def make_page(page, pdf_page, psem): # Prepare everything in parallel @asyncio.coroutine def get_pdf_thumbnail(psem): if page.thumbnail is None: return None return (yield from page.thumbnail.pdf_thumbnail(psem)) @asyncio.coroutine def get_pdf_background(psem): if page.background is None: return None return (yield from page.background.pdf_image(psem)) @asyncio.coroutine def get_pdf_mask(foreground, psem): if foreground.color is not None: return None return (yield from foreground.pdf_mask(psem)) pdf_thumbnail, pdf_background, pdf_foregrounds, pdf_masks = ( yield from asyncio.gather( get_pdf_thumbnail(psem), get_pdf_background(psem), asyncio.gather( *[fg.pdf_image(psem) for fg in page.foreground]), asyncio.gather( *[get_pdf_mask(fg, psem) for fg in page.foreground]))) pdf_page.MediaBox = PdfArray( [0, 0, PdfNumber(page.width), PdfNumber(page.height)]) pdf_page.Group = pdf_group pdf_resources = PdfDict() pdf_colorspace = PdfDict() pdf_colorspace.DefaultRGB = default_rgb_colorspace pdf_resources.ColorSpace = pdf_colorspace pdf_xobject = PdfDict() if pdf_thumbnail is not None: pdf_page.Thumb = pdf_thumbnail im_index = 0 # Save graphics state and scale unity rectangle to page size matrix = TransformationMatrix() matrix.scale(page.width, page.height) before_graphics = ("q\n" + "%s cm\n" % matrix.to_pdf()) after_graphics = "\nQ\n" contents = "" graphics = "" current_color = None if page.color != self._factory.WHITE: if current_color != page.color: current_color = page.color graphics += page.color.to_pdf() + " rg " graphics += ("0 0 1 1 re " + "f\n") if pdf_background is not None: pdf_xobject[PdfName("Im%d" % im_index)] = pdf_background graphics += "/Im%d Do\n" % im_index im_index += 1 for foreground, pdf_foreground, pdf_mask in zip( page.foreground, pdf_foregrounds, pdf_masks): if pdf_mask is not None: pdf_xobject[PdfName("Im%d" % im_index)] = pdf_mask im_index += 1 pdf_xobject[PdfName("Im%d" % im_index)] = pdf_foreground if (foreground.color is not None and current_color != foreground.color): current_color = foreground.color graphics += foreground.color.to_pdf() + " rg " graphics += "/Im%d Do\n" % im_index im_index += 1 if graphics: contents += (before_graphics + graphics.rstrip(" \n") + after_graphics) current_color = None before_text = ("BT\n" + "/F1 1 Tf 3 Tr\n") after_text = "\nET\n" text = "" pdf_annots = [] for t in page.text: if t.text: matrix = TransformationMatrix() # Glyph size is 0.5 x 1 matrix.scale(2 / len(t.text), 1) matrix.translate(-0.5, -0.5) if t.direction == "ltr": pass elif t.direction == "rtl": matrix.translate(0, -1) elif t.direction == "ttb": matrix.rotate(90) matrix.rotate(-t.rotation) matrix.translate(0.5, 0.5) matrix.scale(t.width, t.height) matrix.translate(t.x, t.y) text += "%s Tm %s Tj\n" % ( matrix.to_pdf(), PdfString().from_bytes( t.text.encode("utf-16-be"), bytes_encoding="hex")) if t.external_link is not None or t.internal_link is not None: pdf_annot = PdfDict() pdf_annots.append(pdf_annot) pdf_annot.Type = PdfName.Annot pdf_annot.Subtype = PdfName.Link pdf_annot.Border = [0, 0, 0] pdf_annot.Rect = [ PdfNumber(t.x), PdfNumber(t.y), PdfNumber(t.x + t.width), PdfNumber(t.y + t.height) ] if t.external_link is not None: pdf_a = PdfDict() pdf_annot.A = pdf_a pdf_a.Type = PdfName.Action pdf_a.S = PdfName.URI pdf_a.URI = t.external_link.decode("latin-1") if t.internal_link is not None: pdf_target_page = pdf_pages[t.internal_link[0]] target_x, target_y = t.internal_link[1] pdf_annot.Dest = [ pdf_target_page, PdfName.XYZ, PdfNumber(target_x), PdfNumber(target_y), 0 ] text = text.rstrip(" \n") if text: pdf_resources.Font = pdf_font_mapping contents += (before_text + text + after_text) contents = contents.rstrip(" \n") if contents: pdf_contents = PdfDict() pdf_contents.indirect = True pdf_page.Contents = pdf_contents if COMPRESS_PAGE_CONTENTS: pdf_contents.Filter = [PdfName.FlateDecode] pdf_contents.stream = zlib.compress( contents.encode("latin-1"), 9).decode("latin-1") else: pdf_contents.stream = contents if pdf_annots: pdf_page.Annots = pdf_annots if pdf_xobject: pdf_resources.XObject = pdf_xobject if pdf_resources: pdf_page.Resources = pdf_resources # Report progress nonlocal finished_pages finished_pages += 1 if progress_cb: progress_cb(finished_pages / len(self._pages))
def _pdf_image(self, psem): with TemporaryDirectory(prefix="djpdf-") as temp_dir: # JBIG2Globals are only used in symbol mode # In symbol mode jbig2 writes output to files otherwise # it's written to stdout symbol_mode = self.jbig2_threshold != 1 images_with_shared_globals = [] if symbol_mode and SHARE_JBIG2_GLOBALS: # Find all Jbig2Images that share the same symbol directory for obj in self._factory._cache: if (isinstance(obj, Jbig2Image) and self.compression == obj.compression and self.jbig2_threshold == obj.jbig2_threshold): images_with_shared_globals.append(obj) else: # The symbol directory is not shared with other Jbig2Images images_with_shared_globals.append(self) # Promise all handled Jbig2Images the finished image image_futures = [] my_image_future = None for image in images_with_shared_globals: future = asyncio.Future() asyncio.ensure_future(image._cache.get(future)) image_futures.append(future) if image is self: my_image_future = future # All futures are in place, the lock can be released self._factory._cache_lock.release() self._cache_lock_acquired = False # Prepare everything in parallel @asyncio.coroutine def get_jbig2_images(psem): # Convert images with ImageMagick to bitonal png in parallel yield from asyncio.gather(*[ run_command_async([ CONVERT_CMD, "-alpha", "remove", "-alpha", "off", "-colorspace", "gray", "-threshold", "50%", path.abspath(image.filename), path.abspath(path.join(temp_dir, "input.%d.png" % i))], psem) for i, image in enumerate(images_with_shared_globals)]) cmd = [JBIG2_CMD, "-p"] if symbol_mode: cmd.extend(["-s", "-t", format_number(self.jbig2_threshold, 4)]) for i, _ in enumerate(images_with_shared_globals): cmd.append(path.abspath(path.join(temp_dir, "input.%d.png" % i))) jbig2_images = [] jbig2_globals = None if symbol_mode: yield from run_command_async(cmd, psem, cwd=temp_dir) jbig2_globals = PdfDict() jbig2_globals.indirect = True with open(path.join(temp_dir, "output.sym"), "rb") as f: jbig2_globals.stream = f.read().decode("latin-1") for i, _ in enumerate(images_with_shared_globals): with open(path.join(temp_dir, "output.%04d" % i), "rb") as f: jbig2_images.append(f.read()) else: jbig2_images.append( (yield from run_command_async(cmd, psem, cwd=temp_dir))) return jbig2_images, jbig2_globals @asyncio.coroutine def get_image_mask(image, psem): if image._mask is None: return None return (yield from image._mask.pdf_image(psem)) ((jbig2_images, jbig2_globals), image_masks) = yield from asyncio.gather( get_jbig2_images(psem), asyncio.gather(*[get_image_mask(image, psem) for image in images_with_shared_globals])) for image, jbig2_image, image_mask, image_future in zip( images_with_shared_globals, jbig2_images, image_masks, image_futures): (width, height, xres, yres) = struct.unpack( '>IIII', jbig2_image[11:27]) pdf_image = PdfDict() pdf_image.indirect = True pdf_image.Type = PdfName.XObject pdf_image.Subtype = PdfName.Image pdf_image.Width = width pdf_image.Height = height if image._image_mask: pdf_image.ImageMask = PdfBool(True) else: pdf_image.ColorSpace = PdfName.DeviceGray if image_mask is not None: pdf_image.Mask = image_mask pdf_image.BitsPerComponent = 1 pdf_image.Filter = [PdfName.JBIG2Decode] if symbol_mode: pdf_image.DecodeParms = [{ PdfName.JBIG2Globals: jbig2_globals}] pdf_image.stream = jbig2_image.decode("latin-1") image_future.set_result(pdf_image) return my_image_future.result()
def _pdf_image(self, psem): with TemporaryDirectory(prefix="djpdf-") as temp_dir: # JBIG2Globals are only used in symbol mode # In symbol mode jbig2 writes output to files otherwise # it's written to stdout symbol_mode = self.jbig2_threshold != 1 images_with_shared_globals = [] if symbol_mode and SHARE_JBIG2_GLOBALS: # Find all Jbig2Images that share the same symbol directory for obj in self._factory._cache: if (isinstance(obj, Jbig2Image) and self.compression == obj.compression and self.jbig2_threshold == obj.jbig2_threshold): images_with_shared_globals.append(obj) else: # The symbol directory is not shared with other Jbig2Images images_with_shared_globals.append(self) # Promise all handled Jbig2Images the finished image image_futures = [] my_image_future = None for image in images_with_shared_globals: future = asyncio.Future() asyncio.ensure_future(image._cache.get(future)) image_futures.append(future) if image is self: my_image_future = future # All futures are in place, the lock can be released self._factory._cache_lock.release() self._cache_lock_acquired = False # Prepare everything in parallel @asyncio.coroutine def get_jbig2_images(psem): # Convert images with ImageMagick to bitonal png in parallel yield from asyncio.gather(*[ run_command_async([ CONVERT_CMD, "-alpha", "remove", "-alpha", "off", "-colorspace", "gray", "-threshold", "50%", path.abspath(image.filename), path.abspath(path.join(temp_dir, "input.%d.png" % i)) ], psem) for i, image in enumerate(images_with_shared_globals) ]) cmd = [JBIG2_CMD, "-p"] if symbol_mode: cmd.extend( ["-s", "-t", format_number(self.jbig2_threshold, 4)]) for i, _ in enumerate(images_with_shared_globals): cmd.append( path.abspath(path.join(temp_dir, "input.%d.png" % i))) jbig2_images = [] jbig2_globals = None if symbol_mode: yield from run_command_async(cmd, psem, cwd=temp_dir) jbig2_globals = PdfDict() jbig2_globals.indirect = True with open(path.join(temp_dir, "output.sym"), "rb") as f: jbig2_globals.stream = f.read().decode("latin-1") for i, _ in enumerate(images_with_shared_globals): with open(path.join(temp_dir, "output.%04d" % i), "rb") as f: jbig2_images.append(f.read()) else: jbig2_images.append((yield from run_command_async(cmd, psem, cwd=temp_dir))) return jbig2_images, jbig2_globals @asyncio.coroutine def get_image_mask(image, psem): if image._mask is None: return None return (yield from image._mask.pdf_image(psem)) ((jbig2_images, jbig2_globals), image_masks) = yield from asyncio.gather( get_jbig2_images(psem), asyncio.gather(*[ get_image_mask(image, psem) for image in images_with_shared_globals ])) for image, jbig2_image, image_mask, image_future in zip( images_with_shared_globals, jbig2_images, image_masks, image_futures): (width, height, xres, yres) = struct.unpack('>IIII', jbig2_image[11:27]) pdf_image = PdfDict() pdf_image.indirect = True pdf_image.Type = PdfName.XObject pdf_image.Subtype = PdfName.Image pdf_image.Width = width pdf_image.Height = height if image._image_mask: pdf_image.ImageMask = PdfBool(True) else: # NOTE: DefaultGray color space is required for PDF/A pdf_image.ColorSpace = PdfName.DeviceGray if image_mask is not None: pdf_image.Mask = image_mask pdf_image.BitsPerComponent = 1 pdf_image.Filter = [PdfName.JBIG2Decode] if symbol_mode: pdf_image.DecodeParms = [{ PdfName.JBIG2Globals: jbig2_globals }] pdf_image.stream = jbig2_image.decode("latin-1") image_future.set_result(pdf_image) return my_image_future.result()