从Shell脚本中的Openstack创建命令的输出获取`id`字段

Question

OpenStack commands (eg cinder create , nova boot , glance image-create ) output a table of details about the new volume/instance/image.

$ cinder create --image 3896b01c-6afb-41a4-a207-3db87527be2c --display-name centos7-cloud 30
+---------------------+--------------------------------------+
|       Property      |                Value                 |
+---------------------+--------------------------------------+
|     attachments     |                  []                  |
|  availability_zone  |                 nova                 |
|       bootable      |                false                 |
|      created_at     |      2017-01-12T10:58:00.782361      |
| display_description |                 None                 |
|     display_name    |            centos7-cloud             |
|      encrypted      |                False                 |
|          id         | d1c6369b-73ce-498e-a2ef-2c6cea1d0f90 |
|       image_id      | 3896b01c-6afb-41a4-a207-3db87527be2c |
|       metadata      |                  {}                  |
|         size        |                  30                  |
|     snapshot_id     |                 None                 |
|     source_volid    |                 None                 |
|        status       |               creating               |
|     volume_type     |                 None                 |
+---------------------+--------------------------------------+

I am creating a shell script where the next command refers to the new volume by id (in variable CINDER_ID ).

nova boot --block-device source=volume,id=${CINDER_ID},dest=volume centos

What is the best way to extract the id to a variable?

Answer 1

Following solution using GNU grep which supports look around.

grep -owP 'id.*\|\s\K.*[^\|]+' inputfile
d1c6369b-73ce-498e-a2ef-2c6cea1d0f90

To storing it to a variable:

id=$(grep -owP 'id.*\|\s\K.*[^\|]+' inputfile)

You can try:

 id=$(cinder create --image 3896b01c-6afb-41a4-a207-3db87527be2c --display-name centos7-cloud 30 |grep -owP 'id.*\|\s\K.*[^\|]+' )

Answer 2

Using GNU awk's , gensub-function , you can do something like below. The function in the below logic, removes the leading and trailing whitespaces from columns 2,3 field-separated by |

(checked with writing the table to a file)

awk -F"|" 'gensub(/^[ \t]+|[ \t]+$/,"","g",$2)=="id"{print gensub(/^[ \t]+|[ \t]+$/,"","g",$3);}' file
d1c6369b-73ce-498e-a2ef-2c6cea1d0f90

To store it in a variable,

idValue="$(awk -F"|" 'gensub(/^[ \t]+|[ \t]+$/,"","g",$2)=="id"{print gensub(/^[ \t]+|[ \t]+$/,"","g",$3);}' file)"
printf "%s\n" "$idValue"
d1c6369b-73ce-498e-a2ef-2c6cea1d0f90

You can pipe your command producing the above table to this Awk as

<your-command> | awk '{..}'

Something like,

cinder create --image 3896b01c-6afb-41a4-a207-3db87527be2c --display-name centos7-cloud 30 \
       | awk -F"|" 'gensub(/^[ \t]+|[ \t]+$/,"","g",$2)=="id"{print gensub(/^[ \t]+|[ \t]+$/,"","g",$3);}'
d1c6369b-73ce-498e-a2ef-2c6cea1d0f90

Answer 3

You can use \\s*\\|\\s* as field separator

CINDER_ID=$(cinder create --image 3896b01c-6afb-41a4-a207-3db87527be2c --display-name centos7-cloud 30 |
awk 'BEGIN{FS="\\s*\\|\\s*"}$2=="id"{print $3}')
echo $CINDER_ID

you get

d1c6369b-73ce-498e-a2ef-2c6cea1d0f90

Answer 4

An alternative that I came up with is the following sed

sed -n 's/|\s*id\s*|\s*\(.*\S\)\s*|/\1/p' inputfile
d1c6369b-73ce-498e-a2ef-2c6cea1d0f90

Storing in a variable,

CINDER_ID=$(cinder create --image 3896b01c-6afb-41a4-a207-3db87527be2c \
--display-name centos7-cloud 30 | sed -n 's/|\s*id\s*|\s*\(.*[^\s]\)\s*|/\1/p' )

echo $CINDER_ID
d1c6369b-73ce-498e-a2ef-2c6cea1d0f90

Answer 5

Given that the column values are surrounded by whitespace (in addition to the | column separators), awk 's default field-splitting behavior will do (the | instances then simply count as fields in their own right):

cinder_ID=$(cinder create ... | awk '$2=="id" {print $4}')

^{You could append ; exit ; exit inside the block to stop processing right away; probably not worth it for such a small input set, however.}

Note that I've changed your variable name from CINDER_ID to cinder_ID , because it is better not to use all-uppercase shell-variable names in order to avoid conflicts with environment variables and special shell variables .

Answer 6

I think you do not need any Fancy regular expression matchings here. After you create an image with some name , you can this issue this command to get the image id.

# openstack image list -f value | grep '<Image name>' | awk '{print $1}'

You can do the same for volume.

# openstack volume list -f value | grep '<volumename>' | awk '{print $1}'