Using Tesseract-OCR within the Client
In a previous post I showed how to generate OCR renditions via Powershell. The process worked quite well, and the accuracy is higher than other solutions. After that post I went to upload the powershell scripts to github and decided to re-run each script against a new dataset.
As I ran the OCR script I noticed a few things I did not like about it:
- The script ran fine for hours and then bombed because the search results went stale
- I must remember to run the script after each import of records, or no OCR renditions
- I had to create a custom property to track whether an OCR rendition was generated
To overcome these challenges I'll need to write some code. Time to break out Visual Studio and build a new solution. So let's dive right in!
I opened up Microsoft Visual Studio 2017 and created a new solution with two projects: a C# class library for the add-in, and a C# class library for the Ocr functionality. Here I'm splitting the Ocr functionality into a separate project because in the next post I'll create an event processor plug-in. To make this work I updated the first project to reference the second and set a build dependency between the two.
Next I implemented the ITrimAddIn interface and organized the interface stubs into logical regions, as shown below. I also created a folder named MenuLinks and created two new classes within: UpdateOcrRendition and RemoveOcrRendition. Those classes will expose the menu options to the users within the client.
The two menu link classes look are defined as follows:
using HP.HPTRIM.SDK; namespace CMRamble.Ocr.ClientAddin.MenuLinks { public class UpdateOcrRendition : TrimMenuLink { public const int LINK_ID = 8002; public override int MenuID => LINK_ID; public override string Name => "Update Ocr Rendition"; public override string Description => "Uses the document content to generate OCR text"; public override bool SupportsTagged => true; } }
using HP.HPTRIM.SDK; namespace CMRamble.Ocr.ClientAddin.MenuLinks { public class RemoveOcrRendition : TrimMenuLink { public const int LINK_ID = 8003; public override int MenuID => LINK_ID; public override string Name => "Remove Ocr Rendition"; public override string Description => "Remove any Ocr Renditions"; public override bool SupportsTagged => true; } }
Now in the Add-in class I create a local variable to store the array of MenuLinks, update the Initialise interface stub to instantiate that array, and then force the GetMenuLinks method to return that array....
private TrimMenuLink[] links; public override void Initialise(Database db) { links = new TrimMenuLink[2] { new MenuLinks.UpdateOcrRendition(), new MenuLinks.RemoveOcrRendition() }; } public override TrimMenuLink[] GetMenuLinks() { return links; }
Next up I need to complete the IsMenuItemEnabled method. I do this by switching on the command link ID passed into the method. I compare it to the constant value that backs my Menu Link Id's. If you look closely at the code below, you'll notice that I'm calling "HasOcrRendition" when the link matches my RemoveOcrRendition link. There is no such method in the out-of-the-box .Net SDK. Here I'll be calling a static extension method contained inside the other library. I'm doing this because I know I'll need that same capability (to know if there is an Ocr rendition) across multiple libraries. It also makes the code easier to read.
public override bool IsMenuItemEnabled(int cmdId, TrimMainObject forObject) { switch (cmdId) { case MenuLinks.UpdateOcrRendition.LINK_ID: return forObject.TrimType == BaseObjectTypes.Record && ((HP.HPTRIM.SDK.Record)forObject).IsElectronic; case MenuLinks.RemoveOcrRendition.LINK_ID: return forObject.TrimType == BaseObjectTypes.Record && ((Record)forObject).HasOcrRendition(); default: return false; } }
The last two methods I need to implement within my record add-in are named "ExecuteLink". Here I'll hand the implementation details off to a static class contained within my second project. Doing so makes this code easy to understand and even easier to maintain.
public override void ExecuteLink(int cmdId, TrimMainObject forObject, ref bool itemWasChanged) { HP.HPTRIM.SDK.Record record = forObject as HP.HPTRIM.SDK.Record; if ((HP.HPTRIM.SDK.Record)record != null) { switch (cmdId) { case MenuLinks.UpdateOcrRendition.LINK_ID: RecordController.UpdateOcrRendition(record); break; case MenuLinks.RemoveOcrRendition.LINK_ID: RecordController.RemoveOcrRendition(record); break; default: break; } } } public override void ExecuteLink(int cmdId, TrimMainObjectSearch forTaggedObjects) { switch (cmdId) { case MenuLinks.UpdateOcrRendition.LINK_ID: RecordController.UpdateOcrRenditions(forTaggedObjects); break; case MenuLinks.RemoveOcrRendition.LINK_ID: RecordController.RemoveOcrRenditions(forTaggedObjects); break; default: break; } }
Now I need to build the desired functionality within the solution's second project. To start I'll go ahead and import the tesseract library via the Nuget package manager. As of this post the latest stable version was 3.0.2. Note that I also imported the CM .Net SDK and System.Drawing.
Next I downloaded the latest english language data files and placed them into the required tessdata sub-folder. I also updated the properties of each so that they copy to the output folder if needed.
I decide to now implement the remove ocr rendition feature. One method will work on a single record and a second method will work on a set of tagged objects (same approach as with the Client Addin). To make it super simple I'm not presenting any sort of user interface or options.
#region Remove Ocr Rendition public static bool RemoveOcrRendition(Record record) { return record.RemoveOcrRendition(); } public static void RemoveOcrRenditions(TrimMainObjectSearch forTaggedObjects) { foreach (var result in forTaggedObjects) { HP.HPTRIM.SDK.Record record = result as HP.HPTRIM.SDK.Record; if ((HP.HPTRIM.SDK.Record)record != null) { RemoveOcrRendition(record); } } } #endregion
I again used an extension method, this time naming it "RemoveOcrRendition". I create a new class named "RecordExtensions", mark it static, and implement the functionality. I also add one last extension method that handles the creation of a new ocr rendition. The contents of that class is included below.
using HP.HPTRIM.SDK; namespace CMRamble.Ocr { public static class RecordExtensions { public static void AddOcrRendition(this Record record, string fileName) { if (record.HasOcrRendition()) record.RemoveOcrRendition(); record.ChildRenditions.NewRendition(fileName, RenditionType.Ocr, "Ocr"); } public static bool RemoveOcrRendition(this Record record) { bool removed = false; for (uint i = 0; i < record.ChildRenditions.Count; i++) { RecordRendition rendition = record.ChildRenditions.getItem(i) as RecordRendition; if ((RecordRendition)rendition != null && rendition.TypeOfRendition == RenditionType.Ocr) { rendition.Delete(); removed = true; } } record.Save(); return removed; } public static bool HasOcrRendition(this Record record) { for (uint i = 0; i < record.ChildRenditions.Count; i++) { RecordRendition rendition = record.ChildRenditions.getItem(i) as RecordRendition; if ((RecordRendition)rendition != null && rendition.TypeOfRendition == RenditionType.Ocr) { return true; } } return false; } } }
Now that I have the remove ocr rendition functionality complete I can move onto the update functionality. In order to OCR the file I must first extract it to disk. Then I can extract the text by calling the tesseract library and saving the results back as a new ocr rendition. The code below implements this within the Record Controller class (which is invoked by the addin).
#region Update Ocr Rendition public static bool UpdateOcrRendition(Record record) { bool success = false; string extractedFilePath = string.Empty; string ocrFilePath = string.Empty; try { // get a temp working location on disk var rootDirectory = Path.Combine(Path.GetTempPath(), "cmramble_ocr"); if (!Directory.Exists(rootDirectory)) Directory.CreateDirectory(rootDirectory); // formulate file name to extract, delete if exists for some reason extractedFilePath = Path.Combine(rootDirectory, $"{record.Uri}.{record.Extension}"); ocrFilePath = Path.Combine(rootDirectory, $"{record.Uri}.txt"); FileHelper.Delete(extractedFilePath); FileHelper.Delete(ocrFilePath); // fetch document record.GetDocument(extractedFilePath, false, "OCR", string.Empty); // get the OCR text ocrFilePath = TextExtractor.ExtractFromFile(extractedFilePath); // use record extension method that removes existing OCR rendition (if exists) record.AddOcrRendition(ocrFilePath); record.Save(); success = true; } catch (Exception ex) { } finally { FileHelper.Delete(extractedFilePath); FileHelper.Delete(ocrFilePath); } return success; } public static void UpdateOcrRenditions(TrimMainObjectSearch forTaggedObjects) { foreach (var result in forTaggedObjects) { HP.HPTRIM.SDK.Record record = result as HP.HPTRIM.SDK.Record; if ((HP.HPTRIM.SDK.Record)record != null) { UpdateOcrRendition(record); } } } #endregion
I placed all of the tesseract logic into a new class named TextExtractor. Within that class I have one method that takes a file name and returns the name of a file containing all of the ocr text. If I use tesseract on a PDF though it will give me back the text layers from within the PDF, which defeats my goal. I want tesseract to OCR the images within the PDF.
To accomplish that I used the Xpdf command line utility pdftopng, which extracts all of the images to disk. I then iterate over each image (just like I did within the original powershell script) to generate new OCR content. As each image is processed the results are appended to an ocr text file. That text file is what is returned to the record controller.
using CMRamble.Ocr.Util; using System; using System.Diagnostics; using System.IO; using System.Linq; using Tesseract; namespace CMRamble.Ocr { public static class TextExtractor { /// <summary> /// Exports all images from PDF and then runs OCR over each image, returning the name of the file on disk holding the OCR results /// </summary> /// <param name="filePath">Source file to be OCR'd</param> /// <returns>Name of file containing OCR contents</returns> public static string ExtractFromFile(string filePath) { var ocrFileName = string.Empty; var extension = Path.GetExtension(filePath).ToLower(); if (extension.Equals(".pdf")) { // must break out the original images within the PDF and then OCR those var localDirectory = Path.Combine(Path.GetDirectoryName(filePath), Path.GetFileNameWithoutExtension(filePath)); ocrFileName = Path.Combine(Path.GetDirectoryName(filePath), Path.GetFileNameWithoutExtension(filePath) + ".txt"); FileHelper.Delete(ocrFileName); // call xpdf util pdftopng passing PDF and location to place images Process p = new Process(); p.StartInfo.UseShellExecute = false; p.StartInfo.RedirectStandardOutput = true; p.StartInfo.FileName = "pdftopng"; p.StartInfo.Arguments = $"\"{filePath}\" \"{localDirectory}\""; p.Start(); string output = p.StandardOutput.ReadToEnd(); p.WaitForExit(); // find all the images that were extracted var images = Directory.GetFiles(Directory.GetParent(localDirectory).FullName, "*.png").ToList(); foreach (var image in images) { // spin up an OCR engine and have it dump text to the OCR text file using (var engine = new TesseractEngine(@"./tessdata", "eng", EngineMode.Default)) { using (var img = Pix.LoadFromFile(image)) { using (var page = engine.Process(img)) { File.AppendAllText(ocrFileName, page.GetText() + Environment.NewLine); } } } // clean-up as we go along File.Delete(image); } } return ocrFileName; } } }
All done! I can now compile the add-in and play with it. First I added the menu links to my home ribbon. As you can see below, clicking the remove ocr rendition link changes the number of renditions available.
Along the same line, if I click update ocr rendition then the number of renditions is increased...
In the next post I'll incorporate the same functionality within an event processor plugin, so that all records have their content OCR'd via tesseract.
You can download the full source for this solution here: