need to hack code39 behaviour

[abubelinha]

I see this thread is the only discussion about code39 barcodes.

In the fpdf-based php script I am trying to port to fpdf2-python, there was a very tiny modification of Code39().
So that I could choose the line proportions in generated barcodes (I can't recall exactly how I decided to do this but I think that made barcodes easier to read with my mobile, at least for a particular label size I needed to print some time ago).

This was a hack made by me and I think the equivalent change in python would only affect the beginning of code39() definition:
https://github.com/py-pdf/fpdf2/blob/aa3b1f4a1e16b1d513e5064278b2fb2f5b2747af/fpdf/fpdf.py#L4517C1-L4519C35

Current definition:

def code39(self, text, x, y, w=1.5, h=5):
        """Barcode 3of9"""
        dim = {"w": w, "n": w / 3}

Modified version:

def code39(self, text, x, y, w=1.5, h=5, proportion=3):
        """Barcode 3of9"""
        dim = {"w": w, "n": w / proportion}

I have very low Python programming skills but I think this change wouldn't break compatibility with previous versions.
I doubt this hack can be useful for someone else (I don't see myself improving barcodes technology), but I leave that to you as a possible think to change.

Anyway, I need to do this hack in order to port my php script, so I ask your opinion about the most correct/easiest way to do this in Python.

How can I do this right in my script, without touching the installed fpdf2 package?
So when I upgrade fpdf2 version, my script can keep working and using my modified code39() version.
(at risk of missing only other possible improvements of that specific code39() function which fpdf2 can make in the future)

Thanks a lot in advance for your suggestions
@abubelinha

[gmischler]

I don't have access to ISO/IEC 16388, but all the descriptions of code39 I can find say that the ratio should be between 2:1 and 3:1 (recommended), so I suspect that is exactly what the standard says as well.
There's not really a point in adding a feature to the library for the sole purpose of creating non-standard barcodes.

Of course you're free to subclass FPDF and overwrite the .code39() method for your own purposes. The code you show above should indeed do the trick.
But then, maybe you want to run some new tests first? Even if a phone camera from many years ago may have had difficulties reading a standards-conforming barcode, that will likely not be the case anymore today. Figuring that out might save you from the hassle of creating and maintaining a custom version of fpdf2. Who cares if your barcode is a few mm shorter from now on?

[abubelinha]

Of course you're free to subclass FPDF and overwrite the .code39() method for your own purposes. The code you show above should indeed do the trick.

Yes, that's what I want to do (implementing also the standard ratio limits you told me about).

I'd like to apply the changes within my script, without touching the installed fpdf2 module? So that ...

• if I install other people scripts, their pdf.code39() calls keep working
• if my script calls pdf.code39(), it accepts my new ratio optional parameter
• when I upgrade fpdf2 in the future, all the above keeps working (I shouldn't loose any new fpdf2 features since its code39 function is not expected to be modified, as per your comments)

I guess something like this is the most proper way to do it?

fpdf_modified.py:

from fpdf import FPDF
#from fpdf import check_page # @check_page is a decorator of code39() in FPDF. How do I use it here?

#@check_page   # commented because the above import  doesn't work
def _new_code39(self, text, x, y, w=1.5, h=5, ratio=3):
	"""Barcode 3of9 (modified with non-forced, optional user-defined ratio)"""
	if ratio>3 or ratio<2: # https://en.wikipedia.org/wiki/Code_39
		ratio=3 # use our default to keep within the standard
	dim = {"w": w, "n": w / ratio}
	(...)

FPDF.code39 = _new_code39 # this rewrites the original function so myscript.py can keep using its name

myscript.py:

from fpdf_modified import FPDF # imported from the above script
pdf = FPDF(orientation="P", unit="mm", format="A4")
pdf.add_page()
pdf.code39("*fpdf modified code39*", x=30, y=50, w=1.5, h=20, ratio=2.8)
pdf.output("fpdf-hacked-code39-test.pdf")

Although the above code seems to work, I still have some questions:

1. The original code39() function has a @check_page decorator. What's the proper way of importing/implementing it in my fpdf_modfied.py script? I'm concerned about which things could go wrong if I don't use it.
2. The cod39 example in fpdf2 documentation shows an Arial text below the barcode (so a human can actually read the info in the code). It was like that also in the original PHP Code39() extension.
   But current code39() method does not output that text. Do you know if it was present in any of the previous Python versions of fpdf and later removed, so I don't need to go back to PHP code and figure out how to do that? (I might want to make human-legible codes too as boolean option in my versioned function ... as long as that doesn't break compatibility with other fpdf2 features, like new rotation method).

Thanks a lot again for your time

[gmischler]

Using inheritance and overwriting the method is much simpler than injecting it into the original.

class MyFPDF(FPDF):
    @check_page
    def code39(self, text, x, y, w=1.5, h=5, ratio=3):
        ...

pdf = MyFPDF()

The @check_page decorator simply checks if you already have a page added to the document, since executing the method doesn't make any sense otherwise.

I don't think the Python version ever added the text below the barcode (which means the graphic example in the documentation is incorrect). That should be easy enough to add to your version if you need it.

[abubelinha]

Thanks for the advice @gmischler
Sorry for the dummy questions.
To have the modified class available in a separate fpdf_modified.py file as above, is this a correct way to do it?
(it works, except by @check_page ... but as you put it in your code I guess I am doing something wrong):

fpdf_modified.py:

import fpdf
class MyFPDF(FPDF): 
	# Should I redefine check_page() in this file, outside the class?
	#@check_page # still NOT working (error if I import/invoque it).
	def code39(self, text, x, y, w=1.5, h=5, ratio=3):
		...

myscript.py:

from fpdf_modified import MyFPDF # imported from the above script
pdf = MyFPDF(orientation="P", unit="mm", format="A4")
pdf.add_page()
pdf.code39("*fpdf modified code39*", x=30, y=50, w=1.5, h=20, ratio=2.8)
pdf.output("fpdf-hacked-code39-test.pdf")

[gmischler]

it works, except by @check_page

Since we're using it in a different file, that needs to be a qualified reference: @fpdf.check_page
I didn't actually test my example code before posting it... 😉

[abubelinha]

Thanks. I also posted a wrong line in my last fpdf_modified.py. I was actually using it like this:
from fpdf import FPDF
... instead of
import fpdf

But I must be missing something else. Using both import variants, calling @fpdf.check_page returns an error:

fpdf_modified.py:

	import fpdf # added trying to have access to check_page()
	from fpdf import FPDF
	class MyFPDF(FPDF): 
		@fpdf.check_page
		def code39(self, text, x, y, w=1.5, h=5, fontsize=10, ratio=3):
...

When running myscript.py I get: AttributeError: module 'fpdf' has no attribute 'check_page'

Looks like even with import fpdf is not possible to access check_page() function from fpdf_modified.py?
I don't understand why.

Tried also from fpdf import check_page but it doesn't work either:
ImportError: cannot import name 'check_page' from 'fpdf'

[gmischler]

Oops, I just realized that I've been leading you on the wrong path with check_page.
As it happens, it is not included in the __all__ list of fpdf.py, so you can't import it into another file.

So the simples way is to just leave it away. After all, its only purpose is to give you an easier to understand error message if you try to add some content to a document with no pages added yet.

[abubelinha]

No problem. With this discussion I got examples on how to do things in Python.

Thanks for all your detailed explanations!