rudimeier / bash_ini_parser Goto Github PK

In a script with "set -o errexit" the read_ini will fail. As a work-around, turn off errexit before using read_ini:

set +o errexit
read_ini ~/.script.ini
set -o errexit 

If script is run using "set -e" it crashes on lines 177 ((LINE_NUM++)) and 193 ((SECTIONS_NUM++))
This is because post increment returns value (defaulted at 0) before addition and set -e reads this as false, returning error.

Related stackoverflow: here

Fix is easy enough as lines can be changed to use pre-increment ((++LINE_NUM)) and ((++SECTIONS_NUM))

I want use bash_ini_parser on Embedded Distribution, so What's License of these?

The substitution for the escape of ' not work.

I propose this correction :

diff --git a/read_ini.sh b/read_ini.sh
index 06b1bc8..2ae2a85 100755
--- a/read_ini.sh
+++ b/read_ini.sh
@@ -270,7 +270,7 @@ function read_ini()
                # enclose the value in single quotes and escape any
                # single quotes and backslashes that may be in the value
                VAL="${VAL//\\/\\\\}"
-               VAL="\$'${VAL/\'/\\\'}'"
+               VAL="\$'${VAL//\'/\'}'"
 
                eval "$VARNAME=$VAL"
        done  <"${INI_FILE}"

First off: this is excellent and has saved me a lot of time. Thank you for making it and making it public.

I came across this issue when making the linux equivalent of an existing windows installer.

Works fine after converting the CRLF to LF. This is on Ubuntu 16.04.

My guess is that read only looks for LF and the trailing CR is causing it to fail the section regex test.

Example INI:

[smtp]
host=example.com
port=25
authentication=true
use_tls=true
from_email[email protected]
username=ex
password=expass

Save it with CRLF line endings and read_ini setup.ini -p setup -b 0 will fail with

Error: Invalid line:
1: [smtp]

Bug: LINE_NUM incrementation throws an "exception" when run in bash that ships with Ubuntu 12.04.2. Error was caught when using bash 'trap' mechanism to do exception handling.

Proposed solution: change ((LINE_NUM++)) to LINE_NUM=$(($LINE_NUM + 1))

When I put one variable into a ini file at the end of file, it doesn't read correctly.

Test#5 fails on var3 and 4 with the same issue
expected was ... \ ......
generated only had one \ versus the two in the correct file.

bash -version
4.1.17(1) - release
Current patch level for Solaris 11.3 - patched up as of this date. 2 May 2016.

Sorry, no cut and past using my current VM setup running Solaris (Intel) under VBOX.
if additional info is needed I should be able to reconfigure the VM to allow data / network outside of the box.

Add INI__ALL_SECTIONS can be used in such satution:
Say I want to backup a list of mysql databases, and each of them have a different username and password, and the list of databases can be grow in the future, so I configure all databases in a ini file like this:

[db01]
name=db-host01
username=user1
password=pwd01

[db02]
name=db-host2
username=user2
password=pwd2

And in the backup script I can use some thing like

for db in `echo INI__ALL_SECTIONS`; do 
  name="INI__${db}__name"
  echo "username of $db is ${!name}"
done

So everytime I need add a new database for backup, I only need to add a section in the ini file without change the backup sctipt.

subject changed by Rudi to match the feature request of post #2