Attached is an update to contrib/tablefunc. It implements a new hashed

version of crosstab. This fixes a major deficiency in real-world use of
the original version. Easiest to undestand with an illustration:

Data:
-------------------------------------------------------------------
select * from cth;
  id | rowid |        rowdt        |   attribute    |      val
----+-------+---------------------+----------------+---------------
   1 | test1 | 2003-03-01 00:00:00 | temperature    | 42
   2 | test1 | 2003-03-01 00:00:00 | test_result    | PASS
   3 | test1 | 2003-03-01 00:00:00 | volts          | 2.6987
   4 | test2 | 2003-03-02 00:00:00 | temperature    | 53
   5 | test2 | 2003-03-02 00:00:00 | test_result    | FAIL
   6 | test2 | 2003-03-02 00:00:00 | test_startdate | 01 March 2003
   7 | test2 | 2003-03-02 00:00:00 | volts          | 3.1234
(7 rows)

Original crosstab:
-------------------------------------------------------------------
SELECT * FROM crosstab(
   'SELECT rowid, attribute, val FROM cth ORDER BY 1,2',4)
AS c(rowid text, temperature text, test_result text, test_startdate
text, volts text);
  rowid | temperature | test_result | test_startdate | volts
-------+-------------+-------------+----------------+--------
  test1 | 42          | PASS        | 2.6987         |
  test2 | 53          | FAIL        | 01 March 2003  | 3.1234
(2 rows)

Hashed crosstab:
-------------------------------------------------------------------
SELECT * FROM crosstab(
   'SELECT rowid, attribute, val FROM cth ORDER BY 1',
   'SELECT DISTINCT attribute FROM cth ORDER BY 1')
AS c(rowid text, temperature int4, test_result text, test_startdate
timestamp, volts float8);
  rowid | temperature | test_result |   test_startdate    | volts
-------+-------------+-------------+---------------------+--------
  test1 |          42 | PASS        |                     | 2.6987
  test2 |          53 | FAIL        | 2003-03-01 00:00:00 | 3.1234
(2 rows)

Notice that the original crosstab slides data over to the left in the
result tuple when it encounters missing data. In order to work around
this you have to be make your source sql do all sorts of contortions
(cartesian join of distinct rowid with distinct attribute; left join
that back to the real source data). The new version avoids this by
building a hash table using a second distinct attribute query.

The new version also allows for "extra" columns (see the README) and
allows the result columns to be coerced into differing datatypes if they
are suitable (as shown above).

In testing a "real-world" data set (69 distinct rowid's, 27 distinct
categories/attributes, multiple missing data points) I saw about a
5-fold improvement in execution time (from about 2200 ms old, to 440 ms
new).

I left the original version intact because: 1) BC, 2) it is probably
slightly faster if you know that you have no missing attributes.

README and regression test adjustments included. If there are no
objections, please apply.

Joe Conway

contrib/tablefunc/README.tablefunc

@@ -333,6 +333,125 @@ AS ct(row_name text, category_1 text, category_2 text, category_3 text);
 ==================================================================
 Name
 
+crosstab(text, text) - returns a set of row_name, extra, and
+                      category value columns
+
+Synopsis
+
+crosstab(text source_sql, text category_sql)
+
+Inputs
+
+  source_sql
+
+    A SQL statement which produces the source set of data. The SQL statement
+    must return one row_name column, one category column, and one value
+    column. It may also have one or more "extra" columns.
+
+    The row_name column must be first. The category and value columns
+    must be the last two columns, in that order. "extra" columns must be
+    columns 2 through (N - 2), where N is the total number of columns.
+
+    The "extra" columns are assumed to be the same for all rows with the
+    same row_name. The values returned are copied from the first row
+    with a given row_name and subsequent values of these columns are ignored
+    until row_name changes.
+
+    e.g. source_sql must produce a set something like:
+         SELECT row_name, extra_col, cat, value FROM foo;
+
+             row_name    extra_col   cat    value
+            ----------+------------+-----+---------
+              row1         extra1    cat1    val1
+              row1         extra1    cat2    val2
+              row1         extra1    cat4    val4
+              row2         extra2    cat1    val5
+              row2         extra2    cat2    val6
+              row2         extra2    cat3    val7
+              row2         extra2    cat4    val8
+
+  category_sql
+
+    A SQL statement which produces the distinct set of categories. The SQL
+    statement must return one category column only. category_sql must produce
+    at least one result row or an error will be generated. category_sql
+    must not produce duplicate categories or an error will be generated.
+
+    e.g. SELECT DISTINCT cat FROM foo;
+
+              cat
+            -------
+              cat1
+              cat2
+              cat3
+              cat4
+
+Outputs
+
+  Returns setof record, which must be defined with a column definition
+  in the FROM clause of the SELECT statement, e.g.:
+
+    SELECT * FROM crosstab(source_sql, cat_sql)
+    AS ct(row_name text, extra text, cat1 text, cat2 text, cat3 text, cat4 text);
+
+    the example crosstab function produces a set something like:
+                      <== values  columns ==>
+           row_name   extra   cat1   cat2   cat3   cat4
+           ---------+-------+------+------+------+------
+             row1     extra1  val1   val2          val4
+             row2     extra2  val5   val6   val7   val8
+
+Notes
+
+  1. source_sql must be ordered by row_name (column 1).
+
+  2. The number of values columns is determined at run-time. The 
+     column definition provided in the FROM clause must provide for
+     the correct number of columns of the proper data types.
+
+  3. Missing values (i.e. not enough adjacent rows of same row_name to
+     fill the number of result values columns) are filled in with nulls.
+
+  4. Extra values (i.e. source rows with category not found in category_sql
+     result) are skipped.
+
+  5. Rows with a null row_name column are skipped.
+
+
+Example usage
+
+create table cth(id serial, rowid text, rowdt timestamp, attribute text, val text);
+insert into cth values(DEFAULT,'test1','01 March 2003','temperature','42');
+insert into cth values(DEFAULT,'test1','01 March 2003','test_result','PASS');
+insert into cth values(DEFAULT,'test1','01 March 2003','volts','2.6987');
+insert into cth values(DEFAULT,'test2','02 March 2003','temperature','53');
+insert into cth values(DEFAULT,'test2','02 March 2003','test_result','FAIL');
+insert into cth values(DEFAULT,'test2','02 March 2003','test_startdate','01 March 2003');
+insert into cth values(DEFAULT,'test2','02 March 2003','volts','3.1234');
+
+SELECT * FROM crosstab
+(
+  'SELECT rowid, rowdt, attribute, val FROM cth ORDER BY 1',
+  'SELECT DISTINCT attribute FROM cth ORDER BY 1'
+)
+AS
+(
+       rowid text,
+       rowdt timestamp,
+       temperature int4,
+       test_result text,
+       test_startdate timestamp,
+       volts float8
+);
+ rowid |          rowdt           | temperature | test_result |      test_startdate      | volts  
+-------+--------------------------+-------------+-------------+--------------------------+--------
+ test1 | Sat Mar 01 00:00:00 2003 |          42 | PASS        |                          | 2.6987
+ test2 | Sun Mar 02 00:00:00 2003 |          53 | FAIL        | Sat Mar 01 00:00:00 2003 | 3.1234
+(2 rows)
+
+==================================================================
+Name
+
 connectby(text, text, text, text, int[, text]) - returns a set
     representing a hierarchy (tree structure)
 

contrib/tablefunc/expected/tablefunc.out

@@ -123,6 +123,79 @@ SELECT * FROM crosstab('SELECT rowid, attribute, val FROM ct where rowclass = ''
  test2 | val5 | val6 | val7 | val8
 (2 rows)
 
+--
+-- hash based crosstab
+--
+create table cth(id serial, rowid text, rowdt timestamp, attribute text, val text);
+NOTICE:  CREATE TABLE will create implicit sequence 'cth_id_seq' for SERIAL column 'cth.id'
+insert into cth values(DEFAULT,'test1','01 March 2003','temperature','42');
+insert into cth values(DEFAULT,'test1','01 March 2003','test_result','PASS');
+-- the next line is intentionally left commented and is therefore a "missing" attribute
+-- insert into cth values(DEFAULT,'test1','01 March 2003','test_startdate','28 February 2003');
+insert into cth values(DEFAULT,'test1','01 March 2003','volts','2.6987');
+insert into cth values(DEFAULT,'test2','02 March 2003','temperature','53');
+insert into cth values(DEFAULT,'test2','02 March 2003','test_result','FAIL');
+insert into cth values(DEFAULT,'test2','02 March 2003','test_startdate','01 March 2003');
+insert into cth values(DEFAULT,'test2','02 March 2003','volts','3.1234');
+-- return attributes as plain text
+SELECT * FROM crosstab(
+  'SELECT rowid, rowdt, attribute, val FROM cth ORDER BY 1',
+  'SELECT DISTINCT attribute FROM cth ORDER BY 1')
+AS c(rowid text, rowdt timestamp, temperature text, test_result text, test_startdate text, volts text);
+ rowid |          rowdt           | temperature | test_result | test_startdate | volts  
+-------+--------------------------+-------------+-------------+----------------+--------
+ test1 | Sat Mar 01 00:00:00 2003 | 42          | PASS        |                | 2.6987
+ test2 | Sun Mar 02 00:00:00 2003 | 53          | FAIL        | 01 March 2003  | 3.1234
+(2 rows)
+
+-- this time without rowdt
+SELECT * FROM crosstab(
+  'SELECT rowid, attribute, val FROM cth ORDER BY 1',
+  'SELECT DISTINCT attribute FROM cth ORDER BY 1')
+AS c(rowid text, temperature text, test_result text, test_startdate text, volts text);
+ rowid | temperature | test_result | test_startdate | volts  
+-------+-------------+-------------+----------------+--------
+ test1 | 42          | PASS        |                | 2.6987
+ test2 | 53          | FAIL        | 01 March 2003  | 3.1234
+(2 rows)
+
+-- convert attributes to specific datatypes
+SELECT * FROM crosstab(
+  'SELECT rowid, rowdt, attribute, val FROM cth ORDER BY 1',
+  'SELECT DISTINCT attribute FROM cth ORDER BY 1')
+AS c(rowid text, rowdt timestamp, temperature int4, test_result text, test_startdate timestamp, volts float8);
+ rowid |          rowdt           | temperature | test_result |      test_startdate      | volts  
+-------+--------------------------+-------------+-------------+--------------------------+--------
+ test1 | Sat Mar 01 00:00:00 2003 |          42 | PASS        |                          | 2.6987
+ test2 | Sun Mar 02 00:00:00 2003 |          53 | FAIL        | Sat Mar 01 00:00:00 2003 | 3.1234
+(2 rows)
+
+-- source query and category query out of sync
+SELECT * FROM crosstab(
+  'SELECT rowid, rowdt, attribute, val FROM cth ORDER BY 1',
+  'SELECT DISTINCT attribute FROM cth WHERE attribute IN (''temperature'',''test_result'',''test_startdate'') ORDER BY 1')
+AS c(rowid text, rowdt timestamp, temperature int4, test_result text, test_startdate timestamp);
+ rowid |          rowdt           | temperature | test_result |      test_startdate      
+-------+--------------------------+-------------+-------------+--------------------------
+ test1 | Sat Mar 01 00:00:00 2003 |          42 | PASS        | 
+ test2 | Sun Mar 02 00:00:00 2003 |          53 | FAIL        | Sat Mar 01 00:00:00 2003
+(2 rows)
+
+-- if category query generates no rows, get expected error
+SELECT * FROM crosstab(
+  'SELECT rowid, rowdt, attribute, val FROM cth ORDER BY 1',
+  'SELECT DISTINCT attribute FROM cth WHERE attribute = ''a'' ORDER BY 1')
+AS c(rowid text, rowdt timestamp, temperature int4, test_result text, test_startdate timestamp, volts float8);
+ERROR:  load_categories_hash: provided categories SQL must return 1 column of at least one row
+-- if category query generates more than one column, get expected error
+SELECT * FROM crosstab(
+  'SELECT rowid, rowdt, attribute, val FROM cth ORDER BY 1',
+  'SELECT DISTINCT rowdt, attribute FROM cth ORDER BY 2')
+AS c(rowid text, rowdt timestamp, temperature int4, test_result text, test_startdate timestamp, volts float8);
+ERROR:  load_categories_hash: provided categories SQL must return 1 column of at least one row
+--
+-- connectby
+--
 -- test connectby with text based hierarchy
 CREATE TABLE connectby_text(keyid text, parent_keyid text);
 \copy connectby_text from 'data/connectby_text.data'

contrib/tablefunc/sql/tablefunc.sql

@@ -38,6 +38,61 @@ SELECT * FROM crosstab('SELECT rowid, attribute, val FROM ct where rowclass = ''
 SELECT * FROM crosstab('SELECT rowid, attribute, val FROM ct where rowclass = ''group1'' ORDER BY 1,2;', 3) AS c(rowid text, att1 text, att2 text, att3 text);
 SELECT * FROM crosstab('SELECT rowid, attribute, val FROM ct where rowclass = ''group1'' ORDER BY 1,2;', 4) AS c(rowid text, att1 text, att2 text, att3 text, att4 text);
 
+--
+-- hash based crosstab
+--
+create table cth(id serial, rowid text, rowdt timestamp, attribute text, val text);
+insert into cth values(DEFAULT,'test1','01 March 2003','temperature','42');
+insert into cth values(DEFAULT,'test1','01 March 2003','test_result','PASS');
+-- the next line is intentionally left commented and is therefore a "missing" attribute
+-- insert into cth values(DEFAULT,'test1','01 March 2003','test_startdate','28 February 2003');
+insert into cth values(DEFAULT,'test1','01 March 2003','volts','2.6987');
+insert into cth values(DEFAULT,'test2','02 March 2003','temperature','53');
+insert into cth values(DEFAULT,'test2','02 March 2003','test_result','FAIL');
+insert into cth values(DEFAULT,'test2','02 March 2003','test_startdate','01 March 2003');
+insert into cth values(DEFAULT,'test2','02 March 2003','volts','3.1234');
+
+-- return attributes as plain text
+SELECT * FROM crosstab(
+  'SELECT rowid, rowdt, attribute, val FROM cth ORDER BY 1',
+  'SELECT DISTINCT attribute FROM cth ORDER BY 1')
+AS c(rowid text, rowdt timestamp, temperature text, test_result text, test_startdate text, volts text);
+
+-- this time without rowdt
+SELECT * FROM crosstab(
+  'SELECT rowid, attribute, val FROM cth ORDER BY 1',
+  'SELECT DISTINCT attribute FROM cth ORDER BY 1')
+AS c(rowid text, temperature text, test_result text, test_startdate text, volts text);
+
+-- convert attributes to specific datatypes
+SELECT * FROM crosstab(
+  'SELECT rowid, rowdt, attribute, val FROM cth ORDER BY 1',
+  'SELECT DISTINCT attribute FROM cth ORDER BY 1')
+AS c(rowid text, rowdt timestamp, temperature int4, test_result text, test_startdate timestamp, volts float8);
+
+-- source query and category query out of sync
+SELECT * FROM crosstab(
+  'SELECT rowid, rowdt, attribute, val FROM cth ORDER BY 1',
+  'SELECT DISTINCT attribute FROM cth WHERE attribute IN (''temperature'',''test_result'',''test_startdate'') ORDER BY 1')
+AS c(rowid text, rowdt timestamp, temperature int4, test_result text, test_startdate timestamp);
+
+-- if category query generates no rows, get expected error
+SELECT * FROM crosstab(
+  'SELECT rowid, rowdt, attribute, val FROM cth ORDER BY 1',
+  'SELECT DISTINCT attribute FROM cth WHERE attribute = ''a'' ORDER BY 1')
+AS c(rowid text, rowdt timestamp, temperature int4, test_result text, test_startdate timestamp, volts float8);
+
+-- if category query generates more than one column, get expected error
+SELECT * FROM crosstab(
+  'SELECT rowid, rowdt, attribute, val FROM cth ORDER BY 1',
+  'SELECT DISTINCT rowdt, attribute FROM cth ORDER BY 2')
+AS c(rowid text, rowdt timestamp, temperature int4, test_result text, test_startdate timestamp, volts float8);
+
+
+--
+-- connectby
+--
+
 -- test connectby with text based hierarchy
 CREATE TABLE connectby_text(keyid text, parent_keyid text);
 \copy connectby_text from 'data/connectby_text.data'

contrib/tablefunc/tablefunc.h

@@ -34,6 +34,7 @@
  */
 extern Datum normal_rand(PG_FUNCTION_ARGS);
 extern Datum crosstab(PG_FUNCTION_ARGS);
+extern Datum crosstab_hash(PG_FUNCTION_ARGS);
 extern Datum connectby_text(PG_FUNCTION_ARGS);
 
 #endif   /* TABLEFUNC_H */

contrib/tablefunc/tablefunc.sql.in

@@ -52,6 +52,11 @@ RETURNS setof record
 AS 'MODULE_PATHNAME','crosstab'
 LANGUAGE 'C' STABLE STRICT;
 
+CREATE OR REPLACE FUNCTION crosstab(text,text)
+RETURNS setof record
+AS 'MODULE_PATHNAME','crosstab_hash'
+LANGUAGE 'C' STABLE STRICT;
+
 CREATE OR REPLACE FUNCTION connectby(text,text,text,text,int,text)
 RETURNS setof record
 AS 'MODULE_PATHNAME','connectby_text'