feat: Apple SDK update for version 18.0.0

[ArnabChatterjee20k]

This PR contains updates to the Apple SDK for version 18.0.0.

[greptile-apps]

Greptile Summary

This PR updates the Apple SDK to version 18.0.0, replacing the Theme enum with BrowserTheme, making Presence and PresenceList concrete (non-generic) types with an optional metadata field, and simplifying the data-extraction logic in Document.from(map:) and Row.from(map:).

• BrowserTheme replaces Theme and Presences service methods drop their generic nestedType overloads — both changes are clean breaking-change upgrades.
• The simplified one-liner in Document.from(map:) and Row.from(map:) introduces a regression: when map["data"] is present but is not a [String: Any] (e.g., an array), the as? [String: Any] cast silently returns nil and the entire parent map is passed to JSONDecoder instead of the actual data payload — silently decoding wrong fields or crashing via try!.

Confidence Score: 4/5

Safe to merge once the data-extraction fallback in Document and Row is fixed; all other changes are straightforward.

The simplification of from(map:) in both Document.swift and Row.swift drops the JSONSerialization.isValidJSONObject branch. When the server returns a non-dictionary data value (e.g., an array), the entire parent document map is fed to JSONDecoder as the data payload, producing silently wrong decoded values or a try! crash.

Sources/AppwriteModels/Document.swift and Sources/AppwriteModels/Row.swift — both have the same data-extraction fallback regression.

Important Files Changed

Filename	Overview
Sources/AppwriteModels/Document.swift	Simplified from(map:) data extraction loses the non-dict JSON branch, causing the whole parent map to be used as fallback instead of the actual data value when it is not a dictionary.
Sources/AppwriteModels/Row.swift	Same data-extraction regression as Document.swift — map["data"] as? [String: Any] ?? map silently discards non-dict data payloads in favour of the full parent map.
Sources/AppwriteModels/Presence.swift	Converted from generic Presence<T> to concrete class with optional [String: AnyCodable] metadata; init(from decoder:) correctly uses decodeIfPresent, though from(map:) always produces a non-nil metadata value.
Sources/AppwriteModels/PresenceList.swift	Generic type parameter removed cleanly; uses concrete [Presence] throughout.
Sources/Appwrite/Services/Presences.swift	Generic overloads removed; all service methods simplified to return concrete Presence/PresenceList types.
Sources/AppwriteEnums/BrowserTheme.swift	Renamed from Theme to BrowserTheme; enum cases and conformances unchanged.
Sources/Appwrite/Services/Avatars.swift	theme parameter type updated from AppwriteEnums.Theme to AppwriteEnums.BrowserTheme; no other changes.
Sources/Appwrite/Client.swift	SDK version bumped from 17.2.0 to 18.0.0.

_{Reviews (2): Last reviewed commit: "chore: update Apple SDK to 18.0.0" | Re-trigger Greptile}

[greptile-apps]

When map["metadata"] is absent or nil, the ?? [:] fallback converts it to an empty dictionary, so metadata is always non-nil on the from(map:) path. The init(from decoder:) path uses decodeIfPresent, which correctly returns nil for a missing key. Any caller checking presence.metadata != nil (or using metadata in an if let / guard let) will always see a truthy value when the object was built via from(map:), even when the server returned no metadata.

Suggested change

	metadata: (map["metadata"] as? [String: Any] ?? [:]).mapValues { AnyCodable($0) }
	metadata: (map["metadata"] as? [String: Any]).map { $0.mapValues { AnyCodable($0) } }

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

[greptile-apps]

The new fallback map["data"] as? [String: Any] ?? map silently substitutes the entire parent map whenever map["data"] exists but is not a dictionary (e.g., a JSON array). In the previous code, JSONSerialization.isValidJSONObject(raw) handled that branch correctly and would still serialize the actual data value. Now, if a server ever returns an array-typed data, the entire document map (including $id, $collectionId, etc.) is passed to JSONDecoder instead — either producing a wrong T silently or crashing the try!.

Suggested change

	data: try! JSONDecoder().decode(T.self, from: JSONSerialization.data(withJSONObject: map["data"] as? [String: Any] ?? map, options: []))
	data: try! JSONDecoder().decode(T.self, from: {
	let raw = map["data"]
	if let dict = raw as? [String: Any] {
	return try! JSONSerialization.data(withJSONObject: dict, options: [])
	} else if let raw = raw, JSONSerialization.isValidJSONObject(raw) {
	return try! JSONSerialization.data(withJSONObject: raw, options: [])
	} else {
	return try! JSONSerialization.data(withJSONObject: [:] as [String: Any], options: [])
	}
	}())

[greptile-apps]

Same regression as Document.swift: the ?? map fallback is reached both when data is absent and when data is present but is not a [String: Any] (e.g., an array). In the latter case the entire row map is passed to JSONDecoder instead of the actual data payload, silently decoding wrong data or crashing. The previous implementation correctly handled non-dict JSON via JSONSerialization.isValidJSONObject.

Suggested change

	data: try! JSONDecoder().decode(T.self, from: JSONSerialization.data(withJSONObject: map["data"] as? [String: Any] ?? map, options: []))
	data: try! JSONDecoder().decode(T.self, from: {
	let raw = map["data"]
	if let dict = raw as? [String: Any] {
	return try! JSONSerialization.data(withJSONObject: dict, options: [])
	} else if let raw = raw, JSONSerialization.isValidJSONObject(raw) {
	return try! JSONSerialization.data(withJSONObject: raw, options: [])
	} else {
	return try! JSONSerialization.data(withJSONObject: [:] as [String: Any], options: [])
	}
	}())