hermes-agent/n8n-openai-adapter
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
17,931 Commits 29 Branches 195 Tags
00a0f07824ea7bbca795da1bf8dfa94d17a09526
Commit Graph
100 Commits
Author SHA1 Message Date
José Valim 5a2b24d21f Group Unicode upcase/downcase by prefix (#11310) 
The patch computes byte lookups based on the prefix. For example,
Á, É, etc all have the same prefix <<195>>, so they are lumped
together for lookup and then we just do a byte lookup later. We
tried doing the byte lookup on 64-element tuple (since the byte
is always within 0b10000000 and 0b10111111) but that's slower,
especially because we need to check the byte range for invalid
Unicode, so instead the last byte lookup is a case. Grouping the
top-level lookup makes the cost of a miss 3x cheaper albeit a
hit is 10% more expensive and reduces bytecode size.
 2021-10-14 14:44:13 +02:00
Eksperimental d962ddb5b0 Upgrade Unicode database to v14.0.0 (#11308) 
Closes #11288.
 2021-10-12 18:05:10 +02:00
José Valim 3f5b3f00f1 Depend on Erlang for computing grapheme clusters (#11024) 
Erlang ships with its own embedding of the Unicode
Codebase for a couple releases and this release changes
Elixir to depend on it in order to compute grapheme
clusters.

The Erlang implementation was up to 2x faster in low
codepoints (such as latin1) while the Elixir one could
be faster up to 3x in high codepoints (such as emoji)
so at the end the performance results are roughly the
same. As a benefit, we no longer need to ship our copy
of the grapheme cluster algorithm, which would take up
to 250kB in disk and more than 15 seconds to compile.

Note we still keep our own String downcase and upcase
algorithms, as our version is considerably more efficient
on all cases since it works exclusively with binaries
(more than 5x faster).
 2021-06-01 12:44:55 +02:00
José Valim 0123f8680f Reduce size of Unicode module in 33% by not duplicating codepoints 2021-05-25 18:06:58 +02:00
José Valim 151f11e0b8 Split on \r\n and \n accordingly during unicode compilation, closes #10991 2021-05-20 12:12:42 +02:00
Eksperimental c1e622950c Use proper Unicode reference (#10924) 2021-04-19 17:10:44 +02:00
Enrico Rivarola 218c35a041 Sync unicode data to Unicode 13.0.0 (#10866) 2021-04-03 15:20:05 +02:00
Eksperimental 0beb4f42fe Update Unicode upgrade instructions (#10862) 2021-04-01 22:11:23 +02:00
Eksperimental 37a09feaaa Update Unicode to version 13.0.0 (#10861) 2021-04-01 20:35:52 +02:00
Cẩm Huỳnh ef16468175 Optimize String.codepoints (#10633) 2021-01-08 19:20:27 +01:00
José Valim c36791091f Consider CRLF on Windows, closes #10470 2020-11-01 17:25:06 +01:00
José Valim a1c90e6a89 Only check the mode after matching on the codepoint 
This guarantees the compiler will not emit the code
that is first always checking the mode.
 2020-10-03 10:20:08 +02:00
Alt D. Soy 8ff465fa49 Add :turkic mode option to String case functions (#10376) 
This PR adds support of Turkic languages casing of the letter i according
to Unicode SpecialCasing.txt.
 2020-10-03 09:52:18 +02:00
Eksperimental b0d4288174 Use titlecase in Elixir, Erlang, EEx, Unicode, and Dialyzer (#9557) 2019-11-19 09:10:29 +01:00
José Valim bffb79e553 Update to Unicode 12.1 2019-07-05 11:04:08 +02:00
José Valim 0e6b8732ba Improve docs for built-in data types 2019-07-05 11:04:08 +02:00
José Valim 076dece778 Avoid unicode codepoint computation for lower bytes 2019-07-04 09:38:55 +02:00
Eric Meadows-Jönsson 9b6ec7bae7 Move unreachable function check from xref to group pass (#9168) 2019-06-28 10:25:49 +02:00
Eksperimental d4f8d419e4 Use https instead of http when servers are redirecting (#9094) 
The servers are automatically redirecting from http:// to https://
 2019-05-29 07:37:54 +02:00
José Valim 93fb75d793 Optimize Unicode graphemes 
We only check for Pictographic once we find a ZWJ.
 2019-03-27 01:46:34 +01:00
Eksperimental 99d919e0c8 Use "code point" instead of "codepoint" (#8774) 
That's the term it is used in the Unicode standard.
 2019-02-06 11:21:03 +01:00
Xavier Noria ee5a7fa0a4 Clarify String.next_codepoint/1 re invalid UTF-8 (#8489) 2018-12-08 16:41:57 +01:00
Sun Yaozhu 1967a06abe Fix ZWJ handling in Unicode grapheme clusters (#8361) 2018-11-04 19:56:36 +01:00
José Valim 67b7ffce61 Print number of tests from graphemes tests 2018-10-02 12:25:50 +02:00
José Valim 54cb02c240 Rely on Erlang/OTP normalization which is faster 2018-07-12 13:28:58 +02:00
José Valim c995ba968f Update to Unicode 11 2018-06-06 11:22:36 +02:00
Michał Muskała 722aef3791 Use binary_part/3 instead of :binary.part/3 
binary_part is a guard BIF and is slightly more efficient to use then
:binary.part which does a fully qualified external call.
 2018-04-30 13:22:19 +02:00
José Valim a7eb9adfe9 Consider case ignorable characters on Greek downcasing, closes #7149 
Note this does not impact the runtime cost of other downcasing
operations. The final beam file grew only in 8kb.
 2018-01-09 13:59:13 +01:00
José Valim 44a8f665b1 Improvements to greek handling of downcase, closes #7149 2017-12-26 23:11:17 +01:00
José Valim 6ecfbc546c Optimize String.downcase/1 
Thanks to @michalmuskala for feedback and review.
 2017-12-20 12:58:17 +01:00
José Valim 18288881c2 Provide :greek and :ascii mappings in String upcase/downcase/capitalize 
Closes #7105.
 2017-12-20 11:54:06 +01:00
José Valim b4aac2d7be Do not indent right-side of pipelines in the formatter (#7103) 2017-12-11 21:07:30 +01:00
José Valim b3992aa574 Minimize generated beam size by using ranges 
This approach is less performant when processing
sigma but compiles twice faster and takes two times
less space in disk.
 2017-10-30 18:45:16 +01:00
Ben Olive dfcc88fc33 Handle final sigma unicode exception 
Signed-off-by: José Valim <jose.valim@plataformatec.com.br>
 2017-10-30 18:45:16 +01:00
José Valim c217203558 Improvements to previously formatted files 2017-10-11 22:59:30 +02:00
José Valim 97ad07b7ab Make sure to not break guards on clauses with comments 2017-10-11 21:25:21 +02:00
José Valim 660838b399 Format more files with little formatting requirement 2017-10-11 00:28:10 +02:00
Frank McGeough ba9d08a54b Run formatter on lib/elixir/unicode/tokenizer.ex (#6813) 2017-10-10 18:10:35 +02:00
Guilherme Pasqualino e08df95e86 Run code formatter on lib/elixir/unicode/unicode.ex (#6795) 2017-10-10 12:24:15 +02:00
Michał Muskała c7a7a3625a Add is_binary guard to String.length (#6430) 
This produces a bit more readable message if not a string is passed into `String.length`, then failing in the `next_grapheme_size` function.
 2017-08-03 21:04:12 +02:00
Michał Muskała df33881ceb Optimise binary pattern matching in String.Break (#6340) 
In downcase and upcase pattern match tail as ::bits instead of ::binary. This
saves us a check on each iteration if the tail is a proper binary. We only do
this check in the very beginning.

In trim_leading additionally make sure every head of the recursive function
pattern matches on a binary in first argument so the single binary match context
optimisation can be applied.
 2017-07-16 14:29:45 +02:00
José Valim a311ae403e Update to Unicode 10 (#6293) 2017-07-04 09:57:07 +02:00
Andrea Leopardi 3b365510e5 Use :binary.split/3 in String.Break.split/1 (#6257) 2017-06-25 18:53:11 +02:00
José Valim 3c9b191ec1 Do not rely on on_load because it breaks releases 2017-05-29 15:50:06 +02:00
José Valim 9873e4239f Add non-quoted Unicode atoms and variables (#6158) 
It follows Unicode Annex #31.

See the Unicode Syntax document for a more in-depth reference.
 2017-05-27 22:43:05 +02:00
José Valim 7ca7139bb4 Properly slice titlecase_once binaries 2017-05-19 12:37:05 +02:00
Aleksei Magusev 683a3ad3c5 Correct module name for grapheme tests 2017-02-23 23:58:20 +01:00
José Valim 636c49fb5c Add grapheme tests 2017-02-13 15:44:59 +01:00
José Valim 3e394e8ef0 Incorporate new grapheme rules in Unicode 9 2017-02-13 14:34:04 +01:00
Andrea Leopardi 3604554aea Remove leftover trailing whitespace from the code base 
[ci skip]
 2017-01-13 15:01:47 +01:00