Area Plugin Support

[kavinsharma]

An invoice2data area plugin helps in extracting text on the basis of area coordinates utilizing pdf2text area cropping option. An area plugin is customization to invoice2data to define area cropping with coordinates. Coordinates defined for the template can vary from pdf to pdf.
You just have to add a normal template containing the YAML file in which there are different plugins for fields and tables and you just have to add an area plugin and it works on every pdf.
Just write the field of multiple lines you want to extract and give the coordinates of that field that is (x=? y=? r=? H=? W=?)

x = x coordinate
y = y coordinate
H= height of the area
W= width of the area 
R= resolution of invoice

Area Plugin Options:
Name: field name to map with extracted text
Area: takes dict as input for cropping pdf area and extract text
Regex: Optional parameter, used for further extracting text from the cropped area.
Sample Invoice

Here is a sample of an invoice template including the area plugin that helps you extract multiple lines.

# -*- coding: utf-8 -*-
issuer: The XYZ Company
keywords:
- The XYZ Company
- US Supplier 123
fields:
	amount: TOTAL:\s+(\d+,\d+\.\d\d)
	date: Invoice Date:\s+(\d{1,2}\/\d{1,2}\/\d{4})
	delivery_date: Delivery Date:\s+(\d{1,2}\/\d{1,2}\/\d{4})
	invoice_number: INVOICE:\s+(\w{3}\d{1,8})
	sales_order : Sales Order:\s+(\w{2}\d+)
tables:
	-   start: Line\s+Product\s+Description\s+Quantity
    	end: Prices
    	body: (?P<Line>^\d{2})\s+(?P<Product>\w{2}\-\w{2}\-\w+\-\w+)\s+(?P<Description>\w+\s\w+\s\w+\-\w+)\s+(?P<Quantity>\d+)
area:
    -  name: "Address"
       area: {x: 115,y: 124,r: 300,W: 412, H: 326}
       regex: \w+
     - name: "Bill_to"
       area: {x: 930,y: 416,r: 300,W: 310, H: 294}

options:
	remove_whitespace: false
	currency: USD
	date_formats:
    	- '%d/%m/%Y'
	languages:
    	- en
decimal_separator: '.'

Output :

{'issuer': 'Innomatiq', 'amount': 51500.0, 'date': datetime.datetime(2020, 3, 12, 0, 0), 'invoice_number': 'INV0005', 'Terms': 'Due On Receipt', 'cur
rency': 'USD', 'DESCRIPTION': 'Laptops', 'RATE': '$1,500.00', 'QTY': '1', 'AMOUNT': '$1,500.00', 'Address': 'Invoice INV0005 \r\nInnomatiq\r\n\r\nSuite 2000\r\nPlano, TX\r\n75023\r\n\r\[email protected]', 'Bill_to': 'Dell 7\r\nRound Rock, Tx\r\n78664\r\n(800) 285-1653\r\[email protected]', 'desc': 'Invoice from Innomatiq'}

[rmilecki]

@kavinsharma: I think implementing this feature as separated plugin is non optimal and will cause maintenance problems in a long term. You add another piece of code handling regular expressions. Soon someone will ask for specifying field types (integers, floats, dates). Later someone will ask for handling sums.

I think support for areas should be added to standard fields. Your code should handle extracting areas only and don't care about actual parsing. This should be easy to implement once we managed to polish and merge #308.

I'd suggest working on top of syntax like:

fields:
  foo:
    area: {x: 115,y: 124,r: 300,W: 412, H: 326}
    [parser details]

[kavinsharma]

@rmilecki thanks for the review, I totally agree with you. Let me make these changes and update the PR

[RossK1]

@kavinsharma , any update on reworking this as a parser so it can be merged? This is exactly what I'm looking for to extract addresses from invoices.
Thanks 😄

[erkin98]

any update?

[m3nu]

Needs to pass tests and also a rebase.

[RossK1]

@m3nu, do you still think this would be better as a parser or as a plugin like lines or tables is acceptable?

[m3nu]

Makes sense as plugin, which it already is.

[kavinsharma]

Hi @m3nu @RossK1,
if it makes sense as a plugin, i am fixing the conflicts and working on passing the tests

[RossK1]

Hi @m3nu @RossK1,
if it makes sense as a plugin, i am fixing the conflicts and working on passing the tests

Hey @kavinsharma :) I'm currently working on this too :P
Adding it as an argument to the parser field. Will push the commit in a few minutes here. All credit goes to you for the idea and the base code!

[m3nu]

Added some comments. I see you need to rely on existing modules. We don't really have that anywhere else. Will be better to try reusing what's already there.

[bosd]

This PR is open for a while. What needs to be done to get this merged?

[BenjaminHoegh]

Any news?

[bosd]

Area support has been added in #438
Propably this one can be closed