Vibe coding a python script to create a plain text file with my macOS Photos (Aperture) folder hierarchy

My single greatest Photos.app frustration (I have many) is the inability to search the folder hierarchy. Mine contains hundreds semantically important folder names where the hierarchy is also meaningful. Meaning lost in the catastrophic Aperture to Photos migraiton.

This morning I had an hour free so I asked an ai about available utilities and workarounds. It said there are really no good options, but the Python osxphotos module might be able to traverse the folder hierarchy.

I have dabbled in minor Python coding and I have a half-baked Visual Studio Code environment. So I asked Claude 4.5 in Perplexity (this is not a formal supported coding environment) to write me a script that would use osxphotos to build a text file representation of the hierarchy. I ran whatever it generated.

It took 4-5 tries. I never edited the code myself. The first time there were copious errors, I describe errors and requested a redo. The next two times there were fewer errors, but I only got the top level of the hierarchy. The ai added debug code. It took two more tries of running and reporting errors to get a script that generated the text file I wanted (example):

[Teams and Orgs / MN Special Hockey / MNSH 2006 pre-season] MNSpecialHockey_060317
[Teams and Orgs / MN Special Hockey / MNSH 2022-2023] MSH 2023 Printed
[Teams and Orgs / MN Special Hockey / MNSH Woodbury 2019-2020] MSH Portraits Jan 2020
[Teams and Orgs / MN Special Hockey / MNSH 2008-2009] Nov 2008 MN SH Section 108 Event
[Teams and Orgs / MN Special Hockey / MNSH 2021-2022] Portraits MNSH Woodbury 2022

This is most personally valuable code I have ever "produced" since my days of writing the "medtrans" C program to turn 1990s MEDLINE output into tab delimited importable text.

And I wrote none of it.

I'll be cleaning it up and refining it, but below is the code I have today. It also included album names within a containing folder - I didn't want that but now I find it useful so I'll leave it.

Code

#!/usr/bin/env python3

"""

Export Photos folder and album hierarchy to a text file with tab indentation.

Requires: pip install osxphotos

"""

import osxphotos

from pathlib import Path

def export_folder_hierarchy(output_file="photos_folders.txt"):

"""

Export Photos library folder/album structure to a text file.

Args:

output_file: Path to output text file (default: photos_folders.txt)

"""

# Initialize connection to Photos library

photosdb = osxphotos.PhotosDB()

# Get folders and albums

folders = photosdb.folder_info

albums = photosdb.album_info

print(f"Found {len(folders)} folders and {len(albums)} albums")

# Build maps for folders

folder_map = {f.uuid: {'obj': f, 'children': [], 'type': 'folder'} for f in folders}

# Add albums to the structure

for album in albums:

# Albums have folder_names property which is a list of folder names in the path

if hasattr(album, 'folder_names') and album.folder_names:

# Try to find the parent folder by matching folder names

# folder_names is a list like ['Top Folder', 'Sub Folder']

# We want to match the first (top-level) folder name

top_folder_name = album.folder_names[0] if album.folder_names else None

if top_folder_name:

# Find folder with matching title

parent_folder = None

for folder in folders:

if folder.title == top_folder_name:

parent_folder = folder

break

if parent_folder and parent_folder.uuid in folder_map:

folder_map[parent_folder.uuid]['children'].append({

'obj': album,

'type': 'album',

'title': album.title,

'folder_path': ' / '.join(album.folder_names) if album.folder_names else ''

})

# Recursive function to write hierarchy

def write_item(f, item_data, level=0):

"""Write item and its children with proper indentation."""

indent = '\t' * level

if item_data['type'] == 'folder':

folder = item_data['obj']

f.write(f"{indent}{folder.title}/\n")

# Sort children alphabetically

children = sorted(item_data['children'], key=lambda x: x.get('title', x.get('obj').title).lower())

for child in children:

if child['type'] == 'album':

# Write album with its folder path if nested

folder_path = child.get('folder_path', '')

if folder_path:

f.write(f"{indent}\t[{folder_path}] {child['title']}\n")

else:

f.write(f"{indent}\t{child['title']}\n")

else:

write_item(f, child, level + 1)

# Find root folders

root_folders = [folder_map[f.uuid] for f in folders if f.parent is None]

root_folders = sorted(root_folders, key=lambda x: x['obj'].title.lower())

# Write to output file

with open(output_file, 'w', encoding='utf-8') as f:

f.write("Photos Library Folder Hierarchy\n")

f.write("=" * 50 + "\n\n")

for root in root_folders:

write_item(f, root)

print(f"Folder hierarchy exported to: {Path(output_file).absolute()}")

print(f"Total top-level folders: {len(root_folders)}")

if __name__ == "__main__":

export_folder_hierarchy()

Apple's ai opportunity is context

I use Perplexity as my $20/m answer engine -- and for generic ai tasks. I don't want Perplexity as a longterm ai provider, but I do like being able to experiment with different models. Currently all the leading not-free models are pretty good, but some are more sycophantic than others. I dislike sycophancy; ChatGPT and Sonnet have less of it than Gemini but all are too eager to please (prompts help but one has to be careful not to reveal a preference for a particular response).

For me, at this time, all of the models work significantly better with quite a bit of context. In Perplexity that context is provided in Spaces. Spaces include reference material and the model/prompt settings for the Space.

The great mass of people are not going to do that sort of context work. So vendors are trying to answer questions and apply (lower cost) models without context. Meanwhile they try to scrape together a lot of knowledge about the user from whatever source they get.

Apple's opportunity is they can assemble a lot of context. In my case GBs of information on my main drive, not to mention my calendars, contacts, notes and so on. Apple could ask questions to provide a general default context, such as preference for sycophancy, references to use, web resources, textbooks and so on. 

I consider Apple to be a broken company. I don't think they will be able to get their ai act together under Cook. But if they can, they do have advantages.

Tip: Let your ai tell you what's new and novel in an iOS or macOS release

I like to wait a month (iOS) or six months (macOS) before applying major updates. By the time I apply them all the useful tips and tricks I read along the way are ancient history.

Instead of trying to keep track of these things before the OS is installed wait until you are ready to pull the trigger. Then ask your ($20/m) ai to summarize known issues and interesting new features, tips and tricks. You can provide context as needed (ex: I am an expert user, etc).

PS. Apple got away from providing PDF versions of manuals and user guides -- but if they still did that I'd drop the PDFs into my Perplexity macOS Space.

New lessons on macOS network home backup: WiFi issues and so many bugs

Recently we had a flurry of home computing issues. They feel typical of our time where multiple bugs interact across multiple environments.

In this case there were 3-4 different bugs and failures that impacted our home network and especially backup. The lessons learned were not in my ai responses so they seem worth sharing. The interacting bugs included:

1. Apple hardware bugs in the M1 Air that can make the Air WiFi unreliable. (Could be drivers/software but persistence suggests either a very hard problem or hardware with possible software mitigation.)
2. The odd behavior of Eero wifi
3. Bugs and limitations in macOS SMB networking and in HPFS mean network share filesystems can be corrupted beyond repair.
4. A completely unrelated red herring that turned out to be due to CenturyLink's parent org messing up their DNS configuration.

Such is the nature of our times, where complexity and unsustainable share prices combine to decrease reliability of core systems.

At the core was WiFi instability. Our M1 Air WiFi was constantly fluctuating, leading to constant disconnects. I knew there was something wrong as Carbon Copy Cloner would quite often warn that a backup was being transiently disrupted by the network share disappearing. The very frequent write/read failures presumably led to the HPFS hard drive file system corruption. I had to diagnoses and fix the drive connection before focusing on the underlying WiFi issue.

Things I am doing differently now:

1. Reconfigured physical layout of our Eero base stations so there was a direct "line of sight" short distance between the M1 Air dock location and the Eero base station. The M1 Air needs a much stronger than usual WiFi connection to be stable. The new configuration also offloads some traffic from an overloaded Eero device.
2. We use a Synology Time Machine server as a secondary (not robust) backup. That backup was also corrupted (happens normally anyway -- because bugs, but WiFi issues sped it up). At least in Sequoia if you remove a Synology TM backup destination and then add it back there's an option to replace the original. This is faster than wiping it from the Synology side.
3. I used advanced preferences so Carbon Copy Cloner will dismount the network share after a clone/backup is compete. The less that share is open the better because it's hosted from a MacBook Pro that can be disconnected from the network, and macOS/SMB does not handle that disconnect gracefully.
4. I had configured a user quota for one of our machines that had become too small. Time Machine should have provided guidance about capacity but did not do so in a useful way.

Related

• 2015 - original post on Synology Time Machine
• 2018 - Synology user quota
• 2022 - Eero 6 review

Bug in iOS 18.5 eSIM removal - you have to let it update your contacts

When Tello failed to transfer my CenturyLink number I canceled their service. I then tried to remove their eSIM from my phone. Apple's instructions worked -- but the line name remained even with a deleted eSIM.

It's a bug.

When Apple asks if you want to update any contacts using the number associated with the removed eSIM you have to say yes -- even if there are none. Looks like a simple logic bug. Then the entry will be removed.

Create a macOS desktop shortcut (.inetloc) that deep links to a specific note in Notes.app

Writing this up quickly, I might get back to it later.

I want to have a shortcut to a specific macOS Notes.app note that I can put in a folder.

1. In the note I want to link to hit ⌘K. Search on name of the current note.

2. Do NOT use the note title option. Unclick that and provide your own title. Create link in note.

3. A link is created in the note. Click on that that and then drag it to the desktop. An inetloc file with a deep link is created.

This is what the .inetloc shortcut has in it:

<?xml version="1.0" encoding="UTF-8"?>

<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">

<plist version="1.0">

<dict>

<key>URL</key>

<string>applenotes:note/fcbf1ab3-4779-49d5-b12f-ab33e9809113?ownerIdentifier=_5e4c6731f56ce6c061ce7014d9bc1b2a</string>

</dict>

</plist>

Porting a landline number to a secondary Ting eSIM on iPhone with primary AT&T account

This was much harder than I expected it to be. Below is a simplified account of what eventually succeeded. I actually started out by trying to transfer from CenturyLink to Tello. That port failed. It was impossible to determine why it failed, each company pointed me to the other. I did learn that the porting process is surprisingly manual, there's an actual carrier to carrier voice call involved.  The porting office is kept away from customers to reduce social engineering exploits. Lastly the all important PIN provided by, in this case CenturyLink, is of a form that suggests the carriers don't take it very seriously. Overall I was left with the impression that carriers have fairly successfully sabotaged the landline porting process. I miss having a government that cared about such things.

My best guess as to why the port from CenturyLink to Tello failed is that CenturyLink takes too long to process the request and Tello is too impatient. It took CenturyLink 4-5 days to complete the Port that succeeded, longer than Tello's 3 day limit. Tello's disinterest in the port process makes me wonder how much time they have as an MVNO.

After some glitches I was able to delete my Tello account (they had a bad link that I'm sure they will never fix).

Now to what worked with an preface of why we were among the last to do successfully port a number from a landline to a mobile company. 

Why we ported from a landline in 2025

We are among the last in America to have a $43/m voice only family landline. It persisted because our ancient security system relies on two of the standard four copper landline wires, and the switching costs were comparable to our monthly landline fee. The landline was purely a message phone, it didn't ring.

It was on the list of things to replace but lower priority since minimal savings and added hassles.

Then copper wires started to disappeared in our metro area. I think wire theft may have peaked for the moment, but the copper side of Century Link isn't paying what it would take to maybe keep the wires working. When the elderly techs come by they beg us to leave copper so they can finally rest.

We want to keep the legacy number so I looked into the 2025 porting options. The only thing that's changed since I last looked in 2015 or so is the eSIM. Mastodon friends suggested voip.ms, but it was obvious that we're not their business. (These days we think a lot about how things get managed after we die, and we assume that the kids are having to make decisions. They are not techies.) Also, canceling voip.ms service requires tedious interactions with their outsourced retention people. A huge red flag.

You still can't port from landline directly to Google Voice. The cost of a landline port would be prohibitive for what is still a "free" service. The AIs claim you can't port from CenturyLink VOIP to Google Voice either.

CenturyLink (fiber) would sell us a VOIP phone service for about $30/m that I'm sure would not have a porting issues. This is modest savings from the $43 landline, but we felt we could do better. Also, I'm not impressed with legacy CenturyLink service. (Their fiber service has worked well for us, they behave like separate companies.)

We looked at using an old iPhone as our home phone with its Apple Voice Mail (requires data). AT&T is our ($$) current family carrier and we could add a $30+/m line from them. But of course there are cheaper options.  Back in 2012 we used H2O Wireless for the kids phones. It's rock bottom service quality but if you work at it you should be able to use them for under $100 a year (prepay, no contract). They are still around! Also familiar from the old days: Tello, Hello Mobile, and Ting. All of which will provide new phones that could be left in the kitchen. We could even use one of our old iPhones. (Note Apple vmail requires some data support, so data-free won't work).

Happily there are better options than leaving an old iPhone on a charger. In the past few years eSIMs have started to work fairly well on iPhones. Our current phones can all handle at least 2 (I think up to 8?) eSIMs. So rather than hassle with an old iPhone holding the number, we ported it to an eSIM on Emily's personal iPhone.

I asked the AIs which had the better customer service among H2O Wireless, Tello, Ting, and Hello Mobile. Tello and Ting won easily. I tried Tello and my initial impression was good, but as noted above the port failed. With Tello you get a temporary number and eSIM immediately, then switch numbers if the port works.

After the Tello port failed we removed the eSIM from Emily's iPhone and tried Ting's $10/m plan. I actually sent a referral from an older Ting account I had but I'm not sure the referral credit thing worked.

We again called CenturyLink. Emily's name was on the landline account, so we used her name on the Ting account. We had the correct account number. Emily navigated CenturyLink phone puzzle and got a 4 digit Port code. We submitted the port request with Ting.

There's no way to monitor porting progress from the Ting site but after 5 work days Emily got an email notification.

Then as expected, things got a bit wonky. Ting's documentation on the eSIM assignment process is contradictory and didn't match what happened. We got a link to go to, but the link only displayed a blank page. We tried it from both the desktop and from Safari mobile. When I investigated the Ting account however it seemed like her iPhone was activated.

In fact it appears that Ting was using Apple's eSIM Carrier Activation. Under some conditions, presumably including a port, the eSIM can be pushed directly from the carrier to the target device. No customer interaction needed -- despite Ting's reference to QR codes and the like. (Bit unnerving really.)

After confirming the eSIM was on her phone we configured it. There are several options to look at,  We configured her Vmail response and tested texting and calling. Everything seemed to be in order.