Replacing columns-separated fields having special characters

UNIX for Beginners Q & A
1

Dear colleagues,

**I have tens of files with fields separated by column ':' like below

CCM_JNI_CLASS_PATH=:${CCB_APP_BIN}/CCBG_epsm_opm_wfl_001.jar:${CCB_APP_BIN}/CCBF_fif-api.jar:${CCB_APP_BIN}/CCB_geronimo-jaxws_2.1_spec-1.0.jar:${CCB_APP_BIN}/CCB_geronimo-ws-metadata_2.0_spec-1.1.2.jar:${CCB_APP_BIN}/CCB_jaxb-api-2.1.jar::${CCB_APP_BIN}/CCB_jaxb-impl-2.1.4.jar:${CCB_APP_BIN}/CCB_jaxb-xjc.jar:${CCB_APP_BIN}/CCB_jms-1.1.jar:${CCB_APP_BIN}/CCB_jsr173-1.0_api.jar:${CCB_APP_BIN}/CCB_jsr250-api-1.0.jar:${CCB_APP_BIN}/CCB_jta.jar:${CCB_APP_BIN}/CCB_log4j-1.2.15.jar:${CCB_APP_BIN}/CCB_ojdbc14-1.4.jar:${CCB_APP_BIN}/CCB_saaj-api-1.3.jar

**each file has one occurrence of the path ${CCB_APP_BIN}/CCB_log4j-1.2.15.jar that needs to be replaced by ${CCB_APP_BIN}/CCB_log4j-1.2-api-2.12.1.jar:${CCB_APP_BIN}/CCB_log4j-api-2.12.1.jar:${CCB_APP_BIN}/CCB_log4j-core-2.12.1.jar

**i had difficulties using sed as i could not prevent the special characters interpretation like '$'

** i am on Centos machine:

ccb20b@redmoon CreateEnvironment/CONF/CCB_126.01 $ lsb_release -a
LSB Version:	:base-4.0-amd64:base-4.0-noarch:core-4.0-amd64:core-4.0-noarch
Distributor ID:	CentOS
Description:	CentOS release 6.10 (Final)
Release:	6.10
Codename:	Final

** Appreciate supporting me for required substitution by any possible way you suggest

Many thanks in advance

BR,
Eman

2

Forum guidelines are that you must post your own code, error messages, sample input and out to ask for assistance.

3

You are right - sed only takes RegularExpressions so is not ideal if you want to match plain strings that can contain RE-specific characters.
A better tool for this is awk. Here is a principle solution:

echo 'bla:fromstring:blabla:fromstring' | awk 'BEGIN { RS=":" } $1==from { $1=to } 1' from='fromstring' to='tostring' | paste -sd ":"
bla:tostring:blabla:tostring

The RS (RecordSeparator) is changed from newline to ":".
Because awk has a problem with the last record when ORS (OutputRecordSeparator) is changed to ":" I pipe the newline-separated output to paste that does the output formatting.