Problem statement: We have a QR code that will be stamped on the PDFs which redirects the user to the location of the actual(Digital) document in the application. The requirement around this was that the QR code should be scannable after printing the PDF on a paper without any scale using 300 DPI, the size of the PDF can vary in between A5 and A0. Solution approach: We have printed QR codes of various sizes on papers of various sizes and tested the scannability using multiple devices to determine the sweet spot where the QR is scannable with minimum possible size and using the results while stamping.

The freezing process is done on the PDF to make the assets stamped on it non editable on the Adobe acrobat reader(V8.x). The legacy system was converting all the pages of the PDF to images and merging all the images as pages in the PDF. We implemented in the same way and found out that the process is heavily resource intensive and time consuming and has a lot of quality loss in the output, so we switched the process to using a third party application called Ghostscript to produce the desired output in 1/10 of the time and without any quality loss with lesser resource consumption and deployed it as a microservice.

Data migration from the old system involved in migrating the documents/files along with the data from the database(Total 8 TB approximately). We have prepared a system for the migration which happens in several steps. One of the steps involved is converting the PDF pages to images and this is time and resource consuming and it affects the migration timeline a lot. To counter this, We have deployed an automated solution in an EC2 instance with 16 Core CPU that will migrate all the PDFs that need image conversion to the new system by comparing the job delta between the old and new systems.

Requirement: There is a particular email id where clients will be mailing the details of the project with relevant plan files. Users should be able to view the emails in the inbox as a list and convert them as jobs. The email should be saved as a file of the typeMSG(Microsoft Outlook’s default format) along with attachments in the mail. Solution approach: We are getting the emails using Microsoft Graph APIs and using a third party library to create the MSG files(These files can be imported at a later date in outlook to access the original thread). This system is being run as a scheduled job with an option to refresh on demand.

Requirement: Users will be updating the contents/files of some folders of a particular category using third party applications and the application should handle the files similar to the files uploaded through the application, if PDFs are uploaded, they should be able to be included in the digital plans. Solution approach: We have set up a scheduled job that fetches the current files in the S3 folder in regular intervals or on demand and compares the information in the database to check if any file is added, updated or deleted. Any updates in the files will be saved to the database and if any PDF is updated, we will generate images of the pages and store it in the appropriate folder to load them in model in the user interface.

The main actions in the digital plan lifecycle stamp, merge and freeze consumes heavy resources and time, this will affect the availability of the whole service if many of these actions are being done properly, so we are using microservices to handle this and also we can push these time consuming tasks as background processes freeing up the UI to be used for any other action. To handle heavy traffic, we have set up the microservices in a way that one instance will be launched to serve 1 HTTP request so that the microservice can handle large PDFs simultaneously without any limitations. These microservices can only be accessed through the project’s VPC and will not be exposed to the internet.

Images will be stamped on the PDF using coordinates of lower left corner, width, height and rotation of the image. This allows us to control the position, size and orientation of the stamped images. Even though the placement of images is straightforward, in practice the source PDFs may be having a rotation by default internally when the content of the PDF is displayed in the perfect orientation to the user. This makes the stamping tricky as the origin(0,0) will be shifted to any of the page corners depending on the rotation. Even the width and height of the images will also be interchanged but the width and height of the page will not be affected as the orientation as in portrait or landscape is saved in the PDF’s metadata. To solve the issues introduced by the internal PDF rotation, we have created systems that will adjust the input stamp data for all the images to be stamped by getting the default page rotation so that the final stamping is done as user intended.

To enable a user to place the stamps on the PDFs, we are converting the pages to images without affecting the aspect ratio and loading the images(size and location will be determined based on the conditions that will be discussed further the document) on a model in the website and allowing the user to place other images on it by dragging and dropping them. Once the user places all the images and clicks on the save button, we will prepare the coordinates according to the PDF stamping module by measuring the distance of the image from all the edges.