ARROW-9268: [C++] add string_is{alpnum,alpha...,upper} kernels#7656

Closed

maartenbreddels wants to merge 28 commits intoapache:masterfrom maartenbreddels:ARROW-9268

+826-3

Commits on Jul 12, 2020

ARROW-9268: [C++] add string_is{alpnum,alpha...,upper} kernels

maartenbreddels
authored and
wesm
committed
lint python

maartenbreddels
authored and
wesm
committed
convert arrow scalar to python value

maartenbreddels
authored and
wesm
committed
move utf8proc code to ifdef block

maartenbreddels
authored and
wesm
committed
remove unused variable

maartenbreddels
authored and
wesm
committed
use UTF8PROC_CATEGORY_CN instead of 0

maartenbreddels
authored and
wesm
committed
better version of islower/isupper using trick and more tests

maartenbreddels
authored and
wesm
committed
put tests in place that document what we do not support

maartenbreddels
authored and
wesm
committed
make python tests more robust by ignoring undefined codepoints

maartenbreddels
authored and
wesm
committed
rename string_func_unicode -> utf8_func

maartenbreddels
authored and
wesm
committed
more tests

maartenbreddels
authored and
wesm
committed
some compilers look at unused templates

maartenbreddels
authored and
wesm
committed
misplaced ifdefs

maartenbreddels
authored and
wesm
committed
use lookup table and use isalnum for benchmark

maartenbreddels
authored and
wesm
committed
use utf8 sequence in string (maybe the compiler does not like the literal utf8 char if it does not recognize it?)

maartenbreddels
authored and
wesm
committed
Fix: was not using lut

maartenbreddels
authored and
wesm
committed
performance increase: ~15-20%

maartenbreddels
authored and
wesm
committed
use different high codepoint that does not require unicode 13

maartenbreddels
authored and
wesm
committed
Use 'ArrowType' for template parameter instead of clashing 'StringType'
wesm
committed
compiler bug workaround

maartenbreddels
authored and
wesm
committed
wrong template argument

maartenbreddels
authored and
wesm
committed
change crtp inheritance

maartenbreddels
authored and
wesm
committed
CRTP -> inline lambdas
wesm
committed
Instantiate fewer templates
wesm
committed
Some fixes, try something else to see if it fixes the gcc compilation issue
wesm
committed
Move LUT initialization out of if block
wesm
committed
Compile without utf8proc
wesm
committed
Skip failing test on older utf8proc
wesm
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ARROW-9268: [C++] add string_is{alpnum,alpha...,upper} kernels#7656

Commits on Jul 12, 2020

ARROW-9268: [C++] add string_is{alpnum,alpha...,upper} kernels

lint python

convert arrow scalar to python value

move utf8proc code to ifdef block

remove unused variable

use UTF8PROC_CATEGORY_CN instead of 0

better version of islower/isupper using trick and more tests

put tests in place that document what we do not support

make python tests more robust by ignoring undefined codepoints

rename string_func_unicode -> utf8_func

more tests

some compilers look at unused templates

misplaced ifdefs

use lookup table and use isalnum for benchmark

use utf8 sequence in string (maybe the compiler does not like the literal utf8 char if it does not recognize it?)

Fix: was not using lut

performance increase: ~15-20%

use different high codepoint that does not require unicode 13

Use 'ArrowType' for template parameter instead of clashing 'StringType'

compiler bug workaround

wrong template argument

change crtp inheritance

CRTP -> inline lambdas

Instantiate fewer templates

Some fixes, try something else to see if it fixes the gcc compilation issue

Move LUT initialization out of if block

Compile without utf8proc

Skip failing test on older utf8proc