Merge pull request #3818 from katzyn/underscore

Add support of underscores in numeric literals
h2database · Jun 12, 2023 · fb86669 · fb86669
2 parents 1990523 + d425bae
commit fb86669
Show file tree

Hide file tree

Showing 9 changed files with 300 additions and 99 deletions.
diff --git a/h2/src/main/org/h2/bnf/Bnf.java b/h2/src/main/org/h2/bnf/Bnf.java
@@ -136,6 +136,8 @@ private void parse(Reader reader) throws SQLException, IOException {
         addFixedRule("@open_bracket@", RuleFixed.OPEN_BRACKET);
         addFixedRule("@close_bracket@", RuleFixed.CLOSE_BRACKET);
         addFixedRule("json_text", RuleFixed.JSON_TEXT);
+        Rule digit = ruleMap.get("digit").getRule();
+        ruleMap.get("number").setRule(new RuleList(digit, new RuleOptional(new RuleRepeat(digit, false)), false));
     }
 
     /**

diff --git a/h2/src/main/org/h2/command/Tokenizer.java b/h2/src/main/org/h2/command/Tokenizer.java
diff --git a/h2/src/main/org/h2/res/help.csv b/h2/src/main/org/h2/res/help.csv
@@ -2184,7 +2184,7 @@ A literal value of any data type, or null.
 "
 
 "Literals","Approximate numeric","
-[ + | - ] { { number [ . number ] } | { . number } }
+[ + | - ] { { number [ . [ number ] ] } | { . number } }
 E [ + | - ] expNumber
 ","
 An approximate numeric value.
@@ -2198,6 +2198,7 @@ for negative infinity, use ""CAST('-Infinity' AS dataType)"";
 for ""NaN"" (not a number), use ""CAST('NaN' AS dataType)"".
 ","
 -1.4e-10
+1.111_111E3
 CAST(1e2 AS REAL)
 CAST('NaN' AS DOUBLE PRECISION)
 "
@@ -2269,30 +2270,34 @@ this type, larger values small enough to fit into [BIGINT](https://h2database.co
 type have this type, others also have NUMERIC data type.
 ","
 -1600.05
+134_518.235_114
 "
 
 "Literals","Hex Number","
-[+|-] {0x|0X}{ digit | a-f | A-F } [...]
+[+|-] {0x|0X} { [_] { digit | a-f | A-F } [...] } [...]
 ","
 A number written in hexadecimal notation.
 ","
 0xff
+0x_ABCD_1234
 "
 
 "Literals","Octal Number","
-[+|-] {0o|0O}{ 0-7 } [...]
+[+|-] {0o|0O} { [_] { 0-7 } [...] } [...]
 ","
 A number written in octal notation.
 ","
 0o664
+0o_123_777
 "
 
 "Literals","Binary Number","
-[+|-] {0b|0B}{ 0-1 } [...]
+[+|-] {0b|0B} { [_] { 0-1 } [...] } [...]
 ","
 A number written in binary notation.
 ","
 0b101
+0b_01010101_10101010
 "
 
 "Literals","Int","
@@ -2301,6 +2306,7 @@ A number written in binary notation.
 The maximum integer number is 2147483647, the minimum is -2147483648.
 ","
 10
+65_536
 "
 
 "Literals","GEOMETRY","
@@ -2340,6 +2346,7 @@ JSON X'7472' '7565'
 Long numbers are between -9223372036854775808 and 9223372036854775807.
 ","
 100000
+1_000_000_000
 "
 
 "Literals","Null","
@@ -2351,11 +2358,12 @@ NULL
 "
 
 "Literals","Number","
-digit [...]
+digit [ [_] digit [...] ] [...]
 ","
 The maximum length of the number depends on the data type used.
 ","
 100
+10_000
 "
 
 "Literals","Numeric","
@@ -2369,6 +2377,7 @@ An explicit CAST can be used to change the data type.
 -1600.05
 CAST(0 AS DOUBLE PRECISION)
 -1.4e-10
+999_999_999.999_999
 "
 
 "Literals","String","

diff --git a/h2/src/test/org/h2/test/scripts/datatypes/bigint.sql b/h2/src/test/org/h2/test/scripts/datatypes/bigint.sql
@@ -69,3 +69,24 @@ EXPLAIN VALUES (1L, -2147483648L, 2147483647L, -2147483649L, 2147483648L);
 
 VALUES '9223372036854775807' > 0;
 >> TRUE
+
+SELECT 123_456_789_012_345, 0x_123_456_789_012_A4F, 1000L, 1_000L, 0xFFFFL, 0xFF_FFL;
+> 123456789012345 81985529205303887 1000 1000 65535 65535
+> --------------- ----------------- ---- ---- ----- -----
+> 123456789012345 81985529205303887 1000 1000 65535 65535
+> rows: 1
+
+SELECT 123_456_789_012_345_;
+> exception SYNTAX_ERROR_2
+
+SELECT 0o123_456_700_012_345_;
+> exception SYNTAX_ERROR_2
+
+SELECT 0o123_456_700_012_345__231;
+> exception SYNTAX_ERROR_2
+
+SELECT 1_L;
+> exception SYNTAX_ERROR_2
+
+SELECT 9223372036854775808L;
+> exception SYNTAX_ERROR_2
diff --git a/h2/src/test/org/h2/test/scripts/datatypes/decfloat.sql b/h2/src/test/org/h2/test/scripts/datatypes/decfloat.sql
@@ -290,3 +290,46 @@ SELECT 'NaN' = CAST('NaN' AS DECFLOAT);
 
 SELECT CAST('NaN' AS DOUBLE ) = CAST('NaN' AS DECFLOAT);
 >> TRUE
+
+SELECT 111222E+8_8, 111_222E+1_4;
+> 1.11222E+93 1.11222E+19
+> ----------- -----------
+> 1.11222E+93 1.11222E+19
+> rows: 1
+
+SELECT 111222333444555666777E+8_8, 111_222_333_444_555_666_777E+1_4;
+> 1.11222333444555666777E+108 1.11222333444555666777E+34
+> --------------------------- --------------------------
+> 1.11222333444555666777E+108 1.11222333444555666777E+34
+> rows: 1
+
+SELECT 111222333444555666777.123E+8_8, 111_222_333_444_555_666_777.888_999E+1_4;
+> 1.11222333444555666777123E+108 1.11222333444555666777888999E+34
+> ------------------------------ --------------------------------
+> 1.11222333444555666777123E+108 1.11222333444555666777888999E+34
+> rows: 1
+
+SELECT 1E;
+> exception SYNTAX_ERROR_2
+
+SELECT 1E+;
+> exception SYNTAX_ERROR_2
+
+SELECT 1E-;
+> exception SYNTAX_ERROR_2
+
+SELECT 1E_3;
+> exception SYNTAX_ERROR_2
+
+SELECT 1E+_3;
+> exception SYNTAX_ERROR_2
+
+SELECT 1E+3__3;
+> exception SYNTAX_ERROR_2
+
+SELECT 1E+8_;
+> exception SYNTAX_ERROR_2
+
+SELECT 1.3_E+3__3;
+> exception SYNTAX_ERROR_2
+
diff --git a/h2/src/test/org/h2/test/scripts/datatypes/int.sql b/h2/src/test/org/h2/test/scripts/datatypes/int.sql
@@ -22,3 +22,24 @@ SELECT 0x100, 0o100, 0b100;
 > --- -- -
 > 256 64 4
 > rows: 1
+
+SELECT 100_000, 1_1_1, 0b_1_1, 0o_1_1, 0x_1_1;
+> 100000 111 3 9 17
+> ------ --- - - --
+> 100000 111 3 9 17
+> rows: 1
+
+SELECT 1_;
+> exception SYNTAX_ERROR_2
+
+SELECT _1;
+> exception COLUMN_NOT_FOUND_1
+
+SELECT 1__1;
+> exception SYNTAX_ERROR_2
+
+SELECT 0x__1;
+> exception SYNTAX_ERROR_2
+
+SELECT 0x1_;
+> exception SYNTAX_ERROR_2
diff --git a/h2/src/test/org/h2/test/scripts/datatypes/numeric.sql b/h2/src/test/org/h2/test/scripts/datatypes/numeric.sql
@@ -218,3 +218,24 @@ SELECT CHAR_LENGTH(CAST(A / CAST(B AS NUMERIC(200, 100)) AS VARCHAR)) FROM TEST;
 
 DROP TABLE TEST;
 > ok
+
+SELECT 111_222_333_444_555_666_777_888_999, 111_222_333_444_555_666_777.333_444, 123_456., .333, 345_323.765_329, 1.;
+> 111222333444555666777888999 111222333444555666777.333444 123456 0.333 345323.765329 1
+> --------------------------- ---------------------------- ------ ----- ------------- -
+> 111222333444555666777888999 111222333444555666777.333444 123456 0.333 345323.765329 1
+> rows: 1
+
+SELECT 1_.;
+> exception SYNTAX_ERROR_2
+
+SELECT 1_1._1;
+> exception SYNTAX_ERROR_2
+
+SELECT 9_9.9_;
+> exception SYNTAX_ERROR_2
+
+SELECT 132_134.3__3;
+> exception SYNTAX_ERROR_2
+
+SELECT 111_222_333_444_555_666__777;
+> exception SYNTAX_ERROR_2
diff --git a/h2/src/test/org/h2/test/scripts/testScript.sql b/h2/src/test/org/h2/test/scripts/testScript.sql
@@ -5109,7 +5109,7 @@ SELECT T1.* T2;
 > exception SYNTAX_ERROR_1
 
 select replace('abchihihi', 'i', 'o') abcehohoho, replace('this is tom', 'i') 1e_th_st_om from test;
-> exception SYNTAX_ERROR_1
+> exception SYNTAX_ERROR_2
 
 select monthname(date )'005-0E9-12') d_set fm test;
 > exception SYNTAX_ERROR_1

diff --git a/h2/src/tools/org/h2/build/doc/dictionary.txt b/h2/src/tools/org/h2/build/doc/dictionary.txt
@@ -854,4 +854,4 @@ allotted mismatched wise terminator guarding revolves notion piece submission re
 duplicating unnested hardening sticky massacred
 bck clo cur hwm materializedview udca vol connectionpooldatasource xadatasource
 ampm sssssff sstzh tzs yyyysssss newsequentialid solidus openjdk furthermore ssff secons nashorn fractions
-btrim
+btrim underscores ffl