pg.ddx.io  pgsql-general@postgresql.org mailing list archive  
help / color / mirror / Atom feed
From: Martin Kalcher <martin.kalcher@aboutsource.net>
To: Tom Lane <tgl@sss.pgh.pa.us>
To: John Naylor <john.naylor@enterprisedb.com>
Cc: Thomas Munro <thomas.munro@gmail.com>
Cc: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Subject: [PATCH] Introduce array_shuffle() and array_sample()
Date: Mon, 18 Jul 2022 21:03:35 +0200
Message-ID: <1708d95a-629e-5ac5-deab-d20a8d0b2b0e@aboutsource.net> (raw)
In-Reply-To: <687490.1658152096@sss.pgh.pa.us>
References: <9d160a44-7675-51e8-60cf-6d64b76db831@aboutsource.net>
	<CA+hUKGJFYY6s4nxDRz+PTLNYeUkZJJn_7+iMJoaQeRp5hKz0aQ@mail.gmail.com>
	<b8f44b53-02d6-0be4-42df-cb6f737e6267@aboutsource.net>
	<4073858.1658027909@sss.pgh.pa.us>
	<4077925.1658029046@sss.pgh.pa.us>
	<CA+hUKGLcjrJ9Q2OZAKHN-SQyq_m5QDv8d3WKHYq5HW8Lyozu4w@mail.gmail.com>
	<5ef15c6f-e108-53af-3d82-587fac394f1a@aboutsource.net>
	<463181.1658097987@sss.pgh.pa.us>
	<5b7a1ee9-f942-2502-3a0c-e9fc12511139@aboutsource.net>
	<469954.1658100033@sss.pgh.pa.us>
	<f36445ec-aad9-0531-39f8-ca3a3bb4860e@aboutsource.net>
	<CAFBsxsG_PK3BkhMcfeLmwDS+FBDZKAaUtfrb6KjU79FZ-3H46A@mail.gmail.com>
	<687490.1658152096@sss.pgh.pa.us>

Thanks for all your feedback and help. I got a patch that i consider 
ready for review. It introduces two new functions:

   array_shuffle(anyarray) -> anyarray
   array_sample(anyarray, integer) -> anyarray

array_shuffle() shuffles an array (obviously). array_sample() picks n 
random elements from an array.

Is someone interested in looking at it? What are the next steps?

Martin

Attachments:

0001-Introduce-array_shuffle-and-array_sample.patch (text/x-patch, 9.0K) download From 5498bb2d9f1fab4cad56cd0d3a6eeafa24a26c8e Mon Sep 17 00:00:00 2001 From: Martin Kalcher <martin.kalcher@aboutsource.net> Date: Sun, 17 Jul 2022 18:06:04 +0200 Subject: [PATCH] Introduce array_shuffle() and array_sample() * array_shuffle() shuffles to elements of an array. * array_sample() chooses n elements from an array by random. Shuffling of arrays can already be accomplished with SQL using unnest() and array_agg(order by random()). But that is very slow compared to the new functions. --- doc/src/sgml/func.sgml | 34 ++++++ src/backend/utils/adt/arrayfuncs.c | 163 +++++++++++++++++++++++++++ src/include/catalog/pg_proc.dat | 6 + src/test/regress/expected/arrays.out | 30 +++++ src/test/regress/sql/arrays.sql | 11 ++ 5 files changed, 244 insertions(+) diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml index b6783b7ad0..c676031b4a 100644 --- a/doc/src/sgml/func.sgml +++ b/doc/src/sgml/func.sgml @@ -19395,6 +19395,40 @@ SELECT NULLIF(value, '(none)') ... </para></entry> </row> + <row> + <entry role="func_table_entry"><para role="func_signature"> + <indexterm> + <primary>array_sample</primary> + </indexterm> + <function>array_sample</function> ( <parameter>array</parameter> <type>anyarray</type>, <parameter>n</parameter> <type>integer</type> ) + <returnvalue>anyarray</returnvalue> + </para> + <para> + Returns <parameter>n</parameter> randomly chosen elements from <parameter>array</parameter>. + </para> + <para> + <literal>array_sample(ARRAY[1,2,3,4,5,6], 3)</literal> + <returnvalue>{4,5,1}</returnvalue> + </para></entry> + </row> + + <row> + <entry role="func_table_entry"><para role="func_signature"> + <indexterm> + <primary>array_shuffle</primary> + </indexterm> + <function>array_shuffle</function> ( <type>anyarray</type> ) + <returnvalue>anyarray</returnvalue> + </para> + <para> + Shuffles the elements of the array. + </para> + <para> + <literal>array_shuffle(ARRAY[1,2,3,4,5,6])</literal> + <returnvalue>{4,5,1,3,2,6}</returnvalue> + </para></entry> + </row> + <row> <entry role="func_table_entry"><para role="func_signature"> <indexterm id="function-array-to-string"> diff --git a/src/backend/utils/adt/arrayfuncs.c b/src/backend/utils/adt/arrayfuncs.c index fb167f226a..a6769a8ebf 100644 --- a/src/backend/utils/adt/arrayfuncs.c +++ b/src/backend/utils/adt/arrayfuncs.c @@ -6872,3 +6872,166 @@ trim_array(PG_FUNCTION_ARGS) PG_RETURN_DATUM(result); } + +/* + * Shuffle the elements of an array. + */ +Datum +array_shuffle(PG_FUNCTION_ARGS) +{ + ArrayType *array, + *result; + int16 elmlen; + bool elmbyval; + char elmalign; + Oid elmtyp; + TypeCacheEntry *typentry; + Datum *elms, + elm; + bool *nuls, + nul; + int nelms, + i, + j; + + array = PG_GETARG_ARRAYTYPE_P(0); + nelms = ArrayGetNItems(ARR_NDIM(array), ARR_DIMS(array)); + + /* There is no point in shuffling arrays with less then two elements. */ + if (nelms < 2) + PG_RETURN_ARRAYTYPE_P(array); + + elmtyp = ARR_ELEMTYPE(array); + typentry = (TypeCacheEntry *) fcinfo->flinfo->fn_extra; + + if (typentry == NULL || typentry->type_id != elmtyp) + { + typentry = lookup_type_cache(elmtyp, 0); + fcinfo->flinfo->fn_extra = (void *) typentry; + } + elmlen = typentry->typlen; + elmbyval = typentry->typbyval; + elmalign = typentry->typalign; + + deconstruct_array(array, elmtyp, elmlen, elmbyval, elmalign, + &elms, &nuls, &nelms); + + /* Shuffle elements and nulls using Fisher-Yates shuffle algorithm. */ + for (i = nelms - 1; i > 0; i--) + { + j = random() % (i + 1); + elm = elms[i]; + nul = nuls[i]; + elms[i] = elms[j]; + nuls[i] = nuls[j]; + elms[j] = elm; + nuls[j] = nul; + } + + result = construct_md_array(elms, nuls, + ARR_NDIM(array), ARR_DIMS(array), ARR_LBOUND(array), + elmtyp, elmlen, elmbyval, elmalign); + + pfree(elms); + pfree(nuls); + PG_FREE_IF_COPY(array, 0); + + PG_RETURN_ARRAYTYPE_P(result); +} + +/* + * Choose N random elements from an array. + */ +Datum +array_sample(PG_FUNCTION_ARGS) +{ + ArrayType *array, + *result; + int16 elmlen; + bool elmbyval; + char elmalign; + Oid elmtyp; + TypeCacheEntry *typentry; + Datum *elms, + elm; + bool *nuls, + nul; + int nelms, + rnelms, + rdims[1], + rlbs[1], + i, + j; + + array = PG_GETARG_ARRAYTYPE_P(0); + nelms = ArrayGetNItems(ARR_NDIM(array), ARR_DIMS(array)); + elmtyp = ARR_ELEMTYPE(array); + rnelms = PG_GETARG_INT32(1); + + if (rnelms < 0) + ereport(ERROR, + (errcode(ERRCODE_INVALID_PARAMETER_VALUE), + errmsg("second parameter must not be negative"))); + + /* Return an empty array if the requested sample size is zero. */ + if (rnelms == 0) + { + PG_FREE_IF_COPY(array, 0); + PG_RETURN_ARRAYTYPE_P(construct_empty_array(elmtyp)); + } + + /* + * Return the original array if the requested sample size is greater than + * or equal to its own size. + */ + if (rnelms >= nelms) + PG_RETURN_ARRAYTYPE_P(array); + + typentry = (TypeCacheEntry *) fcinfo->flinfo->fn_extra; + + if (typentry == NULL || typentry->type_id != elmtyp) + { + typentry = lookup_type_cache(elmtyp, 0); + fcinfo->flinfo->fn_extra = (void *) typentry; + } + elmlen = typentry->typlen; + elmbyval = typentry->typbyval; + elmalign = typentry->typalign; + + deconstruct_array(array, elmtyp, elmlen, elmbyval, elmalign, + &elms, &nuls, &nelms); + + /* + * Move rnelms (requested sample size) randomly-chosen elements to the + * beginning of the array, using the Fisher-Yates shuffle algorithm. + */ + for (i = 0; i < rnelms; i++) + { + /* random j such that i <= j < nelms */ + j = (random() % (nelms - i)) + i; + elm = elms[i]; + nul = nuls[i]; + elms[i] = elms[j]; + nuls[i] = nuls[j]; + elms[j] = elm; + nuls[j] = nul; + } + + /* + * Construct a new array from the first rnelms (requested sample size) + * elements of the shuffled array. Even though we are constructing a + * one-dimensional array, we cannot use construct_array(), because it does + * not support NULL elements. + */ + rdims[0] = rnelms; + rlbs[0] = 1; + + result = construct_md_array(elms, nuls, 1, rdims, rlbs, + elmtyp, elmlen, elmbyval, elmalign); + + pfree(elms); + pfree(nuls); + PG_FREE_IF_COPY(array, 0); + + PG_RETURN_ARRAYTYPE_P(result); +} diff --git a/src/include/catalog/pg_proc.dat b/src/include/catalog/pg_proc.dat index 2e41f4d9e8..f51af5ee9f 100644 --- a/src/include/catalog/pg_proc.dat +++ b/src/include/catalog/pg_proc.dat @@ -1681,6 +1681,12 @@ proname => 'arraycontjoinsel', provolatile => 's', prorettype => 'float8', proargtypes => 'internal oid internal int2 internal', prosrc => 'arraycontjoinsel' }, +{ oid => '8464', descr => 'shuffle array', + proname => 'array_shuffle', proisstrict => 't', prorettype => 'anyarray', + proargtypes => 'anyarray', prosrc => 'array_shuffle' }, +{ oid => '8465', descr => 'take samples from array', + proname => 'array_sample', proisstrict => 't', prorettype => 'anyarray', + proargtypes => 'anyarray int4', prosrc => 'array_sample' }, { oid => '764', descr => 'large object import', proname => 'lo_import', provolatile => 'v', proparallel => 'u', diff --git a/src/test/regress/expected/arrays.out b/src/test/regress/expected/arrays.out index ce6f3a65f9..4994e6ac62 100644 --- a/src/test/regress/expected/arrays.out +++ b/src/test/regress/expected/arrays.out @@ -2445,3 +2445,33 @@ SELECT trim_array(ARRAY[1, 2, 3], -1); -- fail ERROR: number of elements to trim must be between 0 and 3 SELECT trim_array(ARRAY[1, 2, 3], 10); -- fail ERROR: number of elements to trim must be between 0 and 3 +-- array_shuffle +SELECT array_agg(a order by a) +FROM +(SELECT unnest(array_shuffle('{NULL,1,2,3,4,5}'::int[]))) v(a); + array_agg +------------------ + {1,2,3,4,5,NULL} +(1 row) + +-- array_sample +SELECT cardinality(array_sample('{NULL,1,2,3,4,5}'::int[], 3)); + cardinality +------------- + 3 +(1 row) + +SELECT array_sample('{1,2,3,4,5}'::int[], 3) <@ '{1,2,3,4,5}'::int[]; + ?column? +---------- + t +(1 row) + +SELECT array_sample('{1,2,3,4,5}'::int[], 10); + array_sample +-------------- + {1,2,3,4,5} +(1 row) + +SELECT array_sample('{1,2,3,4,5}'::int[], -1); -- fail +ERROR: second parameter must not be negative diff --git a/src/test/regress/sql/arrays.sql b/src/test/regress/sql/arrays.sql index f774faf856..0234aafa06 100644 --- a/src/test/regress/sql/arrays.sql +++ b/src/test/regress/sql/arrays.sql @@ -754,3 +754,14 @@ FROM SELECT trim_array(ARRAY[1, 2, 3], -1); -- fail SELECT trim_array(ARRAY[1, 2, 3], 10); -- fail + +-- array_shuffle +SELECT array_agg(a order by a) +FROM +(SELECT unnest(array_shuffle('{NULL,1,2,3,4,5}'::int[]))) v(a); + +-- array_sample +SELECT cardinality(array_sample('{NULL,1,2,3,4,5}'::int[], 3)); +SELECT array_sample('{1,2,3,4,5}'::int[], 3) <@ '{1,2,3,4,5}'::int[]; +SELECT array_sample('{1,2,3,4,5}'::int[], 10); +SELECT array_sample('{1,2,3,4,5}'::int[], -1); -- fail -- 2.37.1

view thread (51+ messages)  latest in thread

Message-ID: <1708d95a-629e-5ac5-deab-d20a8d0b2b0e@aboutsource.net>
Permalink:  ../1708d95a-629e-5ac5-deab-d20a8d0b2b0e@aboutsource.net/
Also on:    postgresql.org/message-id/1708d95a-629e-5ac5-deab-d20a8d0b2b0e@aboutsource.net

 ·  · 

reply

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Reply to all the recipients from your mail client; the link fills in
  the headers below: reply via email
  Some clients, webmail especially, drop In-Reply-To and References from
  a link, and the reply then starts a new thread: use the mbox below.

  To: pgsql-general@postgresql.org
  Cc: martin.kalcher@aboutsource.net, tgl@sss.pgh.pa.us, john.naylor@enterprisedb.com, thomas.munro@gmail.com, pgsql-hackers@lists.postgresql.org
  Subject: Re: [PATCH] Introduce array_shuffle() and array_sample()
  In-Reply-To: <1708d95a-629e-5ac5-deab-d20a8d0b2b0e@aboutsource.net>
  References: <9d160a44-7675-51e8-60cf-6d64b76db831@aboutsource.net> <CA+hUKGJFYY6s4nxDRz+PTLNYeUkZJJn_7+iMJoaQeRp5hKz0aQ@mail.gmail.com> <b8f44b53-02d6-0be4-42df-cb6f737e6267@aboutsource.net> <4073858.1658027909@sss.pgh.pa.us> <4077925.1658029046@sss.pgh.pa.us> <CA+hUKGLcjrJ9Q2OZAKHN-SQyq_m5QDv8d3WKHYq5HW8Lyozu4w@mail.gmail.com> <5ef15c6f-e108-53af-3d82-587fac394f1a@aboutsource.net> <463181.1658097987@sss.pgh.pa.us> <5b7a1ee9-f942-2502-3a0c-e9fc12511139@aboutsource.net> <469954.1658100033@sss.pgh.pa.us> <f36445ec-aad9-0531-39f8-ca3a3bb4860e@aboutsource.net> <CAFBsxsG_PK3BkhMcfeLmwDS+FBDZKAaUtfrb6KjU79FZ-3H46A@mail.gmail.com> <687490.1658152096@sss.pgh.pa.us> <1708d95a-629e-5ac5-deab-d20a8d0b2b0e@aboutsource.net>

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox