Skip to main content

Generating PDF/A From HTML in Meteor

My live-chat app was a folk of project Rocket.Chat which was built with Meteor. The app had a feature that administrative users were able to export the conversations into PDF files. And, they wanted to archive these files for a long time.

I happened to know that PDF/A documents were good for this purpose. It was really frustrated to find a solution with free libraries. Actually, it took me more than two weeks to find a possible approach.

Using Puppeteer to generate a normal PDF and using PDFBox to load and converting the generated PDF into PDF/A compliance.

What is PDF/A?

Here is a definition from Wikipedia:
PDF/A is an ISO-standardized version of the Portable Document Format (PDF) specialized for use in the archiving and long-term preservation of electronic documents. PDF/A differs from PDF by prohibiting features unsuitable for long-term archiving, such as font linking (as opposed to font embedding) and encryption. The ISO requirements for PDF/A file viewers include color management guidelines, support for embedded fonts, and a user interface for reading embedded annotations.

Why PDF/A documents matter

In my point of view, it’s a standard file for archiving thanks to key features:
  • The fonts of document’s content will always be displayed the same on all places, not depending on any devices.
  • Documents are safe. For example, no executable file launches are allowed or no external content references are allowed.
(Find more features at

Several approaches for creating a PDF/A document

At the first glance, I found some approaches as follows:

(1). Paid APIs
  • Aspose PDF: Java-based APIs. It seemed the price was too expensive.
  • PdfTron: Javascript-based APIs. I tried to contact them to know the price but they requested me some information even I could not estimate to answer it.
(2). Free APIs
  • Apache FOP: Java-based APIs. I needed to defined the templates by its own ways (FOP files) instead of HTML and CSS. (I created a hello world app here)
  • PDFBox: Java-based APIs. It looked promising that I could convert a normal PDF document to PDF/A one.
(3). Forking a open source project/building my own PDF/A generating engine
  • pdfkit: Javascript-based project. I needed to know the specification of PDF/A compliance, and to build my own lib for generating PDF/A documents.
Basing on these insights, I decided to go with approach (2).

Apache FOP tryout

I decided to stop this approach. Apache FOP strictly required to provide templates with XSL-FO format which is not supported wisely. Ref:
  • Due to limitation of formatting support, it’s hard to format content of a PDF if its template is complex such as including images, tables, etc..
  • There seemed to be no tools for converting from HTML, CSS into XSL-FO. But my templates were built on HTML, CSS and Meteor and data was bound dynamically.

PDFBox tryout

Basing on the example of creating PDF/A here, I tried to convert an existing normal PDF file into another PDF/A. There were 3 parts to do:
  1. embed/load fonts
  2. include XMP metadata
  3. include color profile
I used these following tools to verify the result:
With filling some mandatory information, I passed the parts 2 and 3 successfully. But, I have really got stuck at part 1 for a long time.

Eventually, I found the root cause by delving deeper into source code of some libraries PDFBox, PDFBox Preflight, node-html-pdf, PhantomJS, etc.. as follows:

I could not successfully generate PDF/A document by converting the existing normal PDFs generated from Meteor because the PDF generated did not embedded the font fully.

My project used node-html-pdf which in turn used PhantomJS to render HTML into PDF. PhantomJS used Qt for writing PDF. The library always embedded the fonts using Subsetting (meaning only the characters were used in the document were embedded). This did not comply with the PDF/A specification.

I couldn’t do anything here because PhantomJS used a very old version of Qt (only Qt v5.x onwards supported PDA/A). The PhantomJS was also discontinued since Mar 2018.

Besides, I could not use PDFBox to embed fonts fully either. There was a method PDType0Font.load for loading fonts but it also embedded the fonts using Subsetting.

Yay! It worked

I discovered that by replacing the process rendering HTML to PDF with the fonts fully embedded.
Using a different library to replace PhantomJS, I found Google Chrome’s Puppeteer.

Then, I could use PDFBox to convert the normal PDF to PDF/A with including XMP metadata and color profile.



  1. I like your blog, I read this blog please update more content on python, further check it once at python online training

  2. what is fantastic post? this is so chock full of useful information I cannot wait to dig deep and start utilizing the resource give me.your exuberance is refreshing.
    Portal Development
    Travel portal development
    Travel white label
    Travel Portal Solution
    B2C Travel Portal
    B2B Travel Portal
    Flight Booking API System
    Flight api integration


Post a Comment

Popular posts from this blog

Applying pipeline “tensorflow_embedding” of Rasa NLU

According to this nice article, there was a new pipeline released using a different approach from the standard one (spacy_sklearn). I wanted to give it a try to see whether it can help with improving bot’s accuracy.

After applying done, I gave an evaluation of “tensorflow_embedding”. It seemed to work better a bit. For example, I defined intents “greet” and “goodbye” with some following messages in my training data.
## intent:greet- Hey! How are you? - Hi! How can I help you? - Good to see you! - Nice to see you! - Hi - Hello - Hi there ## intent:goodbye- Bye - Bye Bye - See you later - Take care - Peace In order to play around with Rasa NLU, I created a project here. You can have a look at this change from this pull request. Yay!

When I entered message “hi bot”, then bot with “tensorflow_embedding” could detect intent “greet” with better confidence scores rather than bot with “spacy_sklearn”. The following are responses after executing curl -X POST localhost:5000/parse -d '{&qu…

Creating a Chatbot with RiveScript in Java

Motivation"Artificial Intelligence (AI) is considered a major innovation that could disrupt many things. Some people even compare it to the Internet. A large investor firm predicted that some AI startups could become the next Apple, Google or Amazon within five years"
- Prof. John Vu, Carnegie Mellon University.

Using chatbots to support our daily tasks is super useful and interesting. In fact, "Jenkins CI, Jira Cloud, and Bitbucket" have been becoming must-have apps in Slack of my team these days.

There are some existing approaches for chatbots including pattern matching, algorithms, and neutral networks. RiveScript is a scripting language using "pattern matching" as a simple and powerful approach for building up a Chabot.
Architecture Actually, it was flexible to choose a programming language for the used Rivescript interpreter like Java, Go, Javascript, Python, and Perl. I went with Java.

Used Technologies and ToolsOracle JDK 1.8.0_151Apache Maven 3.5…

[Snippet] CSS - Child element overlap parent

I searched from somewhere and found that a lot of people says a basic concept for implementing this feature looks like below:
HTML code:
<div id="parent">  <div id="child">  </div> </div> And, CSS:
#parent{   position: relative;   overflow:hidden; } #child{   position: absolute;   top: -1;   right: -1px; }
However, I had a lot of grand-parents in my case and the above code didn't work. Therefore, I needed an alternative. I presumed that my app uses Boostrap and AngularJs, maybe some CSS from them affects mine. I didn't know exactly the problem, but I believed when all CSS is loaded into my browser, I could completely handle it.

I tried to create an example to investigated this problem by Fiddle. Accidentally, I just changed: position: parent; to position: static; for one of parents -> the problem is solved.
Look at my code:

<div class="modal-body dn-placeholder-parent-position"> <div clas…

How I did customize "rasa-nlu-trainer" as my own tool

Check out my implementation here Background I wanted to have a tool for human beings to classify intents and extract entities of texts which were obtained from a raw dataset such as's conversation, Maluuba Frames or here. Then, the output (labeled texts) could be consumed by an NLU tool such as Rasa NLU.

rasa-nlu-trainer was a potential one which I didn't need to build an app from scratch. However, I needed to add more of my own features to fulfill my needs. They were:

1. Loading/displaying raw texts stored by a database such as MongoDB
2. Manually labeling intents and entities for the loaded texts
3. Persisting labeled texts into the database

I firstly did look up what rasa-nlu-trainer's technologies were used in order to see how to implement my mentioned features.
At first glancerasa-nlu-trainer was bootstrapped with Create React App. Create React App is a tool to create a React app with no build configuration, as it said. This tool is also recommended by the …

Styling Sort Icons Using Font Awesome for Primefaces' Data Table

So far, Primefaces has used image sprites for displaying the sort icons. This leads to a problem if we want to make a different style for these icons; for example, I would make the icon "arrow up" more blurry at the first time the table loading because I want to highlight the icon "arrow down". I found a way that I can replace these icons with Font Awesome icons.

We will use "CSS Pseudo-classes" to achieve it. The hardest thing here is that we should handle displaying icons in different cases. There is a case both "arrow up" and "arrow down" showing and other case is only one of these icons is shown.

.ui-sortable-column-icon.ui-icon.ui-icon-carat-2-n-s { background-image: none; margin-left: 5px; font-size: 1.1666em; position: relative; } .ui-sortable-column-icon.ui-icon.ui-icon-carat-2-n-s:not(.ui-icon-triangle-1-s)::before { content: "\f106"; font-family: "FontAwesome"; position: absolut…