fpdf fpdf2 migration documentation

[juleslagarde]

I see you are maintaining a documentation to help the migration from fpdf : #411 (comment)
Thanks a lot for that !

I'm trying to migrate a code base to fpdf2 but the spacing seems different.

Where could I find the documentation that specify the main difference from fpdf ?

[Lucas-C]

Hi @juleslagarde

You will find the documentation you are looking for there: https://py-pdf.github.io/fpdf2/History.html#compatibility-between-pyfpdf-fpdf2

If ever you find that this documentation is missing some important differences, please tell us so that we update it 🙂

[juleslagarde]

Here is how to reproduce my issue.

from fpdf import FPDF, HTMLMixin, FPDF_VERSION

class FPDF_Compat(FPDF, HTMLMixin):
	pass

pdf1 = FPDF_Compat(orientation = 'P', unit = 'mm', format='A4')
pdf1.add_page()
pdf1.write_html(FPDF_VERSION)
pdf1.write_html('<table border=1><tr><th width="100%">Test 1</th></tr></table>')
pdf1.write_html('<table border=1><tr><th width="100%">Test 2</th></tr></table>')
pdf1.output('output.pdf')

result:

I agree that the fpdf2 version is the better choice but I need to find a solution too migrate simply

[Lucas-C]

There is a starting point:

import logging, sys, warnings

warnings.simplefilter("default", DeprecationWarning)
logging.basicConfig(format="%(asctime)s %(filename)s [%(levelname)s] %(message)s",
                    datefmt="%H:%M:%S", level=logging.DEBUG)

from fpdf import FPDF, FPDF_VERSION

print(f"{FPDF_VERSION=}")

# --- compatibility layer START ---
from html import unescape
from html.parser import HTMLParser
HTMLParser.unescape = lambda self, text: unescape(text)

from fpdf.html import HTML2FPDF

class HTML2FPDF_Compat(HTML2FPDF):
    def handle_starttag(self, tag, attrs):
        super().handle_starttag(tag, attrs)
        if tag == 'table':
            self.pdf.ln()
    def handle_endtag(self, tag):
        super().handle_endtag(tag)
        if tag == 'table':
            self.pdf.ln()

class FPDF_Compat(FPDF):
    HTML2FPDF_CLASS = HTML2FPDF_Compat
# --- compatibility layer END ---

pdf1 = FPDF_Compat(orientation = 'P', unit = 'mm', format='A4')
pdf1.add_page()
pdf1.write_html(FPDF_VERSION)
print(f"{pdf1.x=},{pdf1.y=}")
pdf1.write_html('<table border=1><tr><th width="100%">Test 1</th></tr></table>')
pdf1.write_html('<table border=1><tr><th width="100%">Test 2</th></tr></table>')
pdf1.output('issue_1616.pdf')

In fpdf 1.7.2, handle_starttag() & handle_endtag() used to insert vertical space after rendering <table> HTML elements.
This code snippet replicates this behaviour.

Now, 100% compatibility will be hard to achieve.
Even the simple line pdf1.write_html(FPDF_VERSION) behave slightly differently depending on the library used:

• in fpdf 1.7.2, after this call pdf1.x & pdf1.y are positioned on the right of the text just rendered
• in fpdf2, after this call pdf1.x & pdf1.y are positioned at the beginning the next line

There is not simple fix for that change of behaviour, that is deeply related to the new TextRegion class introduced in v2.7.6: https://py-pdf.github.io/fpdf2/TextRegion.html

[Lucas-C]

I'm going to add the following on https://py-pdf.github.io/fpdf2/History.html#compatibility-between-pyfpdf-fpdf2 :

Additionally, HTML rendering is not guaranteed to be identical regarding whitespace,
especially since version 2.7.6 and the new set of classes introduced to manage text flow: https://py-pdf.github.io/fpdf2/TextRegion.html

[Lucas-C]

As a workaround in your case, I would suggest to:

• manage most whitespace changes using tag_styles: https://py-pdf.github.io/fpdf2/HTML.html#styling-html-tags-globally
• use a HTML2FPDF_Compat class, as shown in my snippet, for the other situations, where you increase or decrease pdf.x & pdf.y
• introduce extra lines in your original code to more precisely control pdf.x & pdf.y, in last resort

[juleslagarde]

thanks !
I will try to write a non exhaustive list of all the change I see

[juleslagarde]

Here is a list of changes i see in the write_html function

• table spacing nulled
• force newline at the end (previously consecutive call to write_html could be one the same line)
• spaces inside HTML ignored (previously spaces could be used to indent)
• Table cells must align in width across all rows.
• a table following a bold text will have it's own text in bold (bug?)

here is a new sample code:

from fpdf import FPDF, HTMLMixin, FPDF_VERSION

class FPDF_Compat(FPDF, HTMLMixin):
	pass

pdf1 = FPDF_Compat()
pdf1.add_page()
pdf1.write_html(FPDF_VERSION)
pdf1.write_html('<b>Test5</b><table border=1><tr><th width="100%">Test6</th></tr><tr><td>Test7</td></tr></table>')
pdf1.output(f'_test_fpdf2_{FPDF_VERSION}.pdf')

output: