results frame. The error-indicators is reused for
disabling this functionality.
When nothing is shown, a placeholder text hints the
objective of the results pane.
The preference label for the font size of this pane is
modified for consistency (SQL Log is another widget).
New button and key sequence (ESC) to close the search bar.
Clear selection (by not avoiding the search) when the text to search is
deleted, as expected by @pamtbaau at #1220.
The user is now able of selecting a range of contiguous points in any of
the graphs of the plot, through rectangle selection or click in single
point. This selection triggers automatically the selection of the
corresponding lines in the browsed table. In this way the user can analyse
the plot in detail.
Two other minor improvements: spelling error fixed and the dummy column in
the plot selection table has now transparent foreground (intends to avoid
wondering of the user about what functionality provides this column).
If multiple SQL tabs are opened, show or hide the find bar for all of
them when changing the check state of the button. Also apply the find
bar state to newly opened SQL tabs.
This also fixes a crash which would happen if you click the Find button
prior to opening any SQL tabs, i.e. before opening the first database
file.
The find operations in QScintilla are used for implementing a find frame
in the SQL Editor of the main window. A checkable tool button is added
in the SQL Editor toolbar for showing and hiding this frame.
There are check boxes for Whole Words, Case Sensitive and Regular
Expressions. The QScintilla facilities in the findFirst method are used
for the implementation.
This fixes a regression which recently was introduced by commit
955848f957. Because the status message was
overwritten by a version without the returned row count, you would never
see the intended status message when running SELECT statements in the
Execute SQL tab.
See issue #1211.
Don't use the QSettings class directly. This keeps the code more
consistent and makes it a bit easier to read. It also means that all
parts of the code profit from the settings cache that we have
implemented in the Settings class.
This should improve the detection of read-only query statements vs.
modifiying statements in the Execute SQL tab. The idea is to stop
looking for the SELECT keyword at the beginning of the statement and
instead fully rely on whether SQLite returns any data for this
statement.
See issue #1185.
Remove the valid flag from the SqliteTableModel class and remove its
usage in the Execute SQL tab of the main window. I believe this hasn't
been used for some time now because the main sources of error should
really be noticed before the query is even handed over to the model.
Additionally the valid flag wasn't as realible either anymore because of
the multi-threaded execution of the model queries.
The no collate function warning is triggered in a thread which is used
for loading data. However, the warning is a message box and GUI elements
can only be drawn in the main thread. So the old code would crash. This
is fixed here by jumping to the main thread for showing the message box.
