how to fix the garbled code

[ytcpub]

garbled_26282171_1.pdf

in above pdf, I get some garbled '5\x17\x04\x07\x08;\x02?\x12\x12\x02@9A\x13\x0231\x0231\x02\x111\x02!\x12\x02+\x06\x05\x02;\x02?\x12\x12\x02@9A\x13\x02!\x15\x02\x15!\x02!1\x02\x11!\x02\n\x04\x04\x03', and some normal eg 'on September 16, 2015. © 2015 Am..'
how to get the right content? it seems the encoding problem , but why some is right, and some is \x...

doc = fitz.open('garbled_26282171_1.pdf')
page = doc[0]
txt_blocks = page.get_text("dict", flags=fitz.TEXTFLAGS_TEXT, sort=True)['blocks']

[JorjMcKie]

This is a problem caused by a font. Text extraction can only work if a complete and correct back-translation table from the "glyph" (program that writes the visible character) to the originating Unicode exists.
This needs not to be the case at all!
It therefore is not an encoding problem!
Just missing or incomplete backtranslation information.
Your only way out is OCR-ing that page or the respective text bboxes. To know which they are, make sure to switch off the text extraction flag pymupdf.TEXT_CID_FOR_UNKNOWN_UNICODE (part of all default extraction flags). Your extracted text will then contain � characters for those misses - instead of results of MuPDF's replacement efforts. Pass the respective bbox to OCR and continue with the result. An example script for this approach can be found here.

[ytcpub]

got, thanks very much