Jump to content


  • Posts

  • Joined

  • Last visited

Everything posted by BrendonKoz

  1. Hi Ben, This took me a bit to understand, but I think you're looking to generate an OpenGraph image similar to how GitHub's OG images appear to almost be an HTML page converted to an image? (At first I thought you just wanted to use a template to assign a specific page image and generate the meta text...) Chris Coyier at CSS-Tricks wrote up an overview of a couple different ways to produce something like this. His platform is WordPress using GD, but it could just as easily be applied to ProcessWire and IMagick. There are also services out there to help with this as well. See the comments for even more info. https://css-tricks.com/auto-generated-social-media-images/ I'm in the middle of cleaning, patching/repairing walls, and repainting my guest room after a "friend" destroyed it on me, otherwise I'd offer to get a quick example going for you over the weekend. However, your needs are likely custom to your project and there are any number of various ways to produce this, so hopefully the overview linked will give you a good head start! πŸ™‚
  2. I know I'm super late to this party, but since I've been away for quite some time - but absolutely love ProcessWire - and Ryan mentioned that this roadmap could extend beyond 2021, I thought this was the most appropriate place to comment. Thanks to @teppo's PW Weekly I've been able to, mostly, recount (in summarized form) multiple years of development (with plenty of tangential links, often reading the full blog posts by Ryan) both of PW and in 3rd party development efforts, so I don't feel too terribly left behind. Something a little different not yet discussed: Zooming out and looking at the bigger picture, within regards to the CMS/CMF landscape, I think an area that PW lacks is in marketing of itself, and I don't mean this as a knock on ProcessWire - the product speaks for itself and word of mouth is awesome, and marketing efforts can oftentimes detract from active development time. Marketing isn't exactly a "core feature" of the codebase, but it is something that I think can be code-adjacent. ProcessWire is capable of so, so much, but from just looking at it, or installing it from source, it's hard to see that from the eyes of a first-time user. ProcessWire doesn't use templates like WordPress, but templates are a big business with WordPress. Although some WordPress templates come bundled with custom plugins which starts skirting the line of what ProcessWire does (site profiles), there's a plethora of ready-to-go, custom designed websites with a only the simple task, after installing, of adding content. ProcessWire is more developer focused (which I personally love), but that doesn't mean we can't, or shouldn't, also cater to an audience that may want to see what designs or functionality can be made (almost) immediately available after install. As a simplistic first step here: currently Site Profiles are listed under Modules as its own category, but that's not too intuitive for findability - if this were to have its own landing page and higher level navigation, with feel good images based on the profile (for those that provide them), I think it might help to attract more potential users. On the topic of Site Profiles, since ProcessWire is capable of so much (that many of us aren't even entirely aware of), having some custom built, premium Site Profiles built as premium starter kits (kind of like what my buddy Jack McDade is doing with Statamic) it might be another boon, and additional income source. He's also considering selling branded merchandise, which I think we'd tried before and it didn't take off, but it's easy to do (via something like CafePress) and every little bit helps I suppose? On these notes, should there be - or has there been thought to - an official marketplace via the ProcessWire website for things like these (with a percentage going back to PW)? Pro Module Thoughts: 1. Search: I remember a mention of the front-end search functionality of the current version of the PW website potentially being packaged up into a Pro Module. What's the status on that? I think that - if it could be packaged into a pro module - it would be a fairly popular one. I realize that language issues would be a hurdle and there were strides made towards that endeavor such as with WireWordTools but don't know if the feasibility was broken down once multi-language was considered or if it's still able to be worked on and released. 2. Recurring Dates / RRule I too would have a use for this. Realistically, most businesses could if they want to attempt to accurately list their open hours when they celebrate holidays, as holidays have recurrence, and can have annoying recurrence (like the American Thanksgiving) which is the 4th Thursday of November. API-based things might have a need/want to integrate LazyCron into a non-standard recurrence. My personal usage would be for events - honestly I'm surprised the school websites Ryan's worked on haven't had a need for this. Reporting to Google for JSON-LD on its recurrence may also be helpful to have an API-level access to it as well (I'm not familiar enough to know for sure though). The largest difficulties in handling recurring events is exception management. There are usually two ways to generate recurrence - tie them to a single entry and assign a recurrence rule and extrapolate from that dynamically on call, or use the recurrence rule to generate the actual entries, individually (and usually have them tied to an original). How would exceptions be handled if a single date, or time, needed an adjustment? If one or more were cancelled? If one or more were to be added? If editing the primary edits them all, does it overwrite all others even if individual changes were made? Whether convention or configuration, these are just some potential issues with recurrence. πŸ™‚ Image Reference Field and/or Central Image Manager: It's certainly not necessary to a website being used, created, or maintained - but it would certainly be helpful if it was part of the core, whether optionally (in page settings) or statically (as a field/input setting). Teppo did a great job illustrating why: I'm essentially a technician (title is Web Administrator, but 85% of my job's time is help desk / support), professionally I'm a programmer/web developer, but the business I've worked for has 100+ people who are extremely varied in their technical expertise. 40 or so of them will be expected to have access to the administrative back-end for the primary website I'm (finally) redoing in one way or another, as they will be the content authors/maintainers (I work for a public library, so these are city-level librarians). They would not take the time to find a reference to an image already in the system when they could more easily just upload another completely brand new copy of the same image (let's say it's a 200MB file) multiple times, and resize it to a 240px width. I know that I could create a page that contained a "shared" image repository, or use one of the various modules, but the fact is that it can be very useful for various reasons. Layout Tool / Editor: From my own personal experience, I would agree with bernhard here. I think, based on the 5th time I've read through this topic, that it's been decided that Editor.js, specifically, is not a good fit, but continuing on... I'd done some testing of my own, like bernhard, and found limitations and issues with it. I'd also tried showing it to some basic-skilled coworkers with minimal (but enough) instruction and they struggled, hard. Showing them a CKeditor field (without instruction), however, they were able to do what they wanted and although were confused with certain things, weren't struggling. That said, from what I've seen of Bard and Editor.js, a similar approach can be managed using styling and JS adjustments with the RepeaterMatrix fieldtype, even if it's apparently overkill where storage needs come in. I do like that individual fields are still independently searchable within RepeaterMatrix though, which I think wouldn't be the case with an Editor.js fieldtype solution. How is RepeaterMatrix similar to Editor.js with style and JS changes? Robin S. and monollonom essentially had similar ideas - that the layout is different for Editorjs/Bard type fields, they're more compact and seemingly integrated. From the observation of the front-end GUI interface of Editor.js and Bard, it's doing something fairly similar to what RepeaterMatrix currently does - you create a new item (or start of a block of items) from a predetermined set of options. It's just that the interface of the other two has a different "wow" factor. Click a hoverable + button, choose from the available options, and then go about adding the related content. So if a fieldtype was created that's similar to RepeaterMatrix but stored data in a more efficient way, and was intended to be a more concise and (for lack of a better word) all-enclosed field, I would think using similar markup to RepeaterMatrix with a much more visually minified styling could help here. I'd love to understand more about how the $page-meta() can be used to integrate non-field stored data like bernhard was asking about, because that could be a way for developers to setup layout-like interfaces within custom-made template builders. Actual Page Builder - an Alternative: Fieldtype There are a few open source visual page builders for email templates (ex: GrapeJS). These are well-tested and I'm thinking that for the purposes of single page instances or landing pages, something like that could be used and created into a fieldtype of its own. Remove all of the markup from it afterSave to get only the text content, and place that content in a separate data field in the DB to make it searchable and it's now capable of being a standard page/template in PW. It would be up to the developer to place appropriate limits on it, but beyond that... Just a thought. I β™₯ ProcessWire. (I also love all of the work that's been done since this topic was created, too!)
  3. Although slow and impractical to show all 100 images on a single page in a gallery, would it be as slow and impractical if you used the JS-based lazy-loading for images (don't forget to force container size on the image so there's no re-flow issues during rendering -- ex: https://css-tricks.com/preventing-content-reflow-from-lazy-loaded-images/). It's not necessarily a better solution, just an alternative.
  4. The ProcessWire modules directory link for Fieldtype Leaflet Map Marker is pointing to the master branch's ZIP archive which is the 2.8.1 version. I'm assuming that needs to be updated to point to the proper Github location.
  5. It appears that the issue I'm having with the CSS layout is because I wasn't using AJAX to load a pre-existing template/form, there is no wrapping form field. Even if your module uses InputfieldForm to wrap/render your module's inputs, the form itself is stripped in the page edit admin so that the page template only contains a single form. So...I either need to dynamically wrap my content in a form (with class "InputfieldForm"), or create a template and use AJAX load with the panel. I'm not sure which is simpler. Probably the AJAX for less future maintenance...in which case, I now need to figure out how to have an external template with my module as I can't think of any public modules that do that to study from.
  6. That second bit with the hook is extremely useful - just not in this instance: if the field is included, it's going to be used on the page, and therefore the JS will be required no matter what. However, I'm going to save that bit away for later!!! Thank you! Your function format (and find() call) apparently was what I needed. Using that worked. jQuery(document).on('pw-panel-opened', function (eventData, panelNode) { The iframe, in my case, was not going to work since the form actually exists on the current page. (A semi-unknown/unused aspect of the panel is that you can load content that already exists on the current page using data-panel-id: $button->attr('data-panel-id', 'module-fieldtype-container'); ... when using this method, however, there's no iframe.) Thank you!!!!!!
  7. I think my issue with the columns may be because I'm using the panel to load a form that exists (but is hidden via CSS) on the page-edit screen. Essentially my custom Fieldtype is large enough that it made sense to load the add/edit form somewhere else (modal/panel). It's likely that I'm not wrapping the form that I've created on the page (and grabbing via JS using $button->attr('data-panel-id', 'fieldtype-form');) in a particular class, though I've not yet figured out what that class is. I believe most people that use the panel are targeting an AJAX load to another file/form external from the currently edited page. As for the JS, I'm still trying to figure out how to target only the panel that I just opened and not any panel on the page (i.e.: Page Edit screen's "View" -> "Panel"). I thought I had figured it out with the following, but it only ever runs the ELSE loop. From the panels.js code, it looks like the classes should exist before the custom event trigger is registered, so I'm not sure what's going on. $(document).ready(function() { $(document).on("pw-panel-opened", function(e) { if ($('*[data-panel-id="fieldtype-form"]').hasClass('pw-panel-container-open')) { console.log('book add opened'); } else { console.log('panel opened'); } }); }); Running that test ([data-panel-id="fieldtype-form"]) from the browser's console while open or closed manually gives me the result I expect.
  8. Hi @Zeka, yes I'm developing (or trying to anyway πŸ˜†) a Fieldtype/Inputtype. I'll have to do some more testing with the columns to see what might be the issue I'm having (thanks for showing me your example). I'm also going to upgrade from Master to Dev to see if it makes a difference. I tried your version of the jQuery call and it worked - the first time the panel opened. Successive opens were not getting logged. So it seems I'm targetting the wrong element in my call (as your code example will run any time any panel is activated), but why it doesn't seem to run on all of the actual events is strange. (Edit: Firefox was kindly aggregating similar console.log results for me and I didn't notice.) Hopefully the Dev version will show further success. πŸ˜‰
  9. I tried taking advantage of the panel.js in the "dev" branch and using an "on" method for "pw-panel-opened" in my module but it also failed. $(document).ready(function() { $(document).on("pw-panel-opened", ".InputfieldBooksAddPanel", function(e) { alert('It is open!'); // This never shows up $('#isbn').focus(); // This never receives focus }); }); Also, unrelated, but can panels take advantage of the admin theme's columnWidth for rendering core elements? It doesn't seem to behave the way I'd expect (ex: I can get 50% width for the inputs within the panel, but there is never more than a single column).
  10. Hello all! Has anyone with more JavaScript abilities been able to hook into ProcessWire's jQueryUI Panel functionality in order to handle onload events of the panel itself? I'm trying to simply set focus to the first input element of the form after clicking the button that activates the panel, but my click event does not run (I think due to the return false; in the panel.js as shown below). .../wire/modules/Jquery/JqueryUI/panel.js: // delegate events from toggler to pw-panel-button $toggler.click(function() { $btn.click(); return false; }).on('mouseover', function() { $btn.mouseover(); }).on('mouseout', function() { $btn.mouseout(); });
  11. (Assuming the suggestion is for a normalized way of handling versioning in changelogs for all modules moving forward.) As much as I like the idea of using some dynamic method to keep a version number in check, I'm not sure that a changelog (or markdown file) is the best way to handle that - at least not by simply pulling the first line and expecting that it will contain the updated version number. There are many suggested ways of creating changelogs. (There's even a discussion about best practices mentioning some of those links.) Since there are so many ways to handle changelogs (sometimes organization-specific standards) I'm not sure this could work for everyone in practice (esp. those that are using info.json files for this module data). I definitely think it looks like a terrific idea for your own use case though, and until I thought about it further I got all excited by the prospect! πŸ˜„ I do think it's OK to have a fieldtype/inputtype combo that share the same name also share the same version number.
  12. So, food for thought in terms of REGEX vs DOMDocument: Benefits of using REGEX: Essentially faster/more efficient for processing of the data Doesn't care about valid source structure as it's parsing straight text, not XML nodes Implementation is unlikely to change Detriments of REGEX: Writing a perfect implementation of a REGEX when dealing with HTML to handle all use-cases without experiencing any edge-cases is difficult (might "greedily" match more than intended) It definitely works, but the developer argument is: is it the best (most appropriate) tool for the job? Without a good knowledge of REGEX, harder to understand the underlying code if changes/updates are required Benefits of using DOMDocument: Written specifically for the purposes of this type of task (searching/modifying the DOM) DOMDocument shouldn't ever be "greedy" over what it matches, like REGEX unintentionally tends to do Detriments of DOMDocument: May require valid HTML, but with iterations of HTML, what exactly is considered valid? Would different versions of PHP handle the DOM differently with version differences? Potential of implementation changes. loadHTML() may modify your source - what goes in might not be what comes out Character encodings may cause unforeseen issues (don't they always!) Without a good knowledge of PHP's approach to using DOMDocument, the code process can get rather difficult to understand if changes/updates are required Some further reading from someone else with more thorough testing: https://blog.futtta.be/2014/05/01/php-html-parsing-performance-shootout-regex-vs-dom/ https://blog.futtta.be/2014/04/17/some-html-dom-parsing-gotchas-in-phps-domdocument/ Realistically it's a judgment call. Speed and server efficiency versus (one would hope) better valid modifications/detections. I don't think there's really a right or wrong solution. Some shared hosting servers don't install the DOMDocument PHP extension by default though, so you'd want to check for the existence of the function during your module's install method. P.S. - Thanks for asking the question -- I knew DOMDocument was slower, but haven't compared in awhile. The articles I saw above were an interesting read. πŸ™‚
  13. Is there any established or suggested way of how to handle providing modules that offer functionality from different providers? Example: Let's say the Google Maps and Leaflet Maps modules didn't already exist. Someone thought to create a module, but realized future clients may want one provider over another. Because they're so functionally similar, created a base module that then offers classes (GoogleMap / LeafletMap) to extend the base (Map) class - which can be chosen either in the module configuration or in the field configuration. I've seen a few different modules that offer this functionality over the years - most recently Ryan's Tfa (Two-Factor Authentication) class. Ryan's WireMail and Tfa base classes have a distinct benefit of being part of the core and therefore don't need to be installed or initialized - only the modules that extend them do, so I'm not entirely sure if custom modules requiring the installation of a base module, and then offering extension modules is the best solution. Other solutions are to add folders (or classes) containing required module extensions that will be discoverable by the main module. I think I saw one that used a JSON configuration at one point in my research too. Any thoughts from others who might've handled this for their own custom modules - or have been thinking about this recently?
  14. I appreciate the further glance - though it's most definitely in an unfinished state. When I couldn't get a database save to work I panicked a bit and rewrote some areas (not necessarily finishing them). That said, it's good to make me think about it again - I might've forgotten to go back and fix it up. I still want to provide a few data provider options (manual entry, OpenLibrary, Polaris ILS, etc) through module configuration; need to fix up the interface and figure out accessibility/internationalization concerns, then do some further testing. There's still quite a bit to do yet (I'm primarily stumped at the UI, have a few possible directions I can go), but at least now I can happily move forward! EDIT: Confused myself. Sanitization is currently being done in OpenLibraryBook::set() -- though it'll be a little different once I add various data providers for the book information. I tried checking your website for an Amazon Wishlist or Paypal link (i.e.: "Did I help you? Consider buying me a coffee."). If you want some coffee money, PM me a wishlist or Paypal email. πŸ˜‰
  15. I was honestly expecting a facepalm (missing $, semi-colon, or maybe typehint). Thank you so, so, so, so much!
  16. I began writing this module from scratch, basing it off simple examples of modules, trying to learn the necessary methods required to create a fieldtype. I was WAY off, but learned a lot. Eventually I moved towards examining the FieldtypeEvents example from Ryan and, in many areas, copy/pasting and then adjusting the code (and then trying to determine what it all was doing) method by method, file by file (from his example to my module). I've gotten to the point where I was ready to test the code - I haven't fully fleshed out my visual interface, nor integrated any of the 3rd party API tools this will (can) take advantage of. I also haven't implemented any JS or CSS for the interface. Right now I'm just trying to get it to save and then display any of the data that has been entered into the form fields of a template from the admin. When I save the page, the field is reset to defaults and no feedback message (for the field) is shown at the top of the page upon reload. When using Tracy Debugger from within the processInput method (immediately attempting to debug the $input variable), Tracy caught no data to be shown. If anyone has some time to take a quick look and see what I might've done wrong I'd be super grateful. I've had no forward progress in about 3 days (and like many others was unable to get Xdebug breakpoints working in PHPStorm). 😞 https://github.com/BrendonKoz/FieldtypeBooks
  17. Is it dynamically creating a custom template on-the-fly? Just quickly thinking how it works, though it doesn't really matter. Quite slick indeed.
  18. The purpose of the module is to provide a way to display Leaflet Maps. Leaflet =/= OpenStreetMap. Leaflet is a map display library whereas OpenStreetMaps is the actual map layer data source. OSM can be displayed using Leaflet (popular), OpenLayers, or even Google's map library, so I don't know (for sure) if the purpose of the module was specifically to avoid using Google's technology behind the scenes (which it does anyway as is evidenced from previous discussions of the code). I prefer OpenStreetMaps' imagery and ToS, but prefer Google's geocoder. Like I said though, your fork of the code is your own -- it did not have to follow my personal ideas.
  19. I took a look at it. This is not the same thing that I was trying to accomplish (not that it has to be). These alterations don't use the Google geocoder's returned JSON data, it uses OpenStreetMap. I don't remember how, but I do remember I was able to get the JSON results of the OpenStreetMap geocoded data without any modifications to the module. What I was unable to get was the same values (100% of the time) from Google's geocoder, which contains more information (and often, I've found, more properly formatted and correct) than the OpenStreetMaps version. Because the found addresses for PLACES don't always match ("NYS Public Library" might not list a street address on OpenStreetMap, but Google will include everything), doing a manual lookup with Google's geocoder after-the-fact won't work 100% of the time. (Those were some of my reasons anyway.) As for code: When changing a version number that has companion modules you'd probably want them all to match. Just a suggestion for any future possible patches. LeafletMapMarker.php around line 62: } else if($key == 'raw') { $value = $value; } At first I did that too in my version (I think I removed it in my repo, maybe not) but that's just unnecessary. Still looking through the rest as like I said, I don't fully grasp FieldtypeModule code.
  20. Congratulations on your success. If you're going to make a pull request I'd make it to the original Fieldtype module as that was the only reason I had forked the original module's code to begin with - to have workable code to store the raw return data, and offer a possible pull request.
  21. If the only JSON+LD to be displayed will be event type, I'm personally planning to just hardcode it into the template rather than run another module. I did install that module originally with that intention but it mostly just attempts to simplify the creation of the template code. My `raw` field was used so that the rest of the module's code would be untouched and would continue to run as-is. If I could get it working I'd then add a module config option to offer to always save the raw data (or leave unchecked to not save it; default behavior). That way if the module author wanted to merge changes it wouldn't have any breaking changes. The schema upgrade method would have to be updated though. I didn't notice the other variables that may have worked before -- or maybe I did; if I did, they might've been a carry-over from the fork of Ryan Cramer's GoogleMaps implementation (some code is still the same). I didn't get time to look at this today so the code is no longer fresh in my memory.
  22. I hadn't. I had previously installed TracyDebugger and couldn't figure out how to use it -- I think my problem was as simple as having installed it in the wrong PW instance (it's working now). I'll see if I can find any additional info from that, thank you for the tip! I had not, unfortunately. I was pulled from that task to work on an Omeka (archival CMS of sorts) theme and subsequent issues, and an application form using GIS shapefiles. I hope to be finishing the Omeka thing up today. If I have time later in the day I'll take another look into this. I too don't have much experience working with PW Fieldset modules; I've been trying to figure them out here and there though.
  23. I realize my question is now outside the scope of this module, so I apologize in advance, and also have no expectation of a response -- just hopeful! By looking through this though I think I might be picking up some things with Fieldtype development that I couldn't quite grasp beforehand, so it's still useful (to me). Thank you for the response, @gebeer that was awesome! I made those, and some additional changes, but I've still been unable to get the data to save to the database (I'm instead saving an empty string value). I believe I might need some additional code somewhere, possibly in InputfieldLeafletMapMarker.module's ___processInput method? I tried commenting out the if/else block for whether a JS geocode had been done so that it would (should?) always perform a Google geocode, but that still didn't save to the DB. Any thoughts from anyone on what I might be missing here? I manually altered the database table to include the "raw" column in my field_{$name} field values. Changes (on Github repo removed) : LeafletMapMarker.php [line 38] + $this->set('raw', ''); [line 96] + $this->raw = $json['results'][0]; InputfieldLeafletMapMarker.module [line 310] - "\$page->{$this->name}->zoom" . + "\$page->{$this->name}->zoom\n" . + "\$page->{$this->name}->raw" . FieldtypeLeafletMapMarker.module [line 106] + $marker->raw = $value['raw']; [line 136] - 'zoom' => $marker->zoom/*, + 'zoom' => $marker->zoom, + 'raw' => $marker->raw/*, [line 157] + $schema['raw'] = "TEXT"; // raw google geocode data
  24. Does this module use Google's API at some point in the geocoding process? I noticed that there are calls to: http://maps.googleapis.com/maps/api/geocode/json?sensor=false&address= ...in LeafletMapMarker.php, and calls to... http://nominatim.openstreetmap.org ...in Control.Geocoder.js. I am using this Fieldtype on individual Events pages - it works well and made sense. I'm now trying to go back and add JSON+LD to all of those Events pages which requires a "Place" value (Example: https://developers.google.com/search/docs/data-types/events). Unfortunately the auto-corrected address field from the geocoder ($this->address in Leaflet) is not always something I can easily split into meaningful entities for the JSON+LD Event type's @Place type values. That said, the raw results originally returned from either geocoder (Google's being nicer in this regard, but OSM works too) can be used towards building this out. Based on how this module handles all of its data, is it feasible to extend it somehow to allow saving of the raw geocode data? If so, any ideas on where I might begin to look?
  25. I'd have never looked there. In fact, I know I didn't since I searched for the text string of "permi" in my *.php/*.module files. Dang. Thank you so much! Now I can fix all the other errors and continue on. Thank you!!!!
  • Create New...