//private
//public cq Results { get; set; }
//public Selector selector;
public cq(string html)
{
var dom = new Dom(html).DomContainer;
//this.Elements = new List<DomElement> { dom };
this.Parent = this;
this.Elements = new List<DomElement> { dom };
//this.Document = null;
//this.Results = new cq(this.domEls);
//this.selector = null;
//string Text = this.Dom.Text();
}
public void TestMethod1()
{
//var reader = new StringReader("c");
//var c = reader.Read();
//c = reader.Read();
cq cq = new cq("<div id='my-id'>Hello<div class='class1 class2 class3'> world<ul class='my-class'> Some Text here...</ul></div></div>");
//cq sz2 = cq["#my-id .class2"];
//cq sz3 = sz2[".my-class"];
cq = cq[".class3 ul"];
//Assert.AreEqual(cq.te)
//var sz2 = cq["p div#my-id.class1.class2.class3"];
}
public cq this[string sel]
{
get
{
var selector = new Selector(sel);
List<DomElement> matchEls = new List<DomElement>();
matchEls.AddRange(this.Elements);
foreach(var s in selector.Selectors)
{
matchEls = GetElements(matchEls, s);
}
var sz = new cq(this.Parent, matchEls);
return sz;
}
}
static void Main(string[] args)
{
//var contents = File.ReadAllText(@"C:\Users\tporter\Desktop\Extracts\CSV\artworks.txt");
//var lines = contents.Split(new string[] { "\r\n" }, StringSplitOptions.RemoveEmptyEntries);
// `(.*?)`( .*?) (.*)
//$1$2||
//string header = "id bigint(20)||site varchar(30)||isLive int(1)||record_type int(1)||type varchar(125)||typeSort varchar(255)||typeID bigint(20)||selected_artwork_types Text||stockNumber varchar(255)||stockNumberPrefix varchar(8)||stockNumberNumber int(8)||stockNumberSuffix varchar(8)||artistStockNumber varchar(255)||barcode varchar(255)||artist varchar(255)||artistSort varchar(255)||artistSortTitleYear varchar(255)||artistID bigint(20)||artist_id bigint(20)||title Text||titleSort varchar(255)||artistTitleYear Text||displayDate varchar(255)||isEdition int(1)||editionPublisherOrNotes Text||medium Text||dimensions Text||signedAndDated Text||periodID bigint(20)||series varchar(60)||artistDates varchar(60)||detailsOverride Text||hasWebsiteOverride int(1)||websiteOverride Text||editionDetails Text||editionMasterRecord int(1)||editionNumber int(5)||editionNumberManuallyEntered varchar(60)||editionNumberAsText varchar(60)||editionArtistProofNumber int(5)||editionArtistProof int(1)||editionTotal int(6)||editionArtistProofTotal int(6)||editionID bigint(20)||printedStatus int(1)||marketID int(1)||financial_preset bigint(20)||gallery bigint(20)||consultant Text||location_id bigint(20)||location_code varchar(128)||locationContactID bigint(20)||location Text||location2 Text||locationNotes Text||locationHistory Text||locationUpdateDate date||country_of_origin varchar(60)||lastExaminedBy Text||lastExaminedDate date||currentOwner Text||current_owner_id bigint(20)||currentOwnerNotes Text||currentOwnerUpdateDate date||shippedToContactID bigint(20)||shipped_to_location_id bigint(20)||shipping Text||shippingNotes Text||shippingHistory Text||shippingUpdateDate date||shipperID bigint(20)||shipper varchar(60)||importNumber Text||exportNumber Text||imageData Text||hasImage int(1)||image_last_modified datetime||pictureCredits Text||additionalPictureFilenames Text||additionalPhotoCredits Text||exhibitions Text||exhibitionIDs Text||accountingCurrency bigint(20)||consignedBy varchar(120)||consignedByID bigint(20)||consigneeIsTheArtist int(1)||consignedFromDate date||consignedToDate date||consigned_in_reminder_date date||consignedByPercentageToGallery decimal(24,8)||consignedByNetValue decimal(18,2)||consignedByCalculateFromNet int(1)||consignorContractSigned int(1)||consignorNonStandardContract int(1)||consignmentNotes Text||consignmentReturned int(1)||consignedTo varchar(120)||consignedToID bigint(20)||consignedToFromDate date||consignedToToDate date||consigned_out_reminder_date date||consignedToPercentage decimal(24,8)||consignedToNetValue decimal(18,2)||consignedToCalculateFromNet int(1)||consignmentToNotes Text||consignedToReturned int(1)||loanedTo varchar(120)||loanedToID bigint(20)||loanedToFromDate date||loanedToToDate date||loaned_to_reminder_date date||loanHistory Text||loanedToNotes Text||loanedToReturned int(1)||purchasedFrom varchar(120)||purchasedFromID bigint(20)||purchaseDate date||purchasePrice decimal(18,2)||purchaseCurrency bigint(20)||purchaseTaxRate Text||purchase_price_for_calculations decimal(18,2)||purchaseConversionRate decimal(24,8)||purchasePriceAltCurrencyCurrencyID bigint(20)||purchasePriceAltCurrencyConversionRate decimal(24,8)||purchasePriceAltCurrencyValue decimal(18,2)||purchasePriceInSaleCurrency decimal(18,2)||purchaseShare decimal(18,2)||ownedInShares int(1)||otherOwners Text||otherOwnersShare decimal(18,2)||purchase_paid_date date||purchase_notes Text||production_status int(1)||production_status_date date||productionCostsTotalGallery decimal(18,2)||productionCostsTotalThirdParties decimal(18,2)||productionCostsTotalArtist decimal(18,2)||productionCostsTotal decimal(18,2)||productionCostDetails Text||productionCostsInvoiceText Text||costsIncurredGallery Text||costsIncurredThirdParties Text||costsIncurredArtist Text||costsIncurredArtists Text||costsPaidBy Text||costs_breakdown mediumblob||financialAdditionalInfo Text||notesOnCommissions Text||accounts_version decimal(24,8)||retailPrice decimal(18,2)||retailCurrency bigint(20)||retailPriceOverrideTextAfter int(1)||retailPriceTextAfter varchar(30)||retailTaxRate decimal(24,8)||apply_tax_at_current_rate int(1)||selling_price_for_calculations decimal(18,2)||retailPriceExtrasText varchar(120)||retailPriceExtras decimal(18,2)||retailPriceExtrasPaidByArtist int(1)||retailPriceExtrasAddToRetailPrice int(1)||retailPriceShippingExtrasText varchar(120)||retailPriceShippingExtras decimal(18,2)||retailPriceShippingExtrasPaidByArtist int(1)||retailPriceShippingExtrasAddToRetailPrice int(1)||retailPriceTotalExTax decimal(18,2)||retailPriceTotal decimal(18,2)||retailPriceCalculateFromTotal int(1)||retailPriceUpdateDate date||retailPriceHistory mediumblob||retail_data_json mediumblob||arr_applies int(1)||arr_notes Text||display_price_override varchar(60)||display_price_tax_inclusive_override varchar(60)||display_price_currency_2 varchar(60)||display_price_currency_3 varchar(60)||calculateAccountsAutomatically int(1)||marginPurchaseTaxRate decimal(24,8)||margin decimal(18,2)||importDutyPaid decimal(18,2)||on_ta_report int(1)||ta_reminder_date date||ta_expiry_date date||ta_notes Text||insured int(1)||insuranceValue decimal(18,2)||insuranceValueDate date||insuranceValuedBy varchar(255)||insured_by varchar(255)||insured_by_id bigint(20)||insurance_value_curr1 decimal(18,2)||insurance_value_curr2 decimal(18,2)||insurance_value_curr3 decimal(18,2)||insurance_value_curr4 decimal(18,2)||insurance_value_curr1_date date||insurance_value_curr2_date date||insurance_value_curr3_date date||insurance_value_curr4_date date||insurance_value_curr1_note varchar(120)||insurance_value_curr2_note varchar(120)||insurance_value_curr3_note varchar(120)||insurance_value_curr4_note varchar(120)||insuranceNotes Text||insuranceHistory mediumblob||status int(3)||availability int(3)||availability_detail Text||paymentStatus int(1)||prevStatus int(3)||invoiceID bigint(20)||invoiceDocumentID bigint(20)||invoiceNumber varchar(60)||invoiceName Text||soldTo varchar(120)||soldToID bigint(20)||saleDate date||salePrice decimal(18,2)||saleCurrencyID bigint(20)||saleCurrency varchar(8)||saleDiscount decimal(24,8)||saleDiscountValue decimal(18,2)||invoiceDiscountRate decimal(24,8)||invoiceDiscountValue decimal(18,2)||taxStatus bigint(20)||is_margin_scheme int(1)||saleTax decimal(18,2)||saleTaxRate decimal(24,8)||salePriceExtrasText varchar(120)||salePriceExtras decimal(18,2)||saleInvoiceValue decimal(18,2)||final_sale_price_total decimal(18,2)||threeWaySplit int(1)||removeProductionBeforeDiscount int(1)||saleConversionRate decimal(24,8)||proformaConversionRate decimal(24,8)||taxConversionRate decimal(24,8)||taxProformaConversionRate decimal(24,8)||saleDetails Text||sales_data_json mediumblob||offeredTo varchar(120)||offerPrice varchar(120)||offerNotes Text||offeredToID bigint(20)||offeredDate date||offeredBy varchar(120)||underOffer int(1)||exclusiveOfferPrevAvailability int(3)||offers Text||reservedFor varchar(120)||reservedForID bigint(20)||reservedForUntil date||reserves Text||courtesyOrCreditLine Text||copyrightInfo Text||provenance Text||literature Text||publications Text||publicationIDs Text||collections mediumblob||showOnWebsite int(1)||additional_website_export_item int(1)||showPictureOnWebsite int(1)||showPriceOnWebsite int(1)||websiteCopyrightLine Text||description_for_website mediumtext||replace_artwork_detail_with_video int(1)||use_video_on_private_views int(1)||include_secondary_images_in_private_views int(1)||video_embed_code Text,||add_to_online_shop int(1)||featured_in_shop int(1)||websiteOption1 int(1)||websiteSort int(8)||sort_for_store int(8)||showOnArtistsPage int(1)||in_artists_representative_works int(1)||artists_representative_works_sort int(8)||selected_categories Text||is_defining_work_in_categories Text||represents_category Text||represents_period int(1)||featured_artwork int(1)||main_image Text||secondary_images mediumtext||technical_images mediumtext||show_secondary_images int(1)||main_image_uploaded int(1)||main_image_dimensions varchar(12)||images_hash varchar(60)||mediaFiles Text||auctionHistory Text||press Text||commentaryOrDescription Text||commentaryOrDescriptionLang2 Text||excludeDescriptionFromPresentations int(1)||excludeProvenanceFromPresentations int(1)||excludeLiteratureFromPresentations int(1)||excludePublicationsFromPresentations int(1)||excludeExhibitionsFromPresentations int(1)||publicPresentationPage int(1)||publicPresentationPageUID varchar(60)||condition Text||conditionReportFilepath Text||conditionReportStatus int(3)||conservation Text||materials Text||manufacturer Text||framed int(1)||framing Text||weight varchar(255)||shipping_weight decimal(24,8)||shipping_volumetric_weight decimal(24,8)||shipping_width decimal(24,8)||shipping_height decimal(24,8)||shipping_depth decimal(24,8)||packingAndCrating Text||packingAndCratingNotes Text||availablePhotographs Text||photographyStatus int(1)||slideLabelOverride Text||links Text||notes Text||locked int(1)||isArchive int(1)||ownedPersonally int(1)||financialLocked int(1)||showAllFinancialFields int(1)||certificatesInfo Text||additionalDocumentsInfo Text||searchTerms Text||narrative_search mediumtext||selected_artwork_lists Text||selected_interests Text||custom_int_field1 bigint(20)||custom_int_field2 bigint(20)||custom_int_field3 bigint(20)||custom_int_field4 bigint(20)||custom_dec_field1 decimal(24,12)||custom_dec_field2 decimal(24,12)||sales_points mediumtext||share_sales_points_to_privateviews int(1)||fact_sheet_auto int(1)||fact_sheet_override mediumtext||share_fact_sheet_to_privateviews int(1)||share_fact_sheet_as_qr_code int(1)||privateviews_export_documents mediumtext||attached_high_resolution_images mediumtext,||high_res_image_for_website Text||qr_code_page_url Text||qr_code_image_url Text||qr_code_page_settings Text||pv_by_artist_include Text||pv_by_artist_exclude Text||pv_by_artist_cover_image int(1)||pv_by_artist_always_show_price Text||pv_by_artist_hide_price_by_default int(1)||pv_by_artist_do_not_export_price int(1)||pv_by_artwork_type_include Text||pv_by_artwork_type_exclude Text||pv_by_artwork_type_cover_image int(1)||pv_by_artwork_type_hide_price_by_default int(1)||pv_by_artwork_type_do_not_export_price int(1)||pv_by_artwork_list_include Text||pv_by_artwork_list_exclude Text||pv_by_artwork_list_cover_image int(1)||pv_by_artwork_list_hide_price_by_default int(1)||pv_by_artwork_list_do_not_export_price int(1)||sharing_options Text||import_source Text||importID varchar(40)||imported_json_data mediumblob||importID2 varchar(40)||importID3 varchar(40)||importSoldToID varchar(40)||importedFinancialInformation Text||importPurchaseCurrency varchar(8)||importRetailCurrency varchar(8)||importSaleCurrency varchar(8)||importedArtworkTypes Text||importedInterests Text||importedStatus varchar(30)||importedAvailability varchar(30)||importedTaxStatus varchar(30)||importedConsigneeID varchar(30)||importedNFS int(1)||importedImageData Text||lastSavedBy varchar(50)||lastSaved datetime||createdBy varchar(50)||creationDate date||modificationDate timestamp||revision_id varchar(60)||flRecsCheckedItems Text||_checked_items Text";
string header = "id bigint(20)||site varchar(30)||isLive int(1)||record_type int(1)||type varchar(125)||typeSort varchar(255)||typeID bigint(20)||selected_artwork_types text||stockNumber varchar(255)||stockNumberPrefix varchar(8)||stockNumberNumber int(8)||stockNumberSuffix varchar(8)||artistStockNumber varchar(255)||barcode varchar(255)||artist varchar(255)||artistSort varchar(255)||artistSortTitleYear varchar(255)||artistID bigint(20)||artist_id bigint(20)||title text||titleSort varchar(255)||artistTitleYear text||displayDate varchar(255)||isEdition int(1)||editionPublisherOrNotes text||medium text||dimensions text||signedAndDated text||periodID bigint(20)||series varchar(60)||artistDates varchar(60)||detailsOverride text||hasWebsiteOverride int(1)||websiteOverride text||editionDetails text||editionMasterRecord int(1)||editionNumber int(5)||editionNumberManuallyEntered varchar(60)||editionNumberAsText varchar(60)||editionArtistProofNumber int(5)||editionArtistProof int(1)||editionTotal int(6)||editionArtistProofTotal int(6)||editionID bigint(20)||printedStatus int(1)||marketID int(1)||financial_preset bigint(20)||gallery bigint(20)||consultant text||location_id bigint(20)||location_code varchar(128)||locationContactID bigint(20)||location text||location2 text||locationNotes text||locationHistory text||locationUpdateDate date||country_of_origin varchar(60)||lastExaminedBy text||lastExaminedDate date||currentOwner text||current_owner_id bigint(20)||currentOwnerNotes text||currentOwnerUpdateDate date||shippedToContactID bigint(20)||shipped_to_location_id bigint(20)||shipping text||shippingNotes text||shippingHistory text||shippingUpdateDate date||shipperID bigint(20)||shipper varchar(60)||importNumber text||exportNumber text||imageData text||hasImage int(1)||image_last_modified datetime||pictureCredits text||additionalPictureFilenames text||additionalPhotoCredits text||exhibitions text||exhibitionIDs text||accountingCurrency bigint(20)||consignedBy varchar(120)||consignedByID bigint(20)||consigneeIsTheArtist int(1)||consignedFromDate date||consignedToDate date||consigned_in_reminder_date date||consignedByPercentageToGallery decimal(24,8)||consignedByNetValue decimal(18,2)||consignedByCalculateFromNet int(1)||consignorContractSigned int(1)||consignorNonStandardContract int(1)||consignmentNotes text||consignmentReturned int(1)||consignedTo varchar(120)||consignedToID bigint(20)||consignedToFromDate date||consignedToToDate date||consigned_out_reminder_date date||consignedToPercentage decimal(24,8)||consignedToNetValue decimal(18,2)||consignedToCalculateFromNet int(1)||consignmentToNotes text||consignedToReturned int(1)||loanedTo varchar(120)||loanedToID bigint(20)||loanedToFromDate date||loanedToToDate date||loaned_to_reminder_date date||loanHistory text||loanedToNotes text||loanedToReturned int(1)||purchasedFrom varchar(120)||purchasedFromID bigint(20)||purchaseDate date||purchasePrice decimal(18,2)||purchaseCurrency bigint(20)||purchaseTaxRate text||purchase_price_for_calculations decimal(18,2)||purchaseConversionRate decimal(24,8)||purchasePriceAltCurrencyCurrencyID bigint(20)||purchasePriceAltCurrencyConversionRate decimal(24,8)||purchasePriceAltCurrencyValue decimal(18,2)||purchasePriceInSaleCurrency decimal(18,2)||purchaseShare decimal(18,2)||ownedInShares int(1)||otherOwners text||otherOwnersShare decimal(18,2)||purchase_paid_date date||purchase_notes text||production_status int(1)||production_status_date date||productionCostsTotalGallery decimal(18,2)||productionCostsTotalThirdParties decimal(18,2)||productionCostsTotalArtist decimal(18,2)||productionCostsTotal decimal(18,2)||productionCostDetails text||productionCostsInvoiceText text||costsIncurredGallery text||costsIncurredThirdParties text||costsIncurredArtist text||costsIncurredArtists text||costsPaidBy text||costs_breakdown mediumtext||financialAdditionalInfo text||notesOnCommissions text||accounts_version decimal(24,8)||retailPrice decimal(18,2)||retailCurrency bigint(20)||retailPriceOverrideTextAfter int(1)||retailPriceTextAfter varchar(30)||retailTaxRate decimal(24,8)||apply_tax_at_current_rate int(1)||selling_price_for_calculations decimal(18,2)||retailPriceExtrasText varchar(120)||retailPriceExtras decimal(18,2)||retailPriceExtrasPaidByArtist int(1)||retailPriceExtrasAddToRetailPrice int(1)||retailPriceShippingExtrasText varchar(120)||retailPriceShippingExtras decimal(18,2)||retailPriceShippingExtrasPaidByArtist int(1)||retailPriceShippingExtrasAddToRetailPrice int(1)||retailPriceTotalExTax decimal(18,2)||retailPriceTotal decimal(18,2)||retailPriceCalculateFromTotal int(1)||retailPriceUpdateDate date||retailPriceHistory mediumtext||retail_data_json mediumtext||arr_applies int(1)||arr_notes text||display_price_override varchar(60)||display_price_tax_inclusive_override varchar(60)||display_price_currency_2 varchar(60)||display_price_currency_3 varchar(60)||calculateAccountsAutomatically int(1)||marginPurchaseTaxRate decimal(24,8)||margin decimal(18,2)||importDutyPaid decimal(18,2)||on_ta_report int(1)||ta_reminder_date date||ta_expiry_date date||ta_notes text||insured int(1)||insuranceValue decimal(18,2)||insurance_currency_id int(1)||insuranceValueDate date||insuranceValuedBy varchar(255)||insured_by varchar(255)||insured_by_id bigint(20)||insurance_value_curr1 decimal(18,2)||insurance_value_curr2 decimal(18,2)||insurance_value_curr3 decimal(18,2)||insurance_value_curr4 decimal(18,2)||insurance_value_curr1_date date||insurance_value_curr2_date date||insurance_value_curr3_date date||insurance_value_curr4_date date||insurance_value_curr1_note varchar(120)||insurance_value_curr2_note varchar(120)||insurance_value_curr3_note varchar(120)||insurance_value_curr4_note varchar(120)||insuranceNotes text||insuranceHistory mediumtext||status int(3)||availability int(3)||availability_detail text||paymentStatus int(1)||prevStatus int(3)||invoiceID bigint(20)||invoiceDocumentID bigint(20)||invoiceNumber varchar(60)||invoiceName text||soldTo varchar(120)||soldToID bigint(20)||saleDate date||salePrice decimal(18,2)||saleCurrencyID bigint(20)||saleCurrency varchar(8)||saleDiscount decimal(24,8)||saleDiscountValue decimal(18,2)||invoiceDiscountRate decimal(24,8)||invoiceDiscountValue decimal(18,2)||taxStatus bigint(20)||is_margin_scheme int(1)||saleTax decimal(18,2)||saleTaxRate decimal(24,8)||salePriceExtrasText varchar(120)||salePriceExtras decimal(18,2)||saleInvoiceValue decimal(18,2)||final_sale_price_total decimal(18,2)||threeWaySplit int(1)||removeProductionBeforeDiscount int(1)||saleConversionRate decimal(24,8)||proformaConversionRate decimal(24,8)||taxConversionRate decimal(24,8)||taxProformaConversionRate decimal(24,8)||saleDetails text||sales_data_json mediumtext||offeredTo varchar(120)||offerPrice varchar(120)||offerNotes text||offeredToID bigint(20)||offeredDate date||offeredBy varchar(120)||underOffer int(1)||exclusiveOfferPrevAvailability int(3)||offers text||reservedFor varchar(120)||reservedForID bigint(20)||reservedForUntil date||reserves text||courtesyOrCreditLine text||copyrightInfo text||provenance text||literature text||publications text||publicationIDs text||collections mediumtext||showOnWebsite int(1)||additional_website_export_item int(1)||showPictureOnWebsite int(1)||showPriceOnWebsite int(1)||websiteCopyrightLine text||description_for_website mediumtext||replace_artwork_detail_with_video int(1)||use_video_on_private_views int(1)||include_secondary_images_in_private_views int(1)||video_embed_code text||add_to_online_shop int(1)||featured_in_shop int(1)||websiteOption1 int(1)||websiteSort int(8)||sort_for_store int(8)||showOnArtistsPage int(1)||in_artists_representative_works int(1)||artists_representative_works_sort int(8)||selected_categories text||is_defining_work_in_categories text||represents_category text||represents_period int(1)||featured_artwork int(1)||main_image text||secondary_images mediumtext||images mediumtext||technical_images mediumtext||show_secondary_images int(1)||main_image_uploaded int(1)||main_image_dimensions varchar(12)||images_hash varchar(60)||mediaFiles text||auctionHistory text||press text||commentaryOrDescription text||commentaryOrDescriptionLang2 text||excludeDescriptionFromPresentations int(1)||excludeProvenanceFromPresentations int(1)||excludeLiteratureFromPresentations int(1)||excludePublicationsFromPresentations int(1)||excludeExhibitionsFromPresentations int(1)||publicPresentationPage int(1)||publicPresentationPageUID varchar(60)||condition text||conditionReportFilepath text||conditionReportStatus int(3)||conservation text||materials text||manufacturer text||framed int(1)||framing text||weight varchar(255)||shipping_weight decimal(24,8)||shipping_volumetric_weight decimal(24,8)||shipping_width decimal(24,8)||shipping_height decimal(24,8)||shipping_depth decimal(24,8)||packingAndCrating text||packingAndCratingNotes text||availablePhotographs text||photographyStatus int(1)||slideLabelOverride text||links text||notes text||locked int(1)||isArchive int(1)||ownedPersonally int(1)||financialLocked int(1)||showAllFinancialFields int(1)||certificatesInfo text||additionalDocumentsInfo text||searchTerms text||narrative_search mediumtext||selected_artwork_lists text||selected_interests text||custom_int_field1 bigint(20)||custom_int_field2 bigint(20)||custom_int_field3 bigint(20)||custom_int_field4 bigint(20)||custom_dec_field1 decimal(24,12)||custom_dec_field2 decimal(24,12)||sales_points mediumtext||share_sales_points_to_privateviews int(1)||fact_sheet_auto int(1)||fact_sheet_override mediumtext||share_fact_sheet_to_privateviews int(1)||share_fact_sheet_as_qr_code int(1)||privateviews_export_documents mediumtext||attached_high_resolution_images mediumtext||high_res_image_for_website text||qr_code_page_url text||qr_code_image_url text||qr_code_page_settings text||pv_by_artist_include text||pv_by_artist_exclude text||pv_by_artist_cover_image int(1)||pv_by_artist_always_show_price text||pv_by_artist_hide_price_by_default int(1)||pv_by_artist_do_not_export_price int(1)||pv_by_artwork_type_include text||pv_by_artwork_type_exclude text||pv_by_artwork_type_cover_image int(1)||pv_by_artwork_type_hide_price_by_default int(1)||pv_by_artwork_type_do_not_export_price int(1)||pv_by_artwork_list_include text||pv_by_artwork_list_exclude text||pv_by_artwork_list_cover_image int(1)||pv_by_artwork_list_hide_price_by_default int(1)||pv_by_artwork_list_do_not_export_price int(1)||sharing_options text||import_source text||importID varchar(40)||imported_json_data mediumtext||importID2 varchar(40)||importID3 varchar(40)||importSoldToID varchar(40)||importedFinancialInformation text||importPurchaseCurrency varchar(8)||importRetailCurrency varchar(8)||importSaleCurrency varchar(8)||importedArtworkTypes text||importedInterests text||importedStatus varchar(30)||importedAvailability varchar(30)||importedTaxStatus varchar(30)||importedConsigneeID varchar(30)||importedNFS int(1)||importedImageData text||lastSavedBy varchar(50)||lastSaved datetime||createdBy varchar(50)||creationDate date||modificationDate timestamp||revision_id varchar(60)||flRecsCheckedItems text||_checked_items text||";
var header2 = header.Split(new string[] { "||" }, StringSplitOptions.RemoveEmptyEntries);
var htmlFields = new List<string> { "literature", "provenance", "exhibited", "publications" };
DataTable dt = new DataTable();
foreach(var h in header2)
{
var header3 = h.Split(' ');
var title = header3[0];
var type = header3[1];
Type columnType = null;
if (type.Contains("varchar") || type.Contains("text") || type.Contains("blob"))
columnType = typeof(string);
else if (type.Contains("bigint") || type.Contains("decimal"))
columnType = typeof(double);
else if (type.Contains("int"))
columnType = typeof(int);
else if (type.Contains("date") || type.Contains("time"))
columnType = typeof(DateTime);
else
throw new Exception();
columnType = typeof(string);
dt.Columns.Add(title, columnType);
}
//var contents = File.ReadAllText(@"C:\Users\tporter\Desktop\Extracts\CSV\artworks.txt", Encoding.UTF8);
StreamReader reader = new StreamReader(@"C:\Users\tporter\Desktop\Extracts\CSV\artworks-7-14_2.txt", Encoding.UTF8);
//Regex rx = new Regex("\\\r(?=\n)");
//contents = rx.Replace(contents, "\n");
var fieldCount = header2.Count();
var delimiter = ",";
var newLine = "\r\n";
//var position = 0;
//var xx = (char)53;
//var data = new List<List<string>>();
//var lineData = new List<string>();
var row = new object[fieldCount];
var i = 0;
while(true)
{
if (reader.EndOfStream)
break;
var c = reader.Peek();
//var test = contents[position + 1];
if (c == '"')
{
StringBuilder token = new StringBuilder();
//position++;
reader.Read();
while (true)
{
//c = contents[position++];
c = reader.Read();
if (c == '\\')
{
//token.Append((char)c);
var c2 = reader.Read();
if (c2 == '\r')
{
token.Append("<br />");
}
else if (c2 == '"')
{
token.Append("\"");
}
else
{
token.Append((char)c);
token.Append((char)c2);
}
//var c2 = contents[position++];
//token.Append(c2);
}
else if (c == '"')
{
var cell = token.ToString();
//var head = dt.Columns[i];
//if (htmlFields.Contains(head.ColumnName) && cell.Length > 0)
//{
//}
row[i++] = cell;
//lineData.Add(token.ToString());
//dt.Rows.
break;
}
else
{
token.Append((char)c);
}
}
}
else if (c == ',')
{
//lineData = new List<string>();
reader.Read();
//position++;
continue;
}
else if (c == '\r')
{
reader.Read();
c = reader.Read();
//c = contents[++position];
if (c == '\n')
{
//position++;
//reader.Read();
//if (lineData.Count() != fieldCount)
// throw new Exception();
//data.Add(lineData);
//lineData = new List<string>();
dt.Rows.Add(row);
row = new object[fieldCount];
i = 0;
}
else
{
throw new Exception();
}
}
else if (c == '\\')
{
//c = contents[++position];
reader.Read();
c = reader.Read();
if (c == 'N')
{
reader.Read();
//position++;
//lineData.Add("");
row[i++] = "";
}
else
throw new Exception();
}
}
var columns = File.ReadAllText(@"c:\users\tporter\desktop\artworksHeaders.txt").Split(new string[] { "\r\n" }, StringSplitOptions.RemoveEmptyEntries);
var columns2 = columns.Select(x => x.Split(new string[] { "\t" }, StringSplitOptions.RemoveEmptyEntries));
//List<string> bad = new List<string>();
//var sql = "CREATE TABLE ArtWorks (~~";
foreach(var c in columns2)
{
var first = c.First();
if (dt.Columns.Contains(first))
{
//var thisCol = dt.Columns[first];
dt.Columns[first].ColumnName = c.Last().Replace(" ", "_");
//sql += c.Last().Replace(" ", "_") + " NVARCHAR(MAX) NULL,~~";
}
else
{
//throw new Exception();
//bad.Add(c.First());
}
}
//sql = sql.Substring(0, sql.Length - 2);
//sql += ")";
DataClasses1DataContext db = new DataClasses1DataContext();
db.ExecuteCommand("DELETE FROM ArtWorks");
//var query = (from a in db.ArtWorks_Dirties
// select a);
//db.ArtWorks_Dirties.DeleteAllOnSubmit(query);
//db.SubmitChanges();
var count = dt.Rows.Count;
var artWorks = new List<ArtWork>();
//Parallel.ForEach(dt.AsEnumerable(), (row2) =>
foreach(var row2 in dt.AsEnumerable())
{
var id = row2["id"].ToString();
var aw = new ArtWork();
var properties = typeof(ArtWork).GetProperties();
foreach (var property in properties)
{
if (dt.Columns.Contains(property.Name))
{
if (htmlFields.Contains(property.Name))
{
//if (row2["id"].ToString() == "34")
//{
//}
var prop = row2[property.Name].ToString();
if (prop.Length == 0)
{
property.SetValue(aw, "");
}
else
{
//var prop2 = Regex.Replace(prop, "<!--.*?-->", "");
//CsQuery.CQ el = prop2;
//CsQuery.CQ test = "<p><em>Hello</em><em>World</em></p>";
//test.Select("em").Before("{EM_BEFORE}");
//test.Select("em").After("{EM_AFTER}");
//test.Remove("em");
//var test2 = test.Text().Replace("{EM_BEFORE}", "<em>").Replace("{EM_AFTER}", "</em>");
//el.Select("br");
//.Before("{BREAK}");
//el.Remove("br");
//el.Select("em").Before("{EM_BEFORE}");
//el.Select("em").After("{EM_AFTER}");
//el.Remove("em");
//el.Select("i").Before("{EM_BEFORE}");
//el.Select("i").After("{EM_AFTER}");
//el.Remove("i");
//el.Select("strong").Before("{STRONG_BEFORE}");
//el.Select("strong").After("{STRONG_AFTER}");
//el.Remove("strong");
//el.Select("b").Before("{STRONG_BEFORE}");
//el.Select("b").After("{STRONG_AFTER}");
//el.Remove("b");
//var Text = el.Text();
//var sb = new StringBuilder(Text)
// .Replace("{BREAK}", "<br />")
// .Replace("{EM_BEFORE}", "<em>")
// .Replace("{EM_AFTER}", "</em>")
// .Replace("{STRONG_BEFORE}", "<strong>")
// .Replace("{STRONG_AFTER}", "</strong>");
//if (artWorks.Count() == 2781)
//{
// File.WriteAllText(@"c:\users\tporter\desktop\debug.txt", prop);
//}
cQuery.cq sz = new cQuery.cq(prop);
sz["br"].Before("<br />");
sz["em,i"].Before("<em>");
sz["em,i"].After("</em>");
//cq["i"].Before("<em>");
//cq["i"].After("</em>");
sz["strong,b"].Before("<strong>");
sz["strong,b"].After("</strong>");
//cq["b"].Before("<strong>");
//cq["b"].After("</strong>");
//cq em = cq["em"];
var text = sz.Text();
//var sz2 = cq["em"];
//var sb = new StringBuilder();
property.SetValue(aw, text);
}
}
else
property.SetValue(aw, row2[property.Name]);
}
}
artWorks.Add(aw);
Console.WriteLine(artWorks.Count() + "/" + count);
//db.ArtWorks.InsertOnSubmit(aw);
}
//);
db.ArtWorks.InsertAllOnSubmit(artWorks);
//foreach(DataRow row2 in dt.Rows)
//{
// var aw = new ArtWork();
// var properties = typeof(ArtWork).GetProperties();
// foreach(var property in properties)
// {
// if (dt.Columns.Contains(property.Name))
// {
// if (htmlFields.Contains(property.Name))
// {
// var prop = row2[property.Name].ToString();
// CsQuery.CQ el = prop;
// el.Select("br").Before("{BREAK}");
// el.Remove("br");
// el.Select("em").Before("{EM_BEFORE}");
// el.Select("em").After("{EM_AFTER}");
// el.Remove("em");
// el.Select("i").Before("{EM_BEFORE}");
// el.Select("i").After("{EM_AFTER}");
// el.Remove("i");
// el.Select("strong").Before("{STRONG_BEFORE}");
// el.Select("strong").After("{STRONG_AFTER}");
// el.Remove("strong");
// el.Select("b").Before("{STRONG_BEFORE}");
// el.Select("b").After("{STRONG_AFTER}");
// el.Remove("b");
// //var Text = el.Text();
// string Text = "";
// if (el.Count() > 0)
// {
// Text = el.Children().Text().Replace("{BREAK}", "<br />")
// .Replace("{EM_BEFORE}", "<em>")
// .Replace("{EM_AFTER}", "</em>")
// .Replace("{STRONG_BEFORE}", "<strong>")
// .Replace("{STRONG_AFTER}", "</strong>");
// }
// //else
// //{
// //}
// //Text = Text.Replace("{BREAK}", "<br />")
// // .Replace("{EM_BEFORE}", "<em>")
// // .Replace("{EM_AFTER}", "</em>")
// // .Replace("{STRONG_BEFORE}", "<strong>")
// // .Replace("{STRONG_AFTER}", "</strong>");
// property.SetValue(aw, Text);
// }
// else
// property.SetValue(aw, row2[property.Name]);
// }
// }
// db.ArtWorks.InsertOnSubmit(aw);
//}
db.SubmitChanges();
db.Dispose();
Console.WriteLine("Complete!");
Console.ReadLine();
//dt.Columns.
//foreach(DataColumn col in dt.Columns)
//{
// if (columns2.Where(x => x.First() == col.ColumnName).Count() > 0)
// {
// }
// else
// {
// //throw new Exception();
// }
//}
//ExcelControl ec = new ExcelControl();
//ec.CreateWorkbook(@"C:\Users\tporter\Desktop\Extracts\CSV\artworks - Copy.xlsx", dt);
//int xx = 3;
//new StreamReader(new MemoryStream())
//TextFieldParser parser = new TextFieldParser(@"C:\Users\tporter\Desktop\Extracts\CSV\artworks.txt");
////TextFieldParser parser = new TextFieldParser();
//parser.TextFieldType = FieldType.Delimited;
//parser.SetDelimiters(",");
//parser.HasFieldsEnclosedInQuotes = true;
//while (!parser.EndOfData)
//{
// //Process row
// string[] fields = parser.ReadFields();
// if (fields.Count() != header2.Count())
// throw new Exception();
// for(int i = 0; i < fields.Count(); ++i)
// {
// var header3 = header2[i].Split(' ');
// var title = header3[0];
// var type = header3[1];
// Type columnType = null;
// if (type.Contains("varchar") || type.Contains("Text") || type.Contains("blob"))
// columnType = typeof(string);
// else if (type.Contains("bigint") || type.Contains("decimal"))
// columnType = typeof(double);
// else if (type.Contains("int"))
// columnType = typeof(int);
// else if (type.Contains("date") || type.Contains("time"))
// columnType = typeof(DateTime);
// else
// throw new Exception();
// }
// //foreach (string field in fields)
// //{
// //}
//}
//parser.Close();
}
private cq(cq parent, List<DomElement> els)
{
this.Parent = parent;
this.Elements = els;
}
