View on GitHub


q - Text as Data

Download the last stable version zip file Download the last stable version tar.gz file Download the latest stable version RPM Download the latest stable version Debian package Download the single-file executable Download the single-file executable



Behind the Scenes




`q <flags> <query>`

          Simplest execution is `q "SELECT * FROM myfile"` which prints the entire file.


q allows performing SQL-like statements on tabular text data. Its purpose is to bring SQL expressive power to the Linux command line and to provide easy access to text as actual data.

Query should be an SQL-like query which contains filenames instead of table names (or - for stdin). The query itself should be provided as one parameter to the tool (i.e. enclosed in quotes).

Use -H to signify that the input contains a header line. Column names will be detected automatically in that case, and can be used in the query. If this option is not provided, columns will be named cX, starting with 1 (e.g. q "SELECT c3,c8 from ...").

Use -d to specify the input delimiter.

Column types are auto detected by the tool, no casting is needed.

Please note that column names that include spaces need to be used in the query with back-ticks, as per the sqlite standard.

Query/Input/Output encodings are fully supported (and q tries to provide out-of-the-box usability in that area). Please use -e,-E and -Q to control encoding if needed.

All sqlite3 SQL constructs are supported, including joins across files (use an alias for each table).

See for more details.


q gets one parameter - An SQL-like query.

Any standard SQL expression, condition (both WHERE and HAVING), GROUP BY, ORDER BY etc. are allowed.

JOINs are supported and Subqueries are supported in the WHERE clause, but unfortunately not in the FROM clause for now. Use table aliases when performing JOINs.

The SQL syntax itself is sqlite's syntax. For details look at or search the net for examples.

NOTE: Full type detection is implemented, so there is no need for any casting or anything.

NOTE2: When using the -O output header option, use column name aliases if you want to control the output column names. For example, q -O -H "select count(*) cnt,sum(*) as mysum from -" would output cnt and mysum as the output header column names.


q can also get some runtime flags. The following parameters can be used, all optional:

                -h, --help            show this help message and exit
                -v, --version         Print version

                Input Data Options:
                -H, --skip-header   Skip header row. This has been changed from earlier
                version - Only one header row is supported, and the
                header row is used for column naming
                -d DELIMITER, --delimiter=DELIMITER
                Field delimiter. If none specified, then space is used
                as the delimiter.
                -t, --tab-delimited
                Same as -d <tab>. Just a shorthand for handling
                standard tab delimited file You can use $'\t' if you
                want (this is how Linux expects to provide tabs in the
                command line)
                -e ENCODING, --encoding=ENCODING
                Input file encoding. Defaults to UTF-8. set to none
                for not setting any encoding - faster, but at your own
                -z, --gzipped       Data is gzipped. Useful for reading from stdin. For
                files, .gz means automatic gunzipping
                -A, --analyze-only  Analyze sample input and provide information about
                data types
                -m MODE, --mode=MODE
                Data parsing mode. fluffy, relaxed and strict. In
                strict mode, the -c column-count parameter must
                be supplied as well
                -c COLUMN_COUNT, --column-count=COLUMN_COUNT
                Specific column count when using relaxed or strict
                -k, --keep-leading-whitespace
                Keep leading whitespace in values. Default behavior
                strips leading whitespace off values, in order to
                provide out-of-the-box usability for simple use cases.
                If you need to preserve whitespace, use this flag.

                Output Options:
                -D OUTPUT_DELIMITER, --output-delimiter=OUTPUT_DELIMITER
                Field delimiter for output. If none specified, then
                the -d delimiter is used if present, or space if no
                delimiter is specified
                -T, --tab-delimited-output
                Same as -D <tab>. Just a shorthand for outputing tab
                delimited output. You can use -D $'\t' if you want.
                -O, --output-header
                Output header line. Output column-names are determined
                from the query itself. Use column aliases in order to
                set your column names in the query. For example,
                'select name FirstName,value1/value2 MyCalculation
                from ...'. This can be used even if there was no
                header in the input.
                -b, --beautify      Beautify output according to actual values. Might be
                -f FORMATTING, --formatting=FORMATTING
                Output-level formatting, in the format X=fmt,Y=fmt
                etc, where X,Y are output column numbers (e.g. 1 for
                first SELECT column etc.
                -E OUTPUT_ENCODING, --output-encoding=OUTPUT_ENCODING
                Output encoding. Defaults to 'none', leading to
                selecting the system/terminal encoding

                Query Related Options:
                -q QUERY_FILENAME, --query-filename=QUERY_FILENAME
                Read query from the provided filename instead of the
                command line, possibly using the provided query
                encoding (using -Q).
                -Q QUERY_ENCODING, --query-encoding=QUERY_ENCODING
                query text encoding. Experimental. Please send your
                feedback on this


Table names

The table names are the actual file names that you want to read from. Path names are allowed. Use "-" if you want to read from stdin (e.g. q "SELECT * FROM -")

Multiple files can be concatenated by using one of both of the following ways:

Files with .gz extension are considered to be gzipped and decompressed on the fly.

Parsing Modes

q supports multiple parsing modes:

Output formatting option

The format of F is as a list of X=f separated by commas, where X is a column number and f is a python format:


Example 1: ls -ltrd * | q "select c1,count(1) from - group by c1"

This example would print a count of each unique permission string in the current folder.

Example 2: seq 1 1000 | q "select avg(c1),sum(c1) from -"

This example would provide the average and the sum of the numbers in the range 1 to 1000

Example 3: sudo find /tmp -ls | q "select c5,c6,sum(c7)/1024.0/1024 as total from - group by c5,c6 order by total desc"

This example will output the total size in MB per user+group in the /tmp subtree

Example 4: ps -ef | q -H "select UID,count(*) cnt from - group by UID order by cnt desc limit 3"

This example will show process counts per UID, calculated from ps data. Note that the column names provided by ps are being used as column name in the query (The -H flag activates that option)


Harel Ben-Attia (

@harelba on Twitter

Any feedback/suggestions/complaints regarding this tool would be much appreciated. Contributions are most welcome as well, of course.


Copyright (C) 2012--2014 Harel Ben Attia

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 3, or (at your option) any later version.

This program is distributed in the hope that it will be useful,but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA