How to get the separate quad(quad_like,rect_like) in PDF using search_for method when targeted letters are contiguous and adjacent to each other

[Carnegie-eng]

Hi, I would like to know how to get the separate quad(quad_like,rect_like) in PDF using search_for method when targeted letters are contiguous and adjacent to each other.

Default method of search_for(target) retrunes just 1 quad(quad_like,rect_like) since target letters are contiguous and adjacent to each other. I would like to get the separate quad(quad_like,rect_like).

[Carnegie-eng]

I wonder the quad area by search_for method is changed to be narrow in horizontal. If so, could you let me know how to set the narrow. Thank you.

[JorjMcKie]

Like the text extraction variants, the search method also supports the clip parameter: page.search_for(needle, clip=rect_like).
Using this in combination with page.get_text("dict", clip=...) or "rawdict" may help you bygoing down even to single characters.

In addition, there is the new parameter delimiters in page.get_text("words", clip=rect, delimiters="§$%..."), which allows to specify additional characters to be used as word delimiters: where the normal word extraction returns one single word [email protected], setting delimiters="@." will deliver 4 words "john", "doe", "outlook", "com" - together with their respective coordinates.

Maybe this helps.

[Carnegie-eng]

test_copy.pdf

Thank you for advise. I tried like below to the attached example pdf. Target character is "9".

rect = page.search_for("9", clip=rect_like)
(Result) rect is only 8 area although there is 12 of "9" in the page.

I appreciate if you could advise me how to resolve this issue.

[JorjMcKie]

1-char searches are always problematic of course, because characters in a row may have tiny overlaps (created by the PDF maker). What you always do in similar cases is this:

fitz.TOOLS.set_small_glyph_heights(True)
True
rl=page.search_for("9")
for r in rl:
    page.get_textbox(r)

    
'99'
'99'
'r\n9'
'9'
'9'
'9'
'99'
'99'

The second hickup: if the search algorithm finds multiple adjacent copies of the needle, then one common rectangle is returned as visible above.

[Carnegie-eng]

Thank you for explanation. I need just 1-char, so I will try another aproach. Thank you.