Do you see the same problem when you import to the filesystem instead of a database? Maybe in the latter case, only bookmarks are imported, not files?
I wasn’t until I tried it out. I have to admit that I find the GUI irritating and the results not consistent: importing to the filesystem seemed to work ok whereas importing into the database did not give the results promised by the documentation.
DEVONthink doesn’t change links while downloading (and never did), it only looks for items in your database(s) having the full absolute URL (e.g. after resolving relative links) while browsing and should use them if found. Please send me the database and I’ll have a look at it.
In that case, I repeat my suggestion for the OP to use a command line tool that can change the URLs, eg wget.
The URLs are fine. The question at hand is this part of behavior:
So you open the
all.html page which has a link to
subdir/page.html. When you click that link, DT should “resolve the relative link” to be “the full absolute URL”
http://pat.local/dt-html-import-site/subdir/page.html, and since there’s a document that was downloaded that has that exact URL, use that item.
I emailed you the test database I used.
I seem to be missing something here. Until now, I thought that you wanted to view downloaded HTML pages while you’re offline (cf. your first post here). Now you say that you want DT to “resolve the relative link” (i.e.
subdir/page.html) to the absolute URL (i.e.
But this is, as far as I can tell, the exact behaviour DT shows now. With the obvious caveat that it can’t “resolve” this relative link when the relevant server is offline (either because it is turned off or because you have no connection.
I just tried it with a website here (mind you: one hosted outside of my local network – maybe that’s relevant?) and it works exactly as expected: If WLAN is off, DT displays the HTML pages as they where downloaded. So that seems to work as described in the documentation, at least in this case.
Note Since @pete31 saw the same behaviour as @padillac, also with a local web server, maybe it is related to the fact that the URLs’s domain is
.local? At least in @pete31’s example, I’m certain that the loopback interface will be used. Depending on the network layer, someone/something might take a shortcut there and handle server/connection detection differently then for a “real” interface?
The problem is the local webserver
I apologize for shouting
@BLUEFROG and I see no problem but @pete31 and @padillac do because @BLUEFROG and I apparently tried external websites (i.e. not hosted by a server running on the current machine).
I just tried to replicate @padillac’s setup with a local Apache and saw the exact same behaviour: DT displays the website ok as long as the server is running. If the server is turned off, DT complains about itnot being available. I think that this is inconsistent: It is not relevant on which server the site is hosted.
So maybe @cgrunenberg could have a look into the code and figure out how the local interface (
lo) and the other(s) (
en0 etc.) are handled differently?
Indeed. As described multiple times previously, the process is:
- Use the Download Manager to import a site
- Open one of the imported items
- Click a link, and have DT open the imported document that corresponds to that link, rather than requesting it from the web server.
Does it work when you click a link on one of the pages? That’s the question of this thread. The individual pages all work fine. It’s when clicking on a link from one page to another that DEVONthink does not load the imported document.
Interesting observation, and one that I found plausible. However, I removed the
pat.local entry from
/etc/hosts and restarted the machine, and DT still tries to connect to
pat.local (which now just times out because the domain doesn’t exist).
I will try it with an external site at some point though, to see if it behaves differently. fwiw, @BLUEFROG confirmed that he saw the same issue that I reported.
Yes for an external site.
No for a local stor site.
local is a special domain, so it might be handled differently than other TLDs
Okay! I think we cracked it I changed it to
pat.chicken and it works as expected now. Thank you for digging into this.
Well, the links appear not to work in DTTG, which was the whole point. Oh well. Hopefully DTTG3 adds this functionality.
(If DTTG2 should support this same functionality, please let me know)
The WebKit doesn’t seem to handle local and remote requests the same way but the next release will fix this.
The joys of opaque frameworks
Ok, I read, and reread this entire thread. The solution seems to be pat.chicken
Seriously, I want to do the same as padillac wanted to do,
- Import a website
- Navigate it in DT3 offline (either with no network connection, or with the website down)
I imported a website, subdirectory (complete), Files [all options selected], follow links in subdirectories
When I selected the main page [html], and turned off wifi,I had the same issue, with the page showing “The internet connection appears to be offline”
I am using DevonThink 3, on my iMac. I do not know what he meant by changing from pat.local to pat.chicken - and I do not think that applies for me, though it could
Am I missing the obvious solution?
Which version of macOS do you use and what’s the URL of the page?
So “importing a website for offine use” will probably not do what one would naively expect in many cases nowadays.