How Can I Construct an Efficient CoreData Search, Including Allowing For Preceding and Trailing Characters Here? - ios

Question

Based on straight SQL searches in a previous app, I am adding CoreData searching to a new app. These searches are in a custom dictionary db that the app contains; this function does the work:
public func wordMatcher (pad: Int, word: Array<String>, substitutes : Set<String> ) {
let context = CoreDataManager.shared.persistentContainer.viewContext
var query: Array<String>
var foundPositions : Set<Int> = []
var searchTerms : Array<String> = []
if word.count >= 4 {
for i in 0..<word.count {
for letter in substitutes {
query = word
query[i] = letter
searchTerms.append(query.joined())
let rq: NSFetchRequest<Word> = Word.fetchRequest()
rq.predicate = NSPredicate(format: "name LIKE %#", query.joined())
rq.fetchLimit = 1
do {
if try context.fetch(rq).count != 0 {
foundPositions.insert(i)
break
}
} catch {
}
}
// do aggregated searchTerms search here instead of individual searches?
}
}
}
The NSFetchRequest focuses on one permutation at a time. But I'm accumulating the search string fragments in the array searchTerms because I don't know if it would be more efficient to construct a single query connected with ORs, and I also don't know how to do that in CoreData.
The focus is on the positions in the original term word: I need to indicate if any given location has at least one of the substitutes as a valid fit. So to implement the aggregate searchTerms approach, a FetchRequest would have to happen for each location in the base term.
A second complication is the one referred to in the title of the question. I am using LIKE because the search term in the FetchRequest could be a substring in a longer word. However, the maximum number of letters is 11, and pad is the starting point of the original term in that field of 11 spaces.
So if pad is 3, then I would need to allow for 0..<pad preceding characters. And because there may be trailing characters, I would also want results with 0..<(11 - (pad + word.count)) alphabetic characters after the last letter in the search term.
Regex seems like one way to do this, but I haven't found a clear example of how to do this in this case, and especially with the multiple search terms (if that's the way to go). The limits of SQLite in the previous version forced constructing multiple queries with increasing numbers of "_" underscores to indicate the padding characters; that tended to really explode the number of queries.
BTW, substitutes is limited to an absolute maximum of 9 values, and in practice is usually below 5, so things are a little more manageable.
I would like to get a grip on this, and so if anyone can provide direction or examples that can make this a reasonably efficient function, the help is appreciated greatly.
EDIT:
I've realized that I need a result for each position in the target string, with cases where the leading and trailing spaces also may need to contain a substitute as well.
So I'm moving to this:
public func wordMatcher (pad: Int, word: Array<String>, substitutes : Set<String> ) {
let context = CoreDataManager.shared.persistentContainer.viewContext
var pad_ = pad
var query: Array<String>
var foundPositions : Set<Int> = []
let rq: NSFetchRequest<Word> = Word.fetchRequest()
rq.fetchLimit = 1
let subs = "[\(substitutes.joined())]"
// if word.count >= 4 { // because those locations will be blocked off anyway otherwise
let start = pad > 0 ? -1 : 0
let finish = 11 - (pad + word.count) > 0 ? word.count + 1 : word.count
for i in start..<finish {
query = word
var _pad = 11 - (pad + word.count)
if i == -1 {
query = Array(arrayLiteral: subs) + query
pad_ -= 1
} else if i > word.count {
query.append(subs)
_pad -= 1
} else {
pad_ = pad
query[i] = subs
}
let endPad = _pad > 0 ? "{0,\(_pad)}" : ""
let predMatch = ".\(query.joined())\(endPad)"
print(predMatch)
rq.predicate = NSPredicate(format:"position <= %# AND word MATCHES %#", pad_, predMatch)
do {
if try context.fetch(rq).count != 0 {
foundPositions.insert(i)
}
} catch {
}
// }
}
lFreq = foundPositions
}
This relies on a regex substitution, inserted into the original target string. What I'll have to find out is if this is fast enough at the edge cases, but it may not be critical even in the worst case.
predMatch will end up looking something like "ab[xyx]d{0,3}", and I think I can get rid of the position section by changing it to be "{0,2}ab[xyx]d{0,3}". But I guess I'm going to have to try to find out.