Invoke endwin() *after* calling any other curses functions, such as
refresh(). Calling refresh() after endwin() might restore curses mode
which is a bad thing for a terminal mode initialization routine.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
* Do window preparation and restoring in separate functions
wins_prepare_external() and wins_unprepare_external().
* Use fork_exec() and child_wait() instead of system().
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Can be used to execute an external program. Provides the possibility to
optionally create a pipe to the new process.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Add a todo_write() function that allows one to serialize todo items and
write serialized data to an output stream in a fashion similar to
apoint_write() and event_write().
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
* Sort character width lookup table by character ranges.
* Use binary search instead of linear search for UTF-8 character width
lookups which will speed up utf8_width() (O(log n) instead of O(n)).
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Avoid a segfault that may occur if getstr_fixscr() changes the scroll
offset to something outside the input boundaries (e.g. if a word that is
only partly visible is killed via backward-kill-word).
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
* Rename static showstring() function to getstr_print(), rename
ins_char() to getstr_ins_char() and del_char() to getstr_del_char().
* Refactor out getstring data structure initialization and window
scrolling routines.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Move getstring() related stuff into a separate file as a first step on
our way to UTF-8 support for line editing helpers.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Add utf8_width() and utf8_strwidth() which can be used to calculate the
display width of a single character or a string, respectively. A lookup
table is used to spot double width characters, as well as composing
characters. There currently isn't any code to deal with ambigious
characters.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
There used to be no resize handler in custom_sidebar_config() which
resulted in a messed up screen if the terminal was resized during
sidebar configuration. Doing a simple wins_reset() if the terminal was
resized works around this problem.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Don't display anything rather than segfault if the appointment panel
becomes too small (e.g. during a terminal resize).
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
This one came up after improving resizing behaviour in commit
fec37db06bbbdff79bc9816046a76e0b29e173b4. We used to use outdated width
and height values at various places without noticing it due to the
resize algorithm being called several times on every resize.
This patch ensures we retrieve the new window dimensions *before* doing
any layout calculations.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Use a global flag to record whether the terminal was resized instead of
redrawing everything each time a KEY_RESIZE is read.
Add some additional checks to help_write_pad() as invalid actions may be
passed now due to using signals instead of virtual key presses.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Ensure the relative order of elements with equal keys is maintained when
inserting into a sorted list.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
We forgot to add those to the "_SOURCES" variable in commit
0eb1da8dd7fa3ae9d95e12df6058fb41a5ec1c11 and
d668963e2056027526374098efe2869a0d389b92. Spotted by `make distcheck`.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
* Update copyright dates (use 2004-2011 as date range everywhere).
* Change copyright holder from "Frederic Culot" to "calcurse Development
Team".
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Do this in a fashion similar to what was done in commit
6ff95238766656c5ea9d5c65c35d3aef93499f60.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
* Set all data members to "NULL" in llist_free_inner() after freeing
them. Altough we normally shouldn't continue working with a list that
already went through llist_free_inner(), this will protect against
dangling pointer bugs in case anyone will ever come up with the idea
of doing so.
* Set list head to "NULL" in llist_free(), basically to put the list
into an initialized state.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Rename "days" structure to "excp" which seems to be a better name here.
Use generic linked lists of excp structures instead of using the "days"
structure which again contains a linked list implementation. Do some
cleanups and invocation fixes.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Field seems to be unused. Verify using following spatch:
----
@@ struct recur_apoint *rapt; @@
* rapt->next
----
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Use them instead of "recur_apoint_list" and "next" pointers in
"recur_event" type variables. Includes some code simplifications and
cleanups.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
To be used with llist_fn_match_t callbacks later.
I feel a bit ill adding those functions. This definitely is a hack.
Ultimately, there should be some generic recur_item_inday() function
that accepts both recurring apointments and events (or some wrapper
structure) instead of parameter galeere. This is not the right place to
fix that tho.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Use the new generic list implementation instead of "next" pointers in
day_item type variables. Includes some cleanups and fixes.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Use the new generic list implementation instead of those insane "next"
pointers in events. Includes some cleanups.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Field seems to be unused. Verify using following spatch:
----
@@ struct apoint *apt; @@
* apt->next
----
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Use the new generic list implementation instead of "apoint_list"
everywhere. Simplify stuff and drop unused variables as well.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Add LLIST_TS_* macros in a fashion similar to the already existing
LLIST_* macros. Unlike the non-thread-safe version, these include
LLIST_TS_LOCK and LLIST_TS_UNLOCK which can be used to lock (and unlock)
a list (hence the thread-safety).
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Mostly in preparation to the pending thread-safe list macros. This way,
we have a similar interface to thead-safe and non-thread-safe lists.
This also adds LLIST_FOREACH and LLIST_FIND_FOREACH macros which can be
used as shortcuts when iterating over all list items or a subset of list
items that is accepted by a callback function.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
As discussed on the mailing lists, the various linked list
implementations we currently use at a dozen of different places in the
calcurse source tree are inconvenient and should be replaced by a single
generic solution.
This is a first approach to introduce such a generic implemetation. It
provides following functions:
* llist_init(): Initialize a list.
* llist_free_inner(): Loop through a list and free all items.
* llist_free(): Free the list itself (but not the individual items).
* llist_first(): Get the first item of a list.
* llist_nth(): Get the nth item of a list.
* llist_next(): Get the successor of a list item.
* llist_find_first(): Find an item using a callback function.
* llist_find_next(): Find the next match using a callback function.
* llist_find_nth(): Find the nth item in a list (using a callback).
* llist_get_data(): Get a pointer to the actual data of a list item.
* llist_add(): Add an item at the end of a list.
* llist_add_sorted(): Add an item to a sorted list (using a comparison
callback function).
* llist_remove(): Remove an item from a list.
Linked lists are stored in "llist_t" structures, list items are to be
stored in "llist_item_t" structs.
All of the llist_*() functions either expect a pointer to a llist_t
structure (in case the function operates on the list itself) or a
pointer to a llist_item_t (llist_*_next() and llist_get_data()).
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Read second byte if the first byte is a newline character to ensure the
file doesn't contain any further data.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Using do-while loops seems more appropriate here as "item_time" hasn't
even been read when either of the loops are entered. Spotted by
clang-analyzer.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
Passing a date in format "mm-dd-yy" where short forms are not allowed
would lead to a null pointer dereference here. This one fixes that.
Spotted by clang-analyzer.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
