private void LoadFromJsonRegion(JArray regions, OcrResult result)
{
var regionId = 1;
foreach (var jRegion in regions)
{
var region = new OcrRegion()
{
Id = regionId,
Code = OcrLoaderHelper.GetRegionCode(regionId),
Location = ParseBox(jRegion, result),
Words = result.Words
};
var lineId = 1;
foreach (var jLine in jRegion["lines"])
{
var wordId = 1;
foreach (var jWord in jLine["words"])
{
var word = ParseWord(jWord, result);
result.Words.Add(word);
wordId++;
}
lineId++;
}
result.Regions.Add(region);
regionId++;
}
}
public OcrResult DoParse(string jsonResult, ImageInfo info)
{
var words = new List <OcrWord>();
var json = JObject.Parse(jsonResult);
var lang = default(string);
var result = new OcrResult()
{
Info = info, TextAngle = 0
};
var mainRegion = new OcrRegion()
{
Id = 1,
Code = OcrLoaderHelper.GetRegionCode(1),
Words = words
};
var annotations = json["responses"].Value <JArray>()[0]["textAnnotations"].Value <JArray>();
var wordId = 1;
foreach (var ann in annotations)
{
if (string.IsNullOrWhiteSpace(lang))
{
lang = ann["locale"].Value <string>();
}
if (wordId != 1)
{
var word = new OcrWord()
{
Id = wordId,
Location = GetLocation(ann, info),
Text = ann["description"].Value <string>()
};
words.Add(word);
}
wordId++;
}
mainRegion.Location = new OcrLocation();
mainRegion.Location.X = mainRegion.Words.Min(i => i.Location.X);
mainRegion.Location.Width = mainRegion.Words.Max(i => i.Location.XBound) - mainRegion.Location.X;
mainRegion.Location.Y = mainRegion.Words.Min(i => i.Location.Y);
mainRegion.Location.Height = mainRegion.Words.Max(i => i.Location.YBound) - mainRegion.Location.Y;
result.Regions.Add(mainRegion);
result.Words = words;
return(result);
}
public void TestInitialize()
{
_template = new OcrTemplate
{
CourtId = 1,
DocumentType = "Case Management Statement",
OcrTemplateId = 1
};
var caseNumberRegion = new OcrRegion(1561, 957, 2291, 1158, OcrRegionName.CaseNumber);
var plaintiffRegion = new OcrRegion(150, 830, 1640, 1010, OcrRegionName.Plaintiff);
var attorneyRegion = new OcrRegion(150, 200, 1640, 590, OcrRegionName.Attorney);
var hearingDateRegion = new OcrRegion(150, 1220, 2291, 1350, OcrRegionName.HearingDate);
_template.Regions.Add(caseNumberRegion);
_template.Regions.Add(plaintiffRegion);
_template.Regions.Add(attorneyRegion);
_template.Regions.Add(hearingDateRegion);
}
static void Main(string[] args)
{
var template = new OcrTemplate
{
CourtId = 1,
DocumentType = "Case Management Statement",
OcrTemplateId = 1
};
var caseNumberRegion = new OcrRegion(1561, 957, 2291, 1158, OcrRegionName.CaseNumber);
var plaintiffRegion = new OcrRegion(150, 830, 1640, 1010, OcrRegionName.Plaintiff);
var attorneyRegion = new OcrRegion(150, 200, 1640, 590, OcrRegionName.Attorney);
var hearingDateRegion = new OcrRegion(150, 1220, 2291, 1350, OcrRegionName.HearingDate);
template.Regions.Add(caseNumberRegion);
template.Regions.Add(plaintiffRegion);
template.Regions.Add(attorneyRegion);
template.Regions.Add(hearingDateRegion);
FileInfo pdfFile = new FileInfo("OLPDFS\\43159219.pdf");
var pngFile = CloudOcrService.ConvertPdfToPng(pdfFile);
var results = CloudOcrService.ExtractTemplateFromPdfAsync(pdfFile, template);
foreach (var str in results.Result)
{
}
Console.WriteLine(results.Result.Select(s => s));
foreach (var region in template.Regions)
{
Console.WriteLine(region.RegionName.ToString() + ": " + CloudOcrService.ExtractRegionFromPng(pngFile, region));
}
Console.ReadLine();
}
public Extractor(ConfigOptions options, OcrRegion region, ImageInfo imageInfo)
{
_configOptions = options;
_region = region;
_imageInfo = imageInfo;
}
private string GetRegionText(OcrRegion region)
{
int regionScore = 0;
var rawScores = new List <int>();
var lines = new List <string>();
var sb = new StringBuilder();
foreach (var line in region.Lines)
{
var lineScore = 0;
sb.Clear();
for (int i = 0; i < line.Words.Count; i++)
{
sb.Append(CleanupWord(line.Words[i].Text));
sb.Append(" ");
}
//Calculate line score after clean up the noise chars
var cleanWords = sb.ToString().Split(' ');
for (int i = 0; i < cleanWords.Length; i++)
{
lineScore += GetWordScore(cleanWords[i]);
if (i + 1 < cleanWords.Length)
{
lineScore += GetWordScore2(cleanWords[i], cleanWords[i + 1]);
}
if (i + 2 < cleanWords.Length)
{
lineScore += GetWordScore3(cleanWords[i], cleanWords[i + 1], cleanWords[i + 2]);
}
}
lines.Add(sb.ToString());
rawScores.Add(lineScore);
regionScore += lineScore;
}
sb.Clear();
//Process lines based on region, and surrounding lines
for (int i = 0; i < lines.Count; i++)
{
var lineScore = WeightScore(regionScore) + rawScores[i];
if (i >= 1)
{
lineScore += WeightScore(rawScores[i - 1]);
}
if (i + 1 < lines.Count)
{
lineScore += WeightScore(rawScores[i + 1]);
}
if (lineScore > 0)
{
sb.Append(lines[i]);
}
}
return(sb.ToString());
}