[#MDL-17542] html2text appears to be incompatible with the GPL

Issue Details (XML | Word | Printable)

Key:	MDL-17542
Type:	Bug
Status:	Closed
Resolution:	Fixed
Priority:	Major
Assignee:	Francois Marier
Reporter:	Francois Marier
Votes:	0
Watchers:	2

Operations

Add/Edit UI Mockup to this issue

If you were logged in you would be able to see more operations.

Moodle

html2text appears to be incompatible with the GPL

Created: 07/Dec/08 04:59 AM Updated: 16/Jun/09 10:01 PM

Return to search

Component/s:

Lib

Affects Version/s:

1.6.8, 1.7.6, 1.8.7, 1.9.3

Fix Version/s:

1.8.9, 1.9.5

File Attachments:

html2text.new.php (20 kB)
2.

html2text.php (15 kB)
3.

html2text_replacement18.patch (28 kB)
4.

html2text_replacement19.patch (28 kB)
5.

html2text_replacement20.patch (28 kB)

Issue Links:

Dependency

This issue will help resolve:

~~MDL-18368~~ html2text.php - notice when given a blank/empty string

Relates

This issue has been marked as being related by:

MDL-19447 Since last update german chars in text eMails are displayed wrong

~~MDL-19266~~ Forum posts containing links not sent

URL:	http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=507947
Participants:	Anthony Borrow, Dongsheng Cai, Eloy Lafuente (stronk7), Francois Marier, Helen Foster, Martin Dougiamas and Petr Skoda
Security Level:	None
QA Assignee:	Helen Foster
Difficulty:	Moderate
Resolved date:	22/May/09
Affected Branches:	MOODLE_16_STABLE, MOODLE_17_STABLE, MOODLE_18_STABLE, MOODLE_19_STABLE
Fixed Branches:	MOODLE_18_STABLE, MOODLE_19_STABLE

Description

« Hide

This library was reported in http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=507947

The license for html2text is not included in the code, but rather on the website of the author:

http://www.howtocreate.co.uk/jslibs/termsOfUse.html

The most problematic bit of html2text's license is the requirement to get written permission to distribute:

"However, if you want to include the scripts [...] in bundled or distributed scripts/software (such as Opera or Dashboard widgets, or the Formativ Academic Timetable), you should contact me first and obtain my express written permission."

Then, there is a bit of discrimination for commercial sites through the addition of an "advertising clause":

"Commercial/profit-making public websites:

You should put a note saying that the script was provided for free by http://www.howtocreate.co.uk.

You put a plain text (ie. not script) link to my site on your policy page or other appropriate page, to help me with my search engine rankings."

The Debian developer who filed the bug suggested that we take a look at replacing this library with the one I have attached here. It's got the same name but is under the GPLv2+.

Based on a quick 1.9 grep, that file is only used in lib/weblib.php:html_to_text(), which in turn is only used in format_text_email() of the same library file.

All

Comments

Change History

Version Control

Sort Order:

[ Permalink | « Hide ]

Petr Skoda added a comment - 07/Dec/08 05:47 AM

thanks, going to review it later this week

[ Permalink | « Hide ]

Eloy Lafuente (stronk7) added a comment - 10/Dec/08 03:48 AM

Assigning to Petr

[ Permalink | « Hide ]

Francois Marier added a comment - 16/Dec/08 02:05 PM

Here's a patch which replaces the html2text library with the attached GPL one.

The output is slightly different and we could help with more testing with different HTML inputs.

However, I will need to commit this to the Debian package today as it is considered a release-critical issue for Debian.

[ Permalink | « Hide ]

Martin Dougiamas added a comment - 16/Dec/08 02:34 PM

I'm cool with the patch going into Lenny.

It can also go into HEAD for further testing (but not 1.9.x yet).

[ Permalink | « Hide ]

Martin Dougiamas added a comment - 16/Dec/08 02:39 PM

Passing to Dongsheng to run some performance comparisons (that's the main thing I'm worried about).

[ Permalink | « Hide ]

Dongsheng Cai added a comment - 16/Dec/08 03:22 PM

The new one is faster to process <p></p>
New: 0.0113739967346
Old: 0.183360099792
(500 p tags)

but slower to process <img and <table>
New: 0.755731105804
Old: 0.208492994308
(500 tables, 500 imgs and 500 a tags)

[ Permalink | « Hide ]

Francois Marier added a comment - 16/Dec/08 03:38 PM

> 500 a tags

There is a setting to disable the processing of links (create a list of URLs at the end I think). Perhaps we should disable that to make it faster?

[ Permalink | « Hide ]

Dongsheng Cai added a comment - 16/Dec/08 05:08 PM

Hi, Francois
Another test on links only:
New: 0.126650094986
Old: 0.449314832687

Actually, the new lib faster in everything except <img, the funnest thing is the new lib directly ignore img tags, should we make it support process img tags?

[ Permalink | « Hide ]

Dongsheng Cai added a comment - 16/Dec/08 05:20 PM

I modified this file, now, the new lib wins on all tests

[ Permalink | « Hide ]

Francois Marier added a comment - 17/Dec/08 10:06 AM

Hi Dongsheng,

Here's a new version of the file. There are a few code changes, but the most important one is a security fix for possible code execution (see bottom of http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=508909 for the details).

Can you please run your script against this one as well?

Cheers,
Francois

[ Permalink | « Hide ]

Dongsheng Cai added a comment - 17/Dec/08 10:47 AM

both OOP html2text versions are fater than the proceduring html2text,
new version is a bit slower the the old version, I guess it use more regex in code.

Old OOP: 0.0414950847626

New OOP: 0.055447101593

proceduring: 0.39603304863

the testing text includes 500 links, images, tables and p tags each.

[ Permalink | « Hide ]

Francois Marier added a comment - 17/Dec/08 11:35 AM

New patch against 1.9 with the fixed version of html2text.php

[ Permalink | « Hide ]

Francois Marier added a comment - 17/Dec/08 11:36 AM

New patch against 1.8 with the new version of html2text.php

[ Permalink | « Hide ]

Francois Marier added a comment - 17/Dec/08 11:38 AM

New patch against HEAD with the fixed version of html2text.php

[ Permalink | « Hide ]

Anthony Borrow added a comment - 25/Feb/09 03:13 PM

Francois - Since Martin's concern was with performance and the new function was shown to be faster and because of the licensing issues, should we not also apply the changes to 1.9 and 1.8? I came across this file when looking at another issue (~~MDL-18368~~). Applying the 1.9 patch cleared up the issue I was experiencing in ~~MDL-18368~~ so it would resolve licensing, be faster, and less prone to PHP notices. I see no reason not to apply it to 18STABLE and 19STABLE. Peace - Anthony

[ Permalink | « Hide ]

Anthony Borrow added a comment - 25/Feb/09 03:15 PM

I am re-opening in the hopes that we can apply this to 1.8 and 1.9 and thus resolve ~~MDL-18368~~. Thanks for your consideration. Peace - Anthony

[ Permalink | « Hide ]

Anthony Borrow added a comment - 25/Feb/09 03:18 PM

One last thought/question, if the licensing is an issue for Debian will they not want this fixed in the 1.8 Moodle package?

[ Permalink | « Hide ]

Francois Marier added a comment - 25/Feb/09 03:51 PM

I don't mind applying it to 1.8 and 1.9. There are a few differences in the conversions from HTML to text but as far as I remember, it was a matter of the output being "different" (as opposed to being "wrong").

The Debian packages all use the new code already.

Francois

[ Permalink | « Hide ]

Francois Marier added a comment - 03/Apr/09 07:17 AM

Committed the fix to 1.8 and 1.9 stable.

[ Permalink | « Hide ]

Petr Skoda added a comment - 21/May/09 06:44 PM

looks like we have a serious regression, please review the linked bug report, thanks

[ Permalink | « Hide ]

Francois Marier added a comment - 22/May/09 10:11 AM

I have committed an updated version of this file to CVS, along with a readme describing its origin.

[ Permalink | « Hide ]

Helen Foster added a comment - 16/Jun/09 10:01 PM

Thanks Francois

All content on this web site is made available under the GNU General Public License, unless otherwise stated.