Function get_textpage_ocr

src/utils.py:317–398

Create a Textpage from combined results of normal and OCR text parsing. Args: flags: (int) control content becoming part of the result. language: (str) specify expected language(s). Default is "eng" (English). dpi: (int) resolution in dpi, default 72. full: (bool

(
    page: pymupdf.Page,
    flags: int = 0,
    language: str = "eng",
    dpi: int = 72,
    full: bool = False,
    tessdata: str = None,
)

def get_textpage_ocr(
    page: pymupdf.Page,
    flags: int = 0,
    language: str = "eng",
    dpi: int = 72,
    full: bool = False,
    tessdata: str = None,
) -> pymupdf.TextPage:
    """Create a Textpage from combined results of normal and OCR text parsing.

    Args:
        flags: (int) control content becoming part of the result.
        language: (str) specify expected language(s). Default is "eng" (English).
        dpi: (int) resolution in dpi, default 72.
        full: (bool) whether to OCR the full page image, or only its images (default)
    """
    pymupdf.CheckParent(page)
    tessdata = pymupdf.get_tessdata(tessdata)

    def full_ocr(page, dpi, language, flags):
        zoom = dpi / 72
        mat = pymupdf.Matrix(zoom, zoom)
        pix = page.get_pixmap(matrix=mat)
        ocr_pdf = pymupdf.Document(
                "pdf",
                pix.pdfocr_tobytes(
                    compress=False,
                    language=language,
                    tessdata=tessdata,
                    ),
                )
        ocr_page = ocr_pdf.load_page(0)
        unzoom = page.rect.width / ocr_page.rect.width
        ctm = pymupdf.Matrix(unzoom, unzoom) * page.derotation_matrix
        tpage = ocr_page.get_textpage(flags=flags, matrix=ctm)
        ocr_pdf.close()
        pix = None
        tpage.parent = weakref.proxy(page)
        return tpage

    # if OCR for the full page, OCR its pixmap @ desired dpi
    if full:
        return full_ocr(page, dpi, language, flags)

    # For partial OCR, make a normal textpage, then extend it with text that
    # is OCRed from each image.
    # Because of this, we need the images flag bit set ON.
    tpage = page.get_textpage(flags=flags)
    for block in page.get_text("dict", flags=pymupdf.TEXT_PRESERVE_IMAGES)["blocks"]:
        if block["type"] != 1:  # only look at images
            continue
        bbox = pymupdf.Rect(block["bbox"])
        if bbox.width <= 3 or bbox.height <= 3:  # ignore tiny stuff
            continue
        try:
            pix = pymupdf.Pixmap(block["image"])  # get image pixmap
            if pix.n - pix.alpha != 3:  # we need to convert this to RGB!
                pix = pymupdf.Pixmap(pymupdf.csRGB, pix)