I'm trying to remove the embedded CSS from a html file on Linux Server (Red Hat 6.8). e.g. File 1.htm is as below:
abc
<style type="text/css">
whatever
1
2
3
</style>
def
And what I need is
abc
def
I tried the sed command below
sed -i 's#<style type="text\/css">(.|\n)*<\/style>##g' 1.htm
but it's not working. Could someone shed some light on this? Thanks~
In order to match across lines, you must instruct sed to read the whole file at once.
With GNU sed (Linux) v4.2.2+, the simplest way to do that is to use -z (whose purpose is to read NUL-separated records; in the absence of embedded NULs, the entire file is read).
Also, given your unescaped use of ( and ) as metacharacters, you must activate support for extended regular expressions via the -r option, although you don't strictly need that, because (.|\n*) (which is equivalent to .*) must be replaced with [^<]* in order to potentially match multiple <style> elements individually (.*, because sed regexes are greedy, would match everything up until the last </style> tag in the file, which would malfunction with multiple elements).
sed -z -r -i 's#<style type="text/css">[^<]*</style>\n?##g' 1.htm
Note that I've appended \n? to the regex to ensure that no empty line is left behind by the replacement.
Use of unescaped ? also requires -r.
Since you've chosen # as the s delimiter, you needn't \-escape / chars. in the regex.
With older GNU sed versions, you can use a loop (:a;$!{N;ba}) to read the entire file at once:
sed -r -i ':a;$!{N;ba}; s#<style type="text/css">[^<]*</style>\n?##g' 1.htm
Generally, for a more robust solution, use an HTML/XML-aware tool such as xsltproc (see below).
xsltproc:xsltproc is a third-party utility that comes with macOS and some Linux distributions (e.g., Fedora), and can easily be installed on others (e.g., on Ubuntu, with sudo apt-get install xsltproc).
With the --html option, it is capable of applying XSLT-based transformations to HTML documents too, not just to XML documents.
Here's a sample bash-based solution that demonstrates creating a copy of an HTML document with all <style> elements removed, gratefully adapted from this answer:
# Create a simple sample HTML document with 2 <style> elements at different
# levels of the DOM and save it as "file.html"
cat > file.html <<'EOF'
<html>
<head></head>
<body>
<style type="text/css">
* {
border: 1 solid black;
}
</style>
<p foo='bar'>
abc def
<style type="text/css">
* {
border: 2 dashed blue;
}
</style>
</p>
</body>
</html>
EOF
xsltproc can then apply an XSLT template to the HTML file (normally, such a template is provided as a file as well, but given its brevity, I'm constructing it in memory and providing it like a file via a bash process substitution (<(...))):
# Define the XSLT template that copies all nodes in the document except those
# named "style".
# For an explanation, see https://stackoverflow.com/a/322079/45375
template='<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
<xsl:template match="node()|@*">
<xsl:copy>
<xsl:apply-templates select="node()|@*"/>
</xsl:copy>
</xsl:template>
<xsl:template match="style"/>
</xsl:stylesheet>'
# Invoke xsltproc with the template and the input file.
# --html tells xlstproc to process the file as HTML, both on input and on output.
xsltproc --html <(echo "$template") file.html
The above yields (note how both <style> elements were removed):
<html>
<head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"></head>
<body>
<p foo="bar">
abc def
</p>
</body>
</html>
To replace the input file with the modified copy (to emulate sed -i), use something like:
xsltproc --html <(echo "$template") file.html > /tmp/file.$$ && mv /tmp/file.$$ file.html
sed usually process line by line(1 line at a time). In this case,
We have to read all lines at once. If you are not aware reading all lines at once then you can refer the link.click here.
sed -e ':a' -e 'N' -e '$!ba' -e 's/<style.*<\/style>//' file | awk 'NF'
It will return:
abc
def
Suppose If You have a file like given below:
abc
<style type="text/css">
whatever
1
2
3
</style>
xyz
<style type="text/css">
whatever
1
2
3
</style>
def
<style type="text/css">
whatever
1
2
3
</style>
mno
And you want to print all text outside to tags then
sed '/<style.*>/,/<\/style>/d' file
It will return:
abc
xyz
def
mno
