File: image_parsing.py

package info (click to toggle)
fpdf2 2.8.4-2
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid
  • size: 53,860 kB
  • sloc: python: 39,487; sh: 133; makefile: 12
file content (714 lines) | stat: -rw-r--r-- 25,798 bytes parent folder | download
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
import base64, hashlib, io, zlib
from dataclasses import dataclass
from io import BytesIO
from math import ceil
from urllib.request import urlopen
from pathlib import Path
import logging

try:
    from PIL import Image, TiffImagePlugin
    from PIL import ImageCms, features as PIL_features

    try:
        from PIL.Image import Resampling

        RESAMPLE = Resampling.LANCZOS
    except ImportError:  # For Pillow < 9.1.0
        # pylint: disable=no-member, useless-suppression
        RESAMPLE = Image.ANTIALIAS
except ImportError:
    Image = None
    PIL_features = None

try:
    import numpy
except ImportError:
    numpy = None

from .errors import FPDFException
from .image_datastructures import ImageCache, RasterImageInfo, VectorImageInfo
from .svg import SVGObject


@dataclass
class ImageSettings:
    # Passed to zlib.compress() - In range 0-9 - Default is currently equivalent to 6:
    compression_level: int = -1


LOGGER = logging.getLogger(__name__)
SUPPORTED_IMAGE_FILTERS = ("AUTO", "FlateDecode", "DCTDecode", "JPXDecode", "LZWDecode")
SETTINGS = ImageSettings()

# fmt: off
TIFFBitRevTable = [
    0x00, 0x80, 0x40, 0xC0, 0x20, 0xA0, 0x60, 0xE0, 0x10, 0x90,
    0x50, 0xD0, 0x30, 0xB0, 0x70, 0xF0, 0x08, 0x88, 0x48, 0xC8,
    0x28, 0xA8, 0x68, 0xE8, 0x18, 0x98, 0x58, 0xD8, 0x38, 0xB8,
    0x78, 0xF8, 0x04, 0x84, 0x44, 0xC4, 0x24, 0xA4, 0x64, 0xE4,
    0x14, 0x94, 0x54, 0xD4, 0x34, 0xB4, 0x74, 0xF4, 0x0C, 0x8C,
    0x4C, 0xCC, 0x2C, 0xAC, 0x6C, 0xEC, 0x1C, 0x9C, 0x5C, 0xDC,
    0x3C, 0xBC, 0x7C, 0xFC, 0x02, 0x82, 0x42, 0xC2, 0x22, 0xA2,
    0x62, 0xE2, 0x12, 0x92, 0x52, 0xD2, 0x32, 0xB2, 0x72, 0xF2,
    0x0A, 0x8A, 0x4A, 0xCA, 0x2A, 0xAA, 0x6A, 0xEA, 0x1A, 0x9A,
    0x5A, 0xDA, 0x3A, 0xBA, 0x7A, 0xFA, 0x06, 0x86, 0x46, 0xC6,
    0x26, 0xA6, 0x66, 0xE6, 0x16, 0x96, 0x56, 0xD6, 0x36, 0xB6,
    0x76, 0xF6, 0x0E, 0x8E, 0x4E, 0xCE, 0x2E, 0xAE, 0x6E, 0xEE,
    0x1E, 0x9E, 0x5E, 0xDE, 0x3E, 0xBE, 0x7E, 0xFE, 0x01, 0x81,
    0x41, 0xC1, 0x21, 0xA1, 0x61, 0xE1, 0x11, 0x91, 0x51, 0xD1,
    0x31, 0xB1, 0x71, 0xF1, 0x09, 0x89, 0x49, 0xC9, 0x29, 0xA9,
    0x69, 0xE9, 0x19, 0x99, 0x59, 0xD9, 0x39, 0xB9, 0x79, 0xF9,
    0x05, 0x85, 0x45, 0xC5, 0x25, 0xA5, 0x65, 0xE5, 0x15, 0x95,
    0x55, 0xD5, 0x35, 0xB5, 0x75, 0xF5, 0x0D, 0x8D, 0x4D, 0xCD,
    0x2D, 0xAD, 0x6D, 0xED, 0x1D, 0x9D, 0x5D, 0xDD, 0x3D, 0xBD,
    0x7D, 0xFD, 0x03, 0x83, 0x43, 0xC3, 0x23, 0xA3, 0x63, 0xE3,
    0x13, 0x93, 0x53, 0xD3, 0x33, 0xB3, 0x73, 0xF3, 0x0B, 0x8B,
    0x4B, 0xCB, 0x2B, 0xAB, 0x6B, 0xEB, 0x1B, 0x9B, 0x5B, 0xDB,
    0x3B, 0xBB, 0x7B, 0xFB, 0x07, 0x87, 0x47, 0xC7, 0x27, 0xA7,
    0x67, 0xE7, 0x17, 0x97, 0x57, 0xD7, 0x37, 0xB7, 0x77, 0xF7,
    0x0F, 0x8F, 0x4F, 0xCF, 0x2F, 0xAF, 0x6F, 0xEF, 0x1F, 0x9F,
    0x5F, 0xDF, 0x3F, 0xBF, 0x7F, 0xFF,
]
# fmt: on

LZW_CLEAR_TABLE_MARKER = 256  # Special code to indicate table reset
LZW_EOD_MARKER = 257  # End-of-data marker
LZW_INITIAL_BITS_PER_CODE = 9  # Initial code bit width
LZW_MAX_BITS_PER_CODE = 12  # Maximum code bit width


def preload_image(image_cache: ImageCache, name, dims=None):
    """
    Read an image and load it into memory.

    For raster images: following this call, the image is inserted in `image_cache.images`,
    and following calls to `fpdf.fpdf.FPDF.image()` will re-use the same cached values, without re-reading the image.

    For vector images: the data is loaded and the metadata extracted.

    Args:
        image_cache: an `ImageCache` instance, usually the `.image_cache` attribute of a `FPDF` instance.
        name: either a string representing a file path to an image, an URL to an image,
            an io.BytesIO, or a instance of `PIL.Image.Image`.
        dims (Tuple[float]): optional dimensions as a tuple (width, height) to resize the image
            (raster only) before storing it in the PDF.

    Returns: A tuple, consisting of 3 values: the name, the image data,
        and an instance of a subclass of `ImageInfo`.
    """
    # Identify and load SVG data:
    if str(name).endswith(".svg"):
        try:
            return get_svg_info(name, load_image(str(name)), image_cache=image_cache)
        except Exception as error:
            raise ValueError(f"Could not parse file: {name}") from error
    if isinstance(name, bytes) and _is_svg(name.strip()):
        return get_svg_info(name, io.BytesIO(name), image_cache=image_cache)
    if isinstance(name, io.BytesIO) and _is_svg(name.getvalue().strip()):
        return get_svg_info("vector_image", name, image_cache=image_cache)

    # Load raster data.
    if isinstance(name, str):
        img = None
    elif isinstance(name, Image.Image):
        bytes_ = name.tobytes()
        img_hash = hashlib.new("md5", usedforsecurity=False)  # nosec B324
        img_hash.update(bytes_)
        name, img = img_hash.hexdigest(), name
    elif isinstance(name, (bytes, io.BytesIO)):
        bytes_ = name.getvalue() if isinstance(name, io.BytesIO) else name
        bytes_ = bytes_.strip()
        img_hash = hashlib.new("md5", usedforsecurity=False)  # nosec B324
        img_hash.update(bytes_)
        name, img = img_hash.hexdigest(), name
    else:
        name, img = str(name), name
    info = image_cache.images.get(name)
    if info:
        info["usages"] += 1
    else:
        info = get_img_info(name, img, image_cache.image_filter, dims)
        info["i"] = len(image_cache.images) + 1
        info["usages"] = 1
        info["iccp_i"] = None
        iccp = info.get("iccp")
        if iccp:
            LOGGER.debug(
                "ICC profile found for image %s - It will be inserted in the PDF document",
                name,
            )
            if iccp in image_cache.icc_profiles:
                info["iccp_i"] = image_cache.icc_profiles[iccp]
            else:
                iccp_i = len(image_cache.icc_profiles)
                image_cache.icc_profiles[iccp] = iccp_i
                info["iccp_i"] = iccp_i
            info["iccp"] = None
        image_cache.images[name] = info
    return name, img, info


def _is_svg(bytes_):
    return bytes_.startswith(b"<?xml ") or bytes_.startswith(b"<svg ")


def load_image(filename):
    """
    This method is used to load external resources, such as images.
    It is automatically called when resource added to document by `fpdf.fpdf.FPDF.image()`.
    It always return a BytesIO buffer.
    """
    # if a bytesio instance is passed in, use it as is.
    if isinstance(filename, BytesIO):
        return filename
    if isinstance(filename, Path):
        filename = str(filename)
    # by default loading from network is allowed for all images
    if filename.startswith(("http://", "https://")):
        # disabling bandit & semgrep rules as permitted schemes are whitelisted:
        # nosemgrep: python.lang.security.audit.dynamic-urllib-use-detected.dynamic-urllib-use-detected
        with urlopen(filename) as url_file:  # nosec B310
            return BytesIO(url_file.read())
    elif filename.startswith("data:"):
        return _decode_base64_image(filename)
    with open(filename, "rb") as local_file:
        return BytesIO(local_file.read())


def _decode_base64_image(base64Image):
    "Decode the base 64 image string into an io byte stream."
    frags = base64Image.split("base64,")
    if len(frags) != 2:
        raise NotImplementedError("Unsupported non-base64 image data")
    imageData = frags[1]
    decodedData = base64.b64decode(imageData)
    return BytesIO(decodedData)


def is_iccp_valid(iccp, filename):
    "Checks the validity of an ICC profile"
    try:
        profile = ImageCms.getOpenProfile(BytesIO(iccp))
    except ImageCms.PyCMSError:
        LOGGER.info("Invalid ICC Profile in file %s", filename)
        return False
    color_space = profile.profile.xcolor_space.strip()
    if color_space not in ("GRAY", "RGB"):
        LOGGER.info(
            "Unsupported color space %s in ICC Profile of file %s - cf. issue #711",
            color_space,
            filename,
        )
        return False
    return True


def get_svg_info(filename, img, image_cache):
    svg = SVGObject(img.getvalue(), image_cache=image_cache)
    if svg.viewbox:
        _, _, w, h = svg.viewbox
    else:
        w = h = 0.0
    if svg.width:
        w = svg.width
    if svg.height:
        h = svg.height
    info = VectorImageInfo(data=svg, w=w, h=h)
    return filename, svg, info


def get_img_info(filename, img=None, image_filter="AUTO", dims=None):
    """
    Args:
        filename: in a format that can be passed to load_image
        img: optional `bytes`, `BytesIO` or `PIL.Image.Image` instance
        image_filter (str): one of the SUPPORTED_IMAGE_FILTERS
    """
    if Image is None:
        raise EnvironmentError("Pillow not available - fpdf2 cannot insert images")

    is_pil_img = True
    keep_bytes_io_open = False
    # Flag to check whether a cmyk image is jpeg or not, if set to True the decode array
    # is inverted in output.py
    jpeg_inverted = False
    img_raw_data = None
    if not img or isinstance(img, (Path, str)):
        img_raw_data = load_image(filename)
        img = Image.open(img_raw_data)
        is_pil_img = False
    elif not isinstance(img, Image.Image):
        keep_bytes_io_open = isinstance(img, BytesIO)
        img_raw_data = BytesIO(img) if isinstance(img, bytes) else img
        img = Image.open(img_raw_data)
        is_pil_img = False

    img_altered = False
    if dims:
        img = img.resize(dims, resample=RESAMPLE)
        img_altered = True

    if image_filter == "AUTO":
        # Very simple logic for now:
        if img.format == "JPEG":
            image_filter = "DCTDecode"
        elif img.mode == "1" and PIL_features.check("libtiff"):
            # The 2nd condition prevents from running in a bug sometimes,
            # cf. test_transcode_monochrome_and_libtiff_support_custom_tags()
            image_filter = "CCITTFaxDecode"
        else:
            image_filter = "FlateDecode"

    if img.mode in ("P", "PA") and image_filter != "FlateDecode":
        img = img.convert("RGBA")

    if img.mode not in ("1", "L", "LA", "RGB", "RGBA", "P", "PA", "CMYK"):
        img = img.convert("RGBA")
        img_altered = True

    if img.mode in ("P", "RGBA") and image_filter == "LZWDecode":
        img = img.convert("RGB")
    elif img.mode in ("LA") and image_filter == "LZWDecode":
        img = img.convert("L")

    w, h = img.size
    info = RasterImageInfo()

    iccp = None
    if "icc_profile" in img.info:
        if is_iccp_valid(img.info["icc_profile"], filename):
            iccp = img.info["icc_profile"]

    if img_raw_data is not None and not img_altered:
        # if we can use the original image bytes directly we do (JPEG and group4 TIFF only):
        if img.format == "JPEG" and image_filter == "DCTDecode":
            if img.mode in ("RGB", "RGBA"):
                dpn, bpc, colspace = 3, 8, "DeviceRGB"
            elif img.mode == "CMYK":
                dpn, bpc, colspace = 4, 8, "DeviceCMYK"
                jpeg_inverted = True
            elif img.mode == "L":
                dpn, bpc, colspace = 1, 8, "DeviceGray"
            else:
                raise ValueError(f"Unsupported image mode: {img.mode}")
            img_raw_data.seek(0)
            info.update(
                {
                    "data": img_raw_data.read(),
                    "w": w,
                    "h": h,
                    "cs": colspace,
                    "iccp": iccp,
                    "dpn": dpn,
                    "bpc": bpc,
                    "f": image_filter,
                    "inverted": jpeg_inverted,
                    "dp": f"/Predictor 15 /Colors {dpn} /Columns {w}",
                }
            )
            return info
        # We can directly copy the data out of a CCITT Group 4 encoded TIFF, if it
        # only contains a single strip
        if (
            img.format == "TIFF"
            and image_filter == "CCITTFaxDecode"
            and img.info["compression"] == "group4"
            and len(img.tag_v2[TiffImagePlugin.STRIPOFFSETS]) == 1
            and len(img.tag_v2[TiffImagePlugin.STRIPBYTECOUNTS]) == 1
        ):
            photo = img.tag_v2[TiffImagePlugin.PHOTOMETRIC_INTERPRETATION]
            inverted = False
            if photo == 0:
                inverted = True
            elif photo != 1:
                raise ValueError(
                    f"unsupported photometric interpretation for g4 tiff: {photo}"
                )
            offset, length = ccitt_payload_location_from_pil(img)
            img_raw_data.seek(offset)
            ccittrawdata = img_raw_data.read(length)
            fillorder = img.tag_v2.get(TiffImagePlugin.FILLORDER)
            if fillorder is None or fillorder == 1:
                # no FillOrder or msb-to-lsb: nothing to do
                pass
            elif fillorder == 2:
                # lsb-to-msb: reverse bits of each byte
                ccittrawdata = bytearray(ccittrawdata)
                for i, n in enumerate(ccittrawdata):
                    ccittrawdata[i] = TIFFBitRevTable[n]
                ccittrawdata = bytes(ccittrawdata)
            else:
                raise ValueError(f"unsupported FillOrder: {fillorder}")
            dpn, bpc, colspace = 1, 1, "DeviceGray"
            info.update(
                {
                    "data": ccittrawdata,
                    "w": w,
                    "h": h,
                    "iccp": None,
                    "dpn": dpn,
                    "cs": colspace,
                    "bpc": bpc,
                    "f": image_filter,
                    "inverted": jpeg_inverted,
                    "dp": f"/BlackIs1 {str(not inverted).lower()} /Columns {w} /K -1 /Rows {h}",
                }
            )
            return info

    # garbage collection
    img_raw_data = None

    if img.mode == "1":
        dpn, bpc, colspace = 1, 1, "DeviceGray"
        info["data"] = _to_data(img, image_filter)
    elif img.mode == "L":
        dpn, bpc, colspace = 1, 8, "DeviceGray"
        info["data"] = _to_data(img, image_filter)
    elif img.mode == "LA":
        dpn, bpc, colspace = 1, 8, "DeviceGray"
        alpha_channel = slice(1, None, 2)
        info["data"] = _to_data(img, image_filter, remove_slice=alpha_channel)
        if _has_alpha(img, alpha_channel) and image_filter not in (
            "DCTDecode",
            "JPXDecode",
        ):
            info["smask"] = _to_data(img, image_filter, select_slice=alpha_channel)
    elif img.mode == "P":
        dpn, bpc, colspace = 1, 8, "Indexed"
        info["data"] = _to_data(img, image_filter)
        info["pal"] = img.palette.palette

        # check if the P image has transparency
        if img.info.get("transparency", None) is not None and image_filter not in (
            "DCTDecode",
            "JPXDecode",
        ):
            # convert to RGBA to get the alpha channel for creating the smask
            info["smask"] = _to_data(
                img.convert("RGBA"), image_filter, select_slice=slice(3, None, 4)
            )
    elif img.mode == "PA":
        dpn, bpc, colspace = 1, 8, "Indexed"
        info["pal"] = img.palette.palette
        alpha_channel = slice(1, None, 2)
        info["data"] = _to_data(img, image_filter, remove_slice=alpha_channel)
        if _has_alpha(img, alpha_channel) and image_filter not in (
            "DCTDecode",
            "JPXDecode",
        ):
            info["smask"] = _to_data(img, image_filter, select_slice=alpha_channel)
    elif img.mode == "CMYK":
        dpn, bpc, colspace = 4, 8, "DeviceCMYK"
        info["data"] = _to_data(img, image_filter)
    elif img.mode == "RGB":
        dpn, bpc, colspace = 3, 8, "DeviceRGB"
        info["data"] = _to_data(img, image_filter)
    else:  # RGBA image
        dpn, bpc, colspace = 3, 8, "DeviceRGB"
        alpha_channel = slice(3, None, 4)
        info["data"] = _to_data(img, image_filter, remove_slice=alpha_channel)
        if _has_alpha(img, alpha_channel) and image_filter not in (
            "DCTDecode",
            "JPXDecode",
        ):
            info["smask"] = _to_data(img, image_filter, select_slice=alpha_channel)

    dp = f"/Predictor 15 /Colors {dpn} /Columns {w}"

    if img.mode == "1":
        dp = f"/BlackIs1 true /Columns {w} /K -1 /Rows {h}"

    if not is_pil_img:
        if keep_bytes_io_open:
            img.fp = None  # cf. issue #881
        else:
            img.close()

    info.update(
        {
            "w": w,
            "h": h,
            "cs": colspace,
            "iccp": iccp,
            "bpc": bpc,
            "dpn": dpn,
            "f": image_filter,
            "inverted": jpeg_inverted,
            "dp": dp,
        }
    )
    return info


class temp_attr:
    """
    temporary change the attribute of an object using a context manager
    """

    def __init__(self, obj, field, value):
        self.obj = obj
        self.field = field
        self.value = value

    def __enter__(self):
        self.exists = False
        if hasattr(self.obj, self.field):
            self.exists = True
            self.old_value = getattr(self.obj, self.field)
        setattr(self.obj, self.field, self.value)

    def __exit__(self, exctype, excinst, exctb):
        if self.exists:
            setattr(self.obj, self.field, self.old_value)
        else:
            delattr(self.obj, self.field)


def ccitt_payload_location_from_pil(img):
    """
    returns the byte offset and length of the CCITT payload in the original TIFF data
    """
    # assert(img.info["compression"] == "group4")

    # Read the TIFF tags to find the offset(s) of the compressed data strips.
    strip_offsets = img.tag_v2[TiffImagePlugin.STRIPOFFSETS]
    strip_bytes = img.tag_v2[TiffImagePlugin.STRIPBYTECOUNTS]

    # PIL always seems to create a single strip even for very large TIFFs when
    # it saves images, so assume we only have to read a single strip.
    # A test ~10 GPixel image was still encoded as a single strip. Just to be
    # safe check throw an error if there is more than one offset.
    if len(strip_offsets) != 1 or len(strip_bytes) != 1:
        raise NotImplementedError(
            "Transcoding multiple strips not supported by the PDF format"
        )

    (offset,), (length,) = strip_offsets, strip_bytes

    return offset, length


def transcode_monochrome(img):
    """
    Convert the open PIL.Image imgdata to compressed CCITT Group4 data.

    """
    # Convert the image to Group 4 in memory. If libtiff is not installed and
    # Pillow is not compiled against it, .save() will raise an exception.
    newimgio = BytesIO()

    # we create a whole new PIL image or otherwise it might happen with some
    # input images, that libtiff fails an assert and the whole process is
    # killed by a SIGABRT:
    img2 = Image.frombytes(img.mode, img.size, img.tobytes())

    # Since version 8.3.0 Pillow limits strips to 64 KB. Since PDF only
    # supports single strip CCITT Group4 payloads, we have to coerce it back
    # into putting everything into a single strip. Thanks to Andrew Murray for
    # the hack.
    #
    # Since version 8.4.0 Pillow allows us to modify the strip size explicitly
    tmp_strip_size = (img.size[0] + 7) // 8 * img.size[1]
    if hasattr(TiffImagePlugin, "STRIP_SIZE"):
        # we are using Pillow 8.4.0 or later
        with temp_attr(TiffImagePlugin, "STRIP_SIZE", tmp_strip_size):
            img2.save(newimgio, format="TIFF", compression="group4")
    else:
        # only needed for Pillow 8.3.x but works for versions before that as
        # well
        pillow__getitem__ = TiffImagePlugin.ImageFileDirectory_v2.__getitem__

        def __getitem__(self, tag):
            overrides = {
                TiffImagePlugin.ROWSPERSTRIP: img.size[1],
                TiffImagePlugin.STRIPBYTECOUNTS: [tmp_strip_size],
                TiffImagePlugin.STRIPOFFSETS: [0],
            }
            return overrides.get(tag, pillow__getitem__(self, tag))

        with temp_attr(
            TiffImagePlugin.ImageFileDirectory_v2, "__getitem__", __getitem__
        ):
            img2.save(newimgio, format="TIFF", compression="group4")

    # Open new image in memory
    newimgio.seek(0)
    newimg = Image.open(newimgio)

    offset, length = ccitt_payload_location_from_pil(newimg)

    newimgio.seek(offset)
    return newimgio.read(length)


def _to_lzwdata(img, remove_slice=None, select_slice=None):
    data = bytearray(img.tobytes())

    if remove_slice:
        del data[remove_slice]
    if select_slice:
        data = data[select_slice]

    if img.mode == "1":
        row_size = ceil(img.size[0] / 8)
    else:
        channels_count = len(data) // (img.size[0] * img.size[1])
        row_size = img.size[0] * channels_count

    data_with_padding = bytearray()
    for i in range(0, len(data), row_size):
        data_with_padding.extend(b"\0")
        data_with_padding.extend(data[i : i + row_size])
    data = data_with_padding
    # Start compression

    # The encoder shall begin by issuing a clear-table code:
    result_codes = [LZW_CLEAR_TABLE_MARKER]
    table, next_code, bits_per_code, max_code_value = clear_table()

    current_sequence = b""
    for byte in data:
        next_sequence = current_sequence + bytes([byte])

        if next_sequence in table:
            # Extend current sequence if already in the table
            current_sequence = next_sequence
        else:
            # Output code for the current sequence
            result_codes.append(table[current_sequence])

            # Add the new sequence to the table if there's room
            if next_code <= (1 << LZW_MAX_BITS_PER_CODE) - 1:
                table[next_sequence] = next_code
                next_code += 1
                if next_code > max_code_value and bits_per_code < LZW_MAX_BITS_PER_CODE:
                    bits_per_code += 1
                    max_code_value = (1 << bits_per_code) - 1
            else:
                # If the table is full, emit a clear-table command
                result_codes.append(LZW_CLEAR_TABLE_MARKER)
                table, next_code, bits_per_code, max_code_value = clear_table()

            # Start new sequence
            current_sequence = bytes([byte])

    # Ensure everything actually is encoded
    if current_sequence:
        result_codes.append(table[current_sequence])

    result_codes.append(LZW_EOD_MARKER)

    return pack_codes_into_bytes(result_codes)


def pack_codes_into_bytes(codes):
    """
    Convert the list of result codes into a continuous byte stream, with codes packed as per the code bit-width.
    The bit-width starts at 9 bits and expands as needed.

    """

    (
        _,
        next_code,
        bits_per_code,
        max_code_value,
    ) = clear_table()
    buffer = 0
    bits_in_buffer = 0
    output = bytearray()

    if numpy is not None:
        # Using numpy improves the performance significantly there
        # _cf._ https://github.com/py-pdf/fpdf2/issues/1380
        codes = numpy.array(codes, dtype=numpy.uint32)
    for code in codes:
        buffer = (buffer << bits_per_code) | code
        bits_in_buffer += bits_per_code

        while bits_in_buffer >= 8:
            bits_in_buffer -= 8
            output.append((buffer >> bits_in_buffer) & 0xFF)

        if code == LZW_CLEAR_TABLE_MARKER:
            _, next_code, bits_per_code, max_code_value = clear_table()
        elif code != LZW_EOD_MARKER:
            next_code += 1
            if next_code > max_code_value and bits_per_code < LZW_MAX_BITS_PER_CODE:
                bits_per_code += 1
                max_code_value = (1 << bits_per_code) - 1

    if bits_in_buffer > 0:
        output.append((buffer << (8 - bits_in_buffer)) & 0xFF)

    return bytes(output)


def clear_table():
    """
    Reset the encoding table and coding state to initial conditions.

    """

    table = {bytes([i]): i for i in range(256)}
    next_code = LZW_EOD_MARKER + 1
    bits_per_code = LZW_INITIAL_BITS_PER_CODE
    max_code_value = (1 << bits_per_code) - 1
    return table, next_code, bits_per_code, max_code_value


def _to_data(img, image_filter, **kwargs):
    if image_filter == "FlateDecode":
        return _to_zdata(img, **kwargs)

    if image_filter == "CCITTFaxDecode":
        return transcode_monochrome(img)

    if image_filter == "LZWDecode":
        return _to_lzwdata(img, **kwargs)

    if img.mode == "LA":
        img = img.convert("L")

    if img.mode == "RGBA":
        img = img.convert("RGB")

    if image_filter == "DCTDecode":
        compressed_bytes = BytesIO()
        img.save(compressed_bytes, format="JPEG")
        return compressed_bytes.getvalue()

    if image_filter == "JPXDecode":
        compressed_bytes = BytesIO()
        img.save(compressed_bytes, format="JPEG2000")
        return compressed_bytes.getvalue()

    raise FPDFException(f'Unsupported image filter: "{image_filter}"')


def _to_zdata(img, remove_slice=None, select_slice=None):
    data = bytearray(img.tobytes())
    if remove_slice:
        del data[remove_slice]
    if select_slice:
        data = data[select_slice]
    # Left-padding every row with a single zero:
    if img.mode == "1":
        row_size = ceil(img.size[0] / 8)
    else:
        channels_count = len(data) // (img.size[0] * img.size[1])
        row_size = img.size[0] * channels_count

    data_with_padding = bytearray()
    for i in range(0, len(data), row_size):
        data_with_padding.extend(b"\0")
        data_with_padding.extend(data[i : i + row_size])

    return zlib.compress(data_with_padding, level=SETTINGS.compression_level)


def _has_alpha(img, alpha_channel):
    alpha = bytearray(img.tobytes())[alpha_channel]
    return any(c != 255 for c in alpha)