public static string[] Prepare(string query)
{
List<string> preparedQueries = new List<string>();
string parasitesFile = Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.CommonApplicationData), ConfigurationManager.AppSettings["DataFolder"], "MoreByThisAuthorParasites.txt");
var initialParasites = new List<String>()
{
"(compilador)",
//"a.",
//"and",
//"and others",
"annotated",
"artist",
"artistic",
"associate",
"authors",
"author:",
"author",
//"b.",
//"br",
//"br.",
"bishop",
" by ",
//"c.",
"c.c.v.i.",
"c.h.s.",
"c.i.c.m.",
"c.n.d.",
"c.p.",
"c.pp.s.",
"c.s.c.",
"c.s.j.",
"c.s.sp.",
"c.ss.r.",
"ccvi",
//"chs",
//"cicm",
//"cnd",
"commentary author:",
"commentary",
"compiladors",
"compilador",
"compiler",
"contributors",
"contributor",
//"cp",
"cpps",
//"csc",
"csj",
"cssp",
"cssr",
" csv,",
", csv",
//"d.",
"designers",
"designer",
"directors",
"director",
//"dr",
//"dr.",
//"e.",
"edited",
"editors",
"editor asociado",
"editor",
"editora",
//"f.",
"forewords",
" foreword ",
"foreword",
//"fr",
//"fr.",
"galatians",
"general",
"guide",
//"h.",
//"i.",
"i.h.m.",
"ihm",
"iii",
"illustrators",
"illuminator",
"illustrator",
"introduced",
"introduction",
//"j.",
//"jr",
", jr.",
//"k.",
//"l.",
"lecturers",
"lecturer",
//"m.",
//"m.s.",
//"map",
//"ms",
//"n.",
"notes",
"o carm",
//"ocarm",
//"o.",
"o.c.s.o.",
"o.carm.",
"o.f.m.",
"o.f.m.cap.",
"o.f.m.conv.",
"o.m.i.",
"o.p.",
"o.s.a.",
"o.s.b.",
"o.s.c.",
"o.s.f.",
"o.s.u.",
"obl.s.b.",
"oblsb",
"ocso ",
" ocso",
"ocso",
"ofm cap",
"ofm conv",
//"ofm",
" omi ",
" omi",
", op",
"organizadoras",
" osa",
"osa ",
", osb",
" osb",
"osb ",
//"osc",
"osf ",
" osf",
//"osu",
//"p.",
"phd",
"ph d",
"ph.d.",
//"phd",
"preface",
//"q.",
//"r.",
"r.s.c.j.",
"revised",
"romans",
"rscj",
//"rsm",
//"s.",
"s.a.",
" s.c.",
"s.c. ",
"s.c.",
"s.c.l.",
"s.c.n.",
"s.d.b.",
"s.g.s.",
"s.j.",
"s.m.",
"s.m.a.",
"s.n.d.de n.",
"s.n.j.m.",
"s.p.",
"s.s.",
"s.s.c.",
"s.s.s.",
"s.t.d.",
//"sa",
", sc",
" sc ",
"sch p",
"sch.p.",
//"scl",
//"scn",
//"sdb",
//"sgs",
" sj,",
", sj",
//"sm",
//"sma",
"snd de n",
"sndden",
"snjm",
//"sp",
//"sr",
//"sr.",
", ss",
"ssc",
"sss",
"std",
"study",
"study guide author:",
//"t.",
"text edited",
" the ",
"traductora",
"translated by",
"translated",
"translation",
"translators",
"translator",
//"u.",
//"v.",
//"w.",
" with an ",
" with a ",
" with ",
"written"
//"x.",
//"y",
//"y.",
//"z."
};
if (!File.Exists(parasitesFile))
{
File.WriteAllLines(parasitesFile, initialParasites.ToArray());
}
var parasites = File.ReadAllLines(parasitesFile);
query = query.ToLowerInvariant();
foreach (var parasite in parasites)
{
query = query.Replace(parasite, String.Empty);
}
string splitterFile = Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.CommonApplicationData), ConfigurationManager.AppSettings["DataFolder"], "MoreByThisAuthorSplitter.txt");
var initialSplitter = new[] { ",", " and ", ";", ":", "<br>", "<br/>", "<br />" };
if (!File.Exists(splitterFile))
{
File.WriteAllLines(splitterFile, initialSplitter);
}
var splitter = File.ReadAllLines(splitterFile);
var queryChunks = query.ToLowerInvariant().Split(splitter, StringSplitOptions.RemoveEmptyEntries);
foreach (var chunk in queryChunks)
{
var queryAssembler = new QueryAssembler();
var sanitizedChunk = Regex.Replace(chunk, @"[^\p{Ll}\p{Lu}\p{Lt}\p{Lo}\p{Nd}\p{Pc}]\.", " ").Replace(". ", " ").Trim(' ');
if (String.IsNullOrWhiteSpace(sanitizedChunk))
continue;
queryAssembler.Append(sanitizedChunk);
preparedQueries.Add(queryAssembler.Assemble());
}
return preparedQueries.ToArray();
}
public static string Prepare(string query)
{
var queryChunks = query.Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
var queryAssembler = new QueryAssembler();
bool isExactQueryChunk = false;
string exactQueryChunks = String.Empty;
foreach (var item in queryChunks)
{
if (isExactQueryChunk || item.StartsWith("\""))
{
exactQueryChunks = String.Format("{0} {1}", exactQueryChunks, item);
if (item.EndsWith("\""))
{
queryAssembler.Append(exactQueryChunks);
exactQueryChunks = String.Empty;
isExactQueryChunk = false;
}
else
{
isExactQueryChunk = true;
continue;
}
}
else
queryAssembler.Append(item);
}
return queryAssembler.Assemble();
}
