Incorrect multiline messages initial detection

[piomar123]

Supposedly an initial pattern detection in logs with multiline entries doesn't work properly. Currently if the first few messages include stack traces, the patterns are disabled because the rule matching more than 5/25 lines fails. Example multiline message with a stack trace:

2018-12-28 13:50:42.137  [MAIN] ERROR com.example.myproject.Author: An error occurred
Exception in thread "main" java.lang.IllegalStateException: A book has a null property
        at com.example.myproject.Author.getBookIds(Author.java:38)
        at com.example.myproject.Bootstrap.main(Bootstrap.java:14)
Caused by: java.lang.NullPointerException
        at com.example.myproject.Book.getId(Book.java:22)
        at com.example.myproject.Author.getBookIds(Author.java:36)

Removing stack traces from the first few messages allows to detect patterns correctly for the whole file. The initial detection should ignore lines not matching "Message start pattern".

There is also a similar problem when there is fewer than 5 lines in the log.

Settings: Message pattern: ^(\d{4}-\d\d-\d\d\s\d\d:\d\d:\d\d\.\d{3})\s+[^\[]*\[\s*([^\]]*)\]\s+(\w+)\s+(.*)$ Message start pattern: ^\d Time: yyyy-MM-dd HH:mm:ss.SSS

[HarukaMa]

I have a very similar situation, and decided to play around with it for a while.

After many tries, It seems that the maximum multi-line part (the lines that are not the first line in a multi-line log) in the first 5 log lines must be less than 20 lines. Otherwise the log type detection will fail.

Here is an example of default Monolog format: Pattern: ^\[(.*?)] (\w*)\.(\w*): (.+)$ Start pattern: ^\[\d Time format: yyyy-MM-dd HH:mm:ss

[2020-02-11 23:38:00] local.INFO: Message 1
1
2
3
4
5
6
7
8
9
10
[2020-02-12 02:48:37] local.ERROR: Message 2
11
12
13
[2020-02-12 02:48:37] local.ERROR: Message 3
[2020-02-12 02:48:37] local.ERROR: Message 4
14
15
16
[2020-02-12 02:48:37] local.ERROR: Message 5
17
18
19
[2020-02-11 23:38:00] local.INFO: Message 6
[2020-02-11 23:38:00] local.INFO: Message 7
...

Any more multi-line part in the first 5 messages would cause the detection to fail, but adding new lines to all messages below would be fine.

This is actually quite annoying as there could be logs that begin with some long stack traces in the first few messages, and they would cause the whole file to become not recognized.

EDIT: I then realized that there must be more than 5 messages in the first 25 lines to make the plugin correctly recognize the format. Maybe making the limit more relaxed would be a good choice?

[gren236]

Got the same issue with Monolog. +1 for @HarukaMa 's proposal.

[wayheming]

same problem

[ArtemKar123]

Fixed in #165. Will be released in the next version. Thanks for your feedback!