Powershell -Select String (SLS) 将一行拆分为多行

Question

I've been looking for a way to perform basic grep with Powershell.

I have an input text file but want to extract only some of them For instance,

2020-02-04 05:42:12,132 INFO UselessLog   8=FIX4.4|35=D|49=sender1|56=recipient|....
2020-02-04 05:42:12,134 INFO UselessLog2  8=FIX5.0|35=AB|49=sender2|56=recipient2|....
2020-02-04 05:42:12,136 INFO UselessLog2  8=FIX5.0|35=AB|49=sender3|56=recipient2|....

I'm trying to parse the file with

sls "8=FIX" fixlogs -ca | select -exp line |sls "=sender1" > parseFix

However, when I open the parseFix file, sls just transformed one line into multiple ones; eg, the following two lines represent a single match:

2020-02-04 05:42:12,132 INFO UselessLog   
8=FIX4.4|35=D|49=sender1|56=recipient|...

I'm fairly certain there's a config in Powershell to advise not to split but I've investigated and failed to find the one.

I have no access to Unix/Linux server to do a grep magic also I have checked within StackOverFlow but didn't find anything matching my Question.

Answer 1

You need another select -exp line call after your 2nd sls ( Select-String ) call to make sure that the matching lines are written as-is to the output file:

sls "8=FIX" fixlogs -ca | select -exp line |sls "=sender1" | select -exp line > parseFix

Note: In PowerShell [Core] 7.0+ you can use Select-String -Raw to directly output the matching lines, without the wrapper object that provide metadata about the match that you get by default. That is, there is then no need to extract the lines via separate select -exp Line ( Select-Object -ExpandProperty Line ) calls.

---

Why your output lines were broken apart :

Unlike grep or findstr , Select-String outputs match-information objects (of type Microsoft.PowerShell.Commands.MatchInfo ) that wrap the matching lines, not just the matching lines themselves.

These objects, because they aren't single values such as a string or a number, loosely speaking, are subject to PowerShell's default output formatting when you send them to a file with the > operator, which is effectively an alias for Out-File , and this formatting (the same representation you'd see in the console) involves automatic word wrapping, ie breaking long lines into multiple ones to fit the display width (even though a file has no inherent width; Out-File -Width would allow you to control this width).

Answer 2

There's a lot going on here. Select-string is outputting a object with a custom format view. ">" is the same as "| out-file", which adds extra formatting, like wrapping the line if it thinks it's too long. You didn't show the whole lines of the log but they look pretty long. By the way, out-file defaults to utf16 encoding (PS 5). Set-content seems to work ok and convert the object to a string:

sls 8=FIX fixlogs -ca | select -exp line | sls =sender1 | set-content parsefix

Also, you can make that one regex pattern if the whole thing can be case sensitive or case insensitive.

sls 8=FIX.*=sender1 fixlogs 

fixlogs:1:2020-02-04 05:42:12,132 INFO UselessLog   8=FIX4.4|35=D|49=sender1|56=recipient|....

Or just skip the select:

sls 8=FIX fixlogs -ca | sls =sender1