UTF16 (Swedish grapheme cluster) issues #293

New Issue

2025-12-29T15:28:17+01:00

adam commented

2025-12-29 15:28:17 +01:00

Originally created by @rivera-ernesto on GitHub (Nov 1, 2019).

I have described this issue here https://stackoverflow.com/q/58660170/1049134

Basically I store an identifier string with UTF16 Swedish characters:

"/EXTERNAL/Gemensam RUN/FileCloud Test/Test folder åäö/Test with Swedish characters - åäö.xlsx"

Which then don't match a fetch:

CoreStore.fetchOne(From<Record>().where(\.identifier == identifier)) // Fails

Originally created by @rivera-ernesto on GitHub (Nov 1, 2019). I have described this issue here https://stackoverflow.com/q/58660170/1049134 Basically I store an `identifier` string with UTF16 Swedish characters: `"/EXTERNAL/Gemensam RUN/FileCloud Test/Test folder åäö/Test with Swedish characters - åäö.xlsx"` Which then don't match a fetch: `CoreStore.fetchOne(From<Record>().where(\.identifier == identifier)) // Fails`

adam closed this issue

2025-12-29 15:28:17 +01:00

adam commented

2025-12-29 15:28:18 +01:00

@JohnEstropia commented on GitHub (Nov 5, 2019):

What type is identifier ?

@JohnEstropia commented on GitHub (Nov 5, 2019): What type is `identifier` ?

adam commented

2025-12-29 15:28:19 +01:00

@JohnEstropia commented on GitHub (Nov 5, 2019):

Also, what happens if you use NSPredicate instead?

CoreStore.fetchOne(
    From<Record>().where(NSPredicate(format: "%K == %@", #keyPath(Record.identifier), identifier))
)

@JohnEstropia commented on GitHub (Nov 5, 2019): Also, what happens if you use `NSPredicate` instead? ```swift CoreStore.fetchOne( From<Record>().where(NSPredicate(format: "%K == %@", #keyPath(Record.identifier), identifier)) ) ```

adam commented

2025-12-29 15:28:19 +01:00

@rivera-ernesto commented on GitHub (Nov 6, 2019):

identifier is a Swift string. Found the problem and updated the StackOverflow issue.

In short I had two identifier instances with Swedish characters that are written with different grapheme clusters, so two binary different but both valid representations of the same String.
They are considered equal by Swift but for Core Data (and I guess because of SQLite under it) they are not considered equal, so they don't match the predicate.

It's not a problem of encoding, as both representations are in the same encoding, is a problem of grapheme clusters being valid in more than one way, and no way to "normalize" them that I could find.

Again refer to https://stackoverflow.com/q/58660170/1049134 if you are interested to know more.

@rivera-ernesto commented on GitHub (Nov 6, 2019): `identifier` is a Swift string. Found the problem and updated the StackOverflow issue. In short I had two `identifier` instances with Swedish characters that are written with different grapheme clusters, so two binary different but both valid representations of the same String. They are considered equal by Swift but for Core Data (and I guess because of SQLite under it) they are not considered equal, so they don't match the predicate. It's not a problem of encoding, as both representations are in the same encoding, is a problem of grapheme clusters being valid in more than one way, and no way to "normalize" them that I could find. Again refer to https://stackoverflow.com/q/58660170/1049134 if you are interested to know more.

adam commented

2025-12-29 15:28:19 +01:00

@JohnEstropia commented on GitHub (Nov 8, 2019):

@rivera-ernesto I see. If that's the case I'm not sure this is something that should be "fixed" at the library level. Here are some common practices for such cases:

Store a canonical string in addition to your actual property value and use canonical strings as search strings (See: example1, example2)
Try diacritic-insensitive query operators:

.where(format: "%K CONTAINS[cd] %@", #keyPath(Record.identifier), identifier)

@JohnEstropia commented on GitHub (Nov 8, 2019): @rivera-ernesto I see. If that's the case I'm not sure this is something that should be "fixed" at the library level. Here are some common practices for such cases: 1. Store a canonical string in addition to your actual property value and use canonical strings as search strings (See: [example1](https://stackoverflow.com/questions/17418443/normalize-or-canonicalize-string-for-core-data), [example2](https://stackoverflow.com/questions/16741632/how-to-properly-convert-to-a-canonical-string-for-searching-in-cocoa)) 2. Try diacritic-insensitive query operators: ```swift .where(format: "%K CONTAINS[cd] %@", #keyPath(Record.identifier), identifier) ```

Sign in to join this conversation.

Branches Tags

develop

prototype/swift6

master

prototype/concurrency

minIOS13

prototype/containers

transactionSourceTracking

prototype/swiftui

minIOS11

swift5_3

prototype/propertyWrapperFields

prototype/propertyWrappers

prototype/propertyWrapper

Swift5

throwables

prototype/Swift_4_1

prototype/Swift_4_0

prototype/Swift_3_2

prototype/queryBuilders

temp/develop

corestore4_develop

prototype/mainContextToPSC

master_ios_7_to_9

swift_2_3

gh-pages

_corespotlight-test

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: starred/CoreStore#293