DS7/DSWeb/Areas/MvcShipping/Controllers/OcrController.cs

using DSWeb.Areas.MvcShipping.DB;
using DSWeb.MvcShipping.Comm;
using Newtonsoft.Json;
using org.apache.pdfbox.cos;
using org.apache.pdfbox.pdmodel;
using org.apache.pdfbox.pdmodel.common;
using org.apache.pdfbox.util;
using System;
using System.Collections.Generic;
using System.Drawing;
using System.IO;
using System.Linq;
using System.Runtime.Caching;
using System.Text;
using System.Web;
using System.Web.Mvc;

namespace DSWeb.MvcShipping.Controllers
{
    public class OcrController : Controller
    {
        private static readonly string ocrFilePath = "ocr_temp_files";
        private static readonly string[] allowExts = new string[] { ".pdf", ".xls", ".xlsx", ".doc", ".docx",".rtf" };
        //private static readonly int ImageScale = 1;
        private static readonly int CacheExpire = 5;//缓存到期分钟

        [HttpGet]
        public ActionResult UpOcrAuto(string errMsg)
        {
            ViewData["errMsg"] = errMsg;
            return View();
        }

        [HttpPost]
        public ActionResult UpOcrFile(float scale = 1.5f)
        {
            if (Request.Files.Count > 0)
            {
                var ext = Path.GetExtension(Request.Files[0].FileName).ToLower();
                if (!allowExts.Contains(ext))
                {
                    return RedirectToAction("UpOcrAuto", new { errMsg = "只能上传pdf、excel和word文件！" });
                }

                var ocrDir = Server.MapPath($"~/{ocrFilePath}");
                if (!Directory.Exists(ocrDir))
                {
                    Directory.CreateDirectory(ocrDir);
                }

                var tmpFileName = Guid.NewGuid().ToString();
                var saveFilePathName = Path.Combine(ocrDir, tmpFileName + ext);
                Request.Files[0].SaveAs(saveFilePathName);

                if (ext == ".xls" || ext == ".xlsx")
                {
                    Aspose.Cells.Workbook wb = new Aspose.Cells.Workbook(saveFilePathName);
                    string targetPath = Path.Combine(ocrDir, $"{tmpFileName}.pdf");
                    wb.Save(targetPath, Aspose.Cells.SaveFormat.Pdf);
                }
                else if (ext == ".doc" || ext == ".docx" || ext == ".rtf")
                {
                    Aspose.Words.Document doc = new Aspose.Words.Document(saveFilePathName);
                    string targetPath = Path.Combine(ocrDir, $"{tmpFileName}.pdf");
                    doc.Save(targetPath, Aspose.Words.SaveFormat.Pdf);
                }

                ReadPdf(tmpFileName, scale);

                #region 挂载到单据管理
                var BSNO = Request.Form["BSNO"];
                var CUSTID = Guid.NewGuid().ToString();
                var path = Server.MapPath("../../UploadFiles/MsOpFiles/" + CUSTID);

                if (!Directory.Exists(path))
                {
                    Directory.CreateDirectory(path);
                }

                var size = Request.Files[0].ContentLength;
                var name = Path.GetFileName(Request.Files[0].FileName);

                var usercode = Convert.ToString(Session["USERCODE"]);
                var partname = usercode + DateTime.Now.ToString("yyyyMMddHHmmssfff") + name;
                string filename = path + "\\" + partname;
                string filenameURL = "../../UploadFiles/MsOpFiles/" + CUSTID + "/" + partname;
                if (System.IO.File.Exists(filename))
                {
                    System.IO.File.Delete(filename);
                }
                Request.Files[0].SaveAs(filename);


                ReceiptContext receiptContext = new ReceiptContext();
                Receipt_Doc recDoc = new Receipt_Doc();
                recDoc.GID = Guid.NewGuid().ToString();
                recDoc.BSNO = BSNO;
                recDoc.URL = name;
                recDoc.Driect_URL = filenameURL;
                recDoc.MODIFIEDUSER = Convert.ToString(Session["USERID"]);
                recDoc.MODIFIEDTIME = DateTime.Now;
                recDoc.RECEIPTTYPE = string.Empty;
                recDoc.documentAttachedCode = string.Empty;
                recDoc.documentAttachedNo = string.Empty;
                receiptContext.ReceiptDocs.Add(recDoc);
                receiptContext.SaveChanges();
                #endregion


                return RedirectToAction("SelectRegion", new { fileName = tmpFileName });
            }

            return RedirectToAction("UpOcrAuto");

        }

        [HttpGet]
        public ActionResult SelectRegion(string fileName)
        {
            ViewData["FileName"] = fileName;
            return View();
        }

        [HttpPost]
        public ContentResult GetRegionText(string fileName, int x, int y, int w, int h, float scale = 1.5f)
        {
            if (!MemoryCache.Default.Contains(fileName))
            {
                ReadPdf(fileName, scale);
            }

            dynamic obj = MemoryCache.Default[fileName];
            if (obj != null)
            {
                Rectangle rectMouse = new Rectangle(x, y, w, h);
                List<TextSizeLocation> listTextLocations = obj.Result;
                Size size = obj.Size;
                StringBuilder sb = new StringBuilder();
                int lastY = 0;
                foreach (var item in listTextLocations)
                {
                    int xItem = (int)(item.X * scale);
                    int yItem = (int)(size.Height * scale - item.Y * scale);
                    int wItem = (int)(item.Width * scale);
                    int hItem = (int)(item.Height * scale);

                    Rectangle rect = new Rectangle(xItem, yItem, wItem, hItem);

                    if (rect.IntersectsWith(rectMouse))
                    {
                        if (lastY != 0 && Math.Abs(lastY - item.Y) > 1)
                        {
                            sb.AppendLine();
                        }

                        sb.Append(item.Text);

                        lastY = (int)item.Y;

                    }
                }

                return Content(sb.ToString());
            }

            return Content("");
        }


        [HttpGet]
        public ActionResult GetCanvasImage(string fileName, float scale = 1.5f)
        {
            var ocrDir = Server.MapPath($"~/{ocrFilePath}");
            var imgPath = Path.Combine(ocrDir, $"{fileName}_{scale}.jpg");
            if (!System.IO.File.Exists(imgPath))
            {
                dynamic obj = MemoryCache.Default[fileName];
                if (obj != null)
                {
                    Size size = obj.Size;
                    DrawImg((int)(size.Width * scale), (int)(size.Height * scale), fileName, scale);
                }
            }

            if (System.IO.File.Exists(imgPath))
            {
                return File(imgPath, "image/jpg");
            }

            return null;
        }

        private void ReadPdf(string fileName, float scale)
        {
            var ocrDir = Server.MapPath($"~/{ocrFilePath}");
            var pdfFileName = System.IO.Path.Combine(ocrDir, fileName + ".pdf");

            if (!System.IO.File.Exists(pdfFileName))
            {
                return;
            }

            PDDocument document = PDDocument.load(pdfFileName);
            PrintTextLocations printer = new PrintTextLocations();
            java.util.List allPages = document.getDocumentCatalog().getAllPages();
            PDPage page = (PDPage)allPages.get(0);
            PDStream contents = page.getContents();
            if (contents != null)
            {
                printer.processStream(page, page.findResources(), page.getContents().getStream());

                //处理
                var listT = new List<TextSizeLocation>();
                TextSizeLocation t = null;
                foreach (var item in printer.TextWithLocations)
                {
                    if (t == null)
                    {
                        t = item.Clone();
                        listT.Add(t);
                    }
                    else
                    {

                        if (Math.Abs(t.Width + t.X - item.X) <= item.Width / 2 && Math.Abs(t.Height - item.Height) <= 1) //首尾相接
                        {
                            t.Width += item.Width;
                            t.Text += item.Text;
                        }
                        else
                        {
                            t = item.Clone();
                            listT.Add(t);
                        }
                    }
                }

                var query = from it in listT orderby it.Y descending, it.X select it;
                listT = query.ToList();

                //缓存
                MemoryCache.Default.Remove(fileName);
                var objStore = new { Size = new Size((int)page.findCropBox().getWidth(), (int)page.findCropBox().getHeight()), Result = listT };
                MemoryCache.Default.Add(new CacheItem(fileName, objStore), new CacheItemPolicy() { AbsoluteExpiration = new DateTimeOffset(DateTime.Now.AddMinutes(CacheExpire)) });

                DrawImg((int)(page.findCropBox().getWidth() * scale), (int)(page.findCropBox().getHeight() * scale), fileName, scale);
            }

            document.close();
        }

        private void DrawImg(int width, int height, string fileName, float scale)
        {
            var ocrDir = Server.MapPath($"~/{ocrFilePath}");
            Bitmap bitmap = new Bitmap(width, height);
            Graphics graphics = Graphics.FromImage(bitmap);
            graphics.Clear(Color.White);

            dynamic obj = MemoryCache.Default[fileName];
            if (obj != null)
            {
                List<TextSizeLocation> listTextLocations = obj.Result;
                Size size = obj.Size;

                Random rnd = new Random();
                foreach (var item in listTextLocations)
                {
                    float textSize = GetMatchTextSize(item.Width * scale, item.Height * scale, item.Text);
                    graphics.DrawString(item.Text, new Font("黑体", textSize), Brushes.Black, item.X * scale, bitmap.Height - item.Y * scale);

                }

                var imgPath = Path.Combine(ocrDir, $"{fileName}_{scale}.jpg");
                bitmap.Save(imgPath);
            }
        }

        private float GetMatchTextSize(float width, float height, string str)
        {
            float plus = 0.1f;
            float rtn = 1;
            Bitmap bm = new Bitmap(1, 1);
            Graphics g = Graphics.FromImage(bm);
            for (float i = 1; i < 100; i += plus)
            {
                var font = new Font("黑体", i);
                SizeF size = g.MeasureString(str, font);
                //if (size.Width > width || size.Height > height)
                if (size.Width > width)
                //if (size.Height > height)
                {
                    if (i > 1)
                    {
                        rtn = i - plus;
                    }

                    break;
                }
            }

            return rtn;
        }
    }


    public class TextSizeLocation
    {
        public TextSizeLocation() { }

        public TextSizeLocation(float x, float y, float width, float height, string text)
        {
            this.X = x;
            this.Y = y;
            this.Width = width;
            this.Height = height;
            this.Text = text;
        }

        public float X { get; set; }
        public float Y { get; set; }
        public float Width { get; set; }
        public float Height { get; set; }
        public string Text { get; set; }

        public TextSizeLocation Clone()
        {
            return new TextSizeLocation()
            {
                X = this.X,
                Y = this.Y,
                Width = this.Width,
                Height = this.Height,
                Text = this.Text
            };
        }
    }

    public class PrintTextLocations : PDFTextStripper
    {
        public List<TextSizeLocation> TextWithLocations { get; set; }
        public List<TextPosition> TextPositions { get; set; }

        public PrintTextLocations()
        {
            this.TextWithLocations = new List<TextSizeLocation>();
            this.TextPositions = new List<TextPosition>();
        }

        public override void processStream(PDPage aPage, PDResources resources, COSStream cosStream)
        {
            this.TextWithLocations.Clear();
            this.TextPositions.Clear();
            base.processStream(aPage, resources, cosStream);
        }

        protected override void processTextPosition(TextPosition text)
        {
            this.TextPositions.Add(text);
            this.TextWithLocations.Add(new TextSizeLocation(text.getTextPos().getXPosition(), text.getTextPos().getYPosition(), text.getWidth(), text.getHeight(), text.getCharacter()));
        }
    }
}