Scan images - DO NOT MERGE (#363)

* first attempt

* Filtering works

* Update cleanrepo/CleanRepo/Repo.cs

Co-authored-by: Genevieve Warren <[email protected]>

* Update cleanrepo/CleanRepo/Program.cs

Co-authored-by: Genevieve Warren <[email protected]>

* Update cleanrepo/CleanRepo/Program.cs

Co-authored-by: Genevieve Warren <[email protected]>

* Update cleanrepo/CleanRepo/Program.cs

Co-authored-by: Genevieve Warren <[email protected]>

* Update cleanrepo/CleanRepo/Program.cs

Co-authored-by: Genevieve Warren <[email protected]>

* Update cleanrepo/CleanRepo/Program.cs

Co-authored-by: Genevieve Warren <[email protected]>

* changes as requested

* Update cleanrepo/CleanRepo/Program.cs

Co-authored-by: David Pine <[email protected]>

* Update cleanrepo/CleanRepo/Program.cs

Co-authored-by: David Pine <[email protected]>

* Update cleanrepo/CleanRepo/Program.cs

Co-authored-by: David Pine <[email protected]>

* Update cleanrepo/CleanRepo/Program.cs

Co-authored-by: David Pine <[email protected]>

* Update cleanrepo/CleanRepo/Program.cs

Co-authored-by: David Pine <[email protected]>

* Update cleanrepo/CleanRepo/Program.cs

Co-authored-by: David Pine <[email protected]>

* Update cleanrepo/CleanRepo/Program.cs

Co-authored-by: David Pine <[email protected]>

* Fixed warnings - removed comments

* Moved OCR functionality to bottom of menu, and all together.

* Instructions

* Fix output file logic

* Update cleanrepo/CleanRepo/Program.cs

Co-authored-by: Bill Wagner <[email protected]>

---------

Co-authored-by: Genevieve Warren <[email protected]>
Co-authored-by: David Pine <[email protected]>
Co-authored-by: Bill Wagner <[email protected]>

Author: 4 people

cleanrepo/CleanRepo.csproj

@@ -16,6 +16,7 @@
   </ItemGroup>
   <ItemGroup>
     <PackageReference Include="CommandLineParser" Version="2.9.1" />
+    <PackageReference Include="Tesseract" Version="5.2.0" />
     <PackageReference Include="Microsoft.Build" Version="17.10.4" />
   </ItemGroup>
 </Project>

cleanrepo/Options.cs

@@ -53,6 +53,19 @@ class Options
     [Option("remove-hops", HelpText = "Clean redirection JSON file by replacing targets that are themselves redirected (daisy chains).")]
     public bool RemoveRedirectHops { get; set; }
 
+    [Option("catalog-images-with-text", Default = false, HelpText = "Map images to the markdown/YAML files that reference them, with all text found in images. Must set --ocr-model-directory path.")]
+    public bool CatalogImagesWithText { get; set; }
+
+    [Option("filter-images-for-text", Default = false, HelpText = "Filter images for text. Must set --ocr-model-directory and --filter-text-json-file paths.")]
+    public bool FilterImagesForText { get; set; }
+
+    [Option("ocr-model-directory", HelpText = "Directory that contains the OCR (Tesseract) models for image scanning.")]
+    public string? OcrModelDirectory { get; set; }
+
+    [Option("filter-text-json-file", HelpText = "JSON file of array of strings to filter OCR results with.")]
+    public string? FilterTextJsonFile { get; set; }
+
+
     //[Option("format-redirects", Required = false, HelpText = "Format the redirection JSON file by deserializing and then serializing with pretty printing.")]
     //public bool FormatRedirectsFile { get; set; }
 

cleanrepo/Program.cs

@@ -4,7 +4,10 @@
 using System.Data;
 using System.Diagnostics;
 using System.Text;
+using System.Text.Json;
 using System.Text.RegularExpressions;
+using Tesseract;
+using static System.Net.Mime.MediaTypeNames;
 
 namespace CleanRepo;
 
@@ -26,13 +29,52 @@ static void Main(string[] args)
         //   }
         //
         // ... to avoid hardcoded values in DEBUG preprocessor directives like this:
-        args = new[] { "--orphaned-snippets", "--snippets-directory=c:\\users\\gewarren\\testrepo3\\snippets", "--xml-source=true" };
+        args = new[] {
+                    "--help"
+        };
         //args = new[] { "--orphaned-snippets", "--relative-links", "--remove-hops", "--replace-redirects", "--orphaned-includes", "--orphaned-articles", "--orphaned-images",
         //"--articles-directory=c:\\users\\gewarren\\docs\\docs\\fundamentals", "--media-directory=c:\\users\\gewarren\\docs\\docs\\core",
         //"--includes-directory=c:\\users\\gewarren\\docs\\includes", "--snippets-directory=c:\\users\\gewarren\\docs\\samples\\snippets\\csharp\\vs_snippets_clr",
         //"--docfx-directory=c:\\users\\gewarren\\docs", "--url-base-path=/dotnet", "--delete=false"};
 #endif
 
+        /*
+
+        Display help screen
+
+        args = new[] {
+        "--help"
+         */
+
+
+        /*
+
+        Catalog images with text
+
+        args = new[] {
+        "--catalog-images-with-text", 
+        "--url-base-path=/azure/developer/javascript",
+        "--ocr-model-directory=c:\\Users\\diberry\\repos\\temp\\tesseract\\tessdata_fast",
+        "--articles-directory=c:\\Users\\diberry\\repos\\writing\\docs\\azure-dev-docs-pr-2\\articles",
+        "--media-directory=c:\\Users\\diberry\\repos\\writing\\docs\\azure-dev-docs-pr-2\\articles\\javascript\\media"};
+         */
+
+        /*
+
+        Filter images for text
+
+        args = new[] {
+        "--filter-images-for-text",
+        "--filter-text-json-file=c:\\Users\\diberry\\repos\\filter-text.json",
+        "--url-base-path=/azure/developer/javascript",
+        "--ocr-model-directory=c:\\Users\\diberry\\repos\\temp\\tesseract\\tessdata_fast",
+        "--articles-directory=c:\\Users\\diberry\\repos\\writing\\docs\\azure-dev-docs-pr-2\\articles",
+        "--media-directory=c:\\Users\\diberry\\repos\\writing\\docs\\azure-dev-docs-pr-2\\articles\\javascript\\media"};
+         */
+
+
+
+
         Parser.Default.ParseArguments<Options>(args).WithParsed(RunOptions);
     }
 
@@ -46,6 +88,8 @@ options.FindOrphanedIncludes is false &&
             options.FindOrphanedSnippets is false &&
             options.ReplaceRedirectTargets is false &&
             options.ReplaceWithRelativeLinks is false &&
+            options.CatalogImagesWithText is false &&
+            options.FilterImagesForText is false &&
             options.RemoveRedirectHops is false)
         {
             Console.WriteLine("\nYou didn't specify which function to perform. To see options, use 'CleanRepo.exe -?'.");
@@ -175,8 +219,10 @@ options.ReplaceWithRelativeLinks is false &&
         }
 
         // Catalog images
-        if (options.CatalogImages)
+        if (options.CatalogImages || options.CatalogImagesWithText || options.FilterImagesForText)
         {
+
+
             if (string.IsNullOrEmpty(options.MediaDirectory))
             {
                 Console.WriteLine("\nEnter the path to the directory where you want to catalog media files:\n");
@@ -194,6 +240,11 @@ options.ReplaceWithRelativeLinks is false &&
                 Console.WriteLine($"'{options.MediaDirectory}' is not a child of the docfx.json file's directory '{docFxRepo.DocFxDirectory}'.");
                 return;
             }
+            if (options.CatalogImagesWithText && string.IsNullOrEmpty(options.OcrModelDirectory))
+            {
+                Console.WriteLine($"'--ocr-model-directory' directory was not provided.");
+                return;
+            }
 
             // Add regex to find image refs similar to 'social_image_url: "/dotnet/media/logo.png"'
             // This is done here (dynamically) because it relies on knowing the base path URL.
@@ -203,11 +254,87 @@ options.ReplaceWithRelativeLinks is false &&
             if (docFxRepo._imageRefs is null)
                 docFxRepo._imageRefs = GetMediaFiles(options.MediaDirectory);
 
-            Console.WriteLine($"\nCataloging the images in the '{options.MediaDirectory}' directory...\n");
+            Console.WriteLine($"\nCataloging '{docFxRepo._imageRefs.Count}' images (recursively) in the '{options.MediaDirectory}' directory...\n");
+
+            if (options.CatalogImagesWithText && !string.IsNullOrEmpty(options.OcrModelDirectory))
+            {
+                // Extract hash keys from the dictionary
+                List<string> mediaFilesList = docFxRepo._imageRefs.Keys.ToList();
+
+                // Pass hash keys to ScanMediaFiles
+                docFxRepo._ocrRefs = ScanMediaFiles(mediaFilesList, options.OcrModelDirectory);
+
+
+
+                docFxRepo.OutputImageReferences(true);
+            }
+            else if (options.FilterImagesForText && !string.IsNullOrEmpty(options.OcrModelDirectory))
+            {
+                if (string.IsNullOrEmpty(options.FilterTextJsonFile))
+                {
+                    Console.WriteLine($"\nThe filterTextJsonFile can't be empty when requesting FilterImagesForText.");
+                    return;
+                }
+                if (!File.Exists(options.FilterTextJsonFile))
+                {
+                    Console.WriteLine($"\nThe filterTextJsonFile '{options.FilterTextJsonFile}' doesn't exist.");
+                    return;
+                }
+
+                List<string> searchTerms = [];
+                try
+                {
+                    string jsonContent = File.ReadAllText(options.FilterTextJsonFile);
+
+                    searchTerms = JsonSerializer.Deserialize<List<string>>(jsonContent) ?? new List<string>();
+                }
+                catch (IOException ioEx)
+                {
+                    Console.WriteLine($"\nIO error reading '{options.FilterTextJsonFile}': {ioEx.Message}");
+                }
+                catch (UnauthorizedAccessException uaEx)
+                {
+                    Console.WriteLine($"\nAccess error reading '{options.FilterTextJsonFile}': {uaEx.Message}");
+                }
+                catch (JsonException jsonEx)
+                {
+                    Console.WriteLine($"\nError deserializing '{options.FilterTextJsonFile}': {jsonEx.Message}");
+                }
+                catch (Exception ex) // Fallback for any other unexpected exceptions
+                {
+                    Console.WriteLine($"\nUnexpected error: {ex.Message}");
+                    return;
+                }
+                if (searchTerms.Count == 0)
+                {
+                    Console.WriteLine($"\nNo search terms found in '{options.FilterTextJsonFile}'.");
+                    return;
+                }
+
+                // Extract hash keys from the dictionary
+                List<string> mediaFilesList = docFxRepo._imageRefs.Keys.ToList();
+
+                if(mediaFilesList.Count==0)
+                {
+                    Console.WriteLine($"\nNo media files found.");
+                }
+                // Pass hash keys to ScanMediaFiles
+                Dictionary<string, string> unfilteredResults = ScanMediaFiles(mediaFilesList, options.OcrModelDirectory);
+
+                // Filter results
+                docFxRepo._ocrFilteredRefs = FilterMediaFiles(unfilteredResults, searchTerms);
+
+                docFxRepo.OutputImageReferences(true, true);
+            }
+            else
+            {
+                docFxRepo.OutputImageReferences();
+            }
+
 
-            docFxRepo.OutputImageReferences();
         }
 
+
         // Find orphaned include-type files
         if (options.FindOrphanedIncludes)
         {
@@ -1576,7 +1703,7 @@ private static Dictionary<string, List<string>> GetMediaFiles(string mediaDirect
 
         Dictionary<string, List<string>> mediaFiles = new(StringComparer.InvariantCultureIgnoreCase);
 
-        string[] fileExtensions = ["*.png", "*.jpg", "*.gif", "*.svg"];
+        string[] fileExtensions = [ "*.png", "*.jpg", "*.gif", "*.svg" ]; // Correctly initialize the array
 
         foreach (string extension in fileExtensions)
         {
@@ -1591,6 +1718,62 @@ private static Dictionary<string, List<string>> GetMediaFiles(string mediaDirect
 
         return mediaFiles;
     }
+    /// <summary>
+    /// Returns a dictionary of all .png/.jpg/.gif/.svg files in the directory.
+    /// The search includes the text found in the files.
+    /// </summary>
+    private static Dictionary<string, string> ScanMediaFiles(List<string>? imageFilePaths, string ocrModelDirectory)
+    {
+
+        Dictionary<string, string> ocrDataForFiles = new(StringComparer.InvariantCultureIgnoreCase);
+
+        if (imageFilePaths is null or { Count : 0 })
+        {
+            Console.WriteLine("\nNo .png/.jpg/.gif/.svg files to scan!");
+            return ocrDataForFiles;
+        }
+
+        using var engine = new TesseractEngine(ocrModelDirectory, "eng", EngineMode.Default);
+        foreach (string imageFilePath in imageFilePaths)
+        {
+            using var img = Pix.LoadFromFile(imageFilePath);
+            using Page page = engine.Process(img);
+
+            string text = page.GetText();
+            ocrDataForFiles.Add(imageFilePath, text);
+        }
+        return ocrDataForFiles;
+    }
+
+    // Filter ocrDictionary by filterTerms
+    private static Dictionary<string, List<KeyValuePair<string, string>>> FilterMediaFiles(Dictionary<string, string> ocrDictionary, List<string> filterTerms)
+    {
+        // Sort the filterTerms to ensure the result is sorted by filterTerm
+        filterTerms.Sort();
+
+        Dictionary<string, List<KeyValuePair<string, string>>> filterTermFilesDictionary = [];
+
+        foreach (string filterTerm in filterTerms)
+        {
+            List<KeyValuePair<string, string>> matchedFiles = [];
+
+            foreach (var imageFile in ocrDictionary)
+            {
+                if (imageFile.Value.Contains(filterTerm, StringComparison.OrdinalIgnoreCase))
+                {
+                    // Add both the file path and the text for that file
+                    matchedFiles.Add(new KeyValuePair<string, string>(imageFile.Key, imageFile.Value));
+                }
+            }
+
+            if (matchedFiles.Count > 0)
+            {
+                filterTermFilesDictionary.Add(filterTerm, matchedFiles);
+            }
+        }
+
+        return filterTermFilesDictionary;
+    }
 
     /// <summary>
     /// Gets all *.yml files recursively, starting in the ancestor directory that contains docfx.json.

cleanrepo/README.md

@@ -4,12 +4,13 @@ This command-line tool helps you clean up a DocFx-based content repo. It can:
 
 - Find and delete markdown files that aren't linked from a TOC file.
 - Find and delete orphaned image (.png, .jpg, .gif, .svg) files.
-- Map images to the files that reference them.
+- Map images to the files that reference them and return text from those images.
 - Find and delete orphaned "shared" markdown files (includes).
 - Find and delete orphaned snippet (.cs, .vb, .cpp, .fs, and .xaml) files.
 - Find and replace links to redirected files.
 - Remove daisy chains (or hops) within the redirection files for the docset.
 - Replace site-relative links with file-relative links (includes image links).
+- Filter image list based on strings found in images.
 
 ## Usage
 
@@ -23,6 +24,10 @@ This command-line tool helps you clean up a DocFx-based content repo. It can:
 | --replace-redirects | Find backlinks to redirected files and replace with new target. |
 | --remove-hops | Remove daisy chains within the redirection files for the docset. |
 | --relative-links | Replace site-relative links with file-relative links. |
+| --catalog-images-with-text | Map images to the markdown/YAML files that reference them, with all text found in images. Must set --ocr-model-directory path. |
+| --filter-images-for-text | Filter images for text. Must set --ocr-model-directory and --filter-text-json-file paths. |
+| --ocr-model-directory | Directory that contains the OCR (Tesseract) models for image scanning. |
+| --filter-text-json-file | JSON file of array of strings to filter OCR results with. |
 
 ## Usage examples
 
@@ -43,3 +48,66 @@ This command-line tool helps you clean up a DocFx-based content repo. It can:
   ```
   CleanRepo.exe --orphaned-images
   ```
+
+## Text to image examples
+
+The text-to-image functionality supported in the `--catalog-images-with-text` and `--filter-images-for-text` options is provided by the [Tesseract](https://www.nuget.org/packages/tesseract/) NuGet package. 
+
+### Get the Tesseract models
+
+You must determine which Tesseract models you want to use and install them on your system. Tesseract models are generated per operating system. Tesseract models come in a variety of sizes. You will also need to download the language data files for tesseract 4.0.0 or above from [tesseract-tessdata](https://github.com/tesseract-ocr/tessdata/). Use the `--ocr-model-directory` value to set the path. 
+
+### Catalog images with text
+
+To catalog the images with text:
+
+```console
+CleanRepo --catalog-images-with-text \
+--url-base-path=/azure/developer/javascript \
+--articles-directory=c:\\Users\\diberry\\repos\\writing\\docs\\azure-dev-docs-pr-2\\articles \
+--media-directory=c:\\Users\\diberry\\repos\\writing\\docs\\azure-dev-docs-pr-2\\articles\\javascript\\media
+--ocr-model-directory=c:\\Users\\diberry\\repos\\temp\\tesseract\\tessdata_fast
+```
+
+The output file is prefixed with `ImageFiles-` and looks like: 
+
+```json
+
+```
+
+### Filter images with text
+ 
+
+To file images based on an array of string, use the `--filter-text-json-file` path to the JSON file with the text to filter for:
+
+```json
+["Azure","Microsoft"]
+```
+
+
+```console
+CleanRepo --filter-images-for-text \
+--filter-text-json-file=c:\\Users\\diberry\\repos\\filter-text.json \
+--url-base-path=/azure/developer/javascript \
+--ocr-model-directory=c:\\Users\\diberry\\repos\\temp\\tesseract\\tessdata_fast \
+--articles-directory=c:\\Users\\diberry\\repos\\writing\\docs\\azure-dev-docs-pr-2\\articles \
+--media-directory=c:\\Users\\diberry\\repos\\writing\\docs\\azure-dev-docs-pr-2\\articles\\javascript\\media
+``` 
+
+The output file is prefixed with `FilteredOcrImageFiles-` and looks like: 
+
+```json
+{
+  "Azure": [
+    {
+      "Key": "c:\\Users\\diberry\\repos\\writing\\docs\\azure-dev-docs-pr-2\\articles\\javascript\\media\\visual-studio-code-azure-resources-extension-remove-resource-group.png",
+      "Value": "*J File Edit Selection View Go Run Terminal Help\n\nQa AZURE oo\n\n\u003E FUNCTIONS\n-v RESOURCE GROUPS\n\\ \u0026 Pay-As-You-Go-diberry Y\n|\nEdit Tags...\n\u00A3\nView Properties\nte Open in Portal\nRefresh\n90\n\n \n\n \n\n \n\n \n"
+    }],
+  "Microsoft": [
+    {
+      "Key": "c:\\Users\\diberry\\repos\\writing\\docs\\azure-dev-docs-pr-2\\articles\\javascript\\media\\azure-function-resource-group-management\\azure-portal-function-application-insights-link.png",
+      "Value": "Function App\n\n\u00AE Overview\n\n \n\n| View Application Insights data G)\n\n \n\n \n\n \n\n \n\nActivity log Link to an Application Insights resource\n8. Access control (IAM)\n\u00A9 tes \u00A9 temepiseaieiin yt eb ise ea\n\n@ Diagnose and solve problems\n\n\u00A9 Microsoft Defender for Cloud @ totum Apptzation ihe of check that Applicaton nights OK ard the insramentaion key are removed rm your apliaton,\n\n\u0026 events (preview)\n\nFunctions O) \u00E9sarteg etiam caer toe Gorman Vier Tc home\nApplication Insights. You have the option to disable non-essential data collection, Learn more\n(A) Functions\n\u00A9 App keys\nChange your resource\nB App files\n\n \n\nDeployment\n\n= Deployment slots\n@ Deployment Center\nSettings\n\nHl Configuration\n\n\u0026\u0026 Authentication\n\n\u00AE Application insights\n\n \n"
+    },
+    ]
+}
+```