def _graft_text_layer(*, pdf_base, page_num, text, font, font_key, procset,
rotation, strip_old_text, log):
"""Insert the text layer from text page 0 on to pdf_base at page_num"""
log.debug("Grafting")
if Path(text).stat().st_size == 0:
return
# This is a pointer indicating a specific page in the base file
pdf_text = pikepdf.open(text)
pdf_text_contents = pdf_text.pages[0].Contents.read_bytes()
base_page = pdf_base.pages.p(page_num)
# The text page always will be oriented up by this stage but the original
# content may have a rotation applied. Wrap the text stream with a rotation
# so it will be oriented the same way as the rest of the page content.
# (Previous versions OCRmyPDF rotated the content layer to match the text.)
mediabox = [float(pdf_text.pages[0].MediaBox[v]) for v in range(4)]
wt, ht = mediabox[2] - mediabox[0], mediabox[3] - mediabox[1]
mediabox = [float(base_page.MediaBox[v]) for v in range(4)]
wp, hp = mediabox[2] - mediabox[0], mediabox[3] - mediabox[1]
translate = pikepdf.PdfMatrix().translated(-wt / 2, -ht / 2)
untranslate = pikepdf.PdfMatrix().translated(wp / 2, hp / 2)
corner = pikepdf.PdfMatrix().translated(mediabox[0], mediabox[1])
# -rotation because the input is a clockwise angle and this formula
# uses CCW
rotation = -rotation % 360
rotate = pikepdf.PdfMatrix().rotated(rotation)
# Because of rounding of DPI, we might get a text layer that is not
# identically sized to the target page. Scale to adjust. Normally this
# is within 0.998.
if rotation in (90, 270):
wt, ht = ht, wt
scale_x = wp / wt
scale_y = hp / ht
# log.debug('%r', scale_x, scale_y)
scale = pikepdf.PdfMatrix().scaled(scale_x, scale_y)
# Translate the text so it is centered at (0, 0), rotate it there, adjust
# for a size different between initial and text PDF, then untranslate, and
# finally move the lower left corner to match the mediabox
ctm = translate @ rotate @ scale @ untranslate @ corner
pdf_text_contents = b'q %s cm\n' % ctm.encode(
) + pdf_text_contents + b'\nQ\n'
new_text_layer = pikepdf.Stream(pdf_base, pdf_text_contents)
if strip_old_text:
strip_invisible_text(pdf_base, base_page)
base_page.page_contents_add(new_text_layer, prepend=True)
_update_page_resources(page=base_page,
font=font,
font_key=font_key,
procset=procset)
pdf_text.close()
def _graft_text_layer(
self,
*,
page_num: int,
textpdf: Path,
font: pikepdf.Object,
font_key: pikepdf.Object,
procset: pikepdf.Object,
text_rotation: int,
strip_old_text: bool,
):
"""Insert the text layer from text page 0 on to pdf_base at page_num"""
if Path(textpdf).stat().st_size == 0:
return
# This is a pointer indicating a specific page in the base file
with pikepdf.open(textpdf) as pdf_text:
pdf_text_contents = pdf_text.pages[0].Contents.read_bytes()
base_page = self.pdf_base.pages.p(page_num)
# The text page always will be oriented up by this stage but the original
# content may have a rotation applied. Wrap the text stream with a rotation
# so it will be oriented the same way as the rest of the page content.
# (Previous versions OCRmyPDF rotated the content layer to match the text.)
mediabox = [float(pdf_text.pages[0].MediaBox[v]) for v in range(4)]
wt, ht = mediabox[2] - mediabox[0], mediabox[3] - mediabox[1]
mediabox = [float(base_page.MediaBox[v]) for v in range(4)]
wp, hp = mediabox[2] - mediabox[0], mediabox[3] - mediabox[1]
translate = pikepdf.PdfMatrix().translated(-wt / 2, -ht / 2)
untranslate = pikepdf.PdfMatrix().translated(wp / 2, hp / 2)
corner = pikepdf.PdfMatrix().translated(mediabox[0], mediabox[1])
# -rotation because the input is a clockwise angle and this formula
# uses CCW
text_rotation = -text_rotation % 360
rotate = pikepdf.PdfMatrix().rotated(text_rotation)
# Because of rounding of DPI, we might get a text layer that is not
# identically sized to the target page. Scale to adjust. Normally this
# is within 0.998.
if text_rotation in (90, 270):
wt, ht = ht, wt
scale_x = wp / wt
scale_y = hp / ht
# log.debug('%r', scale_x, scale_y)
scale = pikepdf.PdfMatrix().scaled(scale_x, scale_y)
# Translate the text so it is centered at (0, 0), rotate it there, adjust
# for a size different between initial and text PDF, then untranslate, and
# finally move the lower left corner to match the mediabox
ctm = translate @ rotate @ scale @ untranslate @ corner
base_resources = _ensure_dictionary(base_page, Name.Resources)
base_xobjs = _ensure_dictionary(base_resources, Name.XObject)
text_xobj_name = Name("/" + str(uuid.uuid4()))
xobj = self.pdf_base.make_stream(pdf_text_contents)
base_xobjs[text_xobj_name] = xobj
xobj.Type = Name.XObject
xobj.Subtype = Name.Form
xobj.FormType = 1
xobj.BBox = mediabox
_update_resources(
obj=xobj, font=font, font_key=font_key, procset=[Name.PDF]
)
pdf_draw_xobj = (
(b"q %s cm\n" % ctm.encode()) + (b"%s Do\n" % text_xobj_name) + b"\nQ\n"
)
new_text_layer = pikepdf.Stream(self.pdf_base, pdf_draw_xobj)
if strip_old_text:
strip_invisible_text(self.pdf_base, base_page)
base_page.page_contents_add(new_text_layer, prepend=True)
_update_resources(
obj=base_page, font=font, font_key=font_key, procset=procset
)
def _weave_layers_graft(*, pdf_base, page_num, text, font, font_key, procset,
rotation, log):
"""Insert the text layer from text page 0 on to pdf_base at page_num"""
log.debug("Grafting")
if Path(text).stat().st_size == 0:
return
# This is a pointer indicating a specific page in the base file
pdf_text = pikepdf.open(text)
pdf_text_contents = pdf_text.pages[0].Contents.read_bytes()
if not tesseract.has_textonly_pdf():
# If we don't have textonly_pdf, edit the stream to delete the
# instruction to draw the image Tesseract generated, which we do not
# use.
stream = bytearray(pdf_text_contents)
pattern = b'/Im1 Do'
idx = stream.find(pattern)
stream[idx:(idx + len(pattern))] = b' ' * len(pattern)
pdf_text_contents = bytes(stream)
base_page = pdf_base.pages.p(page_num)
# The text page always will be oriented up by this stage but the original
# content may have a rotation applied. Wrap the text stream with a rotation
# so it will be oriented the same way as the rest of the page content.
# (Previous versions OCRmyPDF rotated the content layer to match the text.)
mediabox = [float(pdf_text.pages[0].MediaBox[v]) for v in range(4)]
wt, ht = mediabox[2] - mediabox[0], mediabox[3] - mediabox[1]
mediabox = [float(base_page.MediaBox[v]) for v in range(4)]
wp, hp = mediabox[2] - mediabox[0], mediabox[3] - mediabox[1]
translate = pikepdf.PdfMatrix().translated(-wt / 2, -ht / 2)
untranslate = pikepdf.PdfMatrix().translated(wp / 2, hp / 2)
# -rotation because the input is a clockwise angle and this formula
# uses CCW
rotation = -rotation % 360
rotate = pikepdf.PdfMatrix().rotated(rotation)
# Because of rounding of DPI, we might get a text layer that is not
# identically sized to the target page. Scale to adjust. Normally this
# is within 0.998.
if rotation in (90, 270):
wt, ht = ht, wt
scale_x = wp / wt
scale_y = hp / ht
log.debug('%r', (scale_x, scale_y))
scale = pikepdf.PdfMatrix().scaled(scale_x, scale_y)
# Translate the text so it is centered at (0, 0), rotate it there, adjust
# for a size different between initial and text PDF, then untranslate
ctm = translate @ rotate @ scale @ untranslate
pdf_text_contents = (b'q %s cm\n' % ctm.encode() + pdf_text_contents +
b'\nQ\n')
new_text_layer = pikepdf.Stream(pdf_base, pdf_text_contents)
base_page.page_contents_add(new_text_layer, prepend=True)
_update_page_resources(page=base_page,
font=font,
font_key=font_key,
procset=procset)