public AsciiLineStructure this[AsciiLineAnalysisOption separationStrategy]
{
get
{
return(_structureForSeparation[separationStrategy]);
}
}
public AsciiLineStructure this[AsciiLineAnalysisOption separationStrategy]
{
get
{
return _structureForSeparation[separationStrategy];
}
}
/// <summary>
/// Evaluates the highest scored separation strategy, and stores the winning separation strategy in <see cref="_highestScoredLineAnalysisOption"/> and the corresponding line structure in <see cref="_highestScoredLineStructure"/>.
/// </summary>
private void EvaluateHighestScoredLineAnalysisOption()
{
// determine, which of the separation strategies results in the topmost total priority (product of number of lines and best line priority)
double maxScore = int.MinValue;
var maxScoredEntry = _lineAnalysisOptionsScoring.First();
foreach (var entry in _lineAnalysisOptionsScoring)
{
double score = (double)entry.Value.NumberOfLines * entry.Value.LineStructure.LineStructureScoring;
if (score > maxScore)
{
maxScore = score;
maxScoredEntry = entry;
}
else if (score == maxScore && entry.Value.NumberOfLines > maxScoredEntry.Value.NumberOfLines)
{
maxScoredEntry = entry;
}
}
_highestScoredLineAnalysisOption = maxScoredEntry.Key;
_highestScoredLineStructure = maxScoredEntry.Value.LineStructure;
}
/// <summary>
/// Determines, which lines are the most fr
/// </summary>
/// <param name="analysisOption"></param>
/// <param name="result"></param>
/// <param name="excludeLineStructureHashes"></param>
/// <param name="maxNumberOfEqualLines"></param>
/// <param name="bestLine"></param>
public static void CalculateScoreOfLineAnalysisOption(AsciiLineAnalysisOption analysisOption, IList <AsciiLineAnalysis> result, HashSet <int> excludeLineStructureHashes, out int maxNumberOfEqualLines, out AsciiLineStructure bestLine)
{
// Dictionary, Key is the hash of the line structure hash, Value is the number of lines that have this hash
var numberOfLinesForLineStructureHash = new Dictionary <int, int>();
bestLine = null;
for (int i = 0; i < result.Count; i++)
{
AsciiLineAnalysis lineResults = result[i];
int lineStructureHash = lineResults[analysisOption].GetHashCode(); // and hash code
if (numberOfLinesForLineStructureHash.ContainsKey(lineStructureHash))
{
numberOfLinesForLineStructureHash[lineStructureHash] = 1 + numberOfLinesForLineStructureHash[lineStructureHash];
}
else
{
numberOfLinesForLineStructureHash.Add(lineStructureHash, 1);
}
}
// determine, which of the line structures is the most frequent one
maxNumberOfEqualLines = 0;
int hashOfMostFrequentStructure = 0;
foreach (var dictEntry in numberOfLinesForLineStructureHash)
{
int lineStructureHash = dictEntry.Key;
if (null != excludeLineStructureHashes && excludeLineStructureHashes.Contains(lineStructureHash))
{
continue;
}
int numberOfLines = dictEntry.Value;
if (maxNumberOfEqualLines < numberOfLines)
{
maxNumberOfEqualLines = numberOfLines;
hashOfMostFrequentStructure = lineStructureHash;
}
} // for each
// search for the maximum priority of those lines with the most frequent structure
int maxPriorityOfMostFrequentLines = 0;
for (int i = 0; i < result.Count; i++)
{
AsciiLineAnalysis lineResults = result[i];
if (hashOfMostFrequentStructure == lineResults[analysisOption].GetHashCode())
{
int prty = lineResults[analysisOption].LineStructureScoring;
if (prty >= maxPriorityOfMostFrequentLines)
{
maxPriorityOfMostFrequentLines = prty;
bestLine = lineResults[analysisOption];
}
} // if
} // for
// if the bestLine is a line with a column count of zero, we should use the next best line
// we achieve this by adding the best hash to a list of excluded hashes and call the function again
if (bestLine != null && bestLine.Count == 0)
{
if (null != excludeLineStructureHashes && !excludeLineStructureHashes.Contains(hashOfMostFrequentStructure))
{
excludeLineStructureHashes.Add(hashOfMostFrequentStructure);
CalculateScoreOfLineAnalysisOption(analysisOption, result, excludeLineStructureHashes, out maxNumberOfEqualLines, out bestLine);
return;
}
else if (null == excludeLineStructureHashes)
{
excludeLineStructureHashes = new HashSet <int>()
{
hashOfMostFrequentStructure
};
CalculateScoreOfLineAnalysisOption(analysisOption, result, excludeLineStructureHashes, out maxNumberOfEqualLines, out bestLine);
return;
}
}
}
/// <summary>
/// Determines, which lines are the most fr
/// </summary>
/// <param name="analysisOption"></param>
/// <param name="result"></param>
/// <param name="maxNumberOfEqualLines"></param>
/// <param name="bestLine"></param>
public static void CalculateScoreOfLineAnalysisOption(AsciiLineAnalysisOption analysisOption, IList <AsciiLineAnalysis> result, out int maxNumberOfEqualLines, out AsciiLineStructure bestLine)
{
CalculateScoreOfLineAnalysisOption(analysisOption, result, null, out maxNumberOfEqualLines, out bestLine);
}
/// <summary>
/// Determines, which lines are the most fr
/// </summary>
/// <param name="analysisOption"></param>
/// <param name="result"></param>
/// <param name="excludeLineStructureHashes"></param>
/// <param name="maxNumberOfEqualLines"></param>
/// <param name="bestLine"></param>
public static void CalculateScoreOfLineAnalysisOption(AsciiLineAnalysisOption analysisOption, IList<AsciiLineAnalysis> result, HashSet<int> excludeLineStructureHashes, out int maxNumberOfEqualLines, out AsciiLineStructure bestLine)
{
// Dictionary, Key is the hash of the line structure hash, Value is the number of lines that have this hash
Dictionary<int, int> numberOfLinesForLineStructureHash = new Dictionary<int, int>();
bestLine = null;
for (int i = 0; i < result.Count; i++)
{
AsciiLineAnalysis lineResults = result[i];
int lineStructureHash = lineResults[analysisOption].GetHashCode(); // and hash code
if (numberOfLinesForLineStructureHash.ContainsKey(lineStructureHash))
numberOfLinesForLineStructureHash[lineStructureHash] = 1 + numberOfLinesForLineStructureHash[lineStructureHash];
else
numberOfLinesForLineStructureHash.Add(lineStructureHash, 1);
}
// determine, which of the line structures is the most frequent one
maxNumberOfEqualLines = 0;
int hashOfMostFrequentStructure = 0;
foreach (var dictEntry in numberOfLinesForLineStructureHash)
{
int lineStructureHash = dictEntry.Key;
if (null != excludeLineStructureHashes && excludeLineStructureHashes.Contains(lineStructureHash))
continue;
int numberOfLines = dictEntry.Value;
if (maxNumberOfEqualLines < numberOfLines)
{
maxNumberOfEqualLines = numberOfLines;
hashOfMostFrequentStructure = lineStructureHash;
}
} // for each
// search for the maximum priority of those lines with the most frequent structure
int maxPriorityOfMostFrequentLines = 0;
for (int i = 0; i < result.Count; i++)
{
AsciiLineAnalysis lineResults = result[i];
if (hashOfMostFrequentStructure == lineResults[analysisOption].GetHashCode())
{
int prty = lineResults[analysisOption].LineStructureScoring;
if (prty >= maxPriorityOfMostFrequentLines)
{
maxPriorityOfMostFrequentLines = prty;
bestLine = lineResults[analysisOption];
}
}// if
} // for
// if the bestLine is a line with a column count of zero, we should use the next best line
// we achieve this by adding the best hash to a list of excluded hashes and call the function again
if (bestLine != null && bestLine.Count == 0)
{
if (null != excludeLineStructureHashes && !excludeLineStructureHashes.Contains(hashOfMostFrequentStructure))
{
excludeLineStructureHashes.Add(hashOfMostFrequentStructure);
CalculateScoreOfLineAnalysisOption(analysisOption, result, excludeLineStructureHashes, out maxNumberOfEqualLines, out bestLine);
return;
}
else if (null == excludeLineStructureHashes)
{
excludeLineStructureHashes = new HashSet<int>() { hashOfMostFrequentStructure };
CalculateScoreOfLineAnalysisOption(analysisOption, result, excludeLineStructureHashes, out maxNumberOfEqualLines, out bestLine);
return;
}
}
}
/// <summary>
/// Determines, which lines are the most fr
/// </summary>
/// <param name="analysisOption"></param>
/// <param name="result"></param>
/// <param name="maxNumberOfEqualLines"></param>
/// <param name="bestLine"></param>
public static void CalculateScoreOfLineAnalysisOption(AsciiLineAnalysisOption analysisOption, IList<AsciiLineAnalysis> result, out int maxNumberOfEqualLines, out AsciiLineStructure bestLine)
{
CalculateScoreOfLineAnalysisOption(analysisOption, result, null, out maxNumberOfEqualLines, out bestLine);
}
/// <summary>
/// Evaluates the highest scored separation strategy, and stores the winning separation strategy in <see cref="_highestScoredLineAnalysisOption"/> and the corresponding line structure in <see cref="_highestScoredLineStructure"/>.
/// </summary>
private void EvaluateHighestScoredLineAnalysisOption()
{
// determine, which of the separation strategies results in the topmost total priority (product of number of lines and best line priority)
double maxScore = int.MinValue;
var maxScoredEntry = _lineAnalysisOptionsScoring.First();
foreach (var entry in _lineAnalysisOptionsScoring)
{
double score = (double)entry.Value.NumberOfLines * entry.Value.LineStructure.LineStructureScoring;
if (score > maxScore)
{
maxScore = score;
maxScoredEntry = entry;
}
else if (score == maxScore && entry.Value.NumberOfLines > maxScoredEntry.Value.NumberOfLines)
{
maxScoredEntry = entry;
}
}
_highestScoredLineAnalysisOption = maxScoredEntry.Key;
_highestScoredLineStructure = maxScoredEntry.Value.LineStructure;
}