Mingwei

@alfredpanda the latest version (updated in the main post) will automatically switch icons to match dark/light mode and adds a user setting for disabling notifications @davidraviv I think? It's a macOS version issue? Realised I've been compiling the previous versions for macOS 13 and above. I think it should work on your Monterey system now? Fingers crossed! I would ask for more info from you but the error log is pointing at something so basic (it can't find a built-in framework) that I don't know what else could be the issue. @c0desurfer Do let me know what device name you're trying to filter out so I can test it on my end

@davidraviv That error message means there's probably some sort of issue with how frameworks are linked to the compiled binary. It's strange because I can't reproduce the error on my end. I'll play around with it and see if I can figure it out, though this might take a while as I'm busy with other stuff right now, and I'll probably need you to help me test it out. @alfredpanda very valid suggestions. I personally like the notifications, but you can easily disable them on your end by deleting the Post Notification action in the workflow (and deleting all lines after the first line in the Run Script action if you have terminal-notifier installed, but I'm guessing you don't). Realised I made icons that only really work in dark mode after releasing it haha, I've been meaning to do something about that for a while but haven't had the time. In the meantime you can change the icons by opening the workflow folder and replacing all icon images with identically named files of your choosing. I got those icons from SF Symbols so you could do the same to create light mode compatible versions.

UPDATE: Finally figured out what the actual problems are, and it's not what I described above (oops!) Documenting it here in case anyone else experiences similar issues: Alfred isn't able to set variables with names that start with a digit, but variable names can contain digits. This might be a system limitation? Added an underscore in front of each numeric variable name, now they can be set without issue. There's no hard limit on environment variable quantities (or at least I didn't hit it). I observed the workflow working when there were less than ~300 items and failing when there were more because I happened to be testing quantities by toggling between showing all processes (which requires running the script filter from root) and showing only user processes. Of course, Alfred is only able to set environment variables in the user scope, which of course were not in my environment when running as root. In retrospect, 2. at least should have been obvious. But I guess I needed to think out loud (and very publicly) on this forum to figure it out haha 😅

Just tried something else, and it seems Alfred might not be able to set variables with numeric names? I created a test workflow that sets a variable with a numeric name (e.g. 12345) using the Arg and Vars utility and tried to output it via Run Script. I received an empty output, just like in the example above: [15:00:39.790] numeric varname test[Keyword] Processing complete [15:00:39.790] numeric varname test[Keyword] Passing output '' to Arg and Vars [15:00:39.794] numeric varname test[Arg and Vars] Processing complete [15:00:39.797] numeric varname test[Arg and Vars] Passing output '' to Run Script [15:00:39.868] numeric varname test[Run Script] Processing complete [15:00:39.886] numeric varname test[Run Script] Passing output '' to Large Type But when I set the variable in the Run Script action itself, as in `12345=hello` or `12345=37283` it works as expected. I then tried making the names of my dynamically generated variables in the original workflow non-numeric, by using process names instead of PIDs. After doing this, I had no issues retrieving the environment variables and calculating cpu usage. This isn't really a solution though, because process names don't identify their processes uniquely and this throws off the calculation whenever there are multiple processes with the same name. I'm also still not able to retrieve environment variables when their number exceeds around 300+, so I think it's likely the problem I was facing was caused by a combination of too many variables & Alfred not being able to set variables with numeric names. Really sorry for the long posts - I realise this is a really niche problem that won't affect most use cases, so I don't want to take up too much time. Just want to document this here, and if I'm not missing any obvious alternate causes / there are no obvious workarounds I'm happy to put this workflow aside.

I'm trying to generate a refreshing list of system processes organised by cpu usage percentage. `ps` would be an obvious way to do this, but I think the cpu% figure it outputs represents cpu time used over the entire runtime of the process. In contrast, I want to get usage time over a short sample period (the rerun interval) so that sudden spikes in cpu usage are reflected, much like in Activity Monitor. To do this I have to calculate the amount of cpu time used by each process between script filter reruns, divided by the rerun interval. At the moment I'm trying to do so by saving the amount of cpu time used by each process in a variable on each script filter run. Since this means one variable for each process, it easily adds up to hundreds of variables. As for why... no particular reason haha. I'm doing this for fun, and to learn more about how system stats are calculated and how APIs like libproc or kinfo_proc work. Likely won't end up using the resulting workflow much, though I can imagine using it in place of activity monitor to kill an unresponsive process if it's faster (so far it is). It's ok if it's not possible, but if there's an alternative way to do this / a mistake I've not found that would be great. I'm not using the variables in a subsequent workflow action, just in subsequent reruns of the script filter. The key is each process's PID, I just retrieve the environment variables in a loop. The swift code retrieving the variables is something like `guard let cpuTime: String = ProcessInfo.processInfo.environment[pid] else { return }` but I'm quite sure it's not the problem, have used it without issue elsewhere. I tried echoing the variables in a Run Script action to see what would happen. I ran `echo ${(P)pid}`, where pid is a variable created by the actioned script filter item containing the process's pid (which is in turn the name of a variable containing the process's cpu usage time). The output in the debugger log seems to vary depending on how many items there are. When there's less than 300 items (roughly, not entirely sure where the boundary is), I get this: [13:51:10.955] Search Processes[Script Filter] Processing complete [13:51:11.021] Search Processes[Script Filter] Passing output '/path/to/some/process' to Run Script [13:51:11.028] Search Processes[Run Script] Processing complete [13:51:11.031] Search Processes[Run Script] Passing output ' ' to Post Notification When there's more, I get this: 13:50:36.827] Search Processes[Script Filter] Processing complete [13:50:36.912] Search Processes[Script Filter] Passing output '/path/to/some/process' to Run Script [13:50:36.914] Search Processes[Run Script] Processing complete [13:50:36.916] Search Processes[Run Script] Passing output '/path/to/some/process ' to Post Notification Not really sure how to interpret either output, but the fact that it just passes the item arg along when the number of variables exceeds a certain number is what made me think the problem might be too many variables.

I'm trying to create a script filter that dynamically generates workflow environment variables on each rerun. The number of variables can get quite large, typically around 300+. I've not been able to call the variables after passing them, but I'm quite sure my script filter JSON is formatted correctly (see below for an example). Is the problem the number of environment variables I'm trying to set? Related to this: Am I right to assume that previously set environment variables are not automatically unset when I pass a new set of variables that do not include them on a subsequent rerun? Example JSON: { "rerun": 2, "variables": { "71114": 369018953, "66776": 7091547666, "528": 1188619262847, "654": 9873755229, "71680": 69060795, "70589": 289630428, "70712": 26009830573, "589": 978183848817, "625": 261194160090, "2390": 12221103447, "71065": 624306639048, "67866": 1229316369, "70427": 9739510787, "752": 252220960370, "71585": 147073393, "64861": 18919027172, "175": 336645919469, "67750": 439291658, "2391": 3801114097862, "71115": 35413996, "688": 39562066662, "67610": 561323856, "70713": 5900752114, "56812": 16642660773, "59248": 791372000125, "754": 102640709518, "629": 907869918301, "69220": 22712486493, "70758": 9072044678, "67609": 257898324, "727": 297295601577, "67611": 4603671846, "64695": 10199393455, "64700": 545706904869, "25863": 40562974624, "728": 9128981472, "71067": 28075681, "2394": 120283103934, "70547": 12186120352, "729": 63155527498, "71163": 33315621, "71281": 132950770, "25696": 22579543291, "67634": 756496911, "67159": 35604950316, "1274": 769329740, "66261": 14347537664, "63774": 1617932835, "70358": 165023076, "71164": 13229162, "67847": 6072538920, "71070": 48162710, "71120": 1914524805, "541": 184061231008, "64675": 2282215683691, "2398": 2769403599, "70595": 36009674597, "542": 207137122531, "50660": 11896548786, "71589": 1690818387, "67848": 112795628765, "71071": 239428198, "71237": 59626195, "2386": 838073975103, "67708": 6421800433, "2399": 73789359479, "71591": 480641121, "544": 137084210155, "71121": 185427787, "640": 62987972683, "68179": 25875047786, "70907": 59574734369, "610": 48842444746, "1309": 1061183502, "2374": 50959080591, "575": 1668668864963, "64252": 62679400156, "2405": 176057209062, "641": 12325618751, "6699": 63826169024, "71451": 8527346476, "546": 935556953066, "66715": 10089254807, "71072": 13750250, "56791": 74720968651, "70316": 6929901282, "2406": 34243853750, "67615": 149195016, "71592": 2192282125, "57433": 1030840303702, "578": 2953770018714, "67588": 650618255, "67923": 659736294, "7254": 403858472, "64677": 425511208300, "2389": 450048076198, "579": 2044050672972, "2407": 8079028895, "549": 15542068733, "71100": 566914210552, "63615": 254528580195, "59327": 61344550069, "65247": 1472468698, "67638": 603198686, "70647": 128110357, "67662": 307527307, "67616": 2862886462, "71593": 118739910, "67852": 343303235, "2378": 1382801477292, "69226": 150935026, "70598": 92359668, "4053": 198430500, "12775": 14241542463, "12021": 67316061669, "67685": 753747513, "66505": 5475011188, "67969": 27296507869, "745": 22203257798, "68349": 652152798, "71074": 129132667, "66953": 5141514814, "70342": 7380274566, "67617": 643576734, "71594": 160901100, "67947": 35562773756, "57435": 450255868, "71690": 57181002, "717": 223271404436, "64679": 31100962480, "70722": 1277240126, "71097": 22322585589, "71102": 115375607, "71220": 493987411, "2520": 13832749697, "59375": 49924713364, "66670": 22680053688, "56794": 30070275434, "64681": 31241928139, "67876": 56835538817, "70555": 318967337, "71053": 4427853611, "942": 172311100438, "71691": 103594999, "70723": 740746034, "67904": 2322232363, "2522": 59046062592, "56795": 10531107289, "12519": 3292170848025, "71623": 134388698, "2548": 1083687076, "65703": 24735747490, "71054": 648608231, "70628": 11969658073, "68353": 1206769001, "976": 13021255813, "560": 23958309655, "2671": 10835283022, "70724": 471055731, "71104": 26813722, "67951": 34219569512, "71099": 4113173217, "64799": 13349943687, "977": 114131122672, "561": 77949149767, "71077": 32713393, "592": 179761929082, "59261": 2133549421, "978": 566097917453, "532": 214254747560, "71055": 83164588, "70629": 90286787, "979": 25269174898, "563": 507513574471, "27461": 6297770146666, "1239": 402998372611, "67622": 4061394924, "564": 141107040468, "71647": 1860042410, "68332": 22486559545, "57440": 110731840109, "67689": 4461036686, "2513": 10727910360, "68399": 598754046, "56797": 5018793480175, "56802": 5734847990, "71625": 144410122, "1769": 28202523634, "15574": 9883062488, "630": 187675748352, "71078": 1054732296, "66627": 8133864114, "70558": 78290879, "67031": 34239921497, "68377": 77729452, "71603": 130996739, "660": 229897277936, "662": 73256705970, "67645": 2044431538, "632": 233805492309, "71080": 83755331, "25924": 68577684607, "70324": 1366048484, "567": 352262973683, "71576": 14309917084, "12713": 3152884701, "57439": 8381979949, "70726": 499825491, "71648": 272559198, "71224": 3061246636, "3458": 1276086185188, "694": 7823815633, "538": 731824456436, "2528": 4143702353336, "66580": 27989508341, "66628": 19311763137, "730": 51429669296, "604": 27340695259, "695": 116833096002, "59263": 1538454473, "67998": 25408810946, "71650": 102437227, "68501": 697833331, "71604": 121105492, "605": 93039320385, "539": 745823049035, "2516": 27843511136, "636": 93843970776, "67670": 13753502157, "70985": 171640800, "12714": 20042237167, "697": 23255572020, "70727": 527869860, "71649": 65015879, "67200": 9899328801, "637": 812864986587, "67624": 1050923313, "67597": 3737231881, "67602": 58622464562, "15576": 407223717, "67033": 407547170095, "71605": 123027526, "609": 26768313813, "66843": 6450138665, "67671": 1479549479, "71272": 3575028719, "71060": 61646917, "70728": 37986537, "70304": 6049005613, "67603": 14329354609, "15577": 4333511684, "70400": 5698458297, "59265": 1127879779, "71606": 699880578, "71059": 3658356876, "67648": 1329105859, "71083": 172683881, "67626": 1208566426, "67956": 14322885874, "25594": 103358736970, "67650": 377107965, "2508": 190516072884, "70541": 22560196613, "3395": 21517105909, "67599": 4592604708, "71607": 11859024291, "25880": 4770156809, "71301": 41804323956, "2509": 19051744558, "71084": 27626525, "68147": 41350042330, "4726": 8508175253, "26002": 133312380803, "1332": 1025289560, "67957": 32476690680, "71062": 21024491, "70518": 135032201, "67651": 84109249, "57445": 879929403, "70542": 3710428168, "25879": 978573829190, "580": 14929076147, "67605": 313814810, "66160": 28226804188, "71608": 314287592, "64691": 4644953039, "582": 2350107662660, "70989": 22832365, "2455": 895469909, "456": 758874992392, "70849": 454783610, "67864": 121632363294, "57446": 111295263660, "553": 114455820583, "939": 39946568444, "523": 374059355049, "66585": 9168789486, "7334": 40670443948, "70403": 7841795359, "67630": 24344729633, "71609": 100989871, "585": 147045994983, "70588": 3997793068, "70711": 93424997, "71633": 396749651, "621": 375226396400, "64858": 1258026862, "586": 424583970545, "67251": 6459717519, "556": 69946306748, "64764": 48049491043, "622": 2564520656895, "70544": 113695255, "67607": 1919992603, "623": 284508299880 }, "items": [...] }

right... should have thought of that 🤦‍♂️. quite new to this whole developing stuff for public use thing. Will definitely work on shipping a script instead then

@vitor Thanks for the detailed explanation! Was going to say I'll keep these tricks in mind for future use - but then realised I'll most likely just end up stealing icons from the extensive collection in your workflow 😅

Will definitely look into that, good point. Would users then have to install xcode command line tools as a pre-requisite though? Did a quick google on how /usr/bin/swift interacts with uncompiled scripts because I wasn't sure and that seems to be the case (example). If that's the case I'd rather include some screenshots / link to the apple support page on getting past the unidentified developer warning. From a security standpoint the difference is minimal I think, given that in both cases the source code can be easily viewed. Other considerations: Would stuff like relative paths get messy if the compiled binary is stored in a temp directory somewhere rather than the workflow directory? And what's the lifespan of the compiled binary - could affect performance, defeating the purpose of using a compiled language. Will try and see how it goes.

@vitor Where/how did you obtain the original icons? I'm so impressed. Tried looking inside system settings.app etc. and still have no idea how you extracted them.

Happy to say mine (the latter one) no longer requires installing additional software, as it now directly uses macos' coreaudio api. Latest version is tested on Ventura

@xgo could you install this pre-release and let me know if it works? Thanks for the tip on airplay! I don't use airplay either so I'd have trouble testing stuff, but I want to try replacing switchaudio-osx with macOS APIs at some point so I might be able to look into it then.

@xgo Error code 2 in the log means the script filter couldn't parse the json coming from switchaudio-osx. I think I've identified the problem, but could I see the output you receive from switchaudio-osx to be sure? Run this in your terminal and paste the output here: switchaudiosource -af json If you have some time, do you mind also downloading v1.0.2 of the workflow and trying that to see if it works? I made some changes in the new version unrelated to the new blocklist feature that could be causing the problem.

@Floating.Point will look into it, should be easy to add.

6 years late, but I just came across this workflow and I really like how well-designed it is, so I've submitted a pull request to add the kill-and-relaunch feature. Unfortunately, I couldn't find any way to do it without changing the output format from XML to JSON (see relevant bug report), which meant dropping support for Alfred 2. The workflow can be downloaded here.

Was editing an old workflow made for Alfred 2 when I encountered this bug. I definitely don't expect this to be fixed since the XML format has long been deprecated, but I figured I'd report it anyway just in case, since it could break a number of old workflows. Alfred version: 4.6 [1266] macOS version: 12.0.1 Replicated it using the example XML from Alfred's support pages. Only 'fruit' is passed on, while 'vegetable' is left out: [00:43:49.179] Logging Started... [00:43:52.017] Advanced Script Filters - Getting Started[Script Filter] Queuing argument '' [00:43:52.094] Advanced Script Filters - Getting Started[Script Filter] Script with argv '' finished [00:43:52.101] Advanced Script Filters - Getting Started[Script Filter] <?xml version="1.0"?> <output> <variables> <variable name="fruit">banana</variable> <variable name="vegetable">carrot</variable> </variables> <items> <item arg="Placeholder item"> <title>Placeholder item</title> </item> </items> </output> [00:43:52.918] Advanced Script Filters - Getting Started[Script Filter] Processing complete [00:43:52.919] Advanced Script Filters - Getting Started[Script Filter] Passing output 'Placeholder item' to Debug [00:43:52.921] Advanced Script Filters - Getting Started[Debug] Output from Script: 'Placeholder item' Variables: { fruit = "banana" }

Thanks for the tip! Will update it

There are probably quite a few sound device selector script filters out there, but the ones I've come across are mostly written in applescript. I rewrote my own in C++ Swift and it runs much faster, so I'm sharing it here in case anyone's interested. Features Directly uses system APIs to list and select sound devices, which makes it very fast and reliable No dependencies - no additional installs required Optionally add devices to a blocklist to hide them from results Optionally disable device selected notifications Installation note On initial run, you'll see a pop-up window that says something like: "sound_filter" / "sound_selector" cannot be opened because the developer cannot be verified. This is a macOS security feature that blocks execution of non-codesigned binaries You're seeing it because this workflow uses pre-compiled binaries rather than interpreted code (e.g. Python) I can't codesign those binaries because I'd need an Apple developer account and that costs money The security risk is that you have to trust that I'm not doing malicious stuff with the binaries (I'm not) (This workflow is open-source and you can compile the binaries yourself, in which case all this is moot, but that's very inconvenient) You have to bypass this security feature to use this workflow. To do so, follow the instructions here. Changelog v2.1.0 Fixed: Icon visibility in light mode (icons automatically change with appearance settings) New: Prettier notifications no longer require terminal-notifier v2.0.0 New: `switchaudio-osx` no longer required New: Significantly faster loading speeds New: Icons now indicate whether a device is selected New: Makes use of Alfred 5 user configuration panel to simplify setup v1.1.1 Fixed: Cannot parse switchaudio-osx JSON in some cases v1.1.0 Added: User-configurable device blocklist (see feature request) Fixed: Issue with user-set path for switchaudio-osx v1.0.2 Minor performance optimisations v1.0.1 Added: In most cases, users no longer need to specify the path of switchaudio-osx (see relevant comment) Fixed: Missing icon images in terminal-notifier notification Download | Source Screenshots

(Just a quick note in case anyone's experiencing a similar problem) The latest update to the Adguard app (2.6.0.1041) seems to be affecting search suggestions. I'm getting the error 'certificate signed by unknown authority' for both Google and Wikipedia searches (haven't tried others). Worked around this by adding an exception for Alfred in Adguard preferences (Preferences > Network > Applications... > uncheck Alfred from the 'Filtered Applications' list).

TLDR: Use this BetterTouchTool preset to display touch bar buttons for triggering Alfred workflows when the Alfred search bar is activated with a keyboard shortcut. Inspired by: Here's what it looks like (workflows not included): It works by: Using BTT to setup a keyboard shortcut that simultaneously activates the Alfred search bar (via applescript) and a touch bar group Setting up an external applescript trigger for the workflows in Alfred and creating buttons for triggering those workflows in the touch bar group Using BTT to create an escape key that also closes BTT touch bar groups It works this way because there's no way to detect the Alfred search bar itself. This does cause some problems: This wouldn't work for you if you don't want the escape key to close groups. A possible workaround is to have the Alfred workflows group to close on a timer instead (see: https://community.folivora.ai/t/auto-closing-touch-bar-groups/5935/3) Left clicking anywhere outside the Alfred search bar when it is open closes the search bar, but not the workflows touch bar group. On the whole though I find that it works quite smoothly. Installation: Requires BetterTouchTool, which costs $6.50 after a 45 day trial (https://folivora.ai/downloads) BTT preset: https://share.folivora.ai/sharedPreset/918335e2-5596-4e0d-a1cb-388e568f8cd0 Example Alfred workflow: http://www.packal.org/workflow/alfred-bettertouchtool-integration