fileA list of files to parse.list of string (required)
regexA list of regex to apply to the file data.list of string (required)
accessorThe accessor to use.string
buffer_sizeMaximum size of line buffer (default 64kb).int


Parses a file with a set of regexp and yields matches as records. The file is read into a large buffer. Then each regular expression is applied to the buffer, and all matches are emitted as rows.

The regular expressions are specified in the Go syntax. They are expected to contain capture variables to name the matches extracted.

For example, consider a HTML file with simple links. The regular expression might be:


To produce rows with a column Link.

The aim of this plugin is to split the file into records which can be further parsed. For example, if the file consists of multiple records, this plugin can be used to extract each record, while parse_string_with_regex() can be used to further split each record into elements. This works better than trying to write a more complex regex which tries to capture a lot of details in one pass.


Here is an example of parsing the /var/lib/dpkg/status files. These files consist of records separated by empty lines:

Package: ubuntu-advantage-tools
Status: install ok installed
Priority: important
Section: misc
Installed-Size: 74
Maintainer: Ubuntu Developers <>
Architecture: all
Version: 17
 /etc/cron.daily/ubuntu-advantage-tools 36de53e7c2d968f951b11c64be101b91
 /etc/update-motd.d/80-esm 6ffbbf00021b4ea4255cff378c99c898
 /etc/update-motd.d/80-livepatch 1a3172ffaa815d12b58648f117ffb67e
Description: management tools for Ubuntu Advantage
 Ubuntu Advantage is the professional package of tooling, technology
 and expertise from Canonical, helping organizations around the world
 manage their Ubuntu deployments.
 Subscribers to Ubuntu Advantage will find helpful tools for accessing
 services in this package.

The following query extracts the fields in two passes. The first pass uses parse_records_with_regex() to extract records in blocks, while using parse_string_with_regex() to further break the block into fields.

SELECT parse_string_with_regex(
     'Architecture:\\s(?P<Architecture>.+)']) as Record
   FROM parse_records_with_regex(
comments powered by Disqus