来自xml解析的awk语句中的substr

Question

Link to the original question: bash script extract XML data into column format and now for a modification and explanation ->

Something within this line of code is not correct and I believe it is with the substr portion and that would be because I don't have a full understanding and would like to learn HOW better to understand it. Yes I have looked at documentation and its not fully clicking. A couple examples as well as an answer would really be helpful.

awk -F'[<>]' 'BEGIN{a["STKPR"]="Prod";a["STKSVBLKU"]="Prod";a["STKSVBLOCK"]="Prod";a["STKSVBLK2"]="Test";} /Name/{name=$3; type=a[substr(name,length(name))]; if (length(type)==0) type="Test";} /SessionHost/+/Host/{print type, name, $3;}'|sort -u

This bit here:

type=a[substr(name,length(name))]; if (length(type)==0) type="Test";

Here is the xml format which each bit is a block for each host that contains the hostname and IP.

<?xml version="1.0"?>
<Connection>
  <ConnectionType>Putty</ConnectionType>
  <CreatedBy>Someone</CreatedBy>
  <CreationDateTime>2014-10-27T11:53:32.0157492-04:00</CreationDateTime>
  <CredentialConnectionID>9F3C3BCF-068A-4927-B996-CA52154CAE3B</CredentialConnectionID>
  <Description>Red Hat Enterprise Linux 5 (64-bit)</Description>
  <Events>
    <OpenCommentPrompt>true</OpenCommentPrompt>
    <WarnIfAlreadyOpened>true</WarnIfAlreadyOpened>
  </Events>
  <Group>PATH/TO/GROUP/NAME</Group>
  <ID>f2007f03-3b33-47d3-8335-ffd84ccc0e6b</ID>
  <MetaInformation />
  <Name>STKSPRDAPP01111</Name>
  <OpenEmbedded>true</OpenEmbedded>
  <PinEmbeddedMode>False</PinEmbeddedMode>
  <Putty>
    <AlwaysAskForPassword>true</AlwaysAskForPassword>
    <Domain>DOMAIN</Domain>
    <FontSize>12</FontSize>
    <Host>10.0.0.111</Host>
    <Port>22</Port>
    <PortFowardingArray />
    <TelnetEncoding>IBM437</TelnetEncoding>
  </Putty>
  <Stamp>85407098-127d-4d3c-b7fa-8f174cb1e3bd</Stamp>
  <SubMode>2</SubMode>
  <TemplateName>SSH-PerUserCreds</TemplateName>
</Connection>

What I want to do is similar to the referenced link above. But here I want to match -->

BEGIN{a["STKPR"]="Prod";a["STKSVBLKU"]="Prod";a["STKSVBLOCK"]="Prod";a["STKSVBLK2"]="Test";

and all of the rest as Test. Best to read the previous post to help make this one more understandable. Thank you.

Answer 1

Because your keys here are of different length, the substr approach is less than optimal. Try:

awk -F'[<>]' '/Name/{n=$3;t="Test"; if(n ~ /^STKPR/) t="Prod"; if (n ~/^STKSVBLKU/) t="Prod"; if (n ~/^STKSVBLOCK/) t="Prod"} /SessionHost/+/Host/{print t, n, $3;}' sample.xml |sort -u
Test STKSPRDAPP01111 10.0.0.111

How It Works

In this case, the type, denoted by t , is set according to a series of if statements. From the above code, they are:

t="Test"
if (n ~ /^STKPR/) t="Prod"
if (n ~ /^STKSVBLKU/) t="Prod" 
if (n ~ /^STKSVBLOCK/) t="Prod"

By setting t="Test" , Test becomes the default: the type will be Test unless another statement matches. If of the following statements looks at the string that begins the host name and, if there is a match, sets type t to a new value. (When a regular expression begins with ^ , that means that what follows must match at the beginning of the string.)

Alternative using fancier regular expressions

Since the above three if statements are all for the Prod type, the three of them could, if you preferred, be rearranged to:

t="Test"
if (n ~ /^STK(PR|SVBLKU|SVBLOCK)/) t="Prod"

(metalcated: Fixed unmatched parentheses bracket)

Answer 2

The substr portion produces a string containing the last character of the string. This is because it is taking a substring of string name starting at the position length(name) going to the end of the string, and because substr is indexed starting at 1.

To match whole strings you can use your variable name rather than processing it with substr .

/Name/ { name=$3; type=a[name]; if (length(type)==0) type="Test"; }