having troubles getting !any() query to work with raven db - linq

Question

I am new to RavenDB and it's been frustrating try to get the Any() LINQ query to work correctly. Here is what my document looks like:
{
"Key": "BKey",
"Text": "B Key",
"IsLocal": false,
"Category": null,
"_destroy": false,
"Translations": [
{
"CultureCode": "es-ES",
"Text": null
},
{
"CultureCode": "ja-JP",
"Text": "Hello"
}
]
}
I would expect the following query to give me all entries that don't have translations for "es-ES":
var nonTranslatedEntries =
_docs.Query<ResourceEntry>().Where( e => e.Translations == null || e.Translations.Count == 0 || !e.Translations.Any(t => t.CultureCode == "es-ES" && t.Text != null))
However, this isn't working. It's bringing back the entries even when a translation for the specified CultureCode exists. It works if I have only have one item inside the translations array. But as soon as I have more than one item inside the translations array, then the query stops working.
As an alternative solution, I tried to do the following:
var translatedEntries = from re in _docs.Query<ResourceEntry>()
where re.Translations.Any(t => t.CultureCode == cultureCode && t.Text != null)
select new {Id = re.Id};
var translatedIds = translatedEntries.ToList().Select(e => e.Id).ToList();
var nonTranslatedEntries =
_docs.Query<ResourceEntry>().Where(e => !e.Id.In(translatedEntryIds));
But that just brings back an empty list.
Any help would be very much appreciated.
Thanks,
Nizar

Answer 1

A static index that will get the job done:
public class Resources_ByTranslation :
AbstractIndexCreationTask<ResourceEntry, Resources_ByTranslation.IndexEntry>
{
public class IndexEntry
{
public string Key { get; set; }
public string CultureCodes { get; set; }
}
public Resources_ByTranslation()
{
Map = entries => from entry in entries
select new {
entry.Key,
CultureCodes = entry.Translations
.Where(x => x.Text != null)
.Select(x => x.CultureCode)
};
}
}
Then query with:
var nonTranslatedEntries =
session.Query<Resources_ByTranslation.IndexEntry, Resources_ByTranslation>()
.Where(x => x.CultureCodes != "es-ES")
.As<ResourceEntry>();
Note that the CultureCodes list is being treated as single string. This is due to how then index matching works internally. It's slightly strange, but it does work.

Answer 2

I ended up fixing my issue as follows:
Upon creation of a resource entry, I go ahead and create all the possible translations and set the translation text to null:
{
"Key": "BKey",
"Text": "B Key",
"IsLocal": false,
"Category": null,
"_destroy": false,
"Translations": [
{
"CultureCode": "es-ES",
"Text": null
},
{
"CultureCode": "ja-JP",
"Text": null
}
]
}
I then created the following index:
public class ResourceEntries_ByCultureCodes : AbstractIndexCreationTask<ResourceEntry>
{
public ResourceEntries_ByCultureCodes()
{
Map = entries => from e in entries
select new
{
e.Id,
e.Key,
e.Text,
e.IsLocal,
e.Category,
_ = e.Translations.Select(t => CreateField(t.CultureCode, t.Text != null, false, true))
};
}
}
That gives me a flat record of translations with the value set to 'true' if a translation exists for a that specific culture. For example, I get back something like this:
{
Id,
Key,
Text,
IsLocal,
Category,
es-ES: <true> or <false> depending upon whether Translation.Text != null
ja-JP:<true> or <false> depending upon whether Translation.Text != null
}
I can then do LuceneQuery as follows to get all entries for "es-ES" that have NOT been translated:
_docs.LuceneQuery<ResourceEntry, ResourceEntries_ByCultureCodes>().WhereEquals("es-ES", false);
I wish there were an easier solution, though.