private int APIGenereateHashValue(APICompany company)
{
var companyName = company.CompanyName;
var country = company.Country;
var website = company.Website ?? " ";
var address = company.Address ?? " ";
var sme = company.SME.ToString();
var categories = "";
foreach (var item in company.SupplyChainCategories)
{
string code = item.SupplyChainCategoryCode.ToString().Split(".")[0];
categories = categories + code.Replace(item.SuperCategory.ToString(), "");
}
var roles = "";
foreach (var item in company.SupplyChainRoles)
{
roles = roles + item.SupplyChainRoleCode.ToString();
}
var totalString = "";
totalString = companyName + country + website + address + sme + categories + roles;
var trimmedAndLowerInvariant = totalString.Trim().ToLowerInvariant();
var spaceRemoved = trimmedAndLowerInvariant.Replace(" ", "");
int hashValue = spaceRemoved.GetHashCode();
return(hashValue);
}
private List <APICompany> ConvertToApiCompany(
List <DataStreamCompany> dataStreamCompanies,
List <SupplyChainRole> SupplyChainRoles,
List <SupplyChainCategory> SuppleChainCategories)
{
var ApiCompanies = new List <APICompany>();
dataStreamCompanies.ForEach(item =>
{
var company = new APICompany();
//Special for Country.
{
if (item.country.Equals("Netherlands"))
{
item.country = "The Netherlands";
}
if (item.country.Equals("UK"))
{
item.country = "United Kingdom";
}
}
company.CompanyName = item.company_name;
company.Country = item.country;
company.Created = Convert.ToDateTime(item.last_updated.date);
company.Website = item.website;
company.SME = item.sme_status;
company.CompanyDirectoryEntryReffNumber = item.entry_reference_number;
company.Status = item.status;
company.SupplyChainCategories = ConvertSupplyChainCategories(item.supply_chain_categories, SuppleChainCategories);
company.SupplyChainRoles = ConvertSupplyChainRoles(item.supply_chain_roles, SupplyChainRoles);
if (item.offices.Any())
{
var office = item.offices.ElementAt(0);
company.Address = office.address ?? "";
company.Latitude = Decimal.Parse(office.lat ?? "0");
company.Longitude = Decimal.Parse(office.lng ?? "0");
//company.Created = DateTime.Now;
}
else
{
company.Address = "";
company.Latitude = 0;
company.Longitude = 0;
//company.Created = DateTime.Now;
}
ApiCompanies.Add(company);
});
return(ApiCompanies);
}
public PotentialDiplicate(DataStreamCompany duplicate, DataStreamCompany match, APICompany match2)
{
this.Duplicate = duplicate;
this.Match = match;
this.Match2 = match2;
}
/// <summary>
/// First reduceses the datastream data by comparing it with itself and removing duplicates.
/// Secund compares with apidata and again removes duplicates if stream data appears in api data.
/// Special: If api data contains Reff key we know for sertain it originates from datastream
/// wee want too keep those for update perpause. We dont consider the api company for matching if it contains reff key.
/// </summary>
/// <param name="dataStreamCompanies"></param>
/// <param name="apiCompanies"></param>
/// <returns></returns>
private List <DataStreamCompany> FindPotentialDuplicates(
List <DataStreamCompany> dataStreamCompanies,
List <APICompany> apiCompanies)
{
var reducedDatastreamCompanies = new List <DataStreamCompany>();
var filteredDataStreamCompanies = new List <DataStreamCompany>();
foreach (var item in dataStreamCompanies)
{
var match = false;
var matchItem = new DataStreamCompany();
//Check the item status. Skips the item if its not one of theses two.
if (!item.status.Equals("Approved") && !item.status.Equals("Partner Added"))
{
continue;
}
//Filteres potential duplicated from datastreamcompanies via comparing it to itself.
foreach (var x in dataStreamCompanies)
{
if (item.entry_reference_number.Equals(x.entry_reference_number))
{
continue;
}
//if countries match wee will go to next step
if (MatchCountries(x.country).Equals(MatchCountries(item.country)))
{
//If name is IN ANY WAY contined within another company we consider it a potential duplicate.
//If true, Item is added to reduced and potentialduplicates.
var xname = x.company_name.Trim().ToLowerInvariant();
var itemname = item.company_name.Trim().ToLowerInvariant();
if (xname.Contains(itemname))
{
match = true;
matchItem = x;
break;
}
else
{
match = false;
}
}
else
{
match = false;
}
}
if (match == false)
{
reducedDatastreamCompanies.Add(item);
}
else
{
this._datastreamPotentialDuplicates.Add(new PotentialDiplicate(item, matchItem, null));
}
}
//filteres portential duplicates away by comparing the reduced list with companies from the API.
//If y has Reff key. It origins from datasteam and we add it to filtered it because its difinitive the same and we might want to update it.
//If y has no reff key and x does not appear in apiCompanies we add it to reduced
reducedDatastreamCompanies.ForEach(x =>
{
var xHash = StreamGenereateHashValue(x);
var match = false;
var skip = false;
var matchItem = new APICompany();
foreach (var y in apiCompanies)
{
if (y.CompanyDirectoryEntryReffNumber == null || y.CompanyDirectoryEntryReffNumber.Equals(""))
{
if (x.company_name.ToLowerInvariant() == y.CompanyName.ToLowerInvariant())
{
var hit = true;
}
//checks for exact match
var yHash = APIGenereateHashValue(y);
if (xHash == yHash)
{
y.CompanyDirectoryEntryReffNumber = x.entry_reference_number;
this._exactHashMatch.Add(y);
skip = true;
break;
}
//if countries match wee will go to next step
if (MatchCountries(x.country).Equals(MatchCountries(y.Country)))
{
//If name is IN ANY WAY contined within another company we consider it a potential duplicate.
//If true, Item is added to reduced and potentialduplicates.
var xname = x.company_name.Trim().ToLowerInvariant();
var yname = y.CompanyName.Trim().ToLowerInvariant();
if (xname.Contains(yname))
{
match = true;
matchItem = y;
break;
}
else
{
match = false;
}
}
else
{
match = false;
}
}
else if (x.entry_reference_number.Equals(y.CompanyDirectoryEntryReffNumber))
{
match = false;
}
}
if (!skip)//we skip this if we have an exact math on hash value.
{
if (match == false)
{
filteredDataStreamCompanies.Add(x);
}
else
{
this._datastreamPotentialDuplicates.Add(new PotentialDiplicate(x, null, matchItem));
}
}
});
return(filteredDataStreamCompanies);
}
/// <summary>
/// Converts stream companies to API company format.
/// </summary>
/// <param name="streamCompanies">Commapny from stream</param>
/// <returns>Formatet companies</returns>
private List <APICompany> ConvertStreamCompanies(
List <DataStreamCompany> streamCompanies,
List <SupplyChainRole> SupplyChainroles,
List <SupplyChainCategory> SuppleChainCategories
)
{
var convertedCompanies = new List <APICompany>();
var filteredStreamCompanies = PrintValues(streamCompanies);
foreach (var item in filteredStreamCompanies)
{
//Check the item status.
if (!item.status.Equals("Approved") && !item.status.Equals("Partner Added"))
{
continue;
}
var company = new APICompany();
//Special for Country.
{
if (item.country.Equals("Netherlands"))
{
item.country = "The Netherlands";
}
if (item.country.Equals("UK"))
{
item.country = "United Kingdom";
}
}
company.CompanyName = item.company_name;
company.Country = item.country;
company.Website = item.website;
company.SME = item.sme_status;
company.CompanyDirectoryEntryReffNumber = item.entry_reference_number;
company.Status = item.status;
company.SupplyChainCategories = ConvertSupplyChainCategories(item.supply_chain_categories, SuppleChainCategories);
company.SupplyChainRoles = ConvertSupplyChainRoles(item.supply_chain_roles, SupplyChainroles);
if (item.offices.Any())
{
//We are getting empty data this must be taken into consieration.
var office = item.offices.ElementAt(0);
try
{
company.Address = office.address;
company.Latitude = Decimal.Parse(office.lat);
company.Longitude = Decimal.Parse(office.lng);
company.Created = DateTime.Now;
}
catch (Exception e)
{
//if(String.IsNullOrWhiteSpace(office.address))
// continue;
company.Address = office.address;
company.Latitude = 0;
company.Longitude = 0;
company.Created = DateTime.Now;
}
}
else
{
company.Address = "";
company.Latitude = 0;
company.Longitude = 0;
company.Created = DateTime.Now;
//continue;
}
convertedCompanies.Add(company);
}
Console.WriteLine(convertedCompanies.Count + " :convertedCompanies");
Console.ReadLine();
return(convertedCompanies);
}
