Document Toolkit and Caching

A large portion of the Document library requires the use of a cache system.

• Documents can contain large numbers of pages containing huge amounts of data. Storing all of this data in physical memory is not feasible in most situations.

• Typically, systems do not have write access to the pages of a document stored in a remote URL, so modifications (annotations, images, or text), cannot be stored in the original document.

• Caching increases the performance of tasks like getting page data such as images or text. Results can be parsed once from the physical file on disk, processed, and then stored in the cache. Subsequent calls to the same data will simply retrieve it from the cache without any extra processing.

• The LEADTOOLS Document Web Service requires the use of a cache. Web methods are session-less by nature and with cache support, the service can use SaveToCache and LoadFromCache to save/load the same document between calls without the need to maintain session states.

Each LEADDocument object contains an ID. This is a unique string value that can be generated automatically by the system (using a GUID generator), or provided by the user and stored in the LEADDocument.DocumentId field. The ID is all that is required to re-construct a document object from a cache using LoadFromCache.

LoadFromUri can be used to create a LEADDocument object that represents a document such as PDF, TIFF, or DOCX stored in a remote URL. The Document is a data structure containing properties such as the mime type, number of pages, size of each page, and other metadata. It does not contain any actual image, SVG, or text data of the pages. The original document data (the PDF, TIFF, or DOCX image) is still stored in the remote URL. This data structure is all that is saved into the cache (by default), and therefore saving and then re-loading a document from the cache is a very fast operation and does not require a large amount of memory. When the user requests an image representation of a page, the document parses it from the original data. This data can also be cached if required, as explained in the "Cache Workflow" section below.

To use caching, an object that implements ObjectCache is initialized once at the start of the application and then passed to the DocumentFactory and LEADDocument methods that require caching. Any cache system that can persist data between application re-starts can be used. Refer to the "Cache System Examples" section below for more information.

Generally, the cache is used in two ways, depending on the type of the application:

• The application uses the cache to get data more quickly from existing documents or to create new documents. The data is neither shared with other applications, nor is it necessary for it to persist between sessions.

  • An example of this is the Desktop LEADTOOLS Document Viewer Demo using the FileCache system. The application obtains a temporary folder and sets it as the cache directory to use FileCache.CacheDirectory. It then uses default CacheItemPolicy objects that do not expire, and the document parts are cached and used. The value of AutoSaveToCache is set to false and each document will delete its own cache entry when it is disposed of and no longer used.

• The application requires the document state to persist between sessions.

  • An example of this is the LEADTOOLS Document Web Service. The service creates an instance of an ObjectCache-derived object at the start of the application, from settings stored in the web.config file. After a document is loaded using LoadFromUri, it is saved in the cache using SaveToCache, and an object containing the document properties (along with its ID) is returned to the JavaScript client. The client can then call the various other methods of the service to obtain pages, images, text, annotations or any other data using this ID. The web methods will call LoadFromCache to re-construct the document from the cache to obtain the desired data before viewing it in the LEADTOOLS Document Viewer.

These types of applications usually store a policy setting in the web.config as well. The setting controls how long items can be stored in the cache before they expire.

Cache Workflow

Cache workflow will describe how the LEADTOOLS Document Web Service uses the cache. The service ships with full source code, and the process can be modified as needed. The project source code is located at:

.NET: [Your installation folder]\Examples\JS\Documents\DocumentViewer\Services\DocumentService

Java: [Your installation folder]\Examples\JS\Documents\DocumentViewer\Services\DocumentServiceJava

The JavaScript client demo is located at:

JavaScript: [Your installation folder]\Examples\JS\Documents\DocumentViewer\Apps\App1\site

TypeScript: [Your installation folder]\Examples\JS\Documents\DocumentViewer\Apps\App1\ts

Application_Start

The global cache object is created and stored in the static _cache variable (accessible to the rest of service code through the ServiceHelper.Cache static property). In the sample implementation, it is a LEADTOOLS FileCache object that stores cache items in the file system (local or as recommended: remote UNC).

The cache eviction policy that determines how long items are kept in the cache is also set up here.

The cache is a persistence system, which means that when the system is restarted, only the cache object is re-created and any non-expired items stored in the cache from previous sessions will still be available.

You can ignore the source code dealing with "Pre-Cache". This deals with special code to pre-cache the LEADTOOLS sample documents used in the demo.

The service can be modified to include replacing the code inside ServiceHelper.CreateCache to initialize and set up a different cache system and set it in the _cache variable.

FactoryController.LoadFromUri

The LoadFromUri is the entry point where the user loads a new document located on a remote URL. The document can be any file format supported by LEADTOOLS such as PDF, TIFF, DOCX, PNG, XLSX, and countless others. It is invoked from the JavaScript Document Viewer client using the "Open URL" menu item.

Ignore the "Pre-Cache" source code Which deals with the sample LEADTOOLS documents used by the demo.

DocumentFactory.LoadFromUri is called, passing the cache object that was previously created and the URL requested by the user in LoadDocumentOptions. This method quickly determines whether the data in the URL contains valid image or document data that is supported, parses the data to determine the number of pages and size of each page, and returns a new LEADDocument object containing the information.

Each new LEADDocument object requires a unique ID. Therefore, if the user did not pass one in LoadDocumentOptions.DocumentId (the value is null), then a new one is created by a GUID generator. If the user wishes to use the user's own ID, the same value is used and it is up to the user to guarantee the uniqueness of this ID. This ID is stored in the LEADDocument.DocumentId property of the created object.

LEADDocument.CacheUri is checked and if it is null, it is set to a value that can be used to obtain the original document data. Refer to the "Under the Hood" section below for more information.

The following properties of the document are set:

• AutoDeleteFromCache is set to false. You do not want it to delete itself from the cache when the .NET or Java object is disposed of.

• AutoDisposeDocuments is set to true: This is useful when the document may contain other child documents in the future. Refer to Creating Documents with LEADTOOLS Document Library for more information.

• AutoSaveToCache is set to false: Not all subsequent operations require re-saving the document to the cache (for instance, obtaining the image data of a page), therefore manual control is used when the document is saved to the cache by setting the value of this property to false to prevent the .NET/Java LEADDocument object from calling SaveToCache when it is disposed of.

• Finally, SaveToCache is called to save the document to the cache. Here, a new cache item is created from the document ID and the data structure required to reconstruct this LEADDocument object is saved into the cache.

For now, think of this as a single operation and a single item: although in reality multiple items are saved into the cache and the original document data (PDF, DOCX, TIFF) may be stored in the cache as well. This is explained in detail in the "Under the Hood" section below.

The JavaScript code will create an instance of a JS LEADDocument object from the JSON data and set it in the viewer. The viewer has all the information needed to construct the skeleton required to view the document. This includes page holders in the correct size in both the view and thumbnails area, the bookmarks tab if supported, and annotation containers. All of these are created but with empty data since the LEADDocument object does not contain any. The viewer is fully functional and the user can scroll and click items that will trigger calls to other methods in the service to obtain the required data.

Multi-user systems that will share the same document ID between different browsers can change the value of LEADDocument.CacheOptions from the default of DocumentCacheOptions.None to store page image, SVG, and text data into the cache and increase performance as explained in the next section.

The service can be modified to include:

• Verifying that the current user has access to the URL and can view the document

• (If the URL can contain just the name of a document), mapping this value to the URL where the actual image data resides in other parts of the system inaccessible from outside the service and passing this new URL to DocumentFactory.LoadFromUri

• Setting LoadDocumentOptions.DocumentId to a value that matches the ID of the document in other parts of the system to automatically map with the LEADTOOLS document ID

• Adding the document ID to a separate database to inform a system that supports sharing documents with multiple IDs that this document is ready for viewing. This feature can also be implemented directly here:

  • First try to load the document from the cache using DocumentFactory.LoadFromCache passing the document ID.
  • If this fails (will return null if the document is not in the cache), then assume that this is the first call to load this document, and continue with the original code.
  • If LoadFromCache succeeds, then the document is already in the cache and it is only necessary to return the LEADDocument object to JavaScript

• Modifying the value of LEADDocument.CacheOptions prior to SaveToCache to increase performance of systems that share the cached document ID among multiple users

PageController methods

The document is constructed and the first page outline is visible but without content. The system determines that the document supports SVG viewing, and requests the document by calling the PageController.GetSvg service method using the document ID and page number.

The service will first try to load the document from ServiceHelper.Cache using DocumentFactory.LoadFromCache with the document ID. This method will only request the small data structure required to re-create the .NET/Java LEADDocument object, and is very fast. From previous discussion, this ID is the only value needed to reconstruct the .NET/Java object.

The DocumentPage.GetSvg method is called with the specified options and the resulting SVG data is streamed back to the JavaScript code, and the .NET/Java object is disposed.

When the LEADDocument object is constructed from the cache, it will use the same settings for AutoSaveToCache and AutoDeleteFromCache: therefore, the document will not save itself back into the cache upon disposal. PageController.GetSvg is considered a read-only method that does not modify the state of the cache object.

The DocumentViewer will generally only call this method a single time per page, and rely on the browser's own caching if requested again (since this is an HTTP GET operation). The method may be called again from the same session only when the browser cache is exhausted. This is performed automatically by the web browser and is outside the control of LEADTOOLS.

The value of LEADDocument.CacheOptions is set to DocumentCacheOptions.None, meaning that only those parts required to reconstruct the document are saved into the cache. The page image, SVG, and text data are not saved into the cache. This minimizes cache size since in almost all cases, the DocumentViewer will never call PageController.GetSvg for a page more than once and the resulting SVG data (which can be large) is never requested from the server again.

Multi-user systems that share the same document ID between different browsers can change the value of LEADDocument.CacheOptions from the default value of DocumentCacheOptions.None to store page image, SVG, and text data into cache to increase performance. For instance, setting the value to All (includes PageSvg) during FactoryController.LoadFromUri above before SaveToCache will instruct the library to store the page data into the cache upon request. The workflow for DocumentPage.GetSvg is as follows:

1. Always: check whether the cache contains data for the key "documentID" + "value_of_pageNumber" + "svg". If found, return it. Naturally the first time this method called for this page, it will not find any data and will go to step 2.

2. Extract the SVG data for the page from the original (document PDF, DOCX, etc.) data. This is almost always a more expensive operation than returning the data directly from the cache.

3. If LEADDocument.CacheOptions of the owner document contains PageSvg, store the SVG data into the cache using the key above.

4. Return the SVG data

Therefore, subsequent calls from other user sessions (or browsers) to obtain the SVG data for the same document and same page will find the data in the cache at the first step and will never extract the data from the original document again.

It is possible the process could reset if the data is evicted from the cache manually or through automatic expiration. If the page SVG key is not found, steps 2-4 will be repeated and the data is re-generated when it is requested the next time.

User-modification to the service can include:

• Redacting or returning an empty page if the user do not have access to view this particular page.

Other Page and DocumentController methods

The other methods of Page and Document controller work in similar fashion to PageController.GetSvg. The document is loaded from the cache, the data is extracted using the .NET/Java LEADDocument object, and returned to JavaScript.

The following methods re-save the LEADDocument object into the cache since they modify the data:

• FactoryController.Decrypt since it will set the password required to read encrypted documents
• PageController.SetAnnotations to save the annotations modified by the user (in preparation to converting the document to other formats)
• FactoryController.SaveToCache to save newly or update virtual documents created through JavaScript into the cache.

Under the Hood

When LEADDocument.SaveToCache is called, the items below are stored in the cache. Calling DocumentFactory.LoadFromCache will succeed if all the values are found in the cache.

This is performed by calling ObjectCache.AddOrGetExisting with regionName equal to the documentID (LEADDocument.DocumentId) and key equal to the value described in the table below. These cache items are always in the cache for a document to be re-constructed (DocumentFactory.LoadFromCache). If the cache system does not support regions (or groups), then it can simply concatenate the value of regionName (the document ID) + key to create a unique cache ID. See the "Caches System Examples" section below.

The original document data (PDF, DOCX, TIFF) is required to parse the document pages data after it has been loaded from the cache. The data is stored in the "DownloadedFile_CacheId" key described below and the value depends on whether the cache system supports external resources.

If the cache supports external resources (ObjectCache.DefaultCacheCapabilities) contains ExtenalResources - such as the default LEADTOOLS FileCache which has access to a file system, then the original document data is downloaded to the physical disk file acting as the store for the cache item. This is performed by calling ObjectCache.GetItemExternalResource and writing the data directly to the file. This can reduce the memory footprint and increase performance.

If the cache does not support external resources (such as the Memory and Ehcache implementation described below), then the original data is stored as a byte[] into the cache item directly.

If client-side PDF rendering support is used with the Documents Service, then direct HTTP access to the original image data is required and must be set in Document.cacheUri JavaScript object. The PDF renderer will use this value to obtain the original data and render the PDF pages directly into the viewer surface and DocumentPage.GetSvg and DocumentPage.GetImage are never called.

The .NET/Java DocumentFactory.LoadFromUri method does not set the value of LEADDocument.CacheUri and leave it to the default value of null prior to returning it to JavaScript. The JavaScript DocumentFactory.loadFromUri method will check if the for value is null and will then replace it the HTTP GET URL required to call the service CacheController.GetDocumentData web method. Refer to source code in the service for more information.

This is the default implementation of the .NET/Java Documents Service for the following reasons:

• Simplifies deployment: The cache can be stored in any location and on any machine. Direct virtual directory access is not required.

• All calls made by JavaScript to obtain document data are routed through one place, namely, the Documents service.

Alternatively, if using a cache system that stores the items in a virtual directory (such as the LEADTOOLS FileCache), then FileCache.CacheVirtualDirectory can be set to the full virtual directory path of the cache items and the .NET/Java DocumentFactory.LoadFromUri will set LEADDocument.CacheUri to the path of the document's original data. Finally, the JavaScript DocumentFactory.loadFromUri method will check for this value, and will not modify it since it is not null.

All the possible cache IDs for a document can be obtained by calling LEADDocument.GetCacheKeys.

When a document is deleted from the cache using DocumentFactory.DeleteFromCache, the cache checks if the system contains DefaultCacheCapabilities.CacheRegions.

• If it does, it is assumed that for performance reasons it should delete a group of items using the region ID for performance. ObjectCache.DeleteRegion is called, passing the document ID as the regionName.

• If it does not support grouping or regions, then ObjectCache.DeleteAll is called with the document ID and all possible keys. It is assumed that the cache system will ignore any key that does not exist, and will only delete the items found.

  Key	Data	Source	Notes
  "downloaded_file"	Either the original URL or byte[] containing the original PDF, DOCX, TIFF, etc data of the document being viewed.	URI passed to DocumentFactory.LoadFromUri	Set during LoadFromUri. Get during LoadFromCache. Must exist. LoadFromCache will check for this and return null if it cannot be found in the cache.
  "annotations_file"	Either the original URL or byte[] containing the optional annotation data.	LoadDocumentOptions.AnnotationsUri passed to DocumentFactory.LoadFromUri	Set during LoadFromUri. Get during LoadFromCache. This is optional and can be null if no annotation file was served with the document.
  "values"	String containing internal data to re-create the Document object.	Created internally during DocumentFactory.LoadFromUri	Set during LoadFromUri. Get during LoadFromCache. Must exist.
  "pages"	String containing internal data to re-create the Document page objects.	Created internally during DocumentFactory.LoadFromUri	Set during LoadFromUri. Get during LoadFromCache. Must exist.
  "bookmarks"	String containing internal data to re-create the Document bookmark objects.	Created internally during DocumentFactory.LoadFromUri	Set during LoadFromUri. Get during LoadFromCache. Must exist.
  "rastercodecs_options"	String containing internal data to re-create the RasterCodecs used to load/save images and SVG.	Created internally during DocumentFactory.LoadFromUri	Set during LoadFromUri. Get during LoadFromCache. Must exist.

The following cache items are added to the cache, depending on DocumentCacheOptions. The format is:

String key = pageId + "_" + itemName

Where pageId is a GUID representing the page.

The following example loads a document from the cache and displays information on all of its cache items:

private static void ShowDocumentCacheInfo(ObjectCache cache, string documentId) 
{ 
   // From "Under the Hood" here: 
   // https://www.leadtools.com/help/sdk/v20/dh/to/document-toolkit-and-caching.html 
   // We know that the document ID is the cache region ID 
   // We also know that a document may contain more than 1 cache item 
   // We also know that all the possible items that belong to a document can be obtained with LEADDocument.GetCacheKeys 
 
   // First load the document from the cache 
   var loadFromCacheOptions = new LoadFromCacheOptions(); 
   loadFromCacheOptions.Cache = cache; 
   loadFromCacheOptions.DocumentId = documentId; 
   using (var document = DocumentFactory.LoadFromCache(loadFromCacheOptions)) 
   { 
      // This is to demonstrate that it is not necessary to keep the global "cache" object around, GetCache 
      // returns it for the document 
      ObjectCache documentCache = document.GetCache(); 
      Assert.IsNotNull(documentCache); 
 
      // Get all the possible cache keys 
      // These keys may or may not exist in the cache, depending on which part of the document was cached. 
      // For instance, if DocumentCacheOptions.PageImage was set in the document, and the image was cached, then there is an item for it 
      Console.WriteLine($"document {documentId} cache policies:"); 
      ISet<string> cacheKeys = document.GetCacheKeys(); 
      foreach (string cacheKey in cacheKeys) 
      { 
         // Does it exist? 
         if (documentCache.Contains(cacheKey, document.DocumentId)) 
         { 
            // Get the policy 
            CacheItemPolicy itemPolicy = documentCache.GetPolicy(cacheKey, document.DocumentId); 
            // This demo sets an absolute expiration, but this is generic code 
            // than can figure it out by examining the values: 
            DateTime absoluteExpiration = itemPolicy.AbsoluteExpiration; 
            TimeSpan slidingExpiration = itemPolicy.SlidingExpiration; 
            DateTime localTimeExpiration; 
 
            if (slidingExpiration != TimeSpan.Zero) 
            { 
               // Has sliding expiration, therefore the expiration will be NOW + sliding 
               localTimeExpiration = DateTime.Now.Add(slidingExpiration); 
            } 
            else 
            { 
               // Absolute expiration is stored in UTC, convert to local 
               localTimeExpiration = absoluteExpiration.ToLocalTime(); 
            } 
 
            // Show the expiration 
            Console.WriteLine($" key:{cacheKey} expiry at {localTimeExpiration}"); 
         } 
      } 
   } 
} 

Cache System Examples

ObjectCache is an abstract class. Derived object can be implemented to add support for caching using external systems. Below are sample implementations.

Note that the .NET Framework contains ASP.NET caching, which does not persist between sessions and so is unsuitable for the Document library.

FileCache

FileCache is the default implementation of ObjectCache. It supports regions, external resources, and virtual directories.

MemoryCache Example

This example shows a simple in-memory cache implementation showing the basics of custom caching. It should not be used in production environment.

MemoryCache

RedisObjectCache Example

This example shows an implementation of Azure Redis Cache to be used with the LEADTOOLS Document Library.

RedisObjectCache

RedisWithBlobsObjectCache Example

This example shows an implementation of Azure Redis Cache and Storage Blobs to be used with the LEADTOOLS Document Library.

RedisWithBlobsObjectCache

Ehcache Example

This examples show an implementation of the popular Java Ehcache system to be used with the LEADTOOLS Document Library.

EhcacheObjectCache

See Also

Document Library Features

Loading Using LEADTOOLS Document Library

Creating Documents with LEADTOOLS Document Library

Uploading Using the Document Library

Document Library Coordinate System

Loading Encrypted Files Using the Document Library

Parsing Text with the Document Library

Barcode Processing with the Document Library

Document Toolkit History Tracking

Document Page Transformation

Using LEADTOOLS Document Viewer

Using LEADTOOLS Document Converters

Document View and Convert Redaction

Related Topics