/// <summary>
/// 通过oTable对象解析表名和内容
/// 表单元格解析为一个字典：键由表的序例号，行号，当前行的列号组成，三个元素间用"_"分隔
/// </summary>
/// <param name="oTable"></param>
public OCRTable(OTable oTable)
{
valueDict = new Dictionary <string, string>();
var tableName = new StringBuilder();
Dictionary <int, int> colCountDict = new Dictionary <int, int>();
var regionIdx = 0;
int maxRow = int.MinValue;
int tableIndex = 0;
foreach (var region in oTable.result.regions)
{
regionIdx++;
if (region.type == "text")
{
foreach (var blk in region.Blocks)
{
var word = blk.Words.Replace("\n", "").Replace(" ", "");
if (regex_ch.IsMatch(word))
{
tableName.Append(word);
}
}
}
else if (region.type == "table")
{
tableIndex++;
foreach (var blk in region.Blocks)
{
var word = blk.Words.Replace("\n", "").Replace(" ", "");
var curRow = 0;
var curCol = 0;
if (blk.Rows.Count > 0)
{
curRow = blk.Rows[0];
}
if (colCountDict.ContainsKey(curRow))
{
curCol = colCountDict[curRow];
}
foreach (var row in blk.Rows)
{
if (!colCountDict.ContainsKey(row))
{
colCountDict.Add(row, 0);
}
colCountDict[row]++;
if (maxRow < row)
{
maxRow = row;
}
}
valueDict.Add(string.Format("{0}_{1}_{2}", tableIndex, curRow, curCol), word);
}
}
}
GuessTableName = tableName.ToString();
RowCount = maxRow;
}
private void processPdf(string pdfPath, System.Text.RegularExpressions.Regex regex, Dictionary <string, string> adjustStrDict, Action stepFunc)
{
var subFolders = Directory.GetDirectories(pdfPath);
foreach (var subFolder in subFolders)
{
processPdf(subFolder, regex, adjustStrDict, stepFunc);
}
var pdfFiles = Directory.GetFiles(pdfPath, "*.pdf");
string dh;
var dict = getFileDict(Path.Combine(pdfPath, "目录.doc"), out dh);
foreach (var pdf in pdfFiles)
{
stepFunc();
string originName = Path.GetFileNameWithoutExtension(pdf);
string tableName = originName;
if (dict.ContainsKey(tableName))
{
tableName = dict[tableName];
}
if (regex.IsMatch(tableName))
{
//将PDF转为图片列表
var imgs = ConvertPdf2Image.Convert(pdf, definition: ConvertPdf2Image.Definition.Four);
var index = 1;
foreach (var img in imgs)
{
var imgBase64 = "";
//将图片转为base64字串，以便传递给华为文字识别API
using (var stream = new MemoryStream())
{
img.Save(stream, System.Drawing.Imaging.ImageFormat.Png);
var bytes = stream.ToArray();
imgBase64 = Convert.ToBase64String(bytes);
}
img.Dispose();
//调用华为OCR接口返回JSON解析串
var jsonString = OCRParser.GetTableJsonStringByBase64(imgBase64);
//反序列化为OTable对象
OTable table = JsonConvert.DeserializeObject <OTable>(jsonString);
OCRTable oCRTable = new OCRTable(table);
//进行解析文本校正
oCRTable.AdjustStringByDict(adjustStrDict);
//var json = JsonConvert.SerializeObject(oCRTable);
var ocrTableString = oCRTable.ToString();
//构造文本文件名
string txtFileName = Path.Combine(pdfPath, string.Format("{0}_{1}{2}", dh, originName, tableName));
if (imgs.Count > 1)
{
txtFileName += "_" + index++;
}
txtFileName += ".txt";
File.WriteAllText(txtFileName, ocrTableString);
}
}
}
}