def iter_images_and_pages(images):
"""This function iterates over a images and also the contained pages. As
OpenCV is not able to handle multipage TIFF files, we use the SDAPS internal
loading method for those."""
for filename in images:
try:
# Check whether this is a TIFF file (ie. try to retrieve the page count)
pages = image.get_tiff_page_count(filename)
is_tiff = True
except AssertionError:
pages = 1
is_tiff = False
for page in xrange(pages):
if not is_tiff:
img = cv2.imread(filename)
else:
# TIFF pages are zero based
surf = image.get_rgb24_from_tiff(filename, page, False)
width = surf.get_width()
height = surf.get_height()
stride = surf.get_stride()
# We need to ensure a sane stride!
np_width = stride / 4
# This converts by doing a copy; first create target numpy array
# We need a dummy alpha channel ...
target = np.empty((height, np_width), dtype=np.uint32)
tmp_surf = cairo.ImageSurface.create_for_data(
target.data, cairo.FORMAT_RGB24, width, height, stride)
cr = cairo.Context(tmp_surf)
cr.set_source_surface(surf)
cr.paint()
del cr
tmp_surf.flush()
del tmp_surf
# Now, we need a bit of reshaping
img = np.empty((height, width, 3), dtype=np.uint8)
# order should be BGR
img[:, :, 2] = 0xff & (target[:, :] >> 16)
img[:, :, 1] = 0xff & (target[:, :] >> 8)
img[:, :, 0] = 0xff & target[:, :]
yield img, filename, page
def iter_images_and_pages(images):
"""This function iterates over a images and also the contained pages. As
OpenCV is not able to handle multipage TIFF files, we use the SDAPS internal
loading method for those."""
for filename in images:
try:
# Check whether this is a TIFF file (ie. try to retrieve the page count)
pages = image.get_tiff_page_count(filename)
is_tiff = True
except AssertionError:
pages = 1
is_tiff = False
for page in xrange(pages):
if not is_tiff:
img = cv2.imread(filename)
else:
# TIFF pages are zero based
surf = image.get_rgb24_from_tiff(filename, page, False)
width = surf.get_width()
height = surf.get_height()
stride = surf.get_stride()
# We need to ensure a sane stride!
np_width = stride / 4
# This converts by doing a copy; first create target numpy array
# We need a dummy alpha channel ...
target = np.empty((height, np_width), dtype=np.uint32)
tmp_surf = cairo.ImageSurface.create_for_data(target.data, cairo.FORMAT_RGB24, width, height, stride)
cr = cairo.Context(tmp_surf)
cr.set_source_surface(surf)
cr.paint()
del cr
tmp_surf.flush()
del tmp_surf
# Now, we need a bit of reshaping
img = np.empty((height, width, 3), dtype=np.uint8)
# order should be BGR
img[:,:,2] = 0xff & (target[:,:] >> 16)
img[:,:,1] = 0xff & (target[:,:] >> 8)
img[:,:,0] = 0xff & target[:,:]
yield img, filename, page
def survey_image(request, slug, filenum, page):
# This function does not open the real SDAPS survey, as unpickling the data
# is way to inefficient.
survey = get_survey_or_404(request, slug, review=True)
image_file = os.path.join(survey.path, "%s.tif" % (filenum,))
if not os.path.exists(os.path.join(survey.path)):
raise Http404
surface = image.get_rgb24_from_tiff(image_file, int(page), False)
if surface is None:
raise Http404
# Create PNG stream and return it
response = HttpResponse(content_type='image/png')
response['Cache-Control'] = 'private, max-age=3600'
surface.write_to_png(response)
return response
def survey_image(request, slug, filenum, page):
# This function does not open the real SDAPS survey, as unpickling the data
# is way to inefficient.
survey = get_survey_or_404(request, slug, review=True)
image_file = os.path.join(survey.path, "%s.tif" % (filenum, ))
if not os.path.exists(os.path.join(survey.path)):
raise Http404
surface = image.get_rgb24_from_tiff(image_file, int(page), False)
if surface is None:
raise Http404
# Create PNG stream and return it
response = HttpResponse(content_type='image/png')
response['Cache-Control'] = 'private, max-age=3600'
surface.write_to_png(response)
return response
def iter_images_and_pages(images):
"""This function iterates over a images and also the contained pages. As
OpenCV is not able to handle multipage TIFF files, we use the SDAPS internal
loading method for those."""
for filename in images:
pages = 1
is_tiff = False
is_pdf = False
try:
# Check whether this is a TIFF file (ie. try to retrieve the page count)
pages = image.get_tiff_page_count(filename)
is_tiff = True
except AssertionError:
pass
if not is_tiff:
try:
gfile = Gio.File.new_for_path(filename)
pdf_doc = Poppler.Document.new_from_gfile(gfile, None, None)
pages = pdf_doc.get_n_pages()
is_pdf = True
except:
# Either not PDF/damaged or poppler not installed properly
pass
for page in xrange(pages):
if is_tiff:
# TIFF pages are zero based
surf = image.get_rgb24_from_tiff(filename, page, False)
img = to_opencv(surf)
elif is_pdf:
# Try to retrieve a single fullpage image, if that fails, render
# document at 300dpi.
THRESH = 10 #pt
pdfpage = pdf_doc.get_page(page)
page_width, page_height = pdfpage.get_size()
images = pdfpage.get_image_mapping()
if len(images) == 1 and (
abs(images[0].area.x1) < THRESH and
abs(images[0].area.y1) < THRESH and
abs(images[0].area.x2 - page_width) < THRESH and
abs(images[0].area.y2 - page_height) < THRESH):
# Assume one full page image, and simply use that.
surf = pdfpage.get_image(images[0].image_id)
else:
# Render page at 300dpi
surf = cairo.ImageSurface(cairo.FORMAT_RGB24, int(300 / 72 * page_width), int(300 / 72 * page_height))
cr = cairo.Context(surf)
cr.scale(300 / 72, 300 / 72)
cr.set_source_rgb(1, 1, 1)
cr.paint()
pdfpage.render_for_printing(cr)
del cr
img = to_opencv(surf)
else:
img = cv2.imread(filename)
yield img, filename, page
def iter_images_and_pages(images):
"""This function iterates over a images and also the contained pages. As
OpenCV is not able to handle multipage TIFF files, we use the SDAPS internal
loading method for those."""
for filename in images:
pages = 1
is_tiff = False
is_pdf = False
try:
# Check whether this is a TIFF file (ie. try to retrieve the page count)
pages = image.get_tiff_page_count(filename)
is_tiff = True
except AssertionError:
pass
if not is_tiff:
try:
gfile = Gio.File.new_for_path(filename)
pdf_doc = Poppler.Document.new_from_gfile(gfile, None, None)
pages = pdf_doc.get_n_pages()
is_pdf = True
except:
# Either not PDF/damaged or poppler not installed properly
pass
for page in xrange(pages):
if is_tiff:
# TIFF pages are zero based
surf = image.get_rgb24_from_tiff(filename, page, False)
img = to_opencv(surf)
elif is_pdf:
# Try to retrieve a single fullpage image, if that fails, render
# document at 300dpi.
THRESH = 10 #pt
pdfpage = pdf_doc.get_page(page)
page_width, page_height = pdfpage.get_size()
images = pdfpage.get_image_mapping()
if len(images) == 1 and (
abs(images[0].area.x1) < THRESH and
abs(images[0].area.y1) < THRESH and
abs(images[0].area.x2 - page_width) < THRESH and
abs(images[0].area.y2 - page_height) < THRESH):
# Assume one full page image, and simply use that.
surf = pdfpage.get_image(images[0].image_id)
else:
# Render page at 300dpi
surf = cairo.ImageSurface(cairo.FORMAT_RGB24, int(300 / 72 * page_width), int(300 / 72 * page_height))
cr = cairo.Context(surf)
cr.scale(300 / 72, 300 / 72)
cr.set_source_rgb(1, 1, 1)
cr.paint()
pdfpage.render_for_printing(cr)
del cr
img = to_opencv(surf)
else:
img = cv2.imread(filename)
yield img, filename, page
def iter_images_and_pages(images):
"""This function iterates over a images and also the contained pages. As
OpenCV is not able to handle multipage TIFF files, we use the SDAPS internal
loading method for those."""
for filename in images:
if not os.path.exists(filename):
raise IOError(errno.ENOENT, _("File does not exist"), filename)
pages = 1
is_tiff = False
is_pdf = False
try:
# Check whether this is a TIFF file (ie. try to retrieve the page count)
pages = image.get_tiff_page_count(filename)
is_tiff = True
except AssertionError:
pass
if not is_tiff:
try:
gfile = Gio.File.new_for_path(filename)
pdf_doc = Poppler.Document.new_from_gfile(gfile, None, None)
pages = pdf_doc.get_n_pages()
is_pdf = True
except:
# Either not PDF/damaged or poppler not installed properly
pass
for page in range(pages):
if is_tiff:
# TIFF pages are zero based
surf = image.get_rgb24_from_tiff(filename, page, False)
img = to_opencv(surf)
elif is_pdf:
# Try to retrieve a single fullpage image, if that fails, render
# document at 300dpi.
THRESH = 10 #pt
pdfpage = pdf_doc.get_page(page)
page_width, page_height = pdfpage.get_size()
images = pdfpage.get_image_mapping()
if len(images) == 1 and (
abs(images[0].area.x1) < THRESH
and abs(images[0].area.y1) < THRESH
and abs(images[0].area.x2 - page_width) < THRESH
and abs(images[0].area.y2 - page_height) < THRESH):
# Assume one full page image, and simply use that.
surf = pdfpage.get_image(images[0].image_id)
else:
dpi = 0
# Try to detect the DPI of the scan
for img in images:
if img.area.y2 - img.area.y1 < page_height / 2:
continue
surf = pdfpage.get_image(img.image_id)
# Calculate DPI from height
dpi_x = round(surf.get_height() /
(img.area.y2 - img.area.y1) * 72)
dpi_y = round(surf.get_width() /
(img.area.x2 - img.area.x1) * 72)
if abs(dpi_x - dpi_y) <= 1:
dpi = max(dpi, dpi_x, dpi_y)
# Fall back to 300dpi for odd values
if dpi < 199 or dpi > 601:
dpi = 300
surf = cairo.ImageSurface(cairo.FORMAT_RGB24,
int(dpi / 72 * page_width),
int(dpi / 72 * page_height))
cr = cairo.Context(surf)
cr.scale(dpi / 72, dpi / 72)
cr.set_source_rgb(1, 1, 1)
cr.paint()
pdfpage.render_for_printing(cr)
del cr
img = to_opencv(surf)
else:
img = cv2.imread(filename)
yield img, filename, page
def iter_images_and_pages(images):
"""This function iterates over a images and also the contained pages. As
OpenCV is not able to handle multipage TIFF files, we use the SDAPS internal
loading method for those."""
for filename in images:
if not os.path.exists(filename):
raise IOError(errno.ENOENT, _("File does not exist"), filename)
pages = 1
is_tiff = False
is_pdf = False
try:
# Check whether this is a TIFF file (ie. try to retrieve the page count)
pages = image.get_tiff_page_count(filename)
is_tiff = True
except AssertionError:
pass
if not is_tiff:
try:
gfile = Gio.File.new_for_path(filename)
pdf_doc = Poppler.Document.new_from_gfile(gfile, None, None)
pages = pdf_doc.get_n_pages()
is_pdf = True
except:
# Either not PDF/damaged or poppler not installed properly
pass
for page in range(pages):
if is_tiff:
# TIFF pages are zero based
surf = image.get_rgb24_from_tiff(filename, page, False)
img = to_opencv(surf)
elif is_pdf:
# Try to retrieve a single fullpage image, if that fails, render
# document at 300dpi.
THRESH = 10 #pt
pdfpage = pdf_doc.get_page(page)
page_width, page_height = pdfpage.get_size()
images = pdfpage.get_image_mapping()
if len(images) == 1 and (
abs(images[0].area.x1) < THRESH and
abs(images[0].area.y1) < THRESH and
abs(images[0].area.x2 - page_width) < THRESH and
abs(images[0].area.y2 - page_height) < THRESH):
# Assume one full page image, and simply use that.
surf = pdfpage.get_image(images[0].image_id)
else:
dpi = 0
# Try to detect the DPI of the scan
for img in images:
if img.area.y2 - img.area.y1 < page_height / 2:
continue
surf = pdfpage.get_image(img.image_id)
# Calculate DPI from height
dpi_x = round(surf.get_height() / (img.area.y2 - img.area.y1) * 72)
dpi_y = round(surf.get_width() / (img.area.x2 - img.area.x1) * 72)
if abs(dpi_x - dpi_y) <= 1:
dpi = max(dpi, dpi_x, dpi_y)
# Fall back to 300dpi for odd values
if dpi < 199 or dpi > 601:
dpi = 300
surf = cairo.ImageSurface(cairo.FORMAT_RGB24, int(dpi / 72 * page_width), int(dpi / 72 * page_height))
cr = cairo.Context(surf)
cr.scale(dpi / 72, dpi / 72)
cr.set_source_rgb(1, 1, 1)
cr.paint()
pdfpage.render_for_printing(cr)
del cr
img = to_opencv(surf)
else:
img = cv2.imread(filename)
yield img, filename, page
