Re^5: Scrubbing XML

They are definitely significant.

value="foobar"
[download]

is not the same as

value="foo bar"
[download]

And XML parsers do not return the same for

value="foo bar"
[download]

and

value="foo  bar"
[download]

You might be thinking of HTML's whitespace collapsing rules. Even then, that's for rendering, and attributes aren't rendered.

Comment on Re^5: Scrubbing XML Select or Download Code

Replies are listed 'Best First'.
Re^6: Scrubbing XML by Your Mother (Archbishop) on Jun 02, 2011 at 16:12 UTC
Well, that's not the meat of what I asked. I asked if newlines were significant. libxml does collapse \n and \r (but apparently not others like \t and \f) to plain spaces in attributes. I'm mostly curious is if this part of the standard or a quirk in libxml's handling. `perl -MXML::LibXML -le'print XML::LibXML->new->parse_string(qq{<root a +ttr="a\nb"/>})->serialize' <?xml version="1.0"?> <root attr="a b"/>` [download]	[reply] [d/l]
Re^7: Scrubbing XML by ikegami (Patriarch) on Jun 02, 2011 at 17:00 UTC
Ah! News to me! If libxml does it, I'm sure it's in the spec. And here it is: 3.3.3 Attribute-Value Normalization. \r\n is is converted to \n. (Done for the entire document.) Entity references (e.g. `é`) are interpolated. \r, \n and \t are converted to spaces. Character references (e.g. `é`) are interpolated.	[reply] [d/l] [select]
Re^8: Scrubbing XML by Jenda (Abbot) on Jun 05, 2011 at 00:34 UTC
Nice. Looks like yet another example of the insanity of the XML specification authors. It's good to know XML parsers MUST corrupt the value of attributes. Really guys ... how many of you knew you have to escape your tabs and newlines when including the data in XML attributes? And which XML generators do that? Jenda Enoch was right! Enjoy the last years of Rome.	[reply]