Improve std::format's width estimation #3903

achabense · 2023-07-23T06:39:01Z

Resolves #3446

This pr introduces a generator for the new _Width_estimate_intervals table. The new project should include EastAsianWidth.txt(link) somewhere.

Here is its raw output; the "Old table:" part can help to confirm that the interval-generating algorithm is correct; the "Was 1, now 2:" and "Was 2, now 1:" part can help to confirm that this implementation is conformant to the standard (by comparing them with those in the annex of the paper)

Output

Old table:
0x1100u, 0x1160u, 0x2329u, 0x232Bu, 0x2E80u, 0x303Fu, 0x3040u, 0xA4D0u, 0xAC00u, 0xD7A4u, 0xF900u, 0xFB00u,
0xFE10u, 0xFE1Au, 0xFE30u, 0xFE70u, 0xFF00u, 0xFF61u, 0xFFE0u, 0xFFE7u, 0x1F300u, 0x1F650u, 0x1F900u, 0x1FA00u,
0x20000u, 0x2FFFEu, 0x30000u, 0x3FFFEu,

New table:
Input path for EastAsianWidth.txt: EastAsianWidth.txt
0x1100u, 0x1160u, 0x231Au, 0x231Cu, 0x2329u, 0x232Bu, 0x23E9u, 0x23EDu, 0x23F0u, 0x23F1u, 0x23F3u, 0x23F4u,
0x25FDu, 0x25FFu, 0x2614u, 0x2616u, 0x2648u, 0x2654u, 0x267Fu, 0x2680u, 0x2693u, 0x2694u, 0x26A1u, 0x26A2u,
0x26AAu, 0x26ACu, 0x26BDu, 0x26BFu, 0x26C4u, 0x26C6u, 0x26CEu, 0x26CFu, 0x26D4u, 0x26D5u, 0x26EAu, 0x26EBu,
0x26F2u, 0x26F4u, 0x26F5u, 0x26F6u, 0x26FAu, 0x26FBu, 0x26FDu, 0x26FEu, 0x2705u, 0x2706u, 0x270Au, 0x270Cu,
0x2728u, 0x2729u, 0x274Cu, 0x274Du, 0x274Eu, 0x274Fu, 0x2753u, 0x2756u, 0x2757u, 0x2758u, 0x2795u, 0x2798u,
0x27B0u, 0x27B1u, 0x27BFu, 0x27C0u, 0x2B1Bu, 0x2B1Du, 0x2B50u, 0x2B51u, 0x2B55u, 0x2B56u, 0x2E80u, 0x2E9Au,
0x2E9Bu, 0x2EF4u, 0x2F00u, 0x2FD6u, 0x2FF0u, 0x2FFCu, 0x3000u, 0x303Fu, 0x3041u, 0x3097u, 0x3099u, 0x3100u,
0x3105u, 0x3130u, 0x3131u, 0x318Fu, 0x3190u, 0x31E4u, 0x31F0u, 0x321Fu, 0x3220u, 0x3248u, 0x3250u, 0xA48Du,
0xA490u, 0xA4C7u, 0xA960u, 0xA97Du, 0xAC00u, 0xD7A4u, 0xF900u, 0xFB00u, 0xFE10u, 0xFE1Au, 0xFE30u, 0xFE53u,
0xFE54u, 0xFE67u, 0xFE68u, 0xFE6Cu, 0xFF01u, 0xFF61u, 0xFFE0u, 0xFFE7u, 0x16FE0u, 0x16FE5u, 0x16FF0u, 0x16FF2u,
0x17000u, 0x187F8u, 0x18800u, 0x18CD6u, 0x18D00u, 0x18D09u, 0x1AFF0u, 0x1AFF4u, 0x1AFF5u, 0x1AFFCu, 0x1AFFDu, 0x1AFFFu,
0x1B000u, 0x1B123u, 0x1B132u, 0x1B133u, 0x1B150u, 0x1B153u, 0x1B155u, 0x1B156u, 0x1B164u, 0x1B168u, 0x1B170u, 0x1B2FCu,
0x1F004u, 0x1F005u, 0x1F0CFu, 0x1F0D0u, 0x1F18Eu, 0x1F18Fu, 0x1F191u, 0x1F19Bu, 0x1F200u, 0x1F203u, 0x1F210u, 0x1F23Cu,
0x1F240u, 0x1F249u, 0x1F250u, 0x1F252u, 0x1F260u, 0x1F266u, 0x1F300u, 0x1F650u, 0x1F680u, 0x1F6C6u, 0x1F6CCu, 0x1F6CDu,
0x1F6D0u, 0x1F6D3u, 0x1F6D5u, 0x1F6D8u, 0x1F6DCu, 0x1F6E0u, 0x1F6EBu, 0x1F6EDu, 0x1F6F4u, 0x1F6FDu, 0x1F7E0u, 0x1F7ECu,
0x1F7F0u, 0x1F7F1u, 0x1F900u, 0x1FA00u, 0x1FA70u, 0x1FA7Du, 0x1FA80u, 0x1FA89u, 0x1FA90u, 0x1FABEu, 0x1FABFu, 0x1FAC6u,
0x1FACEu, 0x1FADCu, 0x1FAE0u, 0x1FAE9u, 0x1FAF0u, 0x1FAF9u, 0x20000u, 0x2FFFEu, 0x30000u, 0x3FFFEu,

Was 1, now 2:
U+231A..U+231B
U+23E9..U+23EC
U+23F0
U+23F3
U+25FD..U+25FE
U+2614..U+2615
U+2648..U+2653
U+267F
U+2693
U+26A1
U+26AA..U+26AB
U+26BD..U+26BE
U+26C4..U+26C5
U+26CE
U+26D4
U+26EA
U+26F2..U+26F3
U+26F5
U+26FA
U+26FD
U+2705
U+270A..U+270B
U+2728
U+274C
U+274E
U+2753..U+2755
U+2757
U+2795..U+2797
U+27B0
U+27BF
U+2B1B..U+2B1C
U+2B50
U+2B55
U+A960..U+A97C
U+16FE0..U+16FE4
U+16FF0..U+16FF1
U+17000..U+187F7
U+18800..U+18CD5
U+18D00..U+18D08
U+1AFF0..U+1AFF3
U+1AFF5..U+1AFFB
U+1AFFD..U+1AFFE
U+1B000..U+1B122
U+1B132
U+1B150..U+1B152
U+1B155
U+1B164..U+1B167
U+1B170..U+1B2FB
U+1F004
U+1F0CF
U+1F18E
U+1F191..U+1F19A
U+1F200..U+1F202
U+1F210..U+1F23B
U+1F240..U+1F248
U+1F250..U+1F251
U+1F260..U+1F265
U+1F680..U+1F6C5
U+1F6CC
U+1F6D0..U+1F6D2
U+1F6D5..U+1F6D7
U+1F6DC..U+1F6DF
U+1F6EB..U+1F6EC
U+1F6F4..U+1F6FC
U+1F7E0..U+1F7EB
U+1F7F0
U+1FA70..U+1FA7C
U+1FA80..U+1FA88
U+1FA90..U+1FABD
U+1FABF..U+1FAC5
U+1FACE..U+1FADB
U+1FAE0..U+1FAE8
U+1FAF0..U+1FAF8

Was 2, now 1:
U+2E9A
U+2EF4..U+2EFF
U+2FD6..U+2FEF
U+2FFC..U+2FFF
U+3040
U+3097..U+3098
U+3100..U+3104
U+3130
U+318F
U+31E4..U+31EF
U+321F
U+3248..U+324F
U+A48D..U+A48F
U+A4C7..U+A4CF
U+FE53
U+FE67
U+FE6C..U+FE6F
U+FF00

Thanks @frederick-vs-ja for pointing out that the old _Width_estimate_intervals table can be overwritten directly 👀

improve generator source code

stl/inc/format

frederick-vs-ja

We should no longer cite N4928, since it didn't contain the changes in P2675R1.
The first C++26 working draft is likely to complete before landing this PR. But I think it makes sense to still cite N4950 since it the final draft of C++23, and MSVC STL is currently citing N4950 in the whole product codes.

stl/inc/format

Co-authored-by: A. Jiang <[email protected]>

…675R1_v2

tools/unicode_properties_parse/format_width_estimate_intervals.cpp

mordante · 2023-07-24T19:03:21Z

tools/unicode_properties_parse/format_width_estimate_intervals.cpp

+// Copyright (c) Microsoft Corporation.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+// The following code generates data for `_Width_estimate_intervals` in <format>.


FYI In libc++ I use a python script that was based on this script.

tools/unicode_properties_parse/format_width_estimate_intervals.cpp

add const

achabense

This is the full list of all remaining issues I'm concerned about. Aside from these issues, should I also offer a test for all affected ranges(those in the annex of P267521) in this pr?

stl/inc/format

tests/libcxx/expected_results.txt

tests/std/tests/P0645R10_text_formatting_utf8/test.cpp

tools/unicode_properties_parse/format_width_estimate_intervals.cpp

achabense · 2023-07-26T18:17:29Z

tools/unicode_properties_parse/format_width_estimate_intervals.cpp

+table_u read_from(ifstream& source) {
+    table_u table;
+
+    // "The unassigned code points in the following blocks default to "W":"


About the citation, the original comment in "EastAsianWidth.txt" for 20000~2FFFFD, 30000~3FFFFD is "All undesignated code points in Planes 2 and 3, whether inside or outside of allocated blocks, default to "W" : "
What's its difference with this sentence? Aren't they of the same meaning?

Also see #3903 (comment) and #3903 (comment)
(My idea is that we can safely keep this part, as it is not harmful, and hopefully for a long time will not become outdated.)

I don't understand the question here - where is this comment coming from, and why is it different from what EastAsianWidth.txt says? Can't we just quote the EastAsianWidth.txt comment verbatim?

Both comments comes from EastAsianWidth.txt; I'm being confused why there should be a different comment for 20000~2FFFFD, 30000~3FFFFD here:

IIUC EastAsianWidth.txt is trying to communicate that not all of Planes 2 & 3 are allocated. In particular, the following ranges are not allocated for any block:

0002A6E0-0002A6FF

0002EE50-0002F7FF

0002FA20-0002FFFD

000323B0-0003FFFD

But I don't think we care about the relationship between planes and blocks. For our purposes, they are just "big areas" and "small areas".

tools/unicode_properties_parse/format_width_estimate_intervals.cpp

add comments for `_Unicode_width_estimate`; add test cases: all 2->1 cases; and some 1->2 cases

tests/std/tests/P0645R10_text_formatting_utf8/test.cpp

barcharcraz

Looks good to me, we should create an issue to port the generator to python.

StephanTLavavej · 2023-08-10T01:09:09Z

I'm mirroring this to the MSVC-internal repo - please notify me if any further changes are pushed.

StephanTLavavej · 2023-08-11T02:25:52Z

Thanks for implementing this C++23 feature and bringing the codebase closer to completeness! ✅ 🚀 😻

frederick-vs-ja · 2023-08-11T07:31:28Z

Thanks for consistently citing N4950 in the whole product code!

achabense added 8 commits July 23, 2023 04:24

add generator

249f79e

update comments

26cb007

improvements

4a7d1a0

update _Width_estimate_intervals and _Unicode_width_estimate

09c253b

Merge branch 'microsoft:main' into P2675R1

fe3b430

review feedback: drop #if _HAS_CXX23

361580d

improve generator source code

Merge branch 'P2675R1' of https://github.com/achabense/STL into P2675R1

e5e162c

nits

077b3fe

achabense requested a review from a team as a code owner July 23, 2023 06:39

achabense commented Jul 23, 2023

View reviewed changes

stl/inc/format Show resolved Hide resolved

add cast

b5a6561

frederick-vs-ja reviewed Jul 23, 2023

View reviewed changes

stl/inc/format Outdated Show resolved Hide resolved

stl/inc/format Outdated Show resolved Hide resolved

citation updates

7750b05

Co-authored-by: A. Jiang <[email protected]>

This comment was marked as resolved.

Sign in to view

StephanTLavavej added format C++20/23 format cxx23 C++23 feature labels Jul 23, 2023

achabense added 2 commits July 23, 2023 17:35

test case fix

a447d99

Merge branch 'P2675R1_v2' of https://github.com/achabense/STL into P2…

197d3f1

…675R1_v2

This comment was marked as resolved.

Sign in to view

achabense commented Jul 24, 2023

View reviewed changes

tools/unicode_properties_parse/format_width_estimate_intervals.cpp Outdated Show resolved Hide resolved

achabense added 2 commits July 24, 2023 15:54

update "format_width_estimate_interval.cpp"

383925f

"std/utilities/format/format.functions/unicode.pass.cpp FAIL"

8ef6bfc

mordante reviewed Jul 24, 2023

View reviewed changes

StephanTLavavej self-assigned this Jul 24, 2023

refactor format_width_estimate_intervals.cpp

359ef00

frederick-vs-ja reviewed Jul 25, 2023

View reviewed changes

This comment was marked as resolved.

Sign in to view

achabense referenced this pull request Jul 26, 2023

remaining issues

0695d65

This comment was marked as resolved.

Sign in to view

verify_impl&VERIFY -> verify;

06cabc2

add const

This comment was marked as resolved.

Sign in to view

achabense commented Jul 26, 2023

View reviewed changes

tools/unicode_properties_parse/format_width_estimate_intervals.cpp Show resolved Hide resolved

StephanTLavavej self-requested a review July 26, 2023 19:32

This comment was marked as resolved.

Sign in to view

StephanTLavavej reviewed Jul 26, 2023

View reviewed changes

tools/unicode_properties_parse/format_width_estimate_intervals.cpp Show resolved Hide resolved

review feedback;

5ac8346

add comments for `_Unicode_width_estimate`; add test cases: all 2->1 cases; and some 1->2 cases

achabense commented Jul 27, 2023

View reviewed changes

tests/std/tests/P0645R10_text_formatting_utf8/test.cpp Outdated Show resolved Hide resolved

achabense commented Jul 27, 2023

View reviewed changes

tests/std/tests/P0645R10_text_formatting_utf8/test.cpp Show resolved Hide resolved

achabense commented Jul 27, 2023

View reviewed changes

tests/std/tests/P0645R10_text_formatting_utf8/test.cpp Show resolved Hide resolved

StephanTLavavej mentioned this pull request Jul 27, 2023

Convert P2675R1 <format> generator from C++ to Python #3908

Closed

Update Standardese citation, drop outdated explanation.

b6080d5

StephanTLavavej approved these changes Jul 27, 2023

View reviewed changes

StephanTLavavej assigned barcharcraz Aug 2, 2023

barcharcraz approved these changes Aug 9, 2023

View reviewed changes

StephanTLavavej unassigned barcharcraz Aug 9, 2023

StephanTLavavej self-assigned this Aug 10, 2023

StephanTLavavej merged commit 84cc12d into microsoft:main Aug 11, 2023

achabense deleted the P2675R1_v2 branch August 11, 2023 02:40

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Improve std::format's width estimation #3903

Improve std::format's width estimation #3903

achabense commented Jul 23, 2023 •

edited

Loading

frederick-vs-ja left a comment

This comment was marked as resolved.

This comment was marked as resolved.

This comment was marked as resolved.

mordante Jul 24, 2023

This comment was marked as resolved.

This comment was marked as resolved.

This comment was marked as resolved.

achabense left a comment •

edited

Loading

achabense Jul 26, 2023 •

edited

Loading

StephanTLavavej Jul 26, 2023

achabense Jul 27, 2023

cpplearner Jul 27, 2023 •

edited

Loading

This comment was marked as resolved.

barcharcraz left a comment

StephanTLavavej commented Aug 10, 2023

StephanTLavavej commented Aug 11, 2023

frederick-vs-ja commented Aug 11, 2023

Improve std::format's width estimation #3903

Improve std::format's width estimation #3903

Conversation

achabense commented Jul 23, 2023 • edited Loading

frederick-vs-ja left a comment

Choose a reason for hiding this comment

This comment was marked as resolved.

This comment was marked as resolved.

This comment was marked as resolved.

mordante Jul 24, 2023

Choose a reason for hiding this comment

This comment was marked as resolved.

This comment was marked as resolved.

This comment was marked as resolved.

achabense left a comment • edited Loading

Choose a reason for hiding this comment

achabense Jul 26, 2023 • edited Loading

Choose a reason for hiding this comment

StephanTLavavej Jul 26, 2023

Choose a reason for hiding this comment

achabense Jul 27, 2023

Choose a reason for hiding this comment

cpplearner Jul 27, 2023 • edited Loading

Choose a reason for hiding this comment

This comment was marked as resolved.

barcharcraz left a comment

Choose a reason for hiding this comment

StephanTLavavej commented Aug 10, 2023

StephanTLavavej commented Aug 11, 2023

frederick-vs-ja commented Aug 11, 2023

achabense commented Jul 23, 2023 •

edited

Loading

achabense left a comment •

edited

Loading

achabense Jul 26, 2023 •

edited

Loading

cpplearner Jul 27, 2023 •

edited

Loading