public void TestCSVWriter()
{
string expectedCsv = UtilsForTesting.LoadCsv("Resources/csv/argentina_diputados_voting_record.csv");
Table table = this.GetTable();
/*
* StringBuilder sb = new StringBuilder();
* (new CSVWriter()).write(sb, table);
* String s = sb.ToString();
* String[] lines = s.Split("\\r?\\n");
* assertEquals(EXPECTED_CSV_WRITER_OUTPUT, lines[0]);
* assertEquals(expectedCsv, s);
*/
using (var stream = new MemoryStream())
using (var sb = new StreamWriter(stream)
{
AutoFlush = true
})
{
(new CSVWriter()).Write(sb, table);
var reader = new StreamReader(stream);
stream.Position = 0;
var s = reader.ReadToEnd().Trim(); // trim to remove last new line
string[] lines = s.Split("\r\n"); // "\\r?\\n"
Assert.Equal(EXPECTED_CSV_WRITER_OUTPUT, lines[0]);
Assert.Equal(expectedCsv, s.Replace("\r\n", "\n"));
}
}
public void TestCSVMultilineRow()
{
string expectedCsv = UtilsForTesting.LoadCsv("Resources/csv/frx_2012_disclosure.csv");
PageArea page = UtilsForTesting.GetAreaFromFirstPage("Resources/frx_2012_disclosure.pdf", new PdfRectangle(double.NaN, double.NaN, double.NaN, double.NaN)); // 53.0f, 49.0f, 735.0f, 550.0f);
SpreadsheetExtractionAlgorithm sea = new SpreadsheetExtractionAlgorithm();
Table table = sea.Extract(page)[0];
StringBuilder sb = new StringBuilder();
(new CSVWriter()).Write(sb, table);
string s = sb.ToString();
Assert.Equal(expectedCsv, s);
/*
* using (var stream = new MemoryStream())
* using (var sb = new StreamWriter(stream) { AutoFlush = true })
* {
* (new CSVWriter()).write(sb, table);
* var reader = new StreamReader(stream);
* stream.Position = 0;
* var s = reader.ReadToEnd().Trim(); // trim to remove last new line
* Assert.Equal(expectedCsv, s);
* }
*/
}
public void TestCSVSerializeTwoTables()
{
string expectedCsv = UtilsForTesting.LoadCsv("Resources/csv/twotables.csv");
List <Table> tables = this.GetTables();
/*
* StringBuilder sb = new StringBuilder();
* (new CSVWriter()).write(sb, tables);
* String s = sb.toString();
* assertEquals(expectedCsv, s);
*/
using (var stream = new MemoryStream())
using (var sb = new StreamWriter(stream)
{
AutoFlush = true
})
{
(new CSVWriter()).Write(sb, tables);
var reader = new StreamReader(stream);
stream.Position = 0;
var s = reader.ReadToEnd().Trim(); // trim to remove last new line
Assert.Equal(expectedCsv, s);
}
}
public void TestCSVSerializeInfinity()
{
string expectedCsv = UtilsForTesting.LoadCsv("Resources/csv/schools.csv");
// top, left, bottom, right // page height = 612
// 53.74f, 16.97f, 548.74f, 762.3f)
PageArea page = UtilsForTesting.GetAreaFromFirstPage("Resources/schools.pdf", new PdfRectangle(16.97, 612 - 548.74, 762.3, 612 - 53.74 - 1)); // remove 1 because add an empty line at the top if not
SpreadsheetExtractionAlgorithm sea = new SpreadsheetExtractionAlgorithm();
Table table = sea.Extract(page)[0];
StringBuilder sb = new StringBuilder();
(new CSVWriter()).Write(sb, table);
string s = sb.ToString();
Assert.Equal(expectedCsv.Trim(), s.Replace("\r\n", "\n"));
/*
* using (var stream = new MemoryStream())
* using (var sb = new StreamWriter(stream) { AutoFlush = true })
* {
* (new CSVWriter()).write(sb, table);
* var reader = new StreamReader(stream);
* stream.Position = 0;
* var s = reader.ReadToEnd().Trim(); // trim to remove last new line
* Assert.Equal(expectedCsv, s);
* }
*/
}
public void TestSpanningCellsToCsv()
{
PageArea page = UtilsForTesting.GetPage("Resources/spanning_cells.pdf", 1);
string expectedCsv = UtilsForTesting.LoadCsv("Resources/csv/spanning_cells.csv");
SpreadsheetExtractionAlgorithm se = new SpreadsheetExtractionAlgorithm();
List <Table> tables = se.Extract(page);
Assert.Equal(2, tables.Count);
StringBuilder sb = new StringBuilder();
(new CSVWriter()).Write(sb, tables);
Assert.Equal(expectedCsv, sb.ToString().Replace("\r\n", "\n").Trim());
}
public void StreamNoGuess1()
{
PageArea page = UtilsForTesting.GetPage("Resources/data.pdf", 1);
// data_stream_noguess.csv was modified for decimal precision
string expectedCsv = UtilsForTesting.LoadCsv("Resources/csv/data_stream_noguess.csv");
BasicExtractionAlgorithm se = new BasicExtractionAlgorithm();
List <Table> tables = se.Extract(page);
StringBuilder sb = new StringBuilder();
(new CSVWriter()).Write(sb, tables[0]);
Assert.Equal(expectedCsv, sb.ToString().Replace("\r\n", "\n"));
}
public void TestSpreadsheetWithNoBoundingFrameShouldBeSpreadsheet()
{
PageArea page = UtilsForTesting.GetAreaFromPage("Resources/spreadsheet_no_bounding_frame.pdf", 1, new PdfRectangle(58.9, 842 - 654.7, 536.12, 842 - 150.56)); // 842 - 150.56)); // 150.56f, 58.9f, 654.7f, 536.12f);
string expectedCsv = UtilsForTesting.LoadCsv("Resources/csv/spreadsheet_no_bounding_frame.csv");
SpreadsheetExtractionAlgorithm se = new SpreadsheetExtractionAlgorithm();
bool isTabular = se.IsTabular(page);
Assert.True(isTabular);
List <Table> tables = se.Extract(page);
StringBuilder sb = new StringBuilder();
(new CSVWriter()).Write(sb, tables[0]);
Assert.Equal(expectedCsv, sb.ToString());
}
public void StreamNoGuess1()
{
// tabula.read_pdf(pdf_path, stream=True, guess=False)
PageArea page = UtilsForTesting.GetPage("Resources/data.pdf", 1);
//
string expectedCsv = UtilsForTesting.LoadCsv("Resources/csv/data_stream_noguess.csv");
BasicExtractionAlgorithm se = new BasicExtractionAlgorithm();
List <Table> tables = se.Extract(page);
StringBuilder sb = new StringBuilder();
(new CSVWriter()).Write(sb, tables[0]);
Assert.Equal(expectedCsv, sb.ToString().Replace("\r\n", "\n"));
}
public void Latice1()
{
PageArea page = UtilsForTesting.GetPage("Resources/data.pdf", 1);
// data_lattice.csv was modified to add the last row, missing in tabula_py
string expectedCsv = UtilsForTesting.LoadCsv("Resources/csv/data_lattice.csv");
SpreadsheetExtractionAlgorithm se = new SpreadsheetExtractionAlgorithm();
List <Table> tables = se.Extract(page);
Assert.Single(tables);
StringBuilder sb = new StringBuilder();
(new CSVWriter()).Write(sb, tables[0]);
Assert.Equal(expectedCsv, sb.ToString().Replace("\r\n", "\n"));
}
public void TestTableWithMultilineHeader()
{
string expectedCsv = UtilsForTesting.LoadCsv("Resources/csv/us-020.csv");
PageArea page = UtilsForTesting.GetAreaFromPage("Resources/us-020.pdf", 2, new PdfRectangle(35.0, 151, 560, 688.5));
BasicExtractionAlgorithm bea = new BasicExtractionAlgorithm();
Table table = bea.Extract(page)[0];
using (var stream = new MemoryStream())
using (var sb = new StreamWriter(stream)
{
AutoFlush = true
})
{
(new CSVWriter()).Write(sb, table);
var reader = new StreamReader(stream);
stream.Position = 0;
var data = reader.ReadToEnd().Replace("\r\n", "\n").Trim(); // trim to remove last new line
Assert.Equal(expectedCsv, data);
}
}
public void TestRealLifeRTL2()
{
string expectedCsv = UtilsForTesting.LoadCsv("Resources/csv/indictb1h_14.csv");
PageArea page = UtilsForTesting.GetAreaFromPage("Resources/indictb1h_14.pdf", 1, new PdfRectangle(120.0, 842 - 622.82, 459.9, 842 - 120.0));
BasicExtractionAlgorithm bea = new BasicExtractionAlgorithm();
Table table = bea.Extract(page)[0];
using (var stream = new MemoryStream())
using (var sb = new StreamWriter(stream)
{
AutoFlush = true
})
{
(new CSVWriter()).Write(sb, table);
var reader = new StreamReader(stream);
stream.Position = 0;
var data = reader.ReadToEnd().Replace("\r\n", "\n").Trim(); // trim to remove last new line
Assert.Equal(expectedCsv, data);
}
}
public void TestSpreadsheetExtractionIssue656()
{
// page height = 482, width 762.3 // 612
// top, left, bottom, right
// 56.925f, 24.255f, 549.945f, 786.555f);
PageArea page = UtilsForTesting.GetAreaFromFirstPage("Resources/Publication_of_award_of_Bids_for_Transport_Sector__August_2016.pdf", new PdfRectangle(24.255, 71, 786.555, 553));
string expectedCsv = UtilsForTesting.LoadCsv("Resources/csv/Publication_of_award_of_Bids_for_Transport_Sector__August_2016.csv");
SpreadsheetExtractionAlgorithm sea = new SpreadsheetExtractionAlgorithm();
List <Table> tables = sea.Extract(page);
Assert.Single(tables);
Table table = tables[0];
StringBuilder sb = new StringBuilder();
(new CSVWriter()).Write(sb, table);
string result = sb.ToString();
Assert.Equal(expectedCsv.Replace("\n", "\r"), result.Replace("\r\n", "\n").Replace("\n", "\r").Trim());
/*
* using (var stream = new MemoryStream())
* using (var sb = new StreamWriter(stream) { AutoFlush = true })
* {
* (new CSVWriter()).write(sb, table);
*
* var reader = new StreamReader(stream);
* stream.Position = 0;
* var s = reader.ReadToEnd().Trim(); // trim to remove last new line
*
* // is there an issue with \r and \n?
* Assert.Equal(expectedCsv.Replace("\n", "\r"), s.Replace("\r\n", "\n").Replace("\n", "\r"));
* }
*/
}
