private List <FormulaItem> BuildFormula(List <string> lines, SearchableContent searchData)
{
List <FormulaItem> formulaItems = new List <FormulaItem>();
foreach (string line in lines)
{
List <string> casArray = helpers.GetCASNumbers(line);
foreach (string cas in casArray)
{
FormulaItem formulaItem = new FormulaItem
{
CASNumber = cas,
OtherInfo = line
};
if (searchData.KnownChemicals.Exists(c => c.CAS == cas))
{
formulaItem.ChemName = searchData.KnownChemicals.Where(c => c.CAS == cas).FirstOrDefault().ChemName;
formulaItem.Score = (int)Scores.HIGH_SCORE;
}
else
{
formulaItem.ChemName = "NAME NOT FOUND";
formulaItem.Score = (int)Scores.MEDIUM_SCORE;
}
formulaItems.Add(formulaItem);
}
}
return(formulaItems);
}
public TextProcessorEngine(IMemoryCache _cache, ServiceSettings _settings)
{
cache = _cache;
settings = _settings;
helpers = new Helpers(cache, settings);
docItemBuilder = new DocItemBuilder(cache, settings);
formulaBuilder = new FormulaBuilder(cache, settings);
searchData = new SearchableContent()
{
KnownChemicals = new List <Component>(),
OtherIdentifiers = new List <OtherIdentifier>(),
SectionHeaders = new List <SectionHeader>(),
SectionList = new List <Section>()
};
searchData.KnownChemicals = cache.GetOrCreate <List <Component> >("knownComponents",
cacheEntry =>
{
return(helpers.GetKnownComponents());
});
searchData.SectionHeaders = cache.GetOrCreate <List <SectionHeader> >("sectionHeaders",
cacheEntry =>
{
return(helpers.GetSectionHeaders());
});
searchData.OtherIdentifiers = cache.GetOrCreate <List <OtherIdentifier> >("otherIdentifiers",
cacheEntry =>
{
return(helpers.GetOtherIdentifiers());
});
}
public DocItem SearchForItem(DocItem docItem, List <string> textlines, SearchableContent searchData)
{
foreach (string searchText in docItem.Terms)
{
int currentLine = 0;
string termType = docItem.Hint.ToLower();
/*
* if (helpers.IsOnlyNumbers(docItem.Section))
* {
* int stopSecton = int.Parse(docItem.Section);
* stopSecton = stopSecton++;
* textlines = helpers.GetSection(textlines, docItem.Section, stopSecton.ToString(), searchData);
* } */
foreach (string line in textlines)
{
string evalLine = line.ToLower();
if (evalLine.Contains(searchText.ToLower()))
{
string restofLine = evalLine.Substring(evalLine.IndexOf(searchText) + searchText.Length + 1).Trim();
if (termType == "number")
{
List <string> evalWords = evalLine.Split(" ").ToList();
evalWords.RemoveAll(x => x.Trim() == string.Empty);
docItem = CheckValue(docItem, searchText, restofLine,
evalWords, line.ToLower(), textlines, currentLine);
}
else if (termType == "text")
{
docItem = CheckText(docItem, restofLine);
}
else if (termType == "yesno")
{
docItem = CheckBool(docItem, restofLine);
}
else
{
docItem = CheckText(docItem, restofLine);
}
}
if (!string.IsNullOrEmpty(docItem.Result))
{
Log.Debug($"[Found DocItem Term]: {searchText} [Score]: {docItem.Score} [Result]: {docItem.Result} ");
return(docItem);
}
else
{
Log.Debug($"[Missed DocItem Term]: {searchText} ");
}
currentLine++;
}
}
return(docItem);
}
public void Index(SearchableContent content)
{
_elasticSearchRepository.EnsureMappingExist();
_elasticSearchRepository.Save(content);
}
public List <string> GetSection(List <string> textlines, string startSection, string stopSection, SearchableContent searchData)
{
//sectionHeaders string are already lower case
List <string> frag = new List <string>();
//if we already parsed the section use it again
if (searchData.SectionList.Any(s => s.Number == startSection))
{
frag = searchData.SectionList.Where(s => s.Number == startSection).First().Content;
return(frag);
}
else
{
List <string> startTextList = searchData.SectionHeaders.Where(x => x.Number == startSection).Select(y => y.Title.ToLower()).ToList();
List <string> stopTextList = searchData.SectionHeaders.Where(x => x.Number == stopSection).Select(y => y.Title.ToLower()).ToList();
frag = GetDocumentFragment(startTextList, stopTextList, textlines);
if (!searchData.SectionList.Any(s => s.Number == startSection))
{
searchData.SectionList.Add(new Section()
{
Number = startSection, Content = frag
});
}
return(frag);
}
}
public List <FormulaItem> GetFormulation(List <string> textlines, string requestGuid, SearchableContent searchData)
{
List <FormulaItem> items = new List <FormulaItem>();
List <string> formulaLines = helpers.GetSection(textlines, "3", "4", searchData);
if (formulaLines.Count > 0)
{
items = BuildFormula(formulaLines, searchData);
}
else
{
if (items.Count == 0)
{
formulaLines = helpers.GetSection(textlines, "2", "3", searchData); //try pre GHS structure
if (formulaLines.Count > 0)
{
items = BuildFormula(formulaLines, searchData);
}
if (items.Count == 0)
{
Log.Information($"No formula section found for request {requestGuid}");
}
}
}
return(items);
}
