encodation;
use code set E for padding as well, saves codeword, props Bue
Jensen (BWIPP PR #279);
add padding after all segments and limit loops to length to make
NS compression work better;
all the above temporary fixes until Bue Jensen's merge request
with the BWIPP PR #279 algorithm
GUI: expand linux "xcb" platform hack for >= Qt 5.1
ISO/IEC 15420:2009 4.3.5 (and BWIPP) and adjust righthand outside
digit to be 4X/2X instead of 5X/3X (ISO/IEC 15420:2009
Annex A.2) away from main symbol so as not to touch add-on
TODO: revisit when standard clarified
encodation in certain cases (and no pessimizations found so far),
props lyngklip (BWIPP);
fix extended char latching when exactly 3 extended chars at end;
count code set C (not digits) in loop deciding when to
shift/latch to extended for better estimate
AZTEC: return warning if ECC < 5% (due to bit-stuffing when version
given); return error if > 22 layers (Zint 26) for Reader
Initialisation symbol requested for better error message
AZTEC/HANXIN/QRCODE: consolidate different ECC data size tables
into one indexed by ECC
DBAR_EXP: check for reduced length <= 77 up front for better error
message
HANXIN: use `malloc()` rather than `z_alloca()` for large binary
array
QRCODE: `ecc_level` now 0-based (not 1-based)
MICROQR: consolidate different version end routines into one
`microqr_end()` and use new `microqr_data` table to simplify code
MICROPDF417: use table for max codewords per column
library: centralize all error messages using new `errtxt()`,
`errtxtf()`, `errtxt_adj()` funcs that protect `symbol->errtxt`
from overflow, & try to make error messages more consistent
thru-out, adding more feedback info to many, & use positional
args "%n$" in prep for l10n (maybe);
`is_sane/is_sane_lookup()` -> `not_sane/not_sane_lookup()`,
returning 1-based position (zero on failure) instead of bool;
`long` ints -> plain `int` (except those dealing with `ftell()`,
`fread()` etc) as depend on int being 32-bits already
GUI: in "grpDATF.ui" use "PlainText" rather than "RichText" for
tracker ratio examples as height of text messing up sometimes
manual: clarify Codablock-F length maximum & add examples
docs: README: pandoc 3.5, Ubuntu 24.04
CMake: use "-Wpedantic" for Clang only as GNU complains about
`errtxtf()` positional args "%n$"
Adapted from OkapiBarcode, with stricter interpretation (only pad
if "+4" totally absent), props Daniel Gredler
gs1: update to latest gs1-syntax-dictionary (linter mm -> mi)
bwipp: update to latest
avoid various weird behaviours
tests: disable GUI-dependent tests (`TestQZint::renderTest()`) if
ZINT_SANITIZE and on Linux for later versions of Qt (5 and 6)
to avoid ASAN "detected memory leaks" errors on exit
CODE49: Better error message on ZINT_ERROR_TOO_LONG
manual: Use floating pt notation for floating pt args on options
backend/tools/data: Remove overlooked "GB2312.TXT" from git
raster: `size2` -> `prev_size`; one line `malloc()`s
C25/CODE128: some code fiddling
tests/PNG: Add some more text examples
from ZXing (props Alex Geller) - about 25-60% slower depending on
data & stack heavy but does improve some outcomes when FNC1s
present (GS1_MODE or manual) although not much else it appears
(the previous algorithm was very good), but has a logical clarity
the other lacked - funcs `c128_dxsmooth()` etc shared with
CODE16K now moved there and renamed as `c16k_`;
also fix extended char latching when exactly 4 extended chars
at end
also manual code set switching now honoured exactly even if
immediate shift required;
manual: make explicit that AI "(00)" and "(01)" prefixes added by
Zint are HRT-only
general: add `extern "C"` wrappers to a few header files
ticket #324, props Jim Shank;
also improve encodation for a few limited cases;
also some code fiddling
BWIPP: update to latest
raster: check for overflow on `size + size2` in `raster_malloc()`
vector: put `malloc()`s on one line for grep ease
docs: pandoc-3.3, clang-tidy-20
`iso3166list` linter and adjusting all others to allow for
multiple optional linters (by checking `data_len` vs `offset`)
library: change invalid `input_mode` reset to return warning;
split func table into 2 - func declarations without prototype
will be error in C23;
make invalid symbology check a separate function using a table
and call near beginning of `ZBarcode_Encode_Segs()`;
in `ZBarcode_BarcodeName()` save some bytes by simplifying
name table and removing "BARCODE_" prefix from entries
output: fix pack logic to use pragma by default (actually more
portable than `__attribute__`)
common.h: remove C99 detection which was artifice of specifying
"-std=", and rejig layout to be more logical
BWIPP: update to latest version
general: change `ZINT_VERSION_BUILD` tests to `#if`s
general: further fiddling with some tables to save a few bytes
CLI: change function arg `optarg` -> `arg` so doesn't shadow
global
general: library now compilable with Comp Cert C (though as it
doesn't support `alloca()` will have multiple memory leaks)
"SetPaths.cmake" (removed) - resolves absolute paths in
"zint-targets.cmake", ticket #318, props John Alexander;
also remove "FindZint.cmake" install as not best practice
(Daniel Pfeifer “Effective CMake"), but retain in tree for now;
also remove APPLE stuff re UNIVERSAL build, obsolete, and
setting of CMAKE_OSX_SYSROOT, seems unnecessary
[f3a982c1dd])
AZTEC/CODE16K/CODEONE/DATAMATRIX/DBAR_EXP/GRIDMATRIX/HANXIN
/MAILMARK_4S/PDF417/MSI_PLESSEY/QRCODE/RMQR/TIF/ULTRA/USPS_IMAIL:
lessen storage of some tables by using least integral size
required (saves ~3K); reformat some tables, comments
AUSPOST/AZTEC: remove unnecessary int casts on array indexing
CODE11/CODE39: move start/stop chars into one entry in tables to
save a few bytes; some reformatting, comments
CODEONE: add `c1_` prefixes
common: more precise compiler/version checks
composite: add `cc_` prefixes; UINT -> unsigned short; use
`sizeof()` instead of `strlen()` for `in_linear_comp` static;
some reformatting, comments
EMF: use table instead of string for `ultra_chars[]`
GIF: remove unnecessary cast
library: use `sizeof()` instead of `strlen()` for static;
add `consts` to `escape_char_process()` & use unsigned for `ch`
DBAR: use `dbar_`, `dbar_ltd_`, `dbar_exp_` prefixes
docs: update pandoc version
inner top/right/bottom/left loops of `az_populate_map()` by
jiggling them around a bit, ticket #317, props Andre Maute
frontend/tests: clean up any directories created in `test_input()`,
ticket #316, props Andre Maute
Previously, width is option 2 and security level is option 1.
Seems like in 54947fb435, the definitions
swapped and that breaks the legacy Qt application using setWidth() and
setSecurityLevel().
I know a better solution is make Qt application use setOption2() and
setOption1() respectively, but it's good to keep these legacy APIs
backward-compatible.
(ticket #181, props Stf Ortm)
CODE128: suppress gcc -Wmaybe-uninitialized false positive (Release
only)
CLI: add "isbn" synonym; save a few bytes using `char[]` rather
than `char *`
allow for old "libpng" (`png_set_scale_16()` not available)
general: Solaris compat
library: use proper function ptr instead of `void *` for function
table; warning suppression "-Wpedantic" -> "-Wstrict-prototypes"
GRIDMATRIX/HANXIN/QRCODE: `xx_define_mode()`: multi-dim `char_modes`
- GUI: scalewindow: fix cropping resolution on initial setup
(`spnResolution` max 1000 -> 254000) and bound X-dim to <= 10 in
`set_maxima()`
- GUI: mainwindow: bound X-dim as above and clamp `m_xdimdpVars`
members on initial load from INI
- GUI: undo `QString::mid()` -> `QString::midRef()` from clazy
& explicitly include "QObject" in "qzint.h" (not Qt6 compatible)
and `out_colour_get_cmyk()`, to lessen chances of rounding
errors (in particular for VC6)
test_large: fix regression in change to test for formatting
`uint64_t` for `printf()`
win32/README: update MinGW/MSYS Qt version
test `ferror()` also in `fm_err()` if `err` zero and file
NetBSD: add instructions and some workarounds (`getopt_long_only()`
in particular)
qzint: check `__GNUC__` version for "-Wstringop-truncation"
suppression
