/// <summary>
/// Get images that are within one "edit" of matching an image in the collection
/// </summary>
public IEnumerable <string> GetApproximateMatchingImages(IEnumerable <string> searchTerms)
{
var fullPathsToImages = new List <string>();
foreach (var term in searchTerms)
{
lock (_padlock)
{
const int kMaxEditDistance = 1;
var itemFormExtractor =
new ApproximateMatcher.GetStringDelegate <KeyValuePair <string, List <string> > >(pair => pair.Key);
var matches =
ApproximateMatcher.FindClosestForms <KeyValuePair <string, List <string> > >(_indexOfWordsToRelativePath,
itemFormExtractor,
term.ToLowerInvariant(),
ApproximateMatcherOptions.None,
kMaxEditDistance);
if (matches != null && matches.Count > 0)
{
foreach (var keyValuePair in matches)
{
fullPathsToImages.AddRange(keyValuePair.Value.Select(GetFullPath));
}
}
}
}
var results = new List <string>();
fullPathsToImages.Distinct().ForEach(p => results.Add(p));
return(_fixPaths(results));
}
/// <summary>
/// Get an list of languages that match the given string in some way (code, name, country)
/// </summary>
public IEnumerable <LanguageInfo> SuggestLanguages(string searchString)
{
if (searchString != null)
{
searchString = searchString.Trim();
}
if (string.IsNullOrEmpty(searchString))
{
yield break;
}
if (searchString == "*")
{
// there will be duplicate LanguageInfo entries for 2 and 3 letter codes
var all_languages = new HashSet <LanguageInfo>(_codeToLanguageIndex.Select(l => l.Value));
foreach (LanguageInfo languageInfo in all_languages.OrderBy(l => l, new ResultComparer(searchString)))
{
yield return(languageInfo);
}
}
else
{
IEnumerable <LanguageInfo> matchOnCode = from x in _codeToLanguageIndex where x.Key.StartsWith(searchString, StringComparison.InvariantCultureIgnoreCase) select x.Value;
List <LanguageInfo>[] matchOnName = (from x in _nameToLanguageIndex where x.Key.StartsWith(searchString, StringComparison.InvariantCultureIgnoreCase) select x.Value).ToArray();
List <LanguageInfo>[] matchOnCountry = (from x in _countryToLanguageIndex where x.Key.StartsWith(searchString, StringComparison.InvariantCultureIgnoreCase) select x.Value).ToArray();
if (!matchOnName.Any())
{
// look for approximate matches
const int kMaxEditDistance = 3;
var itemFormExtractor = new ApproximateMatcher.GetStringDelegate <KeyValuePair <string, List <LanguageInfo> > >(pair => pair.Key);
IList <KeyValuePair <string, List <LanguageInfo> > > matches = ApproximateMatcher.FindClosestForms(_nameToLanguageIndex, itemFormExtractor,
searchString,
ApproximateMatcherOptions.None,
kMaxEditDistance);
matchOnName = (from m in matches select m.Value).ToArray();
}
var combined = new HashSet <LanguageInfo>(matchOnCode);
foreach (List <LanguageInfo> l in matchOnName)
{
combined.UnionWith(l);
}
foreach (List <LanguageInfo> l in matchOnCountry)
{
combined.UnionWith(l);
}
foreach (LanguageInfo languageInfo in combined.OrderBy(l => l, new ResultComparer(searchString)))
{
yield return(languageInfo);
}
}
}
/// <summary>
/// Get an list of languages that match the given string in some way (code, name, country)
/// </summary>
public IEnumerable <LanguageInfo> SuggestLanguages(string searchString)
{
if (searchString != null)
{
searchString = searchString.Trim().ToLowerInvariant();
}
if (string.IsNullOrEmpty(searchString))
{
yield break;
}
else if (searchString == "*")
{
foreach (var l in from x in CodeToLanguageIndex select x.Value)
{
yield return(Set3LetterCode(l));
}
}
else
{
IEnumerable <LanguageInfo> matchOnCode = from x in CodeToLanguageIndex where x.Key.ToLowerInvariant().StartsWith(searchString) select x.Value;
var matchOnName = from x in NameToLanguageIndex where x.Key.ToLowerInvariant().StartsWith(searchString) select x.Value;
if (!matchOnName.Any())
{
// look for approximate matches
const int kMaxEditDistance = 3;
var itemFormExtractor = new ApproximateMatcher.GetStringDelegate <KeyValuePair <string, List <LanguageInfo> > >(pair => pair.Key);
var matches = ApproximateMatcher.FindClosestForms <KeyValuePair <string, List <LanguageInfo> > >(NameToLanguageIndex, itemFormExtractor,
searchString,
ApproximateMatcherOptions.None,
kMaxEditDistance);
matchOnName = from m in matches select m.Value;
}
List <LanguageInfo> combined = new List <LanguageInfo>(matchOnCode);
foreach (var l in matchOnName)
{
combined.AddRange(l);
}
List <LanguageInfo> sorted = new List <LanguageInfo>(combined.Distinct());
sorted.Sort(new ResultComparer(searchString));
foreach (var languageInfo in sorted)
{
yield return(Set3LetterCode(languageInfo));
}
}
}
private IEnumerable <object> GetMatchingPictures(IEnumerable <string> keywords, out bool foundExactMatches)
{
var pictures = new List <string>();
foundExactMatches = false;
foreach (var term in keywords)
{
List <string> picturesForThisKey;
//first, try for exact matches
if (_wordToPartialPathIndex.TryGetValue(term, out picturesForThisKey))
{
pictures.AddRange(picturesForThisKey);
foundExactMatches = true;
}
//then look for approximate matches
else
{
foundExactMatches = false;
var kMaxEditDistance = 1;
var itemFormExtractor = new ApproximateMatcher.GetStringDelegate <KeyValuePair <string, List <string> > >(pair => pair.Key);
var matches = ApproximateMatcher.FindClosestForms <KeyValuePair <string, List <string> > >(_wordToPartialPathIndex, itemFormExtractor,
term,
ApproximateMatcherOptions.None,
kMaxEditDistance);
if (matches != null && matches.Count > 0)
{
foreach (var keyValuePair in matches)
{
pictures.AddRange(keyValuePair.Value);
}
}
}
}
var results = new List <object>();
pictures.Distinct().ForEach(p => results.Add(p));
return(results);
}
private IEnumerable<object> GetMatchingPictures(IEnumerable<string> keywords, out bool foundExactMatches)
{
var pictures = new List<string>();
foundExactMatches = false;
foreach (var term in keywords)
{
List<string> picturesForThisKey;
//first, try for exact matches
if (_wordToPartialPathIndex.TryGetValue(term, out picturesForThisKey))
{
pictures.AddRange(picturesForThisKey);
foundExactMatches = true;
}
//then look for approximate matches
else
{
foundExactMatches = false;
var kMaxEditDistance = 1;
var itemFormExtractor = new ApproximateMatcher.GetStringDelegate<KeyValuePair<string, List<string>>>(pair => pair.Key);
var matches = ApproximateMatcher.FindClosestForms<KeyValuePair<string, List<string>>>(_wordToPartialPathIndex, itemFormExtractor,
term,
ApproximateMatcherOptions.None,
kMaxEditDistance);
if (matches != null && matches.Count > 0)
{
foreach (var keyValuePair in matches)
{
pictures.AddRange(keyValuePair.Value);
}
}
}
}
var results = new List<object>();
pictures.Distinct().ForEach(p => results.Add(p));
return results;
}
/// <summary>
/// Get an list of languages that match the given string in some way (code, name, country)
/// </summary>
public IEnumerable <LanguageInfo> SuggestLanguages(string searchString)
{
if (searchString != null)
{
searchString = searchString.Trim();
}
if (string.IsNullOrEmpty(searchString))
{
yield break;
}
if (searchString == "*")
{
// there will be duplicate LanguageInfo entries for 2 and 3 letter codes and equivalent tags
var all_languages = new HashSet <LanguageInfo>(_codeToLanguageIndex.Select(l => l.Value));
foreach (LanguageInfo languageInfo in all_languages.OrderBy(l => l, new ResultComparer(searchString)))
{
yield return(languageInfo);
}
}
// if the search string exactly matches a hard-coded way to say "sign", show all the sign languages
// there will be duplicate LanguageInfo entries for equivalent tags
else if (new [] { "sign", "sign language", "signes", "langage des signes", "señas", "lenguaje de señas" }.Contains(searchString.ToLowerInvariant()))
{
var parallelSearch = new HashSet <LanguageInfo>(_codeToLanguageIndex.AsParallel().Select(li => li.Value).Where(l =>
l.Names.AsQueryable().Any(n => n.ToLowerInvariant().Contains("sign"))));
foreach (LanguageInfo languageInfo in parallelSearch)
{
yield return(languageInfo);
}
}
else
{
IEnumerable <LanguageInfo> matchOnCode = from x in _codeToLanguageIndex where x.Key.StartsWith(searchString, StringComparison.InvariantCultureIgnoreCase) select x.Value;
List <LanguageInfo>[] matchOnName = (from x in _nameToLanguageIndex where x.Key.StartsWith(searchString, StringComparison.InvariantCultureIgnoreCase) select x.Value).ToArray();
// Apostrophes can cause trouble in lookup. Unicode TR-29 inexplicably says to use
// u2019 (RIGHT SINGLE QUOTATION MARK) for the English apostrophe when it also defines
// u02BC (MODIFIER LETTER APOSTROPHE) as a Letter character. Users are quite likely to
// type the ASCII apostrophe (u0027) which is defined as Punctuation. The current
// data appears to use u2019 in several language names, which means that users might
// end up thinking the language isn't in our database.
// See https://silbloom.myjetbrains.com/youtrack/issue/BL-6339.
if (!matchOnName.Any() && searchString.Contains('\''))
{
searchString = searchString.Replace('\'', '\u2019');
matchOnName = (from x in _nameToLanguageIndex where x.Key.StartsWith(searchString, StringComparison.InvariantCultureIgnoreCase) select x.Value).ToArray();
}
List <LanguageInfo>[] matchOnCountry = (from x in _countryToLanguageIndex where x.Key.StartsWith(searchString, StringComparison.InvariantCultureIgnoreCase) select x.Value).ToArray();
if (!matchOnName.Any())
{
// look for approximate matches
const int kMaxEditDistance = 3;
var itemFormExtractor = new ApproximateMatcher.GetStringDelegate <KeyValuePair <string, List <LanguageInfo> > >(pair => pair.Key);
IList <KeyValuePair <string, List <LanguageInfo> > > matches = ApproximateMatcher.FindClosestForms(_nameToLanguageIndex, itemFormExtractor,
searchString,
ApproximateMatcherOptions.None,
kMaxEditDistance);
matchOnName = (from m in matches select m.Value).ToArray();
}
var combined = new HashSet <LanguageInfo>(matchOnCode);
foreach (List <LanguageInfo> l in matchOnName)
{
combined.UnionWith(l);
}
foreach (List <LanguageInfo> l in matchOnCountry)
{
combined.UnionWith(l);
}
var ordered = combined.OrderBy(l => l, new ResultComparer(searchString));
foreach (LanguageInfo languageInfo in ordered)
{
yield return(languageInfo);
}
}
}
