Add --doctype option to NOT MEDDLE with the doctype found in the input #435

jidanni · 2016-07-09T18:59:42Z

I think it is unconscionable that you have no --doctype option,
including the default (auto) that will LEAVE THE USER'S DOCTYPE ALONE NO
MATTER WHAT.

No matter what he picks, there is a danger, however slight, that the
doctype currently sitting in his input file will get changed under his
feet, behind his back, or whatever you want to call it.

Why don't you add a option to not mess with his doctype, and just print
errors against that doctype, but don't switch it for him. OK?

tidy:
Installed: 1:5.2.0-1.1

jidanni · 2016-07-09T19:09:36Z

OK maybe --doctype auto doesn't in fact change the doctype found. But at least the man page should say that it is guaranteed not to.

Also the way the man page is written, the user might really try "--doctype user".

jidanni · 2016-07-09T19:21:39Z

P.S.,

tidy --doctype '"-//ACME//DTD HTML 3.14159//EN"'``` didn't do anything, and

tidy --doctype: '"-//ACME//DTD HTML 3.14159//EN"'``` made an error.

geoffmcl · 2016-07-10T23:25:58Z

@jidanni thanks for the (evolving) issue...

I certainly agree, the use of --doctype [enum] has some very weird problems...

Enumerate each, and file a separate issue... of course each issue should include tidy libtidy version, sample html, out html, config used, expected html, comments, etc...

And absolutely agree, the present docs help, does imply --doctype user will work! And certainly needs enhancing... a docs issue... please help...

So I agree, you have opened a can of worms... and that is good... separating and isolating the problems, one by one, may lead to a faster solution...

Or do you have a PR for review?

For sure, several --doctype [enum] problems exist, and should be isolated, addressed, and fixed... thanks...

tidy-html5 currently doesn't preserve user-supplied DOCTYPEs in output: <htacg/tidy-html5#435> Signed-off-by: Simon McVittie <smcv@debian.org>

jidanni · 2016-12-08T12:15:39Z

Add a switch to make tidy promise not to change the doctype when in auto
mode. One likes how tidy detects doctypes, but doesn't like it changing
them. Here's my workaround:

$ cat dantidy
#!/bin/sh
# Copyright       : http://www.gnu.org/licenses/gpl.html
# Author          : http://jidanni.org/
# Last Modified On: Thu Dec  8 20:07:16 2016
# Update Count    : 33
set -eu
f=/tmp/dantidytmp_$USER
cat > $f
# Ensure tidy won't change the DOCTYPE it finds:
case $(sed 2q $f) in
    *strict*) DT=strict;;
    *loose*)  DT=loose;;
    '<!DOCTYPE html>'*) DT=html5;;
    *) exit 99;;
esac
perl -C -pwe 's/[^[:ascii:]]/sprintf "\\x{%04x}",ord $&/ge' $f| #my uni2ascii(1)
    tidy --doctype $DT --gnu-emacs yes --indent-spaces 1 --indent auto -utf8 \
	 --tidy-mark no -quiet --fix-backslash no --enclose-block-text yes \
	 --vertical-space yes "$@" |
    perl -C -pwe 's/\\x\{([[:xdigit:]]{4})\}/chr eval "0x$1"/eg' #my ascii2uni(1)
#(not sure if I still need all the uni2ascii stuff.)

geoffmcl · 2016-12-16T18:54:56Z

Also see issue #472, with suggested patch, and a related config issue #468. which needs testing...

Specifically concerning the failure to use the user supplied <fpi> string, marking this as a bug!

brlin-tw · 2016-12-29T04:03:05Z

Provide my case:

I'm working on a few mixed doctype webpages using legacy attributes, and this particular XHTML 1.0 Transitional webpage:

line 23 column 5 - Warning: <table> proprietary attribute "bordercolor"
(..stripped...)
line 339 column 5 - Warning: <table> proprietary attribute "bordercolor"
Info: Doctype given is "-//W3C//DTD XHTML 1.0 Transitional//EN"
Info: Document content looks like HTML Proprietary

the doctype: auto option causes the XHTML 1.0 Transitional DTD being removed and on the second call it adds a XHTML5 DTD, claiming the page looks like XHTML5, and output a bunch of compliance warning since the page is in fact, not XHTML5:

line 2 column 1 - Warning: missing <!DOCTYPE> declaration
(... a bunch of these, stripped)
line 1127 column 5 - Warning: The summary attribute on the <table> element is obsolete in HTML5
line 20 column 3 - Warning: <div> attribute "align" not allowed for XHTML5
(... a bunch of these, stripped)
line 1169 column 13 - Warning: <div> attribute "align" not allowed for XHTML5
line 21 column 5 - Warning: <table> proprietary attribute "bordercolor"
(..stripped...)
line 337 column 5 - Warning: <table> proprietary attribute "bordercolor"
Info: Document content looks like XHTML5
Tidy found 197 warnings and 0 errors!

geoffmcl · 2017-02-28T19:26:42Z

Since we are about to release 5.4, and no work has been done on this, can only move it out to next 5.5.

Although I would have to say it is a little unclear exactly what is expected of tidy, so certainly need further clarification... thanks...

jidanni · 2017-02-28T23:37:28Z

There should be a way for the user to indicate to tidy, via the command line switches, that under no circumstances it is allowed to change the doctype found in the document. And it should use the doctype found when deciding what content is valid or not. Sometimes it is great that tidy decides a different doctype would be more suitable. But sometimes it is not always great.

geoffmcl · 2017-03-01T00:45:17Z

@jidanni I guess I am still very confused... tidy does not change the doctype...

If you leave the --doctype auto, the default, you do not seem to have given us one simple example where tidy changed your existing valid doctype in the output... it may, if none, or is invalid...

So while it is agreed the documentation presently says Tidy will use an educated guess based upon the contents of the document., that seems to only mean it will use your valid doctype in the output, but may add an Info: output suggesting what it found...

So it seems tidy does not decide a different doctype would be more suitable, and change the output doctype, but merely shows, as an Info: type message, of what it would suggest...

So give us an example where tidy did change the doctype, while in auto mode, and maybe we can do something...

Otherwise this seems closed, except for improving the quickref/man tidy description output... and I hope that will be addressed in 5.5, when we will address many message outputs...

Or have I got something really wrong here? Pleas help... thanks...

jidanni · 2017-03-01T00:59:14Z

Fine. Then just add on the man page "The only circumstances tidy will change the doctype found in the document are if ... ".

geoffmcl · 2017-03-01T01:11:57Z

Sounds good... am adding the Docs. and Feature Request tags in the hope that this will get picked up in the full 5.5 documentation review, and removing the Bug tag... thanks...

balthisar · 2017-09-24T17:43:44Z

Added #614 PR to address the documentation issue.

Addresses #435 by updating documentation.

balthisar · 2017-09-26T01:18:02Z

Documentation now live as requested, so I will clear this from the queue.

geoffmcl added the Technical Discussion label Jul 10, 2016

geoffmcl added this to the 5.3 milestone Jul 10, 2016

smcv added a commit to smcv/html-tidy that referenced this issue Jul 22, 2016

t/unicode.t: don't assert that the DOCTYPE is preserved

a2407a2

tidy-html5 currently doesn't preserve user-supplied DOCTYPEs in output: <htacg/tidy-html5#435> Signed-off-by: Simon McVittie <smcv@debian.org>

geoffmcl mentioned this issue Dec 16, 2016

TidyDoctypeMode option documentation missing #472

Closed

geoffmcl added the Bug label Dec 16, 2016

geoffmcl modified the milestones: 5.5, 5.3 Feb 28, 2017

geoffmcl added Docs Feature Request and removed Bug Technical Discussion labels Mar 1, 2017

balthisar added a commit that referenced this issue Sep 24, 2017

Addresses #435 by updating documentation.

965b92a

balthisar added a commit that referenced this issue Sep 26, 2017

Merge pull request #614 from htacg/issue_435

94111a7

Addresses #435 by updating documentation.

balthisar closed this as completed Sep 26, 2017

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add --doctype option to NOT MEDDLE with the doctype found in the input #435

Add --doctype option to NOT MEDDLE with the doctype found in the input #435

jidanni commented Jul 9, 2016

jidanni commented Jul 9, 2016

jidanni commented Jul 9, 2016

geoffmcl commented Jul 10, 2016

jidanni commented Dec 8, 2016

geoffmcl commented Dec 16, 2016

brlin-tw commented Dec 29, 2016 •

edited

geoffmcl commented Feb 28, 2017

jidanni commented Feb 28, 2017 via email

geoffmcl commented Mar 1, 2017

jidanni commented Mar 1, 2017 via email

geoffmcl commented Mar 1, 2017

balthisar commented Sep 24, 2017

balthisar commented Sep 26, 2017

Add --doctype option to NOT MEDDLE with the doctype found in the input #435

Add --doctype option to NOT MEDDLE with the doctype found in the input #435

Comments

jidanni commented Jul 9, 2016

jidanni commented Jul 9, 2016

jidanni commented Jul 9, 2016

geoffmcl commented Jul 10, 2016

jidanni commented Dec 8, 2016

geoffmcl commented Dec 16, 2016

brlin-tw commented Dec 29, 2016 • edited

geoffmcl commented Feb 28, 2017

jidanni commented Feb 28, 2017 via email

geoffmcl commented Mar 1, 2017

jidanni commented Mar 1, 2017 via email

geoffmcl commented Mar 1, 2017

balthisar commented Sep 24, 2017

balthisar commented Sep 26, 2017

brlin-tw commented Dec 29, 2016 •

edited