[layout issue] columns not found properly

in this doc, [docugram.pdf](https://github.com/user-attachments/files/28702508/docugram.pdf) the beginning of the second column appears before the abstract when I run firecrawl/pdf-inspector, and then it resumes at the caption of Figure 1.

```
node test.mjs
TextBased
# DocuGram: Turning Screen Recordings into Documents

## Laurent Denoue Scott Carter Matthew Cooper

### FXPAL FXPAL

FXPAL

3174 Porter Drive 3174 Porter Drive 3174 Porter Drive

Palo Alto, California 94304 USA Palo Alto, California 94304 USA Palo Alto, California 94304 USA

## denoue@fxpal.com carter@fxpal.com cooper@fxpal.com

options, people often need to decontextualize their comments, e.g. in the accompanying email message.

Or export the document into a PDF editor, highlight, and then hope the recipient will also have a PDF reader that understands and renders annotations.

One last resort is for users to record their screen. Although videos can be indexed (see [1] for examples), recipients would typically be shown a video player, making it hard to page through and otherwise read the “document”.

## ABSTRACT

In this paper we describe DocuGram, a novel tool to capture and share documents originating from any application. As users scroll through pages of their document inside the native application (Word, Google Docs, web browser), the system captures and analyses in real-time the rendered video frames and reconstitutes the original document pages into an easy to view HTML-based representation. In addition to detecting and regenerating the document pages, a DocuGram also includes the interactions users had over them, e.g. mouse motions and voice comments. A DocuGram allows users to flexibly share enhanced documents across applications.

## Keywords

Document capture; image processing; video processing; interactive documents.

# 1. INTRODUCTION

We have numerous available choices to edit and view documents, ranging from native applications such as Word and PowerPoint, to online editing tools like Google Docs and Microsoft Office 365. We also have many ways to share these documents, such as by attaching them in email, embedding them in web pages, or simply sending a link of their online location.

But individual source applications often require users to devise custom means to share that document. One could simply attach a Word document by email, hoping the recipients will have Word installed; or one might export the document as a PDF; or generate a sharable link to the document.

While already complex for our own documents, capturing and sharing is even harder for content we don’t author, such as pages from a scanned book hosted on Google Books or a slide deck shown during web conferences. Even the seemingly mundane task of sharing the URL of an online article could cause problems to the recipient (e.g. a pay-wall from the Wall Street Journal).

Furthermore, sharing a document often means leaving out comments and interactions we have over the document. Without

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from Permissions@acm.org.

*DocEng '16, September 12-16, 2016, Vienna, Austria* © 2016 ACM. ISBN 978-1-4503-4438-8/16/09 $15.00 DOI: [http://dx.doi.org/10.1145/2960811.2967154](http://dx.doi.org/10.1145/2960811.2967154)

**Figure 1. A sample DocuGram is a document reconstituted**

```

Start by running the reported node test.mjs with docugram.pdf and compare the extracted text with the document's two-column layout. Trace the PDF text-extraction path that determines reading order, especially around the abstract and Figure 1 caption. Done means the columns are returned in the correct order without losing the surrounding text.

Written by the indexing model from the issue text.