Odd behavior of bash array

Ingen

1/27/24, 3:52 PM

I've recently encountered a strange script behavior and still don't understand y the array behaves this way.

Here is a piece of the script:

for nCDATAReg in $(seq 1 4)
    do
        sTestBuffer1=$(grep -Pzoi '\/\/<\!\[CDATA\[[\s\S]*?\/\/\]\]>' $SOMEFILE | sed  -rz 's/\/\/<\!\[CDATA\[\n(.*)\/\/\]\]>/\1/g' | tr -d '\0' | sed -rz 's/InsertData\(([[:digit:]]+), [[:digit:]]+, \{(.*?)\}\)\;/\1/gm' | sed "${nCDATAReg}q;d")
        printf "\t\t\t\t\tDEBUG: Buffer is: $sTestBuffer1\t"
        aCDATA[0,$nCDATAReg]=$sTestBuffer1
        sTestBuffer2=$(grep -Pzoi '\/\/<\!\[CDATA\[[\s\S]*?\/\/\]\]>' $SOMEFILE | sed  -rz 's/\/\/<\!\[CDATA\[\n(.*)\/\/\]\]>/\1/g' | tr -d '\0' | sed -rz 's/InsertData\(([[:digit:]]+), [[:digit:]]+, \{(.*?)\}\)\;/\2/gm' | sed "${nCDATAReg}q;d")
        aCDATA[1,$nCDATAReg]=$sTestBuffer2
        printf "and second Buffer is: $sTestBufferS\t\n"
        printf "But array now contains: (${aCDATA[0,$nCDATAReg]}) and (${aCDATA[1,$nCDATAReg]})\n"
    done

and file content is:

MORE TEXT ABOVE
</script>
    <script>//<![CDATA[
InsertData(1, 1, {VALUE1});
InsertData(3, 1, {value2});
InsertData(6, 1, {vALUE3});
InsertData(91, 1, {Value4});
//]]></script>    <div class="news-post news-post-style-full"
MORE TEXT BELOW

In theory and in console, its produce a list of the first digit and the content of the string in curly braces. You can see this in DEBUG printf's, the entire regex oneliner is first placed in separate variables and only then in an array cell. But if everything is correct in the variable, like 3 {value2}, then for some reason {value2} is assigned to the array in both cases.

Here is output

DEBUG: Buffer is: 1     and second Buffer is: VALUE1            But array now contains: (VALUE1) and (VALUE1)
DEBUG: Buffer is: 3     and second Buffer is: value2            But array now contains: (value2) and (value2)
DEBUG: Buffer is: 6     and second Buffer is: vALUE3            But array now contains: (vALUE3) and (vALUE3)
DEBUG: Buffer is: 91    and second Buffer is: Value4            But array now contains: (Value4) and (Value4)

What am I missing here? (besides the obvious poor knowledge of bash)

1 + 1

bash

text

regex

html

parsing

glenn jackman

1/27/24, 5:16 PM

Did you declare the array with `declare -A aCDATA` to let bash know it's an _associative_ array? (answer with more explanation forthcoming)

Score:1

Server

glenn jackman

1/29/24, 12:38 AM

Took me a while to get back.

If you don't declare it as an associative array then it's a numerically-indexed array for which the index part is treated as an arithmetic expression. In an arithmetic expression, commas separate sub-expressions and the return value is the result of the last subexpression.

a[0,1]=x is the same as a[1]=x

a[0,1]=foo
a[1,1]=bar
declare -p a      # => declare -a a=([1]="bar")

For an associative array, the index is a string. What you're missing is declare -A arrayname

declare -A aCDATA
aCDATA[0,1]=foo
aCDATA[1,1]=bar
declare -p aCDATA     # => declare -A aCDATA=([0,1]="foo" [1,1]="bar" )

+ 0

Elon Musk

I sit in a Tesla and translated this thread with Ai:

EN: Odd behavior of bash array

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.