From: drh <>
Date: Wed, 14 Aug 2024 17:58:46 +0000 (+0000)
Subject: Improved documentation for the sha3() SQL functions in the shathree.c
X-Git-Tag: version-3.47.0~231
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=78f9ba9774883d56ff6384c5173ff7e1ac42cb88;p=thirdparty%2Fsqlite.git

Improved documentation for the sha3() SQL functions in the shathree.c
extension in ext/misc.

FossilOrigin-Name: 0ce609103eb35cfa26d9b6fa1769679be7e2b9032c37cc8c00a194fe355b06a4
---

diff --git a/ext/misc/shathree.c b/ext/misc/shathree.c
index 46a864af4e..5cb3198805 100644
--- a/ext/misc/shathree.c
+++ b/ext/misc/shathree.c
@@ -15,10 +15,20 @@
 ** Two SQL functions are implemented:
 **
 **     sha3(X,SIZE)
-**     sha3_query(Y,SIZE)
+**     sha3_agg(Y,SIZE)
+**     sha3_query(Z,SIZE)
 **
 ** The sha3(X) function computes the SHA3 hash of the input X, or NULL if
-** X is NULL.
+** X is NULL.  If inputs X is text, the UTF-8 rendering of that text is
+** used to compute the hash.  If X is a BLOB, then the binary data of the
+** blob is used to compute the hash.  If X is an integer or real number,
+** then that number if converted into UTF-8 text and the hash is computed
+** over the text.
+**
+** The sha3_agg(Y) function computes the SHA3 hash of all Y inputs.  Since
+** order is important for the hash, it is recommended that the Y expression
+** by followed by an ORDER BY clause to guarantee that the inputs occur
+** in the desired order.
 **
 ** The sha3_query(Y) function evaluates all queries in the SQL statements of Y
 ** and returns a hash of their results.
@@ -26,6 +36,68 @@
 ** The SIZE argument is optional.  If omitted, the SHA3-256 hash algorithm
 ** is used.  If SIZE is included it must be one of the integers 224, 256,
 ** 384, or 512, to determine SHA3 hash variant that is computed.
+**
+** Because the sha3_agg() and sha3_query() functions compute a hash over
+** multiple values, the values are encode to use include type information.
+**
+** In sha3_agg(), the sequence of bytes that gets hashed for each input
+** Y depends on the datatype of Y:
+**
+**    typeof(Y)='null'         A single "N" is hashed.  (One byte)
+**
+**    typeof(Y)='integer'      The data hash is the character "I" followed
+**                             by an 8-byte big-endian binary of the
+**                             64-bit signed integer.  (Nine bytes total.)
+**
+**    typeof(Y)='real'         The character "F" followed by an 8-byte
+**                             big-ending binary of the double.  (Nine
+**                             bytes total.)
+**
+**    typeof(Y)='text'         The hash is over prefix "Tnnn:" followed
+**                             by the UTF8 encoding of the text.  The "nnn"
+**                             in the prefix is the minimum-length decimal
+**                             representation of the octet_length of the text.
+**                             Notice the ":" at the end of the prefix, which
+**                             is needed to separate the prefix from the
+**                             content in cases where the content starts
+**                             with a digit.
+**
+**    typeof(Y)='blob'         The hash is taken over prefix "Bnnn:" followed
+**                             by the binary content of the blob.  The "nnn"
+**                             in the prefix is the mimimum-length decimal
+**                             representation of the byte-length of the blob.
+**
+** According to the rules above, all of the following SELECT statements
+** should return TRUE:
+**
+**    SELECT sha3(1) = sha3('1');
+**
+**    SELECT sha3('hello') = sha3(x'68656c6c6f');
+**
+**    WITH a(x) AS (VALUES('xyzzy'))
+**      SELECT sha3_agg(x) = sha3('T5:xyzzy')            FROM a;
+**
+**    WITH a(x) AS (VALUES(x'010203'))
+**      SELECT sha3_agg(x) = sha3(x'42333a010203')       FROM a;
+**
+**    WITH a(x) AS (VALUES(0x123456))
+**      SELECT sha3_agg(x) = sha3(x'490000000000123456') FROM a;
+**
+**    WITH a(x) AS (VALUES(100.015625))
+**      SELECT sha3_agg(x) = sha3(x'464059010000000000') FROM a;
+**
+**    WITH a(x) AS (VALUES(NULL))
+**      SELECT sha3_agg(x) = sha3('N') FROM a;
+**
+**
+** In sha3_query(), individual column values are encoded as with
+** sha3_agg(), but with the addition that a single "R" character is
+** inserted at the start of each row.
+**
+** Note that sha3_agg() hashes rows for which Y is NULL.  Add a FILTER
+** clause if NULL rows should be excluded:
+**
+**    SELECT sha3_agg(x ORDER BY rowid) FILTER(x NOT NULL) FROM t1;
 */
 #include "sqlite3ext.h"
 SQLITE_EXTENSION_INIT1
diff --git a/manifest b/manifest
index d30fe768b9..8ae446ea1a 100644
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Fix\san\sassert()\sin\sfts5\sthat\smight\sfail\swith\san\sexpression\slike\s"...\sNOT\s''".
-D 2024-08-14T14:26:02.443
+C Improved\sdocumentation\sfor\sthe\ssha3()\sSQL\sfunctions\sin\sthe\sshathree.c\nextension\sin\sext/misc.
+D 2024-08-14T17:58:46.321
 F .fossil-settings/empty-dirs dbb81e8fc0401ac46a1491ab34a7f2c7c0452f2f06b54ebb845d024ca8283ef1
 F .fossil-settings/ignore-glob 35175cdfcf539b2318cb04a9901442804be81cd677d8b889fcc9149c21f239ea
 F LICENSE.md df5091916dbb40e6e9686186587125e1b2ff51f022cc334e886c19a0e9982724
@@ -415,7 +415,7 @@ F ext/misc/rot13.c 51ac5f51e9d5fd811db58a9c23c628ad5f333c173f1fc53c8491a3603d385
 F ext/misc/scrub.c 2a44b0d44c69584c0580ad2553f6290a307a49df4668941d2812135bfb96a946
 F ext/misc/series.c d96e5aac21658c6b5d54f918ac140460ec7197734c1a4fba806950831a7b1e7a
 F ext/misc/sha1.c 4011aef176616872b2a0d5bccf0ecfb1f7ce3fe5c3d107f3a8e949d8e1e3f08d
-F ext/misc/shathree.c 24e890cb209272379f8d0823fb278d24bd3dca35d5ea565aa0b2794cd36d6197
+F ext/misc/shathree.c 811540cb7eb6f34ccf6ce6f6713c3896d38cbed20d46cb3c7c48d90e3251a7f4
 F ext/misc/showauth.c 732578f0fe4ce42d577e1c86dc89dd14a006ab52
 F ext/misc/spellfix.c c0aa7b80d6df45f7da59d912b38752bcac1af53a5766966160e6c5cdd397dbea
 F ext/misc/sqlar.c a6175790482328171da47095f87608b48a476d4fac78d8a9ff18b03a2454f634
@@ -2204,8 +2204,8 @@ F vsixtest/vsixtest.tcl 6195aba1f12a5e10efc2b8c0009532167be5e301abe5b31385638080
 F vsixtest/vsixtest.vcxproj.data 2ed517e100c66dc455b492e1a33350c1b20fbcdc
 F vsixtest/vsixtest.vcxproj.filters 37e51ffedcdb064aad6ff33b6148725226cd608e
 F vsixtest/vsixtest_TemporaryKey.pfx e5b1b036facdb453873e7084e1cae9102ccc67a0
-P c4d7f9996ce4f5f6d08d83fc3abdb81b6825cfb556318f90c6896736a1da5175
-R 8f6bb730f0c3b63cb61a6b0c78a6a7c9
-U dan
-Z 146061b654867fea67c05bd71b2b625a
+P 317c91edfc69d1d47e233bb6ac13934241742fda1ba03a88fd51bdf98fcbc463
+R 99e8cbe80cd01ac20616f63cdd8dbbda
+U drh
+Z c4d9af30409bcf77d9189a2bdac49004
 # Remove this line to create a well-formed Fossil manifest.
diff --git a/manifest.uuid b/manifest.uuid
index c281d62fd8..e1862fcd06 100644
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-317c91edfc69d1d47e233bb6ac13934241742fda1ba03a88fd51bdf98fcbc463
+0ce609103eb35cfa26d9b6fa1769679be7e2b9032c37cc8c00a194fe355b06a4