Fix: Support rowspans breaking over page jumps (#1460)

[Neyhlo]

Fixes #1460 ---

What changes are being made?

This PR fixes a bug where tables containing large vertically spanned cells (rowspan) would crash when exceeding the page height, and allows them to break across pages cleanly.

Key changes:

1. The restrictive ValueError check in Table.render() that prevented long rows from rendering is removed.
2. Introduced a helper function (_get_span_origin) and logic in _render_table_row to draw the vertical borders for spanned cells on continuation pages (the "ghost borders").

---

Checklist:

• A unit test is covering the code added / modified by this PR
  (Note: We verified this by running existing rowspan tests which fail without the fix.)

• In case of a new feature, docstrings have been added, with also some documentation in the docs/ folder (N/A, this is a bug fix/enhancement to existing feature)

• This PR is ready to be merged

---

Additional comments:

The change in functionality requires updating several golden PDF files in the test suite. Specifically, tests in test_table_rowspan.py and test_table_padding.py fail because the visual output now includes the vertical borders on continuation pages, which changes the MD5 hash.

The reference files (golden files) need to be updated to match the new, correct behavior.

---

By submitting this pull request, I confirm that my contribution is made under the terms of the GNU LGPL 3.0 license.

[andersonhc]

Can you please add some tests for your feature?
/test/table/test_table.py should be the best place for it - and you can use the current tests there as inspiration

[Lucas-C]

Hi @Neyhlo

Without any answer from you, we will soon close this PR.

[Neyhlo]

Hey @Lucas-C

I'm so sorry about my no call no see I was very taken by work.

[ByZay]

@Lucas-C or @andersonhc, would u be able to respond and continue this PR ?

[andersonhc]

@Neyhlo @ByZay I see what you’re aiming for, but I can’t verify the behavior change yet.

• The test doesn’t evidence the change: it never calls pdf.table() or uses rowspans, so it isn’t exercising the new table logic.
• As written, the change doesn’t seem to implement what the PR claims; from the current table code, a rowspan that crosses a page boundary still triggers the same error.

Could you add a focused test that uses pdf.table() with a rowspan that must cross a page brea? That would make it clear whether the new behavior actually works.

[andersonhc]

Suggested change

	from fpdf import FPDF
	class TablePDF(FPDF):
	def active_spanning_table(self, data_list, rows_per_page=25):
	self.set_font("helvetica", size=11)
	col_widths = [60, 60, 60]
	self.set_fill_color(200, 200, 200)
	headers = ["Header 1", "Header 2", "Header 3"]
	def draw_header():
	for i, header in enumerate(headers):
	self.cell(col_widths[i], 10, header, border=1, fill=True)
	self.ln()
	draw_header()
	for group in data_list:
	items = group['items']
	label = group['label']
	for chunk_idx in range(0, len(items), rows_per_page):
	chunk = items[chunk_idx:chunk_idx + rows_per_page]
	if self.get_y() > 250:
	self.add_page()
	draw_header()
	display_label = f"{label}\n" if chunk_idx > 0 else label
	y_start = self.get_y()
	self.multi_cell(col_widths[0], 10, display_label, border=1, align='C')
	y_after_label = self.get_y()
	self.set_xy(self.l_margin + col_widths[0], y_start)
	self.cell(col_widths[1], 10, "", border=1)
	self.cell(col_widths[2], 10, chunk[0], border=1)
	self.ln()
	for item in chunk[1:]:
	if self.get_y() > 270:
	self.add_page()
	draw_header()
	self.multi_cell(col_widths[0], 10, f"{label}\n(suite)", border=1, align='C')
	self.set_xy(self.l_margin + col_widths[0], self.get_y() - 10)
	self.set_x(self.l_margin + col_widths[0])
	self.cell(col_widths[1], 10, "", border=1)
	self.cell(col_widths[2], 10, item, border=1)
	self.ln()
	pdf = TablePDF()
	pdf.add_page()
	data = [
	{"label": "Data 1", "items": [f"Ligne {i}" for i in range(100)]}
	]
	pdf.active_spanning_table(data, rows_per_page=25)
	pdf.output("spantest2.pdf")
	from fpdf import FPDF
	from pathlib import Path
	from test.conftest import assert_pdf_equal
	HERE = Path(__file__).resolve().parent
	def test_table_colspan_break(tmp_path):
	def active_spanning_table(pdf, data_list, rows_per_page=25):
	pdf.set_font("helvetica", size=11)
	col_widths = [60, 60, 60]
	pdf.set_fill_color(200, 200, 200)
	headers = ["Header 1", "Header 2", "Header 3"]
	def draw_header():
	for i, header in enumerate(headers):
	pdf.cell(col_widths[i], 10, header, border=1, fill=True)
	pdf.ln()
	draw_header()
	for group in data_list:
	items = group['items']
	label = group['label']
	for chunk_idx in range(0, len(items), rows_per_page):
	chunk = items[chunk_idx:chunk_idx + rows_per_page]
	if pdf.get_y() > 250:
	pdf.add_page()
	draw_header()
	display_label = f"{label}\n" if chunk_idx > 0 else label
	y_start = pdf.get_y()
	pdf.multi_cell(col_widths[0], 10, display_label, border=1, align='C')
	y_after_label = pdf.get_y()
	pdf.set_xy(pdf.l_margin + col_widths[0], y_start)
	pdf.cell(col_widths[1], 10, "", border=1)
	pdf.cell(col_widths[2], 10, chunk[0], border=1)
	pdf.ln()
	for item in chunk[1:]:
	if pdf.get_y() > 270:
	pdf.add_page()
	draw_header()
	pdf.multi_cell(col_widths[0], 10, f"{label}\n(suite)", border=1, align='C')
	pdf.set_xy(pdf.l_margin + col_widths[0], pdf.get_y() - 10)
	pdf.set_x(pdf.l_margin + col_widths[0])
	pdf.cell(col_widths[1], 10, "", border=1)
	pdf.cell(col_widths[2], 10, item, border=1)
	pdf.ln()
	pdf = FPDF()
	pdf.add_page()
	data = [
	{"label": "Data 1", "items": [f"Ligne {i}" for i in range(100)]}
	]
	active_spanning_table(pdf, data, rows_per_page=25)
	assert_pdf_equal(pdf, HERE / "table_colspan_break.pdf", tmp_path, generate=True)