Page MenuHomeClusterLabs Projects
Diviner User Docs Arcanist User Guide: Script and Regex Linter

Arcanist User Guide: Script and Regex Linter
Phorge Administrator and User Documentation (Application User Guides)

Explains how to use the Script and Regex linter to invoke an existing lint engine that is not integrated with Arcanist.

The Script and Regex linter is a simple glue linter which runs some script on each path, and then uses a regex to parse lint messages from the script's output. (This linter uses a script and a regex to interpret the results of some real linter, it does not itself lint both scripts and regexes).

Configure this linter by setting these keys in your configuration:

  • script-and-regex.script Script command to run. This can be the path to a linter script, but may also include flags or use shell features (see below for examples).
  • script-and-regex.regex The regex to process output with. This regex uses named capturing groups (detailed below) to interpret output.

The script will be invoked from the project root, so you can specify a relative path like scripts/lint.sh or an absolute path like /opt/lint/lint.sh.

This linter is necessarily more limited in its capabilities than a normal linter which can perform custom processing, but may be somewhat simpler to configure.

Script...

The script will be invoked once for each file that is to be linted, with the file passed as the first argument. The file may begin with a -; ensure your script will not interpret such files as flags (perhaps by ending your script configuration with --, if its argument parser supports that).

Note that when run via arc diff, the list of files to be linted does not include binary files, symlinks, deleted files, or directories. These special file types are not supported by this linter.

The script should emit lint messages to stdout, which will be parsed with the provided regex.

For example, you might use a configuration like this:

"script-and-regex.script": "/opt/lint/lint.sh --flag value --other-flag --"

stderr is ignored. If you have a script which writes messages to stderr, you can redirect stderr to stdout by using a configuration like this:

"script-and-regex.script": "sh -c '/opt/lint/lint.sh \"$0\" 2>&1'"

The return code of the script must be 0, or an exception will be raised reporting that the linter failed. If you have a script which exits nonzero under normal circumstances, you can force it to always exit 0 by using a configuration like this:

"script-and-regex.script": "sh -c '/opt/lint/lint.sh \"$0\" || true'"

Multiple instances of the script will be run in parallel if there are multiple files to be linted, so they should not use any unique resources. For instance, this configuration would not work properly, because several processes may attempt to write to the file at the same time:

"script-and-regex.script": "sh -c '/opt/lint/lint.sh --output /tmp/lint.out \"$0\" && cat /tmp/lint.out'"

There are necessary limits to how gracefully this linter can deal with edge cases, because it is just a script and a regex. If you need to do things that this linter can't handle, you can write a phutil linter and move the logic to handle those cases into PHP. PHP is a better general-purpose programming language than regular expressions are, if only by a small margin.

...and Regex

The regex must be a valid PHP PCRE regex, including delimiters and flags.

The regex will be matched against the entire output of the script, so it should generally be in this form if messages are one-per-line:

/^...$/m

The regex should capture these named patterns with (?P<name>...):

  • message (required) Text describing the lint message. For example, "This is a syntax error.".
  • name (optional) Text summarizing the lint message. For example, "Syntax Error".
  • severity (optional) The word "error", "warning", "autofix", "advice", or "disabled", in any combination of upper and lower case. Instead, you may match groups called error, warning, advice, autofix, or disabled. These allow you to match output formats like "E123" and "W123" to indicate errors and warnings, even though the word "error" is not present in the output. If no severity capturing group is present, messages are raised with "error" severity. If multiple severity capturing groups are present, messages are raised with the highest captured severity. Capturing groups like error supersede the severity capturing group.
  • error (optional) Match some nonempty substring to indicate that this message has "error" severity.
  • warning (optional) Match some nonempty substring to indicate that this message has "warning" severity.
  • advice (optional) Match some nonempty substring to indicate that this message has "advice" severity.
  • autofix (optional) Match some nonempty substring to indicate that this message has "autofix" severity.
  • disabled (optional) Match some nonempty substring to indicate that this message has "disabled" severity.
  • file (optional) The name of the file to raise the lint message in. If not specified, defaults to the linted file. It is generally not necessary to capture this unless the linter can raise messages in files other than the one it is linting.
  • line (optional) The line number of the message. If no text is captured, the message is assumed to affect the entire file.
  • char (optional) The character offset of the message.
  • offset (optional) The byte offset of the message. If captured, this supersedes line and char.
  • original (optional) The text the message affects.
  • replacement (optional) The text that the range captured by original should be automatically replaced by to resolve the message.
  • code (optional) A short error type identifier which can be used elsewhere to configure handling of specific types of messages. For example, "EXAMPLE1", "EXAMPLE2", etc., where each code identifies a class of message like "syntax error", "missing whitespace", etc. This allows configuration to later change the severity of all whitespace messages, for example.
  • ignore (optional) Match some nonempty substring to ignore the match. You can use this if your linter sometimes emits text like "No lint errors".
  • stop (optional) Match some nonempty substring to stop processing input. Remaining matches for this file will be discarded, but linting will continue with other linters and other files.
  • halt (optional) Match some nonempty substring to halt all linting of this file by any linter. Linting will continue with other files.
  • throw (optional) Match some nonempty substring to throw an error, which will stop arc completely. You can use this to fail abruptly if you encounter unexpected output. All processing will abort.

Numbered capturing groups are ignored.

For example, if your lint script's output looks like this:

error:13 Too many goats!
warning:22 Not enough boats.

...you could use this regex to parse it:

/^(?P<severity>warning|error):(?P<line>\d+) (?P<message>.*)$/m

The simplest valid regex for line-oriented output is something like this:

/^(?P<message>.*)$/m