Convert perl script to bash

Question

The following script it's about monitoring some logs, the $timestamp is at log at every line that something happens.

Example:
03:19:13.4 Begin summary update for ads.doc.inventory.InventoryItemSummary
03:19:33.9 CronServer:: DailyJob ads.tools.UpdateSummaries@17c5d6cf failed with exception ads.util.AppError: Cannot create UnitName from keys: Each
ads.util.AppError: Cannot create UnitName from keys: Each
at ads.db.DBObjectDefault.createFromKeys(DBObjectDefault.java:42)
at ads.db.DBTable.createFromKeys(DBTable.java:227)
at ads.db.DBValue.getValue(DBValue.java:621)
at ads.dbmanager.DBObjectsManager.initObjects(DBObjectsManager.java:400)
at ads.dbmanager.DBObjectsManager.reload(DBObjectsManager.java:447)
at ads.dbmanager.DBObjectsManager.loadFromStore(DBObjectsManager.java:497)
at ads.doc.inventory.InventoryItemSummary.refreshSince(InventoryItemSummary.java:173)
at ads.db.DBSummaryTable.refreshAll(DBSummaryTable.java:67)
at ads.tools.CronServer$DailyThread.run(CronServer.java:271)
[SOAPException: faultCode=SOAP-ENV:Client; msg=Error opening socket: java.net.ConnectException: Connection refused; targetException=java.lang.IllegalArgumentException: Error opening socket: java.net.ConnectException: Connection refused]
at org.apache.soap.transport.http.SOAPHTTPConnection.send(SOAPHTTPConnection.java:354)
at org.apache.soap.rpc.Call.invoke(Call.java:248)
at ads.support.SupportCall.call(SupportCall.java:56)
at ads.tools.SupportThread.run(SupportThread.java:101)
03:46:42.5 Periodic support request failed: ads.support.SupportException: Error opening socket: java.net.ConnectException: Connection refused
06:31:36.1 Upload failed: java.io.FileNotFoundException: c:/tmp/cygwin1.dll (No such file or directory)
08:01:08.0 connect from /172.22.3.28

I have this perl script

#!/usr/bin/perl
use strict;
use warnings FATAL => qw(all);
my $timestamp = qr/^\d+:\d+:\d+/;
my $block = "";
my $exception = 0;
while (<STDIN>) {
if ($block) {
# $_ is the current input line in perl.
 if ($_ =~ $timestamp) {
 print "\n\n***EXCEPTION***\n$block" if $exception;
 $block = "";
 $exception = 0;
 }
# This isn't actually a second regexp, it's a substring search.
# But you could use a regexp.
 $exception = 1 if (index($_,"Exception") >= 0);
 $block .= $_;
 next;
}
$block = $_ if ($_ =~ $timestamp);
}

and I want to convert it to bash script.

Here's what I've done so far but didn't work, #!/bin/bash

timestamp=^[0-9][0-9]\:[0-9][0-9]\:[0-9][0-9]\.[0-9]
block=""
exception=false
except=.*Exception.*
while read line
do
if [[ -z $block ]] ;
 then
 if [[ $line =~ $timestamp ]]
 then
 if [[ $exception=false ]]
 then
 echo "\n\n***EXCEPTION***\n$block"
 fi
block=""
exception=false
 fi
 if [[ $line =~ $except ]]
 then
 exception=true
 fi
block=$block$line
fi
if [[ $line =~ $timestamp ]]
 then
 block=$line
fi
done

Any help with that?

• suspectus: what exactly is the problem you are having?
• Vince: Bash doesn't handle regular expressions natively. What you propose is a lot more complex and involves using external commands. I wouldn't go so far as to say it's impossible, but it's almost certainly a bad idea.
• C Ts: The output is empty with the bash script!
• Mat: mywiki.wooledge.org/BashGuide/Patterns @Ghodmode.
• C Ts: Yes, exactly this. As input is a log file and I want to print all the block (timestamp to timestamp) if inside the block there is the word exception.

Answer 1

1. you have to quote filename pattern metacharacters like * and [

   timestamp='^[0-9][0-9]:[0-9][0-9]:[0-9][0-9]\.[0-9]'
   except='.*Exception.*'
   

2. the test command (otherwise known as [[) operates on the number of arguments it receives. It is therefore very sensitive to whitespace:

    if [[ $exception = false ]]
   

3. there's a major logic error: you unset the block variable inside the [[ -z $block ]] block: once block is non-zero it can never be unset.

4. Why are you rewriting a working script?

Answer 2

Edit: the following solution is closer to the original script than what I initially wrote (it checks the whole block body, not just the first line, for the word "Exception"):

block=()
newblock() {
 [[ ${block[*]} = *Exception* ]] && printf '%s\n' '' '' '***EXCEPTION***' "${block[@]}"
 block=("$line")
}
while IFS= read -r line; do
 case $line in
 ([0-9][0-9]:[0-9][0-9]:[0-9][0-9].[0-9]\ *)
 newblock
 ;;
 (*)
 block+=("$line")
 ;;
 esac
done
newblock

• C Ts: You're the best, works fine, but I don't understand the block[], (*). Can't understand how it checks the whole block body and how block is defined.
• mirabilos: @CTs [[ ${block[*]} = *Exception* ]] checks the whole block body. block is an array, every array item is a line.
• C Ts: Inside the while loop you insert every line at the array (block+=("$line"))? What is (*) at the while loop? At the newblock method, block=("$line") what exactly does? Sorry for all the questions, but I'm new at bash and first time come across with array!
• mirabilos: (*) belongs to case and is, basically, the equivalent of the default statement in C switch. And x+=(y) appends to the array x, while x=(y) sets the array x empty before appending y.