I would like to convert a (ISO 8601) date from YYYY-MM-DD to something that looks like Mon, 03 Dec 2021 00:00:00 -0600 (RFC-822 / RFC 5322). I have not been able to find questions that address this specific issue.
I have an RSS XML file that looks like this:
bash-5.1$ cat feed.xml
<rss version="2.0">
<channel>
<title>Title string</title>
<link>https://domain/feed.xml</link>
<description>Description string here</description>
<language>en-us</language>
<item>
<title>title string here</title>
<link>link string with https style information</link>
<guid>link string with https style information</guid>
<pubDate>2021-12-03</pubDate>
</item>
<item>
<title>title string here</title>
<link>link string with https style information</link>
<guid>link string with https style information</guid>
<pubDate>2019-08-13</pubDate>
</item>
<item>
<title>title string here</title>
<link>link string with https style information</link>
<guid>link string with https style information</guid>
<pubDate>2018-11-23</pubDate>
</item>
</channel>
</rss>
The output I expect after parsing would be something like this:
...
<pubDate>Fri, 03 Dec 2021 00:00:00 -0600</pubDate>
...
<pubDate>Tue, 13 Aug 2019 00:00:00 -0500</pubDate>
...
<pubDate>Fri, 23 Nov 2018 00:00:00 -0600</pubDate>
...
This output can usually be achieved with the bash date command like so:
bash-5.1$ date -d "2018-11-23" "%a, %d %b %Y %T %z"
Fri, 23 Nov 2018 00:00:00 -0600
I am trying to accomplish this with awk because I learned that command substitution is possible, and I believe I am close, but not quite:
bash-5.1$ awk -F "[><]" -v date="$(date "%a, %d %b %Y %T %z" -d "$3")" '/pubDate/ {print $3date}' feed.xml
2021-12-03Sun, 05 Dec 2021 00:00:00 -0600
2019-08-13Sun, 05 Dec 2021 00:00:00 -0600
2018-11-23Sun, 05 Dec 2021 00:00:00 -0600
It seems like the pattern match executes successfully as well as the date command, but the awk $3 field looks like is not being passed into the shell date command and so the current time is showing instead of the transformed time.
How do I pass the $3 field from the awk pattern match into the date command so that it can convert the date based on the field value?
Any help is greatly appreciated!
CodePudding user response：
As commented, it is recommended to use xml parsing tool speaking generally.
If the xml file is well aligned as the provided example, bash or awk
may work under limited conditions.
With bash:
#!/bin/bash
while IFS= read -r line; do
if [[ $line =~ (.*<pubDate>)([0-9]{4}-[0-9]{2}-[0-9]{2})(</pubDate>.*) ]]; then
datestr=$(date -d "${BASH_REMATCH[2]}" "%a, %d %b %Y %T %z")
line="${BASH_REMATCH[1]}$datestr${BASH_REMATCH[3]}"
fi
echo "$line"
done < feed.xml
The condition $line =~ (.*<pubDate>)([0-9]{4}-[0-9]{2}-[0-9]{2})(</pubDate>.*) matches the <pubdate> line assiging bash variable ${BASH_REMATCH[@] to
the parenthesized substrings. Then line is reconstructed using the reformatted date string.
If gawk which supports mktime() and strftime() functions, you can
also say with gawk:
awk '
{
if (match($0, /^(.*<pubDate>)([0-9]{4})-([0-9]{2})-([0-9]{2})(<\/pubDate>.*)/, a) ) {
ts = mktime(a[2] " " a[3] " " a[4] " 00 00 00") # timestamp since the epoch
datestr = strftime("%a, %d %b %Y %T %z", ts)
$0 = a[1] datestr a[5]
}
} 1' feed.xml
CodePudding user response：
Altering your awk code slightly.
$ awk -v date="$(date "%a, %d %b %Y %T %z")" 'BEGIN {FS=OFS=">"}$1~/pubDate/{split($2,a,"<"); a[1]=date; $2=a[1]"<"a[2]}1' input_file
<rss version=2.0>
<channel>
<title>Title string</title>
<link>https://domain/feed.xml</link>
<description>Description string here</description>
<language>en-us</language>
<item>
<title>title string here</title>
<link>link string with https style information</link>
<guid>link string with https style information</guid>
<pubDate>Mon, 06 Dec 2021 00:00:00 0000</pubDate>
</item>
<item>
<title>title string here</title>
<link>link string with https style information</link>
<guid>link string with https style information</guid>
<pubDate>Mon, 06 Dec 2021 00:00:00 0000</pubDate>
</item>
<item>
<title>title string here</title>
<link>link string with https style information</link>
<guid>link string with https style information</guid>
<pubDate>Mon, 06 Dec 2021 00:00:00 0000</pubDate>
</item>
</channel>
</rss>
sed can also be used if the variable is set.
$ date="$(date "%a, %d %b %Y %T %z")"
$ sed "/pubDate/ s/....-..-../$date/" input_file