Meow is a log analysis and processing tool written in Rust that allows (or will allow) operations such as filtering (like grep), highlighting and text replacing. Think of it as if you could make cat speak: it will read from stdin, apply the commands supplied, and emit an output.

The project started as a way to scratch my itch for log processing and to learn Rust at the same time.

meow command1 [command2] ... [commandN]

These are the available commands and their syntax. All the patterns are case insensitive regexes:

• Filtering: Filters the line and only prints it if it contains text matching the specified regular expression. Every filter command will be highlighted in a different color.
  • Syntax: fc:regex. Also just the regex, without the fc: header
  • Examples: fc:sourcebuffer, sourcebuffer. 'fc:[0-9][0-9]*'.
• Filtering without highlighting: Same as above, but without highlighting the matched string.
  • Syntax: fn:regex
  • Examples: fn:memdump, 'fn:[.]cpp'
• Negative filter: Selects only the lines that don't match the regex filter.
  • Syntax: n:regex
  • Example: n:audio
• Substitution: Replaces one pattern for another.
  • Syntax: 's:#regex#replacement_text'
  • Examples: s:#pattern#replacement, 's:/(?<adjective>big|small)/${adjective}ish' (Any delimiter character is supported. See the syntax for capture groups here)
• Time filter: Assuming the lines start with a timestamp (eg: 0:01:10.881123150), selects only the lines between the target start and end timestamps. Specifying multiple time filters will generate matches that fit on any of the time ranges. Overlapping ranges should work, but better don't use them.
  • Syntax: ft:[begin_timestamp]-[end_timestamp]
  • Examples: ft:0:00:24.787450146-0:00:24.790741865, ft:0:00:24.787450146-, ft:-0:00:24.790741865, ft:-
• Highlight threads: Highlights each thread in a GStreamer log in a different color.
  • Syntax: ht: (without parameters)

Note that all these commands are executed in order, so you can easily refine the behaviour by carefully choosing commands in the right order. For instance sourcebuffer h:true h:false 'h:[0-9]:[0-9:.]*[0-9]' n:enqueue will select only the lines containing the "sourcebuffer" word (highlighting the word). Then on those selected lines, highlight the words "true" and "false", as well as any timestamp that may appear in the line (in 3 different colors). If any of the lines contains the "enqueue" word, they will be discarded and not shown.

Some features that might be interesting for the future are:

• Parenthesis support for the expressions, or at least some stopper ("!", ";"...) for "sibling expressions" (expressions having the same priority), so that any new expression after the stopper isn't considered as "sibling", but as something that happens after the previous sibling expressions have been applied (eg: instead of fc:a OR fc:b OR fc:c, being able to have (fc:a OR fc:b) AND fc:c).
• Loading a config file where preexisting expressions can be defined (with an alias) and then the alias used in the command line, sort of as a library of expressions (eg: for transforming WebKit {458333/1000000 = 0.458333} timestamps into something more manageable such as 0.458333).
• More flexible syntax for thread highlighting and time filtering, not always tied to the third and first line tokens, respectively.