module Jumplinks

[tpr]

Well you could place the description right next to the "From" cell content. If it's sufficiently different than the content before it then it should be usable, plus adding text-overflow: ellipsis you can limit the length (plus adding a "title" html tag to see all on hover).

opacity: 0.66;
font-size: 0.92em;
display: inline-block;
font-style: italic;
margin: 0 0 0 1em;
max-width: 200px;
overflow: hidden;
white-space: nowrap;
vertical-align: middle;
text-overflow: ellipsis;

However I would still prefer the full-line descriptions even if the inconsistent look.

A tooltip would be better than nothing but I'm not quite a fond of that.

[Mike Rockett]

1 hour ago, Macrura said:

hopefully it's not too much extra work

It's not about extra work, it's really about not bloating the module. Perhaps I should put it to vote?

[Mike Rockett]

47 minutes ago, tpr said:

Well you could place the description right next to the "From" cell content.

That could potentially work. However, in order to prevent too much text-clutter, perhaps a lighter colour would work, which would then darken on hover...

[tpr]

Darken on hover is a good idea. You could make somehow the descriptions optional too and let the user decide whether he wants the clutter or not

[Mike Rockett]

2 minutes ago, tpr said:

Darken on hover is a good idea. You could make somehow the descriptions optional too and let the user decide whether he wants the clutter or not

When the descriptions are on the bottom, row height becomes inconsistent. On the right, however, the opposite is true, and the clutter is not as bad.

I'm going to play around with this a little.

[Mike Rockett]

Update: Haven't played with it yet - will do so this weekend.

Quick question: I'd like to see very basic analytics of the new module's usage - would anyone be against me installing a basic install/uninstall tracker that communicates with my server? It would literally just send a simple payload containing the domain name and the date on which it was installed or uninstalled. This information would be kept private, but I could potentially put a badge in the documentation to indicate active installs - that's a big maybe though... I would prefer to keep all data 100% private.

[tpr]

Is it possible to redirect all 404 hits to the homepage? I've just tried ".*" -> "/" and it appends the full url like this:

http://domain.com/http://domain.com/

Tracker: I think most devs wouldn't like this feature but I'm not against it, provided it's OFF by default.

[Mike Rockett]

9 hours ago, tpr said:

Is it possible to redirect all 404 hits to the homepage? I've just tried ".*" -> "/" and it appends the full url like this:

http://domain.com/http://domain.com/

Tracker: I think most devs wouldn't like this feature but I'm not against it, provided it's OFF by default.

I see - this should definitely be possible (using {all}) and page:1 or a relative/absolute link to your domain root. I will take a look at this when I'm back from work this afternoon.

[Mike Rockett]

Looks like a genuine bug fell in here somewhere - all redirects with Page IDs appear to be broken on my side. Investigating now.

[Mike Rockett]

Unfortunately, I'm not able to determine why the httpUrl is being appended to the root URL. I'll need to look into this further, but it has something to do with preg_replace_callback which is at fault here. The method in question is supposed iterate through each source wildcard, applying any replacements to the destination. But as their are no wildcards in the destination, it should be changing anything. When you supply / as your destination, Jumplinks converts this to page:1 so it can get the full URL to that page. As it loops over the source, it's somehow adding that URL twice to the destination.

I need to head out now, but will be back in a bit to investigate.

I do know that Jumplinks 2 is not succeptible to this as it uses a completely different method, powered by FastRoute. I'll try to continue work on that Jumplinks 2 today. (To everyone who's been waiting in anticipation for it, I'm so sorry it's taken so long.)

[Mike Rockett]

Have pushed a fix; please update to 1.5.46 and let me know if it works.

[tpr]

Seems to work fine, thanks!

[owzim]

Hi @Mike Rockett, this module is awesome, thanks!

I have this collection thing going on:

example source: project?id=1&foo=bar

So I set it up like this:
source: project?id={id}&foo=bar
dest: projects/{id|projects}

Works!

But, when &foo=bar is omitted this does not work:
project?id=1

How to make anything after id optional, I tried brackets, but it does not work.

source: project?id={id}[anything here, but optional]

Thanks!

[Mike Rockett]

@owzim - Thanks. Jumplinks v1 doesn't support explicit optional trailing parts, so you'll need to create a separate jumplink for now if you want to capture foo=bar. However, if it doesn't matter what needs to trail the first parameter, then you can simply use project?id={id}{all} and leave the destination as-is.

v2 will support this. You'll be able to wrap the optional part in square brackets and Jumplinks will add the optional part to the end of the destination if it appears in the source, and it'll also ensure the URL is structured properly. In other words, you'll be able to set it up like this:

Source: project?id={id}[&foo=bar]

Request: project?id=4 --> project/project-name
Request: project?id=4&foo=bar --> project/project-name?foo=bar

If the source is project?id={id}[!&foo=bar], however, then the optional part will not be added to the destination.

I haven't actually worked on the feature yet, so it may change; but this is how I plan on doing it. (Speaking of which, forgot to add this to the roadmap!)

[owzim]

Thanks @Mike Rockett, the implicit option with {all} will suffice.

I haven't dug too deep into all the options, but it would be great if mapping could be done without collections:

source: project?id={someCustomField}
dest: projects/{someCustomField}

So that someCustomField would be a field of projects/ child pages, like $pages->get('/projects/')->children('someCustomField={value}');

This way one could simply migrate the ids of the old site over.

source: project?id={oldID}
dest: projects/{oldID}

$pages->get('/projects/')->children('oldID={id}');

 

[Mike Rockett]

@owzim - This is what destination selectors are for. In your case:

source: project?id={id}
dest: [[parent=/projects/, oldID={id}]]

Anything inside double-square-brackets is a selector, and returns the full httpUrl of the page. You can also append a query string to the destination for the purposes of analytics, if you wish.

(The syntax is changing in v2 to @[selector].)

[owzim]

A classic case of RTFM %#~*  

Thanks @Mike Rockett, this is brilliant.

[Mike Rockett]

3 hours ago, owzim said:

A classic case of RTFM %#~*  

I can be guilty of the same, hehe! You're welcome.

[Mike Rockett]

Have been doing some work on v2, Currently, the new manager looks like this the image below.

Descriptions: I've opted to show the optional descriptions underneath the source, and there will also be an option to not show them if it starts looking too cluttered. Originally, I'd said that they should be shown to the side of the source, but that makes it difficult and tedious to prevent datatables from running of the side of the page. With that said, my recommendation is to only use descriptions where they are really necessary -- like if you need to use the description for the purposes of searching a large data-set and you can't remember the source. At the end of the day, this feature won't be used my many and, for those who don't use it, it won't get in your way.

Parameter Highlighting: You'll also notice that parameters in the jumplink destinations are now highlighted - this makes it easy for those who don't know 100% what's going on (like someone who has taken over a site) to learn quickly (tooltips provided). This feature can be disabled, though I'm not sure if it should be disabled by default. Please vote in reply -- if there's no feedback, I'll leave it enabled by default. Lastly, I don't think it's necessary to highlight source parameters (too flashy), but I do plan to add a slight underline to those for the purposes of tooltips, which will only really be handy for aliases such as segment.

Slug alias: Speaking of aliases, I've added a slug alias, which strictly captures parts of a URI that are slugs; that is, a word, optionally followed by a single hyphen, rinse, repeat. Segment, however, is not strict and allows any amount of consecutive hyphens. In the new documentation, I'll recommend the use of slug.

[owzim]

Hi @Mike Rockett, I can't get the CSV import on 1.5.46 to work:

Tried:

foo,bar
baz,qux

foo;bar
baz;qux

foo,bar,,
baz,qux,,

foo;bar;;
baz;qux;;

In the cases without trailing delimiters I get:

Notice: Undefined index: source
Notice: Undefined index: destination
ProcessJumplinks: SQLSTATE[22007]: Invalid datetime format: 1292 Incorrect datetime value: '-0001-11-30 00:00:00' for column 'date_start' at row 1

With trailing delimiters, the jumplinks are created but source is empty and destination is a string: page:1.

I made sure, no empty lines are present.

[Mike Rockett]

Thanks for spotting that @owzim; looks like an essential line in the importer mysteriously vanished (in this commit that's a year old; only spotted now). More than likely a mistake on my side, but I do remember phpfmt was playing up at some point. Anyways, adding the line back - will bump shortly.

[Mike Rockett]

Bumped to 1.5.47. Important update for anyone needing CSV. Though it doesn't look like anyone has used it since January last year ? Thanks again @owzim!

[tpr]

Well I would use the import feature but haven't found an export feature to move my jumplinks around. Is there any or will it be in v2?

[Mike Rockett]

@tpr - Jumplinks 2 will be able to export all of its data, including config, via JSON or a gzipped archive containing JSON data. It will also be able to import from CSV and JSON files generated by other systems. When importing from JSON, a specific format will be required, but the importer will ask for fieldnames where they don't match the destination fieldnames. So If you provide {{"from": "/Somedir/file.ext", "to": "/awesome/path"}}, it will work correctly without asking questions, but if from or to or any other field does not use those names, then the importer will ask what the names are so it can read those instead.

----

@All - Initially, I wanted to have the PHP version requirement match that of Processwire, but I would now like to stick to the PW version recommendation, which is currently 5.5+. This allows me to use slightly newer components. However, 5.5 is at end of life, and only 5.6+ is currently supported. In the interests of users with hosting providers who still offer 5.5 as the newest version, Jumplinks will work for you. That said, using a host that provides unsupported software is not recommended. Many providers are moving over to 7.0 and 7.1, so the Jumplinks PHP version requirement become 5.6 when 7.0 active support comes to an end in 10 months. Then I'll follow that format in the future, meaning that 7.0 will be required when 7.1 active support comes to an end in 1 year and 10 months.

It's either that or I run a poll to see what version of PHP folks are using, and then make a decision from there.

For those interested: http://php.net/supported-versions.php

[owzim]

Thanks for the update @Mike Rockett. After the update: without trailing delimiters (foo;bar) I get:

ProcessJumplinks: SQLSTATE[22007]: Invalid datetime format: 1292 Incorrect datetime value: '-0001-11-30 00:00:00' for column 'date_start' at row 1

With trailing delimiters (foo;bar;;) I get:

Notice: Undefined index: date_end
ProcessJumplinks: SQLSTATE[22007]: Invalid datetime format: 1292 Incorrect datetime value: '-0001-11-30 00:00:00' for column 'date_start' at row 1

This only happens if I input more than one line. If it's only one line, it works. If it's only one line, both values separated by the delimiter land in the source field, destination stays empty.

Is there a reason, why you rolled your own CSV parser instead of using str_getcsv?