Pattern matching » History » Version 8
    « Previous - 
    Version 8/13
    (diff) - 
    Next » - 
    Current version
    
    digitalcircuit, 09/13/2018 04:36 AM 
    More formatting polish
    
Pattern matching¶
WORK IN PROGRESS
- Table of contents
- Pattern matching
Upgrade notice¶
This changed in Quassel 0.13.  Follow the migration guide if you're upgrading from an earlier version.
When using the client and core with different versions, some ignore rules will be processed differently.
Breakdown of usage¶
Quassel supports several types of pattern matching for ignore and highlight rules.
Ignore Rules¶
| Regular expression | Item | Pattern matching type | 
|---|---|---|
| Unchecked | Ignore Rule | Wildcard | 
| Scope | Multiple wildcard | |
| Checked | Ignore Rule | Regular expression | 
| Scope | Multiple wildcard | 
For backwards compatibility, the Regular expression checkbox only affects Ignore Rule and not Scope.
Highlight Rules¶
| RegEx | Item | Pattern matching type | 
|---|---|---|
| Unchecked | Rule | Phrase | 
| Sender | Multiple wildcard | |
| Channel | Multiple wildcard | |
| Checked | Rule | Regular expression | 
| Sender | Regular expression | |
| Channel | Regular expression | 
This applies to both Local Highlights and Remote Highlights (or if using Monolithic, Legacy Highlights and Highlights).
Types¶
Phrase¶
This matches exactly what you type.
word
- Matches
	- word
- A word.
- has a word in it
 
- Does not match
	- wording
 
Changed in Quassel 0.13:
Spaces are matched, too, allowing you to be more specific.
spaced
There's a leading space in the example above.
- Matches
	- , spaced
- is spaced out
 
- Does not match
	- ;spaced
 
Wildcard¶
Changed in Quassel 0.13:
This uses simple wildcard substitution with * and ? to match any characters or a single character, respectively.  The backslash character \ allows matching a literal *, ?, or \.
*Asking questions\? Nope?
- Matches
	- Asking questions? Nopea
- Basking questions? Nope.
 
- Does not match
	- Asking questions. Nope.
- Asking questions? Nopes.
 
Implicit wildcard is supported, too, allowing you to match everything except for what you specify.
!*filter*
- Matches
	- Everything, unless it contains something from the below list
 
- Does not match
	- filter
- filtering
- #nofilter yo
 
Exclamation points can be escaped at the beginning to match something that starts with ! (this is not required elsewhere).
\!*filter*
- Matches
	- !filter
- !yes filtering
 
- Does not match
	- filter
 
The \ character can be escaped, too, in case you're matching something that starts with \!.
\\!*filter*
- Matches
	- \!filter
- \!yes filtering
 
- Does not match
	- filter
- !filter
 
Multiple wildcard¶
This allows matching multiple wildcard patterns all in one rule by separating each wildcard with semicolons ;.
alpha; beta*
- Matches
	- alpha
- betas
- beta!
 
- Does not match
	- alphamore
- nonbeta
 
Changed in Quassel 0.13:
More complex wildcard patterns are possible.  Use \ to escape special characters.
Alice!*; Bob!*@example.com; Carol*!*; !Caroline!* Dan!*; escaped \; separator; \!not-inverted; \\!slash-prefixed
For ignore rule Scope, you can use ; and line breaks interchangeably.
- Matches
	- Alice![...]
- Bob![...]@example.com
- Carol[...]![...]except as noted below
- Dan![...]
- escaped ; separator
- !not-inverted
- \!slash-prefixed
 
- Does not match
	- Caroline![...]
- @Malice![...]
- John!
 
Implicit wildcard is supported, too, allowing you to match everything except for what you specify.
!Announce*!*; !Wheatley!aperture@*
- Matches
	- Everything, unless it contains something from the below list
 
- Does not match
	- Announce[...]![...]
- Wheatley!aperture[...]@
 
Regular expression¶
This matches using the full regular expression language, best for complex rules and those already familiar with how regular expressions work.  Quassel offers a non-standard ! prefix to invert the rule; if not desired, escape as \!.
The web provides many resources for testing and verifying regular expressions, including https://regex101.com/ .
Some differences exist from Perl due to design choices and limitations of the Qt framework. These are documented on the Qt documentation site for QRegularExpression with Qt 5, and QRegExp for Qt 4. If you're not sure, you're likely using a Qt 5 build.
simple.\*escape-match.*
- Matches
	- simpleA*escape-match
- simpleA*escape-matchBBBB
 
- Does not match
	- not above
- simpleA*escape-mat
- simple*escape-match
- simpleABBBBescape-matchBBBB
 
Inverted rules are supported, allowing you to match everything except for what you specify.
!invert.\*escape-match.*
- Matches
	- Everything, unless it contains something from the below list
 
- Does not match
	- invertA*escape-match
- invertA*escape-matchBBBB
 
Changed in Quassel 0.13:
Exclamation points can be escaped at the beginning to match something that starts with ! (this is not required elsewhere).
\!simple.\*escape-match.*
- Matches
	- !simpleA*escape-matchBBBB
 
- Does not match
	- simpleA*escape-matchBBBB
 
The \ character can be escaped, too, in case you're matching something that starts with \!.
\\!simple.\*escape-match.*
- Matches
	- \!simpleA*escape-matchBBBB
 
- Does not match
	- !simpleA*escape-matchBBBB
 
Migrating to Quassel 0.13¶
	Quassel 0.13 introduced new behavior for pattern matching to provide more flexibility and to improve performance.  Unfortunately, due to the complexity of automatically migrating (patches welcome!), you might have to manually migrate some ignore and highlight rules.
If you're writing new rules, you can start at the breakdown of usage.
Ignore Rules¶
The wildcard matching mode now allows inverting the match with !, and escaping * and ? with \, so you can match a phrase containing question marks or asterisks.  However, this means \ must now be escaped as \\, and ! at the start must be escaped as \!.
| Before | After | Remarks | 
|---|---|---|
| !literal-matching! | \!literal-matching! | Only need to escape !if it's at the start | 
| slashes \ within | slashes \\ within | |
| \! prefixed | \\! prefixed | 
The regular expression format now allows for escaping the first ! in case you want to literally match it, e.g. \!regex.*.
If your client and core version do not match, they will process some ignore rules differently.
If you've upgraded your Quassel client to 0.13 and upgrading the Quassel core to 0.13 is not yet feasible, you can workaround this for ignore rules that start with ! or contain \ by migrating them as above.  For Permanent ignore rules, the core will keep pattern matching according to pre-0.13 behavior, which likely won't cause issues.  If it does, you can temporarily change the affected ignore rules to Dynamic.
Highlight Rules¶
The RegEx checkbox now affects the Channel column, too, switching between multiple wildcard and regular expressions as described above.
Your existing highlight rules are automatically migrated to RegEx if you've specified anything in the Channel column.  Any affected rules will surround the phrase like so, (^\|\W)original phrase(\W\|$).
If you want to make use of the new multiple wildcard matching mode, uncheck RegEx and set Channel to a multiple wildcard pattern.
Example of undoing automatic RegEx conversion to use new wildcard mode
| RegEx | Item | Pattern | 
|---|---|---|
| Checked | Rule | (^\|\W)highlight phrase(\W\|$) | 
| Channel | #quassel(|-*) | |
| Unchecked | Rule | highlight phrase | 
| Channel | #quassel; #quassel-* | 
With the Remote Highlights feature in Quassel 0.13, the Sender column behaves the same way as the Channel column, switching modes according to the RegEx checkbox.
Technical details¶
For all the technical details and source code changes, take a look at Quassel pull request #415.  MultiPhrase is not covered here as it's only used inside Quassel for nickname matching.