private static string ParseNormalSearchTerms(string query, SandoQueryDescription sandoQueryDescription)
{
query = Regex.Replace(query, InvalidCharactersRegex, " ");
//ensure non-split query terms are added too.
var nonSplitTerms = query.Split(' ');
foreach (var nonSplit in nonSplitTerms)
{
if (!String.Empty.Equals(nonSplit))
{
sandoQueryDescription.SearchTerms.Add(nonSplit.Trim());
}
}
var wordSplitter = new WordSplitter();
var searchTerms = wordSplitter.ExtractWords(query).Where(w => !String.IsNullOrWhiteSpace(w));
foreach (var searchTerm in searchTerms)
{
sandoQueryDescription.SearchTerms.Add(searchTerm.Trim());
query = query.Replace(searchTerm, " ");
}
sandoQueryDescription.SearchTerms = sandoQueryDescription.SearchTerms.Distinct().ToList();
return(query);
}
private static string ParseLocationFilters(string query, SandoQueryDescription sandoQueryDescription)
{
var matches = Regex.Matches(query, LocationFilterRegex, RegexOptions.IgnoreCase);
foreach (Match match in matches)
{
var matchedFilter = match.Groups["filter"].Value;
var matchedLocation = match.Groups["location"].Value;
if (!String.IsNullOrWhiteSpace(matchedLocation))
{
sandoQueryDescription.Locations.Add(matchedLocation);
}
query = query.Replace(matchedFilter, " ");
}
return(query);
}
private static string ParseLiteralSearchTerms(string query, SandoQueryDescription sandoQueryDescription)
{
var matches = Regex.Matches(query, LiteralSearchRegex);
foreach (Match match in matches)
{
var matchedLiteral = match.Groups["literal"].Value;
var matchedLiteralContent = match.Groups["literalcontent"].Value;
if (!String.IsNullOrWhiteSpace(matchedLiteralContent))
{
sandoQueryDescription.LiteralSearchTerms.Add(matchedLiteral);
}
query = query.Replace(matchedLiteral, " ");
}
query = query.Replace("\"", String.Empty);
return(query);
}
public SandoQueryDescription Parse(string query)
{
var sandoQueryDescription = new SandoQueryDescription {
OriginalQuery = query
};
var parseFunctions = GetQueryParseFunctions();
var analyzedQuery = query;
foreach (var function in parseFunctions)
{
if (String.IsNullOrWhiteSpace(analyzedQuery))
{
return(sandoQueryDescription);
}
analyzedQuery = function(analyzedQuery, sandoQueryDescription);
}
return(sandoQueryDescription);
}
private static string ParseFileExtensionFilters(string query, SandoQueryDescription sandoQueryDescription)
{
var matches = Regex.Matches(query, FileExtensionFilterRegex, RegexOptions.IgnoreCase);
foreach (Match match in matches)
{
var matchedFilter = match.Groups["filter"].Value;
var matchedFileExtension = match.Groups["fileext"].Value.ToLower();
if (matchedFilter.StartsWith("-"))
{
sandoQueryDescription.FileExtensions.Add("-" + matchedFileExtension);
}
else
{
sandoQueryDescription.FileExtensions.Add(matchedFileExtension);
}
query = query.Replace(matchedFilter, " ");
}
return(query);
}
private static string ParseProgramElementTypeFilters(string query, SandoQueryDescription sandoQueryDescription)
{
var matches = Regex.Matches(query, ProgramElementTypeFilterRegex, RegexOptions.IgnoreCase);
foreach (Match match in matches)
{
var matchedFilter = match.Groups["filter"].Value;
var matchedProgramElementType = match.Groups["type"].Value.ToLower();
if (matchedFilter.StartsWith("-"))
{
sandoQueryDescription.ProgramElementTypes.Add("-" + matchedProgramElementType);
}
else
{
sandoQueryDescription.ProgramElementTypes.Add(matchedProgramElementType);
}
query = query.Replace(matchedFilter, " ");
}
return(query);
}
private static string ParseNormalSearchTerms(string query, SandoQueryDescription sandoQueryDescription)
{
query = Regex.Replace(query, InvalidCharactersRegex, " ");
var wordSplitter = new WordSplitter();
//add unsplit words
var splitTerms = query.Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
foreach (string term in splitTerms)
{
sandoQueryDescription.SearchTerms.Add(term.Trim());
}
var searchTerms = wordSplitter.ExtractWords(query).Where(w => !String.IsNullOrWhiteSpace(w));
foreach (var searchTerm in searchTerms)
{
sandoQueryDescription.SearchTerms.Add(searchTerm.Trim());
query = query.Replace(searchTerm, " ");
}
sandoQueryDescription.SearchTerms = sandoQueryDescription.SearchTerms.Distinct().ToList();
return(query);
}
private static string ParseProgramElementTypeFilters(string query, SandoQueryDescription sandoQueryDescription)
{
var matches = Regex.Matches(query, ProgramElementTypeFilterRegex, RegexOptions.IgnoreCase);
foreach (Match match in matches)
{
var matchedFilter = match.Groups["filter"].Value;
var matchedProgramElementType = match.Groups["type"].Value.ToLower();
if (matchedFilter.StartsWith("-"))
sandoQueryDescription.ProgramElementTypes.Add("-" + matchedProgramElementType);
else
sandoQueryDescription.ProgramElementTypes.Add(matchedProgramElementType);
query = query.Replace(matchedFilter, " ");
}
return query;
}
private static string ParseNormalSearchTerms(string query, SandoQueryDescription sandoQueryDescription)
{
query = Regex.Replace(query, InvalidCharactersRegex, " ");
var wordSplitter = new WordSplitter();
//add unsplit words
var splitTerms = query.Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
foreach (string term in splitTerms)
{
sandoQueryDescription.SearchTerms.Add(term.Trim());
}
var searchTerms = wordSplitter.ExtractWords(query).Where(w => !String.IsNullOrWhiteSpace(w) && !splitTerms.Contains("-"+w)).OrderByDescending(s => s.Length);
foreach (var searchTerm in searchTerms)
{
sandoQueryDescription.SearchTerms.Add(searchTerm.Trim());
query = query.Replace(searchTerm, " ");
}
sandoQueryDescription.SearchTerms = sandoQueryDescription.SearchTerms.Distinct().ToList();
return query;
}
private static string ParseLocationFilters(string query, SandoQueryDescription sandoQueryDescription)
{
var matches = Regex.Matches(query, LocationFilterRegex, RegexOptions.IgnoreCase);
foreach (Match match in matches)
{
var matchedFilter = match.Groups["filter"].Value;
var matchedLocation = match.Groups["location"].Value;
if (!String.IsNullOrWhiteSpace(matchedLocation))
{
sandoQueryDescription.Locations.Add(matchedLocation);
}
query = query.Replace(matchedFilter, " ");
}
return query;
}
private static string ParseLiteralSearchTerms(string query, SandoQueryDescription sandoQueryDescription)
{
var matches = Regex.Matches(query, LiteralSearchRegex);
foreach (Match match in matches)
{
var matchedLiteral = match.Groups["literal"].Value;
var matchedLiteralContent = match.Groups["literalcontent"].Value;
if (!String.IsNullOrWhiteSpace(matchedLiteralContent))
{
sandoQueryDescription.LiteralSearchTerms.Add(matchedLiteral);
}
query = query.Replace(matchedLiteral, " ");
}
query = query.Replace("\"", String.Empty);
return query;
}
private static string ParseFileExtensionFilters(string query, SandoQueryDescription sandoQueryDescription)
{
var matches = Regex.Matches(query, FileExtensionFilterRegex, RegexOptions.IgnoreCase);
foreach (Match match in matches)
{
var matchedFilter = match.Groups["filter"].Value;
var matchedFileExtension = match.Groups["fileext"].Value.ToLower();
if (matchedFilter.StartsWith("-"))
sandoQueryDescription.FileExtensions.Add("-" + matchedFileExtension);
else
sandoQueryDescription.FileExtensions.Add(matchedFileExtension);
query = query.Replace(matchedFilter, " ");
}
return query;
}
public SandoQueryDescription Parse(string query)
{
var sandoQueryDescription = new SandoQueryDescription {OriginalQuery = query};
var parseFunctions = GetQueryParseFunctions();
var analyzedQuery = query;
foreach (var function in parseFunctions)
{
if (String.IsNullOrWhiteSpace(analyzedQuery))
return sandoQueryDescription;
analyzedQuery = function(analyzedQuery, sandoQueryDescription);
}
return sandoQueryDescription;
}
private static string ParseNormalSearchTerms(string query, SandoQueryDescription sandoQueryDescription)
{
query = Regex.Replace(query, InvalidCharactersRegex, " ");
//ensure non-split query terms are added too.
var nonSplitTerms = query.Split(' ');
foreach (var nonSplit in nonSplitTerms)
{
if(!String.Empty.Equals(nonSplit))
sandoQueryDescription.SearchTerms.Add(nonSplit.Trim());
}
var wordSplitter = new WordSplitter();
var searchTerms = wordSplitter.ExtractWords(query).Where(w => !String.IsNullOrWhiteSpace(w));
foreach (var searchTerm in searchTerms)
{
sandoQueryDescription.SearchTerms.Add(searchTerm.Trim());
query = query.Replace(searchTerm, " ");
}
sandoQueryDescription.SearchTerms = sandoQueryDescription.SearchTerms.Distinct().ToList();
return query;
}
