Postgres: \copy syntax error in .sql file - postgresql

Question

I'm trying to write a script that copies data from a crosstab query to a .csv file in Postgres 8.4. I am able to run the command in the psql command line but when I put the command in a file and run it using the -f option, I get a syntax error.
Here's an example of what I'm looking at (from this great answer):
CREATE TEMP TABLE t (
section text
,status text
,ct integer
);
INSERT INTO t VALUES
('A', 'Active', 1), ('A', 'Inactive', 2)
,('B', 'Active', 4), ('B', 'Inactive', 5)
, ('C', 'Inactive', 7);
\copy (
SELECT * FROM crosstab(
'SELECT section, status, ct
FROM t
ORDER BY 1,2'
,$$VALUES ('Active'::text), ('Inactive')$$)
AS ct ("Section" text, "Active" int, "Inactive" int)
) TO 'test.csv' HEADER CSV
I then run this and get the following syntax error:
$ psql [system specific] -f copy_test.sql
CREATE TABLE
INSERT 0 5
psql:copy_test.sql:12: \copy: parse error at end of line
psql:copy_test.sql:19: ERROR: syntax error at or near ")"
LINE 7: ) TO 'test.csv' HEADER CSV
^
A similar exercise doing just a simple query without crosstab works without incident.
What is causing the syntax error and how can I copy this table to a csv file using script file?

Answer 1

psql thinks your first command is just \copy ( and the lines below that are from another unrelated statement. Meta-commands aren't spread on multiple lines, because newline is is a terminator for them.
Relevant excerpts from psql manpage with some emphasis added:
Meta-Commands
Anything you enter in psql that begins with an unquoted backslash is a
psql meta-command that is processed by psql itself. These commands
make psql more useful for administration or scripting. Meta-commands
are often called slash or backslash commands.
....
....(skipped)
Parsing for arguments stops at the end of the line, or when another
unquoted backslash is found. An unquoted backslash is taken as the
beginning of a new meta-command. The special sequence \\ (two
backslashes) marks the end of arguments and continues parsing SQL
commands, if any. That way SQL and psql commands can be freely mixed
on a line. But in any case, the arguments of a meta-command cannot
continue beyond the end of the line.
So the first error is that \copy ( failing, then the lines below are interpreted as an independent SELECT which looks fine until line 7 when there is a spurious closing parenthesis.
As told in the comments, the fix would be to cram the whole meta-command into a single line.

Answer 2

As with this answer, create a multi-line VIEW with a single-line \copy command, e.g.:
CREATE TEMP TABLE t (
section text
,status text
,ct integer
);
INSERT INTO t VALUES
('A', 'Active', 1), ('A', 'Inactive', 2)
,('B', 'Active', 4), ('B', 'Inactive', 5)
, ('C', 'Inactive', 7);
CREATE TEMP VIEW v1 AS
SELECT * FROM crosstab(
'SELECT section, status, ct
FROM t
ORDER BY 1,2'
,$$VALUES ('Active'::text), ('Inactive')$$)
AS ct ("Section" text, "Active" int, "Inactive" int);
\copy (SELECT * FROM v1) TO 'test.csv' HEADER CSV
-- optional
DROP VIEW v1;

Answer 3

The answers listed here explain the reasoning quite clearly. Here is a small hack that allows you to have your sql contain multiple lines and work with psql.
# Using a file
psql -f <(tr -d '\n' < ~/s/test.sql )
# or
psql < <(tr -d '\n' < ~/s/test.sql )
# Putting the SQL using a HEREDOC
cat <<SQL | tr -d '\n' | \psql mydatabase
\COPY (
SELECT
provider_id,
provider_name,
...
) TO './out.tsv' WITH( DELIMITER E'\t', NULL '', )
SQL

Answer 4

According to the psql documentation:
-f filename
--file filename
Use the file filename as the source of commands instead of reading commands interactively. After the file is processed, psql terminates. This is in many ways equivalent to the internal command \i.
If filename is - (hyphen), then standard input is read.
Using this option is subtly different from writing psql < filename. In general, both will do what you expect, but using -f enables some nice features such as error messages with line numbers. There is also a slight chance that using this option will reduce the start-up overhead. On the other hand, the variant using the shell's input redirection is (in theory) guaranteed to yield exactly the same output that you would have gotten had you entered everything by hand.
This would be one of those cases where the -f option treats your input differently from the command line. Removing your newlines worked, redirecting the original file to psql's stdin would likely have worked as well.