foxreplace
foxreplace copied to clipboard
Woundorf
Auto Substitution on page load causes some functions of a webpage to break, while manual start of the substittution works
What steps will reproduce the problem?
1. Create four filter rules (in one group) for a webpage like this:
On Tab "URLs" Add Urls:
http://www.kwick.de/forum/topic/
www.kwick.de/forum/topic/
I just added my exported rules. Alternatively here is the description:
On Tab "Substitution" add the following rules (all with match case enabled and
of type Regular Expression):
Rule 1:
Purpose: insert a fake Metatag in front of each <tr> entry which will be
deleted using Rule 3 and Rule 4
Regexp:
(<tr *itemprop="comment"[^>]*>[\s\S]*?)(</tr|<meta *itemprop="creator"
content="([^"]*))
Substitution:
<meta type="TROLLCHECK" istroll="no" creator="$3"/>$1$2
Rule 2:
Purpose: "set" the attribute istroll to true if a special user name appears
(the example uses just a random user name no troll, but a quite productive one
which is good for testing purposes):
Regexp:
(<meta type="TROLLCHECK" istroll=)[^>]*"Tesla_Trooper"
Substitution:
$1"yes"
Rule 3:
Purpose: Delete Troll postings (the thr-entries below the fake-metatags with
istroll="yes")
Regexp:
<meta type="TROLLCHECK" istroll="yes"[\s\S]*?</tr>
Substitution: (empty)
Rule 4:
Purpose: remove the fake metatags for all all other tr entries
Regexp:
<meta type="TROLLCHECK"[^>]*>
Substitution (empty)
2. Now open the following page
http://www.kwick.de/forum/topic/437293/wissenschaftliche-fragen/
3. Navigate to a page on which the user Tesla_Trooper posted something (you can
see it on the left) using the index buttons on the bottom of the page. At the
moment there are two postings of this user on page 2 they are from around 25.
june .2013, 22:27. As new articles are written they might move of course. But
any article of this user will do. You can of course also alter rule 2 to mark
any other users postings.
4. While autoreplace is switched off, start the replacement manually when you
are on a page with postings of this user.
5. The Articles of this user should dissapear. Press on the index buttons to
switch to another page. They should work.
6. Reload the page, to make the articles "visible" again. Now switch on
autoreplace. Set the cursor in the adress bar of the browser and hit enter.
7. Check if the navigation buttons work and if the postings of the user were
removen. The removal works this way, but very often (but not quite always) the
navigation buttons on the bottom of the page (i mean the links 1, 2, 3 -->
which can be used to select a page to jump to) won't work anymore.
I never experienced this behaviour when I started the substitution manually.
What is the expected output? What do you see instead?
1. The substitution should always work (but this is also what happens, when the
adress is entered in the adress bar).
2. The page functions should still work (navigation links), but in some cases
they don't work anymore.
What version of the product are you using? On what operating system?
I don't know which version of FoxReplace I am using, but I just installed it
during the last weekend.
I am using Ubuntu Linux (64 bit version) and Firefox 21
Please provide any additional information below.
Original issue reported on code.google.com by [email protected] on 26 Jun 2013 at 9:30
Attachments:
I just tested it once more. With the same effect (links are not working
anymore). I recognized that the links are indeed there. At least I can use the
"copy link" function (I use a german version of firefox, maybe it is named
differently) and paste the link in an editor to verify it and it is really the
correct address.
Btw. the substitution is not done automatically when you click on such a link
(e.g. if you open the page without having autoreplace switched on, then switch
it on and click on a (index-)link on the bottom, the replacement is not
performed, ony when you enter the address in the addressbar. But thats another
toppic I guess.
Original comment by [email protected] on 26 Jun 2013 at 10:04
Hi, I can reproduce your problem. It seems that the page is not fully reloaded
when you click in the pagination links. For example, if you select the text
"Meet new people" in the left or in the top yellow bar (I don't know if it
appears when logged in) and then navigate to another page, the text keeps
selected, when after a full load it shouldn't. So there may be some kind of
strange ajax going on, because the links look like normal links, as you say.
Original comment by [email protected] on 27 Jun 2013 at 9:01
I need to clarify, I can reproduce the problem about auto-replace not working,
but the pagination links have worked every time.
Original comment by [email protected] on 27 Jun 2013 at 9:41
Hi Marc, thank you for the analysis. You mean, the pagination links work, when
the two options are not checked (link- and script replacement), right?
Yes, thats what I also experienced. Btw. the problem with the broken links
seems only to happen, when Replace scripts is checked (with just replace URLs
it also seems to work).
Is the missing auto-replace also page specific? Isn't there always a
Page-Loaded event if a new page is displayed in the browse?
Original comment by [email protected] on 28 Jun 2013 at 8:10
Yes, I had both options unchecked. If I check "replace scripts" I can reproduce
the problem of the links not working. There is a JavaScript error logged in the
error console in that case. I suppose that it is the same problem as in issue
81 or a similar one.
Autoreplace doesn't work because the page is not fully loaded, only a fragment
is loaded with AJAX. This can be seen in the web console (Firefox or Tools menu
-> Web developer -> Web console, or Ctrl+Shift+K):
1. Load any page of the forum
2. Push the Clean button in the console
3. Switch to another page using one of the links at the bottom.
4. In the console you will see the GET petition with a link like this:
http://www.kwick.de/forum/topic/437293/wissenschaftliche-fragen/5/?__env=ajax&_=
1372408596938
If you follow the link you can see that the response is a JSON object with the
new content of the page, which is then inserted in place of the previous page
in the DOM tree.
So, the problem with autoreplace is really the same as in issue 13.
Original comment by [email protected] on 28 Jun 2013 at 8:51
Hi Marc,
thank you very much for analyzing the problem. I just checked like you describe
it, you are right. But why do the links in the page source look like ordinary
links? is the html code replaced and the orignal version is just there for
compatibility reasons (with browsers that dont support some functions)?
I guess thats no good news for me, is it?
Do you plan to include a solution for ajax? I guess then either the regular
expressions would have to be applied either just to the loaded fragment, or to
the source of the whole page after it has been altered, or does the browser not
really update the source code of the page it is uploading unless the user opens
the page source view?
J.
Original comment by [email protected] on 28 Jun 2013 at 8:22
I suppose the links are there for when JavaScript is disabled. I suppose it is
somehow intercepted when a link is clicked and then it is converted to the
other link, but I don't know this is done (I'm not an expert on AJAX).
I would like to include a solution for AJAX (issue 13), but I don't know how to
solve it for now.
Original comment by [email protected] on 29 Jun 2013 at 2:16
I found the same problem with a page that is constantly running the linux
"tail" command. ie auto replace does not work but manual does. Unfortunately it
is a password protected page so I cannot share it here.
Original comment by [email protected] on 24 Sep 2013 at 12:22