Hard trimming PDFs?

[willus]

As you figured out, when headers and footers are truly embedded into a PDF procedural stream, most PDF utility apps will not physically remove them as this would require parsing and interpreting the PDF stream and figuring out which part of it displays graphics outside the bounding/clipping box and removing those instructions. This is very difficult to do cleanly and reliably. The real issue is that calibre does not correctly pay attention to the PDF clipping box when converting the PDF to EPUB, as it should.

[Shohreh]

Indeed, Calibre cropping everything that sits outside the mediabox would have solved the issue easily.

[Shohreh]

To remove header + footer on all the pages:

doc = fitz.open("input.pdf")
#To find mediabox of page 25 in input file: cpdf -page-info input.pdf 25

WIDTH = doc[0].mediabox.width
HEIGHT = doc[0].mediabox.height
rect_header = fitz.Rect(0,0,WIDTH,50) #left,top, right,bottom
rect_footer = fitz.Rect(0,770,WIDTH,790)

numpages = doc.page_count
for index in range(numpages):
	page = doc[index]
	page.add_redact_annot(rect_header)
	page.add_redact_annot(rect_footer)
	page.apply_redactions()
doc.save("redacted.pdf")

#ebook-convert.exe redacted.pdf redacted.epub

[Shohreh]

On Windows, through its Measurement tool enabled/disabled through the "m" key, the free application SumatraPDF lets you see the mouse coordinates as you move it across the PDF.

To find the left,top and right,bottom coordinates of the section you want to remove, simple move the mouse, and type the coordinates into the script.