def clip_page(cls, page:fitz.Page, bbox:fitz.Rect, zoom:float=3.0):
'''Clip pixmap according to bbox from page.'''
# improve resolution
# - https://pymupdf.readthedocs.io/en/latest/faq.html#how-to-increase-image-resolution
# - https://github.com/pymupdf/PyMuPDF/issues/181
image = page.getPixmap(clip=bbox, matrix=fitz.Matrix(zoom, zoom)) # type: fitz.Pixmap
return cls.to_raw_dict(image, bbox)
def clip_page(cls,
page: fitz.Page,
bbox: fitz.Rect = None,
zoom: float = 3.0):
'''Clip page pixmap (without text) according to `bbox` (entire page by default).
'''
# hide text before clip the image only
# render Tr: set the text rendering mode
# - 3: neither fill nor stroke the text -> invisible
# read more:
# - https://github.com/pymupdf/PyMuPDF/issues/257
# - https://www.adobe.com/content/dam/acom/en/devnet/pdf/pdfs/pdf_reference_archives/PDFReference.pdf
doc = page.parent
for xref in page._getContents():
stream = doc._getXrefStream(xref).replace(b'BT', b'BT 3 Tr') \
.replace(b'Tm', b'Tm 3 Tr') \
.replace(b'Td', b'Td 3 Tr')
doc._updateStream(xref, stream)
# improve resolution
# - https://pymupdf.readthedocs.io/en/latest/faq.html#how-to-increase-image-resolution
# - https://github.com/pymupdf/PyMuPDF/issues/181
bbox = page.rect if bbox is None else bbox & page.rect
image = page.getPixmap(clip=bbox,
matrix=fitz.Matrix(zoom,
zoom)) # type: fitz.Pixmap
return cls.to_raw_dict(image, bbox)
def clip_page(cls,
page: fitz.Page,
bbox: fitz.Rect = None,
zoom: float = 3.0):
"""Clip page pixmap (without text) according to ``bbox``.
Args:
page (fitz.Page): pdf page to extract.
bbox (fitz.Rect, optional): Target area to clip. Defaults to None, i.e. entire page.
zoom (float, optional): Improve resolution by this rate. Defaults to 3.0.
Returns:
dict: Raw dict of the extracted pixmap.
"""
# hide text before clip the image only
# render Tr: set the text rendering mode
# - 3: neither fill nor stroke the text -> invisible
# read more:
# - https://github.com/pymupdf/PyMuPDF/issues/257
# - https://www.adobe.com/content/dam/acom/en/devnet/pdf/pdfs/pdf_reference_archives/PDFReference.pdf
doc = page.parent
for xref in page.get_contents():
stream = doc.xrefStream(xref).replace(b'BT', b'BT 3 Tr') \
.replace(b'Tm', b'Tm 3 Tr') \
.replace(b'Td', b'Td 3 Tr')
doc.updateStream(xref, stream)
# improve resolution
# - https://pymupdf.readthedocs.io/en/latest/faq.html#how-to-increase-image-resolution
# - https://github.com/pymupdf/PyMuPDF/issues/181
bbox = page.rect if bbox is None else bbox & page.rect
image = page.getPixmap(clip=bbox,
matrix=fitz.Matrix(zoom,
zoom)) # type: fitz.Pixmap
return cls.to_raw_dict(image, bbox)
def render_fitz_page(
page: fitz.Page, zoom: float, pixel_ratio: float, clip: fitz.Rect = None
) -> QtGui.QPixmap:
scale_ratio = zoom * pixel_ratio
pix = page.getPixmap(matrix=fitz.Matrix(scale_ratio, scale_ratio), clip=clip)
mode = "RGBA" if pix.alpha else "RGB"
img = Image.frombytes(mode, [pix.width, pix.height], pix.samples)
img = ImageQt.ImageQt(img)
img.setDevicePixelRatio(pixel_ratio)
return img
def to_image(self, page: fitz.Page):
'''Convert to image block dict if this is a vector graphic paths.'''
bbox = self.bbox
image = page.getPixmap(clip=bbox)
return {
'type': 1,
'bbox': tuple(bbox),
'ext': 'png',
'width': bbox.width,
'height': bbox.height,
'image': image.getImageData(output="png")
}
def extract_vector_graphics(cls,
page: fitz.Page,
exclude_areas: list,
clip_image_res_ratio: float = 3.0):
"""Detect and extract vector graphics by clipping associated page area.
Args:
page (fitz.Page): pdf page to extract images.
exclude_areas (list): A list of bbox-like ``(x0, y0, x1, y1)`` area to exclude,
e.g. raster image area, table area.
clip_image_res_ratio (float, optional): Resolution ratio of clipped bitmap. Defaults to 3.0.
Returns:
list: A list of extracted and recovered image raw dict.
.. note::
Contours for vector graphics are detected first with ``opencv-python``.
"""
# find contours
contours = cls._detect_svg_contours(page, exclude_areas)
# filter contours
fun = lambda a, b: a.bbox & b.bbox
groups = contours.group(fun)
# clip images
images = []
for group in groups:
bbox = group.bbox
pix = page.getPixmap(clip=bbox,
matrix=fitz.Matrix(clip_image_res_ratio,
clip_image_res_ratio))
raw_dict = cls._to_raw_dict(pix, bbox)
images.append(raw_dict)
return images
def _clip_page(cls,
page: fitz.Page,
bbox: fitz.Rect = None,
zoom: float = 3.0):
"""Clip page pixmap (without text) according to ``bbox``.
Args:
page (fitz.Page): pdf page to extract.
bbox (fitz.Rect, optional): Target area to clip. Defaults to None, i.e. entire page.
zoom (float, optional): Improve resolution by this rate. Defaults to 3.0.
Returns:
fitz.Pixmap: The extracted pixmap.
"""
# hide text
cls._hide_page_text(page)
# improve resolution
# - https://pymupdf.readthedocs.io/en/latest/faq.html#how-to-increase-image-resolution
# - https://github.com/pymupdf/PyMuPDF/issues/181
bbox = page.rect if bbox is None else bbox & page.rect
return page.getPixmap(clip=bbox,
matrix=fitz.Matrix(zoom,
zoom)) # type: fitz.Pixmap
