MB_CUR_MAX is the maximum number of bytes in a multibyte character
for the current locale, and might not be a constant expression.
MB_LEN_MAX is the maximum number of bytes in a multibyte character
for any locale, and always expands to a constant-expression.
* sprinkle const, static
* account for lineno in unput
* Add an EMPTY string that is used when a non-const empty string is needed.
* make inputFS static and dynamically allocated
* Simplify and in the process avoid -Wwritable-strings
* make fs const to avoid -Wwritable-strings
An input/output error indicates a fatal condition, even if it
occurs when closing a file. Awk should not return success on I/O
error, but treat I/O errors as it already treats write errors.
Test case:
$ (trap '' PIPE; awk 'BEGIN { print "hi"; }'; echo "E $?" >&2) | :
awk: i/o error occurred while closing /dev/stdout
source line number 1
E 2
The test case pipes a line into a dummy command that reads no
input, with SIGPIPE ignored so we rely on awk's own I/O checking.
No write error is detected, because the pipe is buffered; the
broken pipe is only detected as an I/O error on closing stdout.
Before this commit, "E 0" was printed (indicating status 0/success)
because an I/O error merely produced a warning. A shell script
was unable to detect the I/O error using the exit status.
On case-insensitive file systems (i.e.: macOS), T.concat and
t.concat are the same file, so these conflicted. This commit
renames T.concat to avoid the conflict.
Further simplify printf % parsing by eating the length specifiers
during the copy phase, and substitute 'j' when finalizing the format.
Add some more tests for this.
* More cleanups:
- sprinkle const
- add a macro (setptr) that cheats const to temporarily NUL terminate strings
remove casts from allocations
- use strdup instead of strlen+strcpy
- use x = malloc(sizeof(*x)) instead of x = malloc(sizeof(type of *x)))
- add -Wcast-qual (and casts through unitptr_t in the two macros we
cheat (xfree, setptr)).
* More cleanups:
- add const
- use bounded sscanf
- use snprintf instead of sprintf
* More cleanup:
- use snprintf/strlcat instead of sprintf/strcat
- use %j instead of %l since we are casting to intmax_t/uintmax_t
* Merge the 3 copies of the code that evaluated array strings with separators
and convert them to keep track of lengths and use memcpy instead of strcat.
* Fix formats for 32 bit machines broken by previous commit.
We use intmax_t to provide maximum range for both 32 and 64 bit machines.
* More cleanups:
- sprinkle const
- add a macro (setptr) that cheats const to temporarily NUL terminate strings
remove casts from allocations
- use strdup instead of strlen+strcpy
- use x = malloc(sizeof(*x)) instead of x = malloc(sizeof(type of *x)))
- add -Wcast-qual (and casts through unitptr_t in the two macros we
cheat (xfree, setptr)).
* More cleanups:
- add const
- use bounded sscanf
- use snprintf instead of sprintf
* More cleanup:
- use snprintf/strlcat instead of sprintf/strcat
- use %j instead of %l since we are casting to intmax_t/uintmax_t
* Merge the 3 copies of the code that evaluated array strings with separators
and convert them to keep track of lengths and use memcpy instead of strcat.
- sprinkle const
- add a macro (setptr) that cheats const to temporarily NUL terminate strings
remove casts from allocations
- use strdup instead of strlen+strcpy
- use x = malloc(sizeof(*x)) instead of x = malloc(sizeof(type of *x)))
- add -Wcast-qual (and casts through unitptr_t in the two macros we
cheat (xfree, setptr)).
As it is never dereferenced in the n == -1 case it shouldn't cause any
problems. However, UBSAN complains about this, so it is required to run
the tests when compiling with -fsanitize=undefined.
When matching a caret, the expression `f->gototab[s][c] = f->curstat;` in
cgoto() will index the 2D-array gototab with [s][261]. However, gototab
is declared as being of size [NSTATES][NCHARS], so [32][259]. Therefore,
this assignment will write to the state for character 0x1.
I'm not sure how to create a regression test for this, but increasing the
array size to HAT+1 values fixes the error and the tests still pass.
I found this issue while running awk on a CHERI system with sub-object
protection enabled. On x86, this can be reproduced by compiling awk
with -fsanitize=undefined.
