/// <summary>
/// Learns a program to extract a single region using two examples in two different files.
/// Learning from different files is similar to learning with multiple examples from a single file.
/// Demonstrates how to learn with examples from different files.
/// </summary>
private static void LearnRegionUsingMultipleFiles()
{
var session = new RegionSession();
StringRegion input1 = RegionSession.CreateStringRegion("Carrie Dodson 100");
StringRegion input2 = RegionSession.CreateStringRegion("Leonard Robledo 75");
session.AddConstraints(
new RegionExample(input1, input1.Slice(7, 13)), // "Carrie Dodson 100" => "Dodson"
new RegionExample(input2, input2.Slice(8, 15)) // "Leonard Robledo 75" => "Robledo"
);
RegionProgram topRankedProg = session.Learn();
if (topRankedProg == null)
{
Console.Error.WriteLine("Error: Learning fails!");
return;
}
StringRegion testInput = RegionSession.CreateStringRegion("Margaret Cook 320"); // expect "Cook"
StringRegion output = topRankedProg.Run(testInput);
if (output == null)
{
Console.Error.WriteLine("Error: Extracting fails!");
return;
}
Console.WriteLine("\"{0}\" => \"{1}\"", testInput, output);
}
/// <summary>
/// Learns a program to extract a region with both positive and negative examples.
/// Demonstrates the use of negative examples.
/// </summary>
private static void LearnRegionWithNegativeExamples()
{
var session = new RegionSession();
StringRegion input =
RegionSession.CreateStringRegion("Carrie Dodson 100\nLeonard Robledo NA\nMargaret Cook 320");
StringRegion[] records = { input.Slice(0, 17), input.Slice(18, 36), input.Slice(37, 54) };
// Suppose we want to extract "100", "320".
session.AddConstraints(
new RegionExample(records[0], records[0].Slice(14, 17)), // "Carrie Dodson 100" => "100"
new RegionNegativeExample(records[1], records[1]) // no extraction in "Leonard Robledo NA"
);
// Extraction.Text will find a program whose output does not OVERLAP with any of the negative examples.
RegionProgram topRankedProg = session.Learn();
if (topRankedProg == null)
{
Console.Error.WriteLine("Error: Learning fails!");
return;
}
foreach (StringRegion record in records)
{
string output = topRankedProg.Run(record)?.Value ?? "null";
Console.WriteLine("\"{0}\" => \"{1}\"", record, output);
}
}
/// <summary>
/// Learns all region programs that satisfy the examples (advanced feature).
/// Demonstrates access to the entire program set.
/// </summary>
private static void LearnAllRegionPrograms()
{
var session = new RegionSession();
StringRegion input = RegionSession.CreateStringRegion("Carrie Dodson 100");
session.AddConstraints(new RegionExample(input, input.Slice(14, 17))); // "Carrie Dodson 100" => "Dodson"
ProgramSet allPrograms = session.LearnAll().ProgramSet;
IEnumerable <ProgramNode> topKPrograms = allPrograms.TopK(RegionLearner.Instance.ScoreFeature, 3);
var i = 0;
StringRegion[] otherInputs =
{
input, RegionSession.CreateStringRegion("Leonard Robledo NA"),
RegionSession.CreateStringRegion("Margaret Cook 320")
};
foreach (ProgramNode programNode in topKPrograms)
{
Console.WriteLine("Program {0}:", ++i);
var program = new RegionProgram(programNode, ReferenceKind.Parent);
foreach (StringRegion str in otherInputs)
{
StringRegion r = program.Run(str);
Console.WriteLine(r == null ? "null" : r.Value);
}
}
}
/// <summary>
/// Learns top-ranked 3 region programs.
/// Demonstrates access to lower-ranked programs.
/// </summary>
private static void LearnTop3RegionPrograms()
{
var session = new RegionSession();
StringRegion input = RegionSession.CreateStringRegion("Carrie Dodson 100");
session.AddConstraints(new RegionExample(input, input.Slice(14, 17))); // "Carrie Dodson 100" => "Dodson"
IEnumerable <RegionProgram> topKPrograms = session.LearnTopK(3);
var i = 0;
StringRegion[] otherInputs =
{
input, RegionSession.CreateStringRegion("Leonard Robledo NA"),
RegionSession.CreateStringRegion("Margaret Cook 320")
};
foreach (RegionProgram prog in topKPrograms)
{
Console.WriteLine("Program {0}:", ++i);
foreach (StringRegion str in otherInputs)
{
var r = prog.Run(str);
Console.WriteLine(r != null ? r.Value : "null");
}
}
}
/// <summary>
/// Learns to serialize and deserialize Extraction.Text program.
/// </summary>
private static void SerializeProgram()
{
var session = new RegionSession();
StringRegion input = RegionSession.CreateStringRegion("Carrie Dodson 100");
session.AddConstraints(new RegionExample(input, input.Slice(7, 13))); // "Carrie Dodson 100" => "Dodson"
RegionProgram topRankedProg = session.Learn();
if (topRankedProg == null)
{
Console.Error.WriteLine("Error: Learning fails!");
return;
}
string serializedProgram = topRankedProg.Serialize();
RegionProgram deserializedProgram = Loader.Instance.Region.Load(serializedProgram);
StringRegion testInput = RegionSession.CreateStringRegion("Leonard Robledo 75"); // expect "Robledo"
StringRegion output = deserializedProgram.Run(testInput);
if (output == null)
{
Console.Error.WriteLine("Error: Extracting fails!");
return;
}
Console.WriteLine("\"{0}\" => \"{1}\"", testInput, output);
}
/// <summary>
/// Learns a program to extract a single region from a file.
/// </summary>
private static void LearnRegion()
{
var session = new RegionSession();
StringRegion input = RegionSession.CreateStringRegion("Carrie Dodson 100");
// Only one example because we extract one region from one file.
// Position specifies the location between two characters in the file. It starts at 0 (the beginning of the file).
// An example is identified by a pair of start and end positions.
session.AddConstraints(new RegionExample(input, input.Slice(7, 13))); // "Carrie Dodson 100" => "Dodson"
RegionProgram topRankedProg = session.Learn();
if (topRankedProg == null)
{
Console.Error.WriteLine("Error: Learning fails!");
return;
}
StringRegion testInput = RegionSession.CreateStringRegion("Leonard Robledo 75"); // expect "Robledo"
StringRegion output = topRankedProg.Run(testInput);
if (output == null)
{
Console.Error.WriteLine("Error: Extracting fails!");
return;
}
Console.WriteLine("\"{0}\" => \"{1}\"", testInput, output);
}
public static async Task <StructureExtractor> TrainExtractorAsync(IEnumerable <Tuple <string, uint, uint> > examples, IEnumerable <string> noneLabeledExamples = null)
{
if (null == examples || !examples.Any())
{
throw new AggregateException($"{nameof(examples)} must not be null or empty");
}
var regionSession = new RegionSession();
foreach (var example in examples)
{
var stringRegion = new StringRegion(example.Item1, Semantics.Tokens);
var field = stringRegion.Slice(example.Item2, example.Item3);
regionSession.AddConstraints(new RegionExample(stringRegion, field));
}
if (noneLabeledExamples?.Any() == true)
{
regionSession.AddInputs(noneLabeledExamples.Select(e => new StringRegion(e, Semantics.Tokens)));
}
var program = await regionSession.LearnAsync();
if (null == program)
{
throw new Exception("No program found");
}
return(new StructureExtractor(program));
}
/// <summary>
/// Learns a program to extract a single region using another region that appears after it as reference (i.e.,
/// succeeding sibling region).
/// Demonstrates how sibling referencing works.
/// </summary>
private static void LearnRegionReferencingSucceedingSibling()
{
var session = new RegionSession();
StringRegion input =
RegionSession.CreateStringRegion("Carrie Dodson 100\nLeonard Robledo 75\nMargaret Cook 320");
StringRegion[] records = { input.Slice(0, 17), input.Slice(18, 36), input.Slice(37, 54) };
StringRegion[] numbers = { input.Slice(14, 17), input.Slice(34, 36), input.Slice(51, 54) };
// Suppose we want to extract the first name w.r.t the number
session.AddConstraints(
new RegionExample(numbers[0], records[0].Slice(0, 6)), // "Carrie" => "100"
new RegionExample(numbers[1], records[1].Slice(18, 25)) // "Leonard" => "75"
);
RegionProgram topRankedProg = session.Learn();
if (topRankedProg == null)
{
Console.Error.WriteLine("Error: Learning fails!");
return;
}
foreach (StringRegion number in numbers)
{
string output = topRankedProg.Run(number)?.Value ?? "null";
Console.WriteLine("\"{0}\" => \"{1}\"", number, output);
}
}
/// <summary>
/// Learns a program to extract a region and provides other references to help find the intended program.
/// Demonstrates the use of additional references.
/// </summary>
private static void LearnRegionWithAdditionalReferences()
{
var session = new RegionSession();
StringRegion input =
RegionSession.CreateStringRegion("Carrie Dodson 100\nLeonard Robledo 75\nMargaret Cook ***");
StringRegion[] records = { input.Slice(0, 17), input.Slice(18, 36), input.Slice(37, 54) };
// Suppose we want to extract "100", "75", and "***".
session.AddConstraints(new RegionExample(records[0], records[0].Slice(14, 17)));
// "Carrie Dodson 100" => "100"
// Additional references help Extraction.Text observe the behavior of the learnt programs on unseen data.
// In this example, if we do not use additional references, Extraction.Text may learn a program that extracts the first number.
// On the contrary, if other references are present, it knows that this program is not applicable on the third record "Margaret Cook ***",
// and promotes a more applicable program.
session.AddInputs(records.Skip(1));
RegionProgram topRankedProg = session.Learn();
if (topRankedProg == null)
{
Console.Error.WriteLine("Error: Learning fails!");
return;
}
foreach (StringRegion record in records)
{
string output = topRankedProg.Run(record)?.Value ?? "null";
Console.WriteLine("\"{0}\" => \"{1}\"", record, output);
}
}
public string RunSingle(string programAsString, string input)
{
var program = Loader.Instance.Region.Load(programAsString);
var inputRegion = RegionSession.CreateStringRegion(input);
var result = program.Run(inputRegion);
return(result.ToString());
}
public static StringRegion RegionFromFile(string path)
{
if (fileCache.ContainsKey(path))
{
return(fileCache[path]);
}
string text = Util.NormalizeBuildLogString(File.ReadAllText(path));
StringRegion region = RegionSession.CreateStringRegion(text);
fileCache[path] = region;
return(region);
}
public static List <StringRegion> LoadBenchmark(string filename, out StringRegion document)
{
string content = File.ReadAllText(filename);
Match[] examples = ExampleRegex.Matches(content).Cast <Match>().ToArray();
document = RegionSession.CreateStringRegion(content.Replace("}", "").Replace("{", ""));
var result = new List <StringRegion>();
for (int i = 0, shift = -1; i < examples.Length; i++, shift -= 2)
{
int start = shift + examples[i].Index;
int end = start + examples[i].Length;
result.Add(document.Slice((uint)start, (uint)end));
}
return(result);
}
public override void Run()
{
if (!Directory.Exists(ReportDirPath))
{
Directory.CreateDirectory(ReportDirPath);
}
Console.WriteLine($"Learning Extraction.Text region program for {SubBenchmarkName}");
Success = false;
while (!Success && TryGetNextExample(out RegionExample nextConstraint))
{
var session = new RegionSession();
UsedExamples.Add(nextConstraint);
session.Constraints.Add(UsedExamples);
Program = session.Learn();
Success = Program != null && AllExamples.All(e => Valid(e, Program));
}
RecordResult();
}
/// <summary>
/// Learns a program to extract a region using positive examples and the matching regular expression.
/// Demonstrates the possibility to give other constraint (regex) to Extraction.Text.
/// This is an advanced feature.
/// </summary>
private static void LearnRegionWithRegexes()
{
StringRegion input =
RegionSession.CreateStringRegion("Carrie Dodson 100\nLeonard Robledo NA\nMargaret Cook 320");
StringRegion[] records = { input.Slice(0, 17), input.Slice(18, 36), input.Slice(37, 54) };
// Suppose we want to extract the number out of a record
var examples = new[]
{
new RegionExample(records[0], records[0].Slice(14, 17)), // "Carrie Dodson 100" => "100"
};
Regex lookBehindRegex = new Regex("\\s");
Regex lookAheadRegex = null;
Regex matchingRegex = new Regex("\\d+");
IEnumerable <RegionProgram> topRankedPrograms = RegionLearner.Instance.LearnTopK(examples,
RegionLearner.Instance.ScoreFeature,
1,
null,
default(ProgramSamplingStrategy),
null,
lookBehindRegex,
matchingRegex,
lookAheadRegex).TopPrograms;
RegionProgram topRankedProg = topRankedPrograms.FirstOrDefault();
if (topRankedProg == null)
{
Console.Error.WriteLine("Error: Learning fails!");
return;
}
foreach (StringRegion record in records)
{
string output = topRankedProg.Run(record)?.Value ?? "null";
Console.WriteLine("\"{0}\" => \"{1}\"", record, output);
}
}
public string LearnSingle(List <TextExtractExample> textExtractExamples)
{
var session = new RegionSession();
var regionExamples = new List <RegionExample>();
foreach (var textExtractExample in textExtractExamples)
{
var inputRegion = RegionSession.CreateStringRegion(textExtractExample.text);
var textExtractSelection = textExtractExample.selections.First(); // at most only one example is added per string region
if (textExtractSelection != null)
{
var exampleRegion = inputRegion.Slice((uint)textExtractSelection.startPos, (uint)textExtractSelection.endPos);
var regionExample = new RegionExample(inputRegion, exampleRegion);
regionExamples.Add(regionExample);
}
}
session.AddConstraints(regionExamples);
var program = session.Learn();
return(program.Serialize());
}
/// <summary>
/// Learns a program to extract a multiple regions using two examples in two different files.
/// Learning multiple regions is similar to learning single regions, it's just doing same task again for different regions.
/// </summary>
private static void LearnMultipleRegionsUsingMultipleFiles(List <string> paths, List <Dictionary <string, string> > regionsToLearn)
{
List <StringRegion> inputs = new List <StringRegion>();
for (int i = 0; i < paths.Count; i++)
{
string s = File.ReadAllText(paths[i]);
inputs.Add(RegionSession.CreateStringRegion(s));
}
int trainingDocumentCount = 2;
List <string> fieldsToLearn = regionsToLearn[0].Keys.ToList();
Dictionary <string, RegionSession> sessionPerField = new Dictionary <string, RegionSession>();
foreach (string field in fieldsToLearn)
{
RegionSession session = new RegionSession();
for (int i = 0; i < trainingDocumentCount; i++)
{
string output = regionsToLearn[i][field];
uint start = inputs[i].IndexOfRelative(output).Value;
uint end = (uint)(start + output.Length);
RegionExample example = new RegionExample(inputs[i], inputs[i].Slice(start, end));
session.AddConstraints(example);
}
sessionPerField.Add(field, session);
}
Dictionary <string, RegionProgram> programPerField = new Dictionary <string, RegionProgram>();
foreach (var fieldSessionPair in sessionPerField)
{
RegionProgram program = fieldSessionPair.Value.Learn();
if (program == null)
{
Console.Error.WriteLine("Error: Learning fails for Field : " + fieldSessionPair.Key);
}
else
{
programPerField.Add(fieldSessionPair.Key, program);
}
}
//testing
StreamWriter outputWriter = new StreamWriter(@"..\..\output.txt");
outputWriter.WriteLine(string.Join("\t|\t", programPerField.Keys));
for (int i = trainingDocumentCount; i < inputs.Count; i++)
{
List <string> values = new List <string>();
foreach (var fieldProgramPair in programPerField)
{
string value = fieldProgramPair.Value.Run(inputs[i])?.Value;
values.Add(value);
}
outputWriter.WriteLine(string.Join("\t|\t\t", values));
}
outputWriter.Flush();
outputWriter.Close();
}
public static void extract(List <string> inputlistString)
{
RegionSession RegionSession = new RegionSession();
string regexQuantity = @"[\d*(\.|\,)?\d*]+\s?(tons|mts|mt|ton|kg|kgs|dwt|mtons)";
string [] records = { "- QUANTITY : 25,000 MTS (BAGS 25KGS)", //0
"15,000 mts +/- 10 % CAN (Axan)", //1
"16,000 mts +/- 10 % CAN-27", //2
"16,000 mts CAN-27", //3
"5,100 mts min/max Nitrabor", //4
"5,100 mts Nitrabor", //5
"6.5 tons", //6
"7 tons", //7
"- Qty : 15,000MT (+/-5% MOLCO)", //8
"7.5 tons",
" 6,600 mts +/- 10 % moloo NPK (19-04-19)",
" 6,600 mts NPK",
" 7,000 mts CAN (Axan)",
" 8,000 mts CAN (Axan)",
" 5000 TONS OPTION 4000 OPTION 2000 TONS DJIBOUTI",
" 9000 TONS MOMBASA",
"(2000mt for kwangyang, 3000mt for ulsan)",
"- 2500 TONS LOAD",
"- Quantity: 20,00mt - 2 pct moloo calcined petcoke",
"- Q’ty: 2 lots x 41,000MT +/- 10% MOLOO",
"- Unit Weight 30.5 Tons.",
"-Q’TTY : ABT. 4,000 – 5,000MT",
"10 m - UW 22 tons",
"150 x 100 x 20 cm = uw 150 kg (x5)",
"18.5 tons",
"19 tons",
"20,000 - 35,000 DWT .",
"25.000 tons metcoke",
"3.500 TONS WOODPULP",
"30.000tons",
"33,500 MT ON 10.65M SSW",
"4.4 tons",
"40-50,000 dwt",
"40.000tons",
"41,000 MTONS",
"44/4570 mts profiles",
"4500/5K TONS CORN",
"5 HO / HA / TPC 44.5 MT",
"5 tons",
"50.000tons",
"5000mtons wric",
"570 x 130 x 130 cm = 3 mt",
"5K TONS CORN",
"6 tons",
"7000MTS, (basis 4cr x 30mts min) / 12000MTS.",
"as p/c fr grd 25ts tng",
"Cargo quantity : 10.000 tons.",
"CARGO&QTTY: ABT50,000MT COAL IN BULK 10PCT MOLCO",
"DWCC: 2000-7000 ts",
"grd sid min 4 x 25 ts cranes",
"Max 20 years / min 25mt cranes",
"Max. 28 M.Tons.",
"No. 1 5,00 m x 4,05 m x 3,95 m = 8.645 kg",
"No. 2 6,30 m x 4,70 m x 3,16 m = 11.995 kg",
"one monthly shipment of 40.000 mt 5%moloo",
"penang(4k mt) +port kelang (24k mt) to luoyuan",
"QTY",
"Qty :20000 MT",
"QTY : 42000 MTS +- 10% MOL CHARTER OPTION",
"Qty (mt)",
"qty - 30,000 mt +/- 2% moloo metalurgical calcined alumina powder in blk",
"Qty 20 – max 40k Rockphosphate (as sole or part cargo)",
"Qty 30K min max - BF Coke , (SF is 65 wog)",
"QTY 5,500 CBM 10% MOLCO (NEED HOLD CAPA MIN 14,000 CBM)",
"QTY : 20,000 MT +/- 10%",
"QTY : 60,000 MT +/- 5%",
"QTY : MIN 27,000 MT - MAX 30,000 MT",
"qty : 30000 +- 10% molco",
"Qty : 50,000 MT +/- 10%",
"Qty : 55000 mt +\\-10",
"Qty above 25k - 40k for 2 port discharge Hazira + Kandla",
"Qty break up - 25,000 mt at Pipavav and 30,000 mt at Porbandar",
"Qty is 6,000GMT & needs 23,000CBM & above!",
"Qty is 7,000GMT (est. 5,200 to 5,300DBMT) & needs 25,000CBM & above!",
"Qty upto 25k for 1 port discharge Hazira",
"Qty: 10,000BDMT +/-10% MOLOO (S/f: 5.4)",
"Qty: 20000MT",
"Qty: 26000 TO 33000 TONS at owners option",
"Qty: 30 kt and 15kt ±10% Moloo",
"Qty: 35,000mt",
"Qty: 35,000mt=0A=",
"QTY: 40,000-45000 MT LIMESTONE",
"qty: 9,000 mt",
"Qty: Firm Quantity-1X (45,000-60,000) MT MOLOO",
"Qty:35,000mt",
"Qty:35,000mt=0A=",
"QTYS : 25,000 ~ 30,000 MT / SHIPMENT (IN BULK OR 1,000KG BIG BAG)",
"Quantity",
"Quantity : 7,000mt – 8,000mt (inchopt)",
"Quantity :50-55,000MT in Bulk ( need geared vsl )",
"quantity : 25000 mt (+/- 5% charter option ) , other details ",
"QUANTITY : 30 – 40,000 M/TONS (ABOUT) TO BE FINALIZED",
"quantity : 25,000 mt +/-10% moloo sand",
"Quantity : 10,000 OR 15,000 Mts",
"QUANTITY : 25000 MT (+/- 5% Charter Option ) , OTHER DETAILS INCLUDING",
"Quantity 30,000 mt +/-5%",
"Quantity (MT)",
"Quantity - 10,000mt +/-10% MOLOO",
"Quantity - 100,000mt +/-10% MOLOO",
"Quantity - 17,000 BDMT+/-10% MOLOO",
"Quantity - 45,000mt +/-10% MOLOO",
"Quantity - 5,000 MT +/-10% MOLOO",
"Quantity - 55,000mt +/-10% MOLOO",
"Quantity 12000 mt +/- 10 Pct choption",
"quantity : (250.000 )mt .",
"Quantity : 10,000 mt, Or 11,000 mt, Or 15,000 Mts, All in Choptn",
"Quantity : 15000mt+/-5% molcopt, Steel scrap for melting in bulk, SF 2.0 abt wog.",
"Quantity : 200,000 Tons",
"Quantity : 200,000 Tons of Aggregate",
"quantity : 40000 +/-10% rock phosphate",
"Quantity : 7000-8,000mt 2% MOLCO",
"Quantity : About 10,000 mt, Or 11,000 mt, Or 15,000 Mts, All in Choptn ",
"Quantity : About 11,000 mt, Or 15,000 Mts, in Choptn",
"QUANTITY : LOADABLE QTY 50,000 MT + 10% (supramax or ultramax only)",
"QUANTITY AS DETERMINED BY THE DRAFT SURVEY WITH THE FLWG REMARKS ONLY:",
"QUANTITY INCREASED",
"Quantity is 42\'500 - 52\'000 MT in Charterer\'s option.",
"quantity requested, if greater.",
"QUANTITY TO BE LOADED AT STOCKTON AND LB IS CHOPT.",
"Quantity:",
"Quantity: 40,000 MT +/- 10% Moloo",
"Quantity: 50,000 MT +/- 10% Moloo",
"Quantity: abt 27,000mt Bulk Harmless Fertilizers",
"Quantity: abt 28,000mt Bulk Harmless Fertilizers.",
"QUANTITY: 1,000 MTS , 5 MTS / COIL",
"Quantity: 1,641 pipes (97 bends) / 5,050 mt / 5,522 cbm 5% molchopt",
"Quantity: 1,656 pipes (114 bends) / 5,069 mt / 5,654 cbm 5% molchopt",
"Quantity: 20,000-40,000mt BHF",
"Quantity: 20,000mts (+/-10%)",
"Quantity: 21.000mt 10% more or less",
"Quantity: 25,000 mts 10pct moloo",
"QUANTITY: 25,000 MTS MIN/MAX",
"Quantity: 25,000 MTS Min/Max (IDEALLY)",
"Quantity: 25.000 MT",
"Quantity: 25000 MT in Bulk",
"Quantity: 25000-27500 MTs in Bulk",
"quantity: 30,000 mt +/- 10% chopt",
"Quantity: 30000mt (ready in the port)",
"Quantity: 30000mt (ready in the port)=0A=",
"quantity: 35,000mt 10% moloo",
"Quantity: 40,000 – 45,000 GMT",
"Quantity: 50,000MT/10% owners option Thermal Coal",
"Quantity: 55,000MT/10% owners option",
"Quantity: 58,000MT/10% owners option",
"Quantity: 8000 Mton (+/ - 5 % molco)",
"QUANTITY: 9,300MT, TRY LESS IF NEEDED",
"QUANTITY: MIN 11,000MT +5% CHOPT",
"Quantity: Min 25.000 - Max 30.000 MT in Charterers option",
"quantity: min 30000 mt + 10pct oo",
"QUANTITY: MIN 41000/ MAX 44000 MT IN OO",
"Quantity: Total 24000mt 10% molchopt Steel scrap",
"TOTAL QUANTITY: ABOUT MIN 150,000 MTS (6 SHIPMENTS) - 350,000MTS +/- 10% PER YEAR IN CHOPT.",
"TTL QUANTITY 250.000 MTS",
"TTL W 900 MTONS",
"TTL W 900MTONS",
"VPT- 8000 MT/day",
"Weight : 1.578 tons/bag",
"Weight_1 : 41000",
"Weight_2 : 41000" };
StringRegion inputRegion = RegionSession.CreateStringRegion(records[0]);
StringRegion inputRegion2 = RegionSession.CreateStringRegion(records[1]);
StringRegion inputRegion3 = RegionSession.CreateStringRegion(records[2]);
StringRegion inputRegion4 = RegionSession.CreateStringRegion(records[3]);
StringRegion inputRegion5 = RegionSession.CreateStringRegion(records[4]);
StringRegion inputRegion6 = RegionSession.CreateStringRegion(records[5]);
StringRegion inputRegion7 = RegionSession.CreateStringRegion(records[6]);
StringRegion inputRegion8 = RegionSession.CreateStringRegion(records[7]);
StringRegion inputRegion9 = RegionSession.CreateStringRegion(records[8]);
StringRegion inputRegion10 = RegionSession.CreateStringRegion(records[9]);
Console.WriteLine(inputRegion.Slice(13, 23));
Console.WriteLine(inputRegion2.Slice(0, 11));
Console.WriteLine(inputRegion3.Slice(0, 11));
IEnumerable <Match> matches = Regex.Matches(@"- QUANTITY : 25,000 MTS (BAGS 25KGS)", regexQuantity, RegexOptions.IgnoreCase);
foreach (var match in matches)
{
Console.WriteLine(match.Value);
}
var examples = new[] {
new RegionExample(inputRegion, inputRegion.Slice(14, 23)),
new RegionExample(inputRegion2, inputRegion2.Slice(1, 11)),
new RegionExample(inputRegion3, inputRegion3.Slice(1, 11)),
};
// RegionSession.Constraints.Add(
// new RegionExample(inputRegion2, new[] {
// inputRegion2.Slice(1, 6), // input => "25,000"
// inputRegion2.Slice(8, 11), // input => "MTS"
// })
// );
// RegionSession.Constraints.Add(
// new RegionExample(inputRegion3, new[] {
// inputRegion3.Slice(1, 6), // input => "25,000"
// inputRegion3.Slice(8, 11), // input => "MTS"
// })
// );
// RegionSession.Constraints.Add(
// new RegionExample(inputRegion4, new[] {
// inputRegion4.Slice(1, 6), // input => "25,000"
// inputRegion4.Slice(8, 11), // input => "MTS"
// })
// );
// RegionSession.Constraints.Add(
// new RegionExample(inputRegion5, new[] {
// inputRegion5.Slice(1, 5), // input => "25,000"
// inputRegion5.Slice(7, 10), // input => "MTS"
// })
// );
// RegionSession.Constraints.Add(
// new RegionExample(inputRegion6, new[] {
// inputRegion6.Slice(1, 5), // input => "25,000"
// inputRegion6.Slice(7, 10), // input => "MTS"
// })
// );
// RegionSession.Constraints.Add(
// new RegionExample(inputRegion7, new[] {
// inputRegion7.Slice(1, 3), // input => "25,000"
// inputRegion7.Slice(5, 8), // input => "MTS"
// })
// );
// RegionSession.Constraints.Add(
// new RegionExample(inputRegion8, new[] {
// inputRegion8.Slice(1, 1), // input => "25,000"
// inputRegion8.Slice(3, 6), // input => "MTS"
// })
// );
// RegionSession.Constraints.Add(
// new RegionExample(inputRegion9, new[] {
// inputRegion9.Slice(9, 14), // input => "25,000"
// inputRegion9.Slice(15, 16), // input => "MTS"
// })
// );
//IEnumerable<RegionProgram> topRankedPrograms = RegionLearner.Instance.LearnTopK(RegionExample, null, 1, regexQuantity);;
// if ( RegionSession.Learn()!= null)
// {
// Console.WriteLine(RegionSession.Learn());
// }
// List <StringRegion> inputlist =new List<StringRegion> ();
// foreach (var input in inputlistString)
// {
// inputlist.Add(RegionSession.CreateStringRegion(input));
// }
// foreach (var result in topRankedProg.Run(inputlist))
// {
// foreach (var item in result)
// {
// Console.WriteLine(item);
// }
// }
IEnumerable <RegionProgram> topRankedPrograms = RegionLearner.Instance.LearnTopK(examples,
RegionLearner.Instance.ScoreFeature,
1,
null,
default(ProgramSamplingStrategy),
null,
new Regex(regexQuantity)).TopPrograms;
RegionProgram topRankedProg = topRankedPrograms.FirstOrDefault();
if (topRankedProg == null)
{
Console.Error.WriteLine("Error: Learning fails!");
return;
}
foreach (string input in inputlistString)
{
StringRegion inputStringRegion = RegionSession.CreateStringRegion(input);
string output = topRankedProg.Run(inputStringRegion)?.Value ?? "null";
Console.WriteLine("\"{0}\" => \"{1}\"", inputStringRegion, output);
}
return;
}
internal void ReifyFromString(string inputString)
{
var s = inputString.Replace("\r\n", "\n");
_stringRegion = RegionSession.CreateStringRegion(s).Slice(Start, End);
}
