-- The path spec of the shredder (docs/COL-SHRED.md): canonical forms and errors
CREATE EXTENSION pg_sift;
CREATE EXTENSION pg_sift_test;
\set VERBOSITY default
-- the canonical form: one flat item per column, declaration order, "." separators
SELECT sift._test_shred_spec('attributes->locale text, metrics.* float4, items[]{category text, sku text, kind text, qty int4, price float4}, notes[]{type text, author_kind text}');
                                                                                     _test_shred_spec                                                                                      
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
 attributes.locale text, metrics.* float4, items[].category text, items[].sku text, items[].kind text, items[].qty int4, items[].price float4, notes[].type text, notes[].author_kind text
(1 row)

SELECT sift._test_shred_spec('a.b int4, a.c text, a.d.e bool');
        _test_shred_spec        
--------------------------------
 a.b int4, a.c text, a.d.e bool
(1 row)

SELECT sift._test_shred_spec('A INT, b SmallInt, c BIGINT, d Real, e DOUBLE PRECISION, f double, g boolean, h Text');
                           _test_shred_spec                           
----------------------------------------------------------------------
 A int4, b int2, c int8, d float4, e float8, f float8, g bool, h text
(1 row)

-- groups are sugar for flat paths and merge with them
SELECT sift._test_shred_spec('items[].sku text, items[]{qty int4}, items[].price float4');
                     _test_shred_spec                     
----------------------------------------------------------
 items[].sku text, items[].qty int4, items[].price float4
(1 row)

SELECT sift._test_shred_spec('a[]{b[]{c int4, d text}, e float8}, f int8');
                   _test_shred_spec                   
------------------------------------------------------
 a[].b[].c int4, a[].b[].d text, a[].e float8, f int8
(1 row)

SELECT sift._test_shred_spec('m[][] float4, t[] text(64), w.*{x int4, y[] text}');
                   _test_shred_spec                   
------------------------------------------------------
 m[][] float4, t[] text(64), w.*.x int4, w.*.y[] text
(1 row)

SELECT sift._test_shred_spec('a.*.* int4');
 _test_shred_spec 
------------------
 a.*.* int4
(1 row)

SELECT sift._test_shred_spec('a.* [] text');
 _test_shred_spec 
------------------
 a.*[] text
(1 row)

-- quoted keys, separators and white space
SELECT sift._test_shred_spec(' "a.b" -> "c d" text ,"say ""hi""" int4');
          _test_shred_spec           
-------------------------------------
 "a.b"."c d" text, "say ""hi""" int4
(1 row)

SELECT sift._test_shred_spec(E'a\n->\tb[ ]\n  text');
 _test_shred_spec 
------------------
 a.b[] text
(1 row)

SELECT sift._test_shred_spec('naïve int4, ключ text');
   _test_shred_spec    
-----------------------
 naïve int4, ключ text
(1 row)

SELECT sift._test_shred_spec('$id int8, 2024 text');
  _test_shred_spec   
---------------------
 $id int8, 2024 text
(1 row)

-- the fingerprint is stable for the canonical text
SELECT length(sift._test_shred_spec('a int4')) , sift._test_shred_spec('a int4') = sift._test_shred_spec(' a   INT4 ');
 length | ?column? 
--------+----------
      6 | t
(1 row)

-- errors, with the position
SELECT sift._test_shred_spec('');
ERROR:  sift shred paths: expected a key at the start of a path at position 1
SELECT sift._test_shred_spec('a');
ERROR:  sift shred paths: expected a type after the path (bool, int2, int4, int8, float4, float8, text or astext) at position 2
SELECT sift._test_shred_spec('a nope');
ERROR:  sift shred paths: unknown type "nope" (bool, int2, int4, int8, float4, float8, text or astext) at position 3
SELECT sift._test_shred_spec('a int4,');
ERROR:  sift shred paths: expected a key at the start of a path at position 8
SELECT sift._test_shred_spec('a int4 b text');
ERROR:  sift shred paths: expected "," or the end of the paths at position 8
SELECT sift._test_shred_spec('a int4 junk');
ERROR:  sift shred paths: expected "," or the end of the paths at position 8
SELECT sift._test_shred_spec('a.b.');
ERROR:  sift shred paths: expected a key at position 5
SELECT sift._test_shred_spec('a - b int4');
ERROR:  sift shred paths: unexpected "-" (a separator is "." or "->"; quote keys that contain it) at position 3
SELECT sift._test_shred_spec('a[ int4');
ERROR:  sift shred paths: expected "[]" at position 2
SELECT sift._test_shred_spec('"a int4');
ERROR:  sift shred paths: unterminated quoted key at position 1
SELECT sift._test_shred_spec('a int4 #');
ERROR:  sift shred paths: unexpected character "#" at position 8
SELECT sift._test_shred_spec('"" int4');
ERROR:  sift shred paths: empty key at position 1
SELECT sift._test_shred_spec('* int4');
ERROR:  sift shred paths: expected a key at the start of a path at position 1
SELECT sift._test_shred_spec('[] int4');
ERROR:  sift shred paths: expected a key at the start of a path at position 1
SELECT sift._test_shred_spec('a text(');
ERROR:  sift shred paths: expected the length in bytes at position 8
SELECT sift._test_shred_spec('a text(x)');
ERROR:  sift shred paths: the length of a text is a number of bytes at position 8
SELECT sift._test_shred_spec('a text(0)');
ERROR:  sift shred paths: the length of a text is between 1 and 1048576 bytes at position 8
SELECT sift._test_shred_spec('a text(2000000)');
ERROR:  sift shred paths: the length of a text is between 1 and 1048576 bytes at position 8
SELECT sift._test_shred_spec('a text(10');
ERROR:  sift shred paths: expected ")" at position 10
SELECT sift._test_shred_spec('a int4(10)');
ERROR:  sift shred paths: expected "," or the end of the paths at position 7
SELECT sift._test_shred_spec('a {b int4}');
ERROR:  sift shred paths: a group of keys needs an array (key[]) or a wildcard (key.*) before it at position 3
SELECT sift._test_shred_spec('a[] {}');
ERROR:  sift shred paths: expected a key at position 6
SELECT sift._test_shred_spec('a[] {b int4');
ERROR:  sift shred paths: expected "}" at position 12
SELECT sift._test_shred_spec('a[] {b int4} c');
ERROR:  sift shred paths: expected "," or the end of the paths at position 14
SELECT sift._test_shred_spec('a int4, a text');
ERROR:  sift shred paths: path "a" is declared twice at position 9
SELECT sift._test_shred_spec('a int4, a.b text');
ERROR:  sift shred paths: path "a" is used as a value and as an object at position 9
SELECT sift._test_shred_spec('a.b text, a int4');
ERROR:  sift shred paths: path "a" is used as an object and as a value at position 11
SELECT sift._test_shred_spec('a.* int4, a.b int4');
ERROR:  sift shred paths: path "a" is used as a wildcard and as an object at position 11
SELECT sift._test_shred_spec('a.b int4, a.* int4');
ERROR:  sift shred paths: path "a" is used as an object and as a wildcard at position 11
SELECT sift._test_shred_spec('a[] int4, a.b int4');
ERROR:  sift shred paths: path "a" is used as an array and as an object at position 11
SELECT sift._test_shred_spec('a[] int4, a.* int4');
ERROR:  sift shred paths: path "a" is used as an array and as a wildcard at position 11
SELECT sift._test_shred_spec('a[].b int4, a[] int4');
ERROR:  sift shred paths: path "a[]" is used as an object and as a value at position 14
SELECT sift._test_shred_spec('a[]{b int4, b text}');
ERROR:  sift shred paths: path "a[].b" is declared twice at position 13
-- limits
SELECT sift._test_shred_spec('a1.b2.c3.d4.e5.f6.g7.h8.i9.j10.k11.l12.m13.n14.o15.p16 int4');
                      _test_shred_spec                       
-------------------------------------------------------------
 a1.b2.c3.d4.e5.f6.g7.h8.i9.j10.k11.l12.m13.n14.o15.p16 int4
(1 row)

SELECT sift._test_shred_spec('a1.b2.c3.d4.e5.f6.g7.h8.i9.j10.k11.l12.m13.n14.o15.p16.q17 int4');
ERROR:  sift shred paths: a path has at most 16 steps at position 56
SELECT sift._test_shred_spec((SELECT string_agg(format('c%s int4', i), ', ') FROM generate_series(1, 256) i)) IS NOT NULL AS ok256;
 ok256 
-------
 t
(1 row)

SELECT sift._test_shred_spec((SELECT string_agg(format('c%s int4', i), ', ') FROM generate_series(1, 257) i));
ERROR:  sift shred paths: more than 256 columns at position 2709
SELECT sift._test_shred_spec((SELECT string_agg(format('g%s[] int4', i), ', ') FROM generate_series(1, 63) i)) IS NOT NULL AS ok63;
 ok63 
------
 t
(1 row)

SELECT sift._test_shred_spec((SELECT string_agg(format('g%s[] int4', i), ', ') FROM generate_series(1, 64) i));
ERROR:  sift shred paths: more than 63 arrays and wildcards at position 748
SELECT sift._test_shred_spec(repeat('a int4,', 20000) || 'b int4');
ERROR:  sift shred paths: the spec has more than 65536 bytes
HINT:  Declare fewer paths, or a wildcard path for the keys of an object.
DROP EXTENSION pg_sift_test;
DROP EXTENSION pg_sift;