How to fix bad results on some pages that look like a Form

I have a pdf with a page that resembles a form or report. this is what I get in md:

## EXAMPLE:

Project No.:

STP-0059-03(023)/100556301

Contractor:

Sam Finley, Inc.

County:

Lauderdale

Material:

907-304-B

Granular Material, Class 5, Group D

Lbs

0000011 TRUCK NUMBER

789

907-304-B

0028552 QUANTITY

STP-0059-03(023)/100556 SAMFINLEYINC LAUDERDALECOUNTY

6GRANMATL CL5GPD TON

PROJECTID

PAYITEM

GROSS

- [x] LBS.

76920

052114

- [ ] GALS.

TARE

- [ ] CU. YD..

30420

DATE

- [ ] MSTRE

TEMP

TIME

CHECKEDBYPOINTOFLOADING:

°F

- [ ] A.M.

- [ ] P.M.

JohnDoe

TEMP

TIME

RECEIVEDBYPOINTOFDELIVERY:

JamesWilliams

°F

- [ ] A.M.

- [ ] P.M.

POINTOFDELIVERYSTATION:

TOSTATION:

MSTRECONTENT

0.058 *

REMARKS:

No. 333601

Moistureobtainedat1:15P.M.

PROJECTENGINEER'SORIGINAL

Construction Manual

Start by reproducing the reported conversion with the PDF page shown in issue #3653 and trace the document-parsing path that produces the displayed Markdown. Compare the output with the form's structure; done means the page's fields, labels, and checked values are represented more accurately, though the issue does not specify an exact expected output or test location.

Written by the indexing model from the issue text.