Single-row/column tables are skipped

[c01o]

related: #252

Problem

Single-row or single-column tables should be interpreted as single-line text, but they are simply being ignored.

Codes to Reproduce

# cell 1
import reportlab
from reportlab.lib import colors
from reportlab.lib.pagesizes import letter
from reportlab.platypus import SimpleDocTemplate, Table, TableStyle

def create_pdf(output_path, data = None):
    if data is None:
        data = []

    pdf = SimpleDocTemplate(output_path, pagesize=letter)

    table = Table(data)
    table.setStyle(TableStyle([('GRID', (0, 0), (-1, -1), 1, colors.black)]))

    pdf.build([table])

# cell 2
import pymupdf
import pymupdf4llm

def parse_with_pymupdf(pdf_path):
    _pdf = pymupdf.open(pdf_path)
    print(f"--pymupdf--\n{_pdf[0].get_text()}")


def parse_with_pymupdf4llm(pdf_path):
    _content = pymupdf4llm.to_markdown(pdf_path)
    print(f"--pymupdf4llm {pymupdf4llm.__version__}--\n{_content}")

# cell 3
testcases = [
    ['1x1.pdf', [["0,0/1x1"]]],
    ['1x2.pdf', [["0,0/1x2", "0,1/1x2"]]],
    ['2x1.pdf', [["0,0/2x1"], ["1,0/2x1"]]],
    ['2x2.pdf', [["0,0/2x2", "0,1/2x2"], ["1,0/2x2", "1,1/2x2"]]]
]

from pathlib import Path

testdir = Path('./test')
testdir.mkdir(exist_ok=True)

for name, data in testcases:
    pdf_path = testdir / name
    
    print(name + ':')
    create_pdf(str(pdf_path), data)
    parse_with_pymupdf(pdf_path)
    parse_with_pymupdf4llm(pdf_path)
    print('-----------------------------')

Result

In PDFs with tables smaller than 2x2, the table content is being ignored.

1x1.pdf:
--pymupdf--
0,0/1x1

--pymupdf4llm 0.0.21--
-----


-----------------------------
1x2.pdf:
--pymupdf--
0,0/1x2
0,1/1x2

--pymupdf4llm 0.0.21--
-----


-----------------------------
2x1.pdf:
--pymupdf--
0,0/2x1
1,0/2x1

--pymupdf4llm 0.0.21--
-----


-----------------------------
2x2.pdf:
--pymupdf--
0,0/2x2
0,1/2x2
1,0/2x2
1,1/2x2

--pymupdf4llm 0.0.21--
|0,0/2x2|0,1/2x2|
|---|---|
|1,0/2x2|1,1/2x2|


-----


-----------------------------

Version

0.0.21

[JorjMcKie]

Please provide a reproducing file!

[c01o]

The script above has create_pdf() function which generates reproducing files. Is it not enough?

[JorjMcKie]

Ah, sorry, didn't understand that.

[JorjMcKie]

The problem here is insufficient logic to classify vector graphics as "insignificant" (= just appearance sugar).
I hope this fix will be more precisely cover this.

[JorjMcKie]

Fixed with v0.0.22.