Technical FAQs

How accurate is PrizmDoc Viewer’s auto-redaction process? Is there any estimate of the percentage of matches that it will fail to redact?

If you’re performing auto-redaction as described in our documentation, then all matches for the input regular expression will be redacted in the final document. However, it’s difficult to express that confidence in the form of a percentage. PrizmDoc is subjected to a suite of automatic testing to ensure that its services are behaving as intended, including redactions. We never release a new version unless it passes 100% of those tests.

That being said, the primary way that inaccuracy could enter the system is if you’re attempting to redact scanned documents that have been OCRed. In this case, it’s a matter of your OCR software’s accuracy, rather than the accuracy of PrizmDoc’s redaction process. If the searchable text of the document doesn’t accurately reflect the visible text on the page (for example, if a smudged “discovery” is incorrectly recognized as “disccvery”), then the auto-redaction will be unable to recognize it due to the incorrect input it has been given. This isn’t a problem if the documents were not OCRed.

In PrizmDoc Viewer, can I delete a saved annotation collection? When I make annotations on a document, I can save them and give them a name. Then if I come back to the document later, I can access my saved annotations, add new ones, and delete individual ones, but I don’t see a way to delete the collections themselves, is this possible in PrizmDoc?

Deleting the annotation references is possible, though our Viewer sample does not have a built-in way to delete them. This is so individual users cannot delete annotations meant for a larger audience. You can add the feature to your project by using our Markup Layers API:

DELETE /MarkupLayers/u{viewingSessionId}/{layerRecordId}

We are using the Webforms Samples and want to know how to enable the “Always User Raster” feature. Which file needs to be modified and how should it be modified?

In order to enable the “Always Use Raster” option for our Webforms sample, you will need to do the following:

• Modify the Default.aspx file
• Update the Central Configuration file

How To:

1. In the Sample, before modifying, make a backup of the following file for safekeeping:

\Prizm\Samples\dotnet\webforms\full-viewer-sample\Default.aspx

2. In the original file, modify the code that is located close to line 115:

\Prizm\Samples\dotnet\webforms\full-viewer-sample\Default.aspx

Line to Replace:

} else {
        sessionData.source = getDocumentSource(query.document || '');
        }

Replace with:

} else {
        sessionData.source = getDocumentSource(query.document || '');
        sessionData.render = {
          html5: {
             alwaysUseRaster: true
          }
       };
}

3. Next, to update the Central Configuration file you will need to stop the Prizm Services.

4. Make a backup of the prizm\prizm-services-config.yml file.

5. Edit the .yml file and remove the leading # for the following line and add true to the array so that it looks like the following:

viewing.sessionConstraints.render.alwaysUseRaster.allowedValues:
[true, false]

6. Restart the Prizm Services.

Can European/CJK character sets be auto-redacted in PrizmDoc?

In general, if PrizmDoc is able to extract text from the document, it is able to do a redaction. There are no known issues with auto/manual redaction except OCR’d “text over image” documents. For these documents, image text may have a different location or height/width in comparison to the text recognition result included in that document as well. A redaction will be applied to the text, but the underlying text on the image can still be visible.

For reporting purposes, we want to rename the markup layer JSON files to help identify which tenant made the markup. For example:

c1fea5868607402c1b664272e5b82aeb7fff7390_0_0hg91gTqSCeg_P9B2Ko7BaH-YxkaazjH.json

rename to:

accusoft_c1fea5868607402c1b664272e5b82aeb7fff7390_0_0hg91gTqSCeg_P9B2Ko7BaH-YxkaazjH.json.

Is this possible?

The actual filename for an annotation .json file contains an encrypted value of the file path and document name. This is what is used to associate the JSON markup layer file with the document.

If you make changes to the first part of the file, PrizmDoc will no longer be able to associate those markups to the actual file.

As a alternative, you can set up a separate PAS instance for each tenant, which will provide a different markup layer for each tenant.

Or you can programmatically handle the markups and store them specifically for each tenant. The following articles outline our Markup JSON specifications:

• PrizmDoc Markup JSON specification:
  https://help.accusoft.com/PrizmDoc/latest/HTML/webframe.html#markup-json-specification.html
• Use the JSON Markup Schema: https://help.accusoft.com/PrizmDoc/latest/HTML/webframe.html#marks-json-schema-usage.html

Does viewing HTML in PrizmDoc Viewer allow JavaScript execution or local file access? Can PrizmDoc Viewer block externally-referenced content from being rendered?

When viewing HTML in PrizmDoc, JavaScript and local file access are disabled.

Additionally, you may configure the security.htmlRendering.blockExternalContent setting found in PrizmDoc’s Central Configuration file. When rendering any source document which uses HTML content, this setting controls whether or not externally-referenced content, such as images and iframes, will be blocked. This option affects any source document file type which uses HTML, including HTML, EML, and MSG.

I want to load an HTML document in PrizmDoc with UTF-8 encoding. Can this be done automatically in the product?

Currently, no. We have a parameter for .txt files which does that (detailed here), but this “textFileEncoding” intentionally only works for .txt, not .html files. There is a feature request for this:

https://ideas.accusoft.com/ideas/PDV-I-546

In the meantime, this can be fixed manually by adding charset = “utf-8” to the meta tag of the HTML document. One POC way this might be done programmatically is below in Python 3.7 (need obvious polishing like checking for the tag already existing, multiple “meta” tags, etc):

with open(filename, "r") as file:
    content = file.read()

index = content.find("meta") + len("meta")

new_content = content[:index] + " charset=\"utf-8\" " + content[index:]

with open(filename, "w") as file:
    file.write(new_content)

We are using PrizmDoc and may be adding more end users, so we wanted to verify whether there is a limit to the number of viewing sessions that can be active at one time.

PrizmDoc does not specifically limit the number of viewing sessions that can be active at any given time. However, when opening a viewing session with a document, a conversion does take place. The number of conversions that can be handled simultaneously is based on the hardware constraints of your server.

Various factors, such as the size of documents, the number of conversions, and the hardware being utilized may impact the performance of your PrizmDoc server.

We provide server sizing and performance documentation that is designed to provide guidance and approximate server size based on the number of conversions you plan to do per minute.

Please refer to our documentation page:

https://help.accusoft.com/PrizmDoc/latest/HTML/webframe.html#prizmdoc-server-sizing-servers.html

We want to use one PAS instance for multiple customers while ensuring their calls and data are separated so that other customers cannot access them. Is there a way PrizmDoc can accomplish this?

How To:

Use the following steps to set up the multi-tenancy feature:

1. Add a new line inside of /usr/share/prizm/pas/pcc.nix.yml (Linux) or C:\Prizm\pas\pcc.win.yml (Windows) in the following format:

undocumentedFeature.multiTenancy: “enabled”

2. After enabling this feature, all requests to PAS must include the header accusoft-tenant-id. The following shows an example request for creating a new viewing session:

POST /ViewingSession HTTP/1.1
Host: localhost:3000
Content-Type: application/json
accusoft-tenant-id: myUniqueTenantId
cache-control: no-cache
Postman-Token: 5edd698a-5e4f-46d2-b93a-42cc57371dce
  {
  "source": {
  "type": "document",
  "fileName": "1040ez.pdf"
   }
  }

NOTE: All Tenant Ids are converted to lowercase and must be unique between tenants. This means that in terms of the application both “MyTenant” and “mytenant” are equivalent and would be able to see the same files. Tenant Ids are not generated for the application and must be generated and handled by the integration components. Tenant Ids are also strictly alphanumeric at this time.

File storage including Documents, markupXml, formDefinitions and markupLayerRecords will now be appended with the
{tenantId} as shown in the above example with a Documents configuration of /usr/share/prizm/Samples/Documents. The request would attempt to create a viewing session from the following file: /usr/share/prizm/Samples/Documents/myuniquetenantid/1040ez.pdf

Viewing Package data stored in the database will have the tenantId included in the composite index as well as include an accusoftTenantId column.

When doing a text search in a document in PrizmDoc, the highlight color for the search results is a light blue background. Is there a way to change the highlight color for search results in the PrizmDoc Viewer?

You can change the highlight color specifically in the PrizmDoc Viewer file viewer.js file located in the viewer-assets\js folder of your project/application.

You will need to locate 2 instances of the search term “highlightColor: undefined” in the code block below located close to line 4400. The actual line number will vary depending on the version of viewer.js you are using. You can replace “undefined” with a hexadecimal color value which you can look up from the following site: https://htmlcolorcodes.com/

// This is a request for a new searchQuery triggered by the search input field.
// Generate new search terms, and save them globally.
prevMatchingOptions = _.clone(matchingOptions);

if (matchingOptions.exactPhrase) {
    // We need to match the exact string, as is
    if (queryString.length) {
        searchTerms.push({
            searchTerm: queryString,
            highlightColor: undefined,
            searchTermIsRegex: false,
            contextPadding: 25,
            matchingOptions: matchingOptions
        });
    }
} else {
    // Split up multiple words in the string into separate search term objects
    var queryArr = queryString.split(' ');
    queryArr = _.unique(queryArr);
    _.forEach(queryArr, function(query){
        if (query.length) {
            searchTerms.push({
                searchTerm: query,
                highlightColor: undefined,
                searchTermIsRegex: false,
                contextPadding: 25,
                matchingOptions: matchingOptions
            });
        }
    });
}