chore(arrow-cast): support white space prefixed parse for ints and floats

[Rich-T-kid]

Which issue does this PR close?

• Closes half of arrow-cast numeric parsers fail to parse whitespace-padded strings #9538.

Rationale for this change

currently parsing ints and floats with white space prefiing the number caused the cast to fail

assert_eq!(Float64Type::parse("\n\t\n\t\n40.5123"), Some(40.5123));
assert_eq!(Float64Type::parse(" 1.5"), Some(1.5));
assert_eq!(Int32Type::parse(" 3"), Some(3));
assert_eq!(Int32Type::parse("          30"), Some(30));

What changes are included in this PR?

this PR adds a check to the string before parsing. it checks the first byte, if its not an ascii digit it will trim the left side of the string.

Performance was a consideration to avoid regressions here — see #9537 (comment). In the worst case where a string doesn't need trimming (no leading whitespace), the cost is just one slice operation to grab the first byte plus a range comparison to check if it's an ASCII digit.

benchmarks show no significant change

Are these changes tested?

yes. I created a test with 12 assertiosn to validate behavoir

Are there any user-facing changes?

yes. arrow-rs/datafusion cast operations will no longer fail when a number is prefixed with white space.

[Rich-T-kid]

cc @alamb @aryan-212 @Rafferty97 since you had interest in #9538 & #9537

[Rich-T-kid]

supporting post-fix white space should be pretty similar as well

[aryan-212]

@Rich-T-kid Since we are only concerned with ascii whitespaces, you can use trim_ascii_start() instead of trim_start() which is catered towards unicode whitespaces. Would be even faster, perhaps.
wdyt?

[Rich-T-kid]

@Rich-T-kid Since we are only concerned with ascii whitespaces, you can use trim_ascii_start() instead of trim_start() which is catered towards unicode whitespaces. Would be even faster, perhaps. wdyt?

@aryan-212 thanks for the suggestion, ill update the PR

[Jefffrey]

the screenshotted benchmarks are for string -> f32, but it seems f32/f16 paths are not changed only f64. i think we need to add to f32/f16 and rerun those benchmarks

[Rich-T-kid]

the screenshotted benchmarks are for string -> f32, but it seems f32/f16 paths are not changed only f64. i think we need to add to f32/f16 and rerun those benchmarks

whoops 😅. I forgot to add this to the f32/16 branches.

@Jefffrey should be good now

[Jefffrey]

run benchmark cast_kernels
env:
BENCH_FILTER: utf8 to f32

[adriangbot]

🤖 Arrow criterion benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

group               main                                   rich-T-kid_allow-white-space-aware-parse
-----               ----                                   ----------------------------------------
cast utf8 to f32    1.01      5.7±0.04µs        ? ?/sec    1.00      5.7±0.03µs        ? ?/sec

Resource Usage

base (merge-base)

Metric	Value
Wall time	10.0s
Peak memory	7.4 MiB
Avg memory	5.7 MiB
CPU user	7.6s
CPU sys	0.0s
Peak spill	0 B

branch

Metric	Value
Wall time	15.0s
Peak memory	7.3 MiB
Avg memory	5.0 MiB
CPU user	10.0s
CPU sys	0.0s
Peak spill	0 B

---

File an issue against this benchmark runner

[Rich-T-kid]

cast utf8 to f32 1.01 5.7±0.04µs ? ?/sec 1.00 5.7±0.03µs ? ?/sec

nice

[Jefffrey]

run benchmark cast_kernels
env:
BENCH_FILTER: utf8 to f32

[adriangbot]

🤖 Arrow criterion benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

group               main                                   rich-T-kid_allow-white-space-aware-parse
-----               ----                                   ----------------------------------------
cast utf8 to f32    1.00      5.7±0.03µs        ? ?/sec    1.00      5.7±0.03µs        ? ?/sec

Resource Usage

base (merge-base)

Metric	Value
Wall time	10.0s
Peak memory	7.4 MiB
Avg memory	5.7 MiB
CPU user	7.8s
CPU sys	0.0s
Peak spill	0 B

branch

Metric	Value
Wall time	10.0s
Peak memory	7.3 MiB
Avg memory	5.3 MiB
CPU user	7.0s
CPU sys	0.0s
Peak spill	0 B

---

File an issue against this benchmark runner

[Jefffrey]

thanks @Rich-T-kid & @aryan-212

[alamb]

Nice!