.TH "HXNORMALIZE" "1" "10 Jul 2011" "7.x" "HTML-XML-utils"
.SH NAME
hxnormalize \- pretty-print an HTML file
.SH SYNOPSIS
.B hxnormalize
.RB "[\| " \-x " \|]"
.RB "[\| " \-X " \|]"
.RB "[\| " \-e " \|]"
.RB "[\| " \-d " \|]"
.RB "[\| " \-s " \|]"
.RB "[\| " \-L " \|]"
.RB "[\| " \-i
.IR indent " \|]"
.RB "[\| " \-l
.IR line\-length " \|]"
.RB "[\| " \-c
.IR commentmagic " \|]"
.RI "[\| " file-or-URL " \|]"
.SH DESCRIPTION
.LP
The
.B hxnormalize
command pretty-prints an HTML or XML file, and also tries to fix small
HTML errors. The output is the same file, but with a maximum line length
and with optional indentation to indicate the nesting level of each
line.
.SH OPTIONS
The following options are supported:
.TP 10
.B \-x
Applies XML conventions: empty elements are written with a slash at
the end (e.g., <IMG\ />) and, if the input is HTML, any \(oq<\(cq and
\(oq&\(cq inside <style> and <script> elements are escaped as
\(oq&lt;\(cq and \(oq&amp;\(cq. (The input is assumed to be HTML
unless the
.B \-X
option is present.) Implies
.BR \-e .
.TP
.B \-e
Always inserts endtags, even if HTML does not require them (for
example: </p> and </li>).
.TP
.B \-X
Makes
.B hxnormalize
assume the input is well-formed XML. It does not try to infer omitted
HTML tags, does not assume elements such as <img> and <br> are empty,
and does not treat \(oq<\(cq and \(oq&\(cq inside <style> and <script>
as normal characters.
.TP
.B \-d
Omit the DOCTYPE from the output.
.TP
.BI \-i " indent"
Set the number of spaces to indent each nesting level. Default is 2.
Not all elements cause an indent. In general, elements that can occur
in a block environment are started on a new line and cause an indent,
but inline elements, such as EM and SPAN do not cause an indent.
.TP
.BI \-l " line\-length"
Sets the maximum length of lines.
.B hxnormalize
will wrap lines so that all lines are as long as possible, but no
longer than this length. Default is 72. Words that are longer than the
line length will not be broken, and will extend past this length. A
\(oqword\(cq is a sequence of characters delimited by white space.) The
content of the STYLE, SCRIPT and PRE elements will not be
line-wrapped.
.TP
.B \-s
Omit <span> tags that don't have any attributes.
.TP
.B \-L
Remove redundant \(oqlang\(cq and \(oqxml:lang\(cq attributes. (I.e.,
those whose value is the same as the language inherited from the
parent element.)
.TP
.BI \-c " commentmagic"
Comments are normally placed right after the preceding text. That is
usually correct for short comments, but some comments are meant to be
on a separate line.
.I commentmagic
is a string and when that string occurs inside a comment,
.B hxnormalize
will output an empty line before that comment. E.g. \fB\-c "===="\fR
can be used to put all comments that contain \(oq====\(cq on a separate
line, preceded by an empty line. By default, no comments are treated
that way.
.SH OPERANDS
The following operand is supported:
.TP 10
.I file-or-URL
The name or URL of an HTML file. If absent, standard input is read
instead.
.SH "EXIT STATUS"
The following exit values are returned:
.TP 10
.B 0
Successful completion.
.TP
.B > 0
An error occurred in the parsing of the HTML file.
.B hxnormalize
will try to correct the error and produce output anyway.
.SH ENVIRONMENT
To use a proxy to retrieve remote files, set the environment variables
.B http_proxy
and
.BR ftp_proxy "."
E.g.,
.B http_proxy="http://localhost:8080/"
.SH BUGS
.LP
The error recovery for incorrect HTML is primitive.
.LP
.B hxnormalize
will not omit an endtag if the white space after it could possibly be
significant. E.g., it will not remove the first </p> from
\(oq<div><p>text</p> <p>text</p></div>\(cq.
.LP
.B hxnormalize
can currently only retrieve remote files over HTTP. It doesn't handle
password-protected files, nor files whose content depends on HTTP
\(oqcookies.\(cq
.LP
When converting from XML to HTML (option
.B \-X
without option
.BR \-x ),
any pairs of \(OQ<![CDATA[\(CQ and \(oq]]>\(cq are removed and
character entities &lt; &gt; &quot; &apos; and &amp; are expanded (to
\(oq<\(cq, \(oq>\(cq, \(oq"\(cq, \(oq'\(cq and \(oq&\(cq,
respectively), but any other character entities are not expanded. To
expand other character entities, pipe the input through
.BR hxunent (1)
first.
.LP
To limit lines to a given number of characters,
.B hxnormalize
breaks lines at spaces (or inside tags). Some writing systems do not
use spaces between words and thus
.B hxnormalize
may not be able to break lines, except at already existing line
breaks.
.LP
To make short lines longer,
.B hxnormalize
will combine lines and replace a line break by a space, except in
writing systems that do not put spaces between words, where the line
break is replaced by nothing.
.B hxnormalize
currently only does the latter for Japanese, Chinese, Korean,
Khmer and Thai. (The text must be correctly marked up with
\(oqlang\(cq or \(oqxml:lang\(cq.)
.SH "SEE ALSO"
.BR asc2xml (1),
.BR xml2asc (1),
.BR hxunent (1),
.BR UTF-8 " (RFC 2279)"