From 2101cd1bf4793e3f613dffcefcaa0cffff0d9c4c Mon Sep 17 00:00:00 2001 From: Andrew Murray Date: Sun, 5 Jan 2025 20:24:51 +1100 Subject: [PATCH] Allow saving multiple frames as BigTIFF --- Tests/test_file_tiff.py | 12 +++++-- src/PIL/TiffImagePlugin.py | 69 +++++++++++++++++++++++--------------- 2 files changed, 51 insertions(+), 30 deletions(-) diff --git a/Tests/test_file_tiff.py b/Tests/test_file_tiff.py index dedd48c20b3..30fccbad3e9 100644 --- a/Tests/test_file_tiff.py +++ b/Tests/test_file_tiff.py @@ -117,10 +117,16 @@ def test_bigtiff(self, tmp_path: Path) -> None: def test_bigtiff_save(self, tmp_path: Path) -> None: outfile = str(tmp_path / "temp.tif") - hopper().save(outfile, big_tiff=True) + im = hopper() + im.save(outfile, big_tiff=True) - with Image.open(outfile) as im: - assert im.tag_v2._bigtiff is True + with Image.open(outfile) as reloaded: + assert reloaded.tag_v2._bigtiff is True + + im.save(outfile, save_all=True, append_images=[im], big_tiff=True) + + with Image.open(outfile) as reloaded: + assert reloaded.tag_v2._bigtiff is True def test_seek_too_large(self) -> None: with pytest.raises(ValueError, match="Unable to seek to frame"): diff --git a/src/PIL/TiffImagePlugin.py b/src/PIL/TiffImagePlugin.py index 61eb1524311..aad3495985b 100644 --- a/src/PIL/TiffImagePlugin.py +++ b/src/PIL/TiffImagePlugin.py @@ -962,13 +962,16 @@ def tobytes(self, offset: int = 0) -> bytes: result = self._pack("Q" if self._bigtiff else "H", len(self._tags_v2)) entries: list[tuple[int, int, int, bytes, bytes]] = [] - offset += len(result) + len(self._tags_v2) * (20 if self._bigtiff else 12) + 4 + + fmt = "Q" if self._bigtiff else "L" + fmt_size = 8 if self._bigtiff else 4 + offset += ( + len(result) + len(self._tags_v2) * (20 if self._bigtiff else 12) + fmt_size + ) stripoffsets = None # pass 1: convert tags to binary format # always write tags in ascending order - fmt = "Q" if self._bigtiff else "L" - fmt_size = 8 if self._bigtiff else 4 for tag, value in sorted(self._tags_v2.items()): if tag == STRIPOFFSETS: stripoffsets = len(entries) @@ -1024,7 +1027,7 @@ def tobytes(self, offset: int = 0) -> bytes: ) # -- overwrite here for multi-page -- - result += b"\0\0\0\0" # end of entries + result += self._pack(fmt, 0) # end of entries # pass 3: write auxiliary data to file for tag, typ, count, value, data in entries: @@ -2043,20 +2046,21 @@ def setup(self) -> None: self.offsetOfNewPage = 0 self.IIMM = iimm = self.f.read(4) + self._bigtiff = b"\x2B" in iimm if not iimm: # empty file - first page self.isFirst = True return self.isFirst = False - if iimm == b"II\x2a\x00": - self.setEndian("<") - elif iimm == b"MM\x00\x2a": - self.setEndian(">") - else: + if not iimm in PREFIXES: msg = "Invalid TIFF file header" raise RuntimeError(msg) + self.setEndian("<" if iimm.startswith(II) else ">") + + if self._bigtiff: + self.f.seek(4, os.SEEK_CUR) self.skipIFDs() self.goToEnd() @@ -2076,11 +2080,13 @@ def finalize(self) -> None: msg = "IIMM of new page doesn't match IIMM of first page" raise RuntimeError(msg) - ifd_offset = self.readLong() + if self._bigtiff: + self.f.seek(4, os.SEEK_CUR) + ifd_offset = self._read(8 if self._bigtiff else 4) ifd_offset += self.offsetOfNewPage assert self.whereToWriteNewIFDOffset is not None self.f.seek(self.whereToWriteNewIFDOffset) - self.writeLong(ifd_offset) + self._write(ifd_offset, 8 if self._bigtiff else 4) self.f.seek(ifd_offset) self.fixIFD() @@ -2126,18 +2132,20 @@ def setEndian(self, endian: str) -> None: self.endian = endian self.longFmt = f"{self.endian}L" self.shortFmt = f"{self.endian}H" - self.tagFormat = f"{self.endian}HHL" + self.tagFormat = f"{self.endian}HH" + ("Q" if self._bigtiff else "L") def skipIFDs(self) -> None: while True: - ifd_offset = self.readLong() + ifd_offset = self._read(8 if self._bigtiff else 4) if ifd_offset == 0: - self.whereToWriteNewIFDOffset = self.f.tell() - 4 + self.whereToWriteNewIFDOffset = self.f.tell() - ( + 8 if self._bigtiff else 4 + ) break self.f.seek(ifd_offset) - num_tags = self.readShort() - self.f.seek(num_tags * 12, os.SEEK_CUR) + num_tags = self._read(8 if self._bigtiff else 2) + self.f.seek(num_tags * (20 if self._bigtiff else 12), os.SEEK_CUR) def write(self, data: Buffer, /) -> int: return self.f.write(data) @@ -2185,13 +2193,17 @@ def rewriteLastShort(self, value: int) -> None: def rewriteLastLong(self, value: int) -> None: return self._rewriteLast(value, 4) + def _write(self, value: int, field_size: int) -> None: + bytes_written = self.f.write( + struct.pack(self.endian + self._fmt(field_size), value) + ) + self._verify_bytes_written(bytes_written, field_size) + def writeShort(self, value: int) -> None: - bytes_written = self.f.write(struct.pack(self.shortFmt, value)) - self._verify_bytes_written(bytes_written, 2) + self._write(value, 2) def writeLong(self, value: int) -> None: - bytes_written = self.f.write(struct.pack(self.longFmt, value)) - self._verify_bytes_written(bytes_written, 4) + self._write(value, 4) def close(self) -> None: self.finalize() @@ -2199,24 +2211,27 @@ def close(self) -> None: self.f.close() def fixIFD(self) -> None: - num_tags = self.readShort() + num_tags = self._read(8 if self._bigtiff else 2) for i in range(num_tags): - tag, field_type, count = struct.unpack(self.tagFormat, self.f.read(8)) + tag, field_type, count = struct.unpack( + self.tagFormat, self.f.read(12 if self._bigtiff else 8) + ) field_size = self.fieldSizes[field_type] total_size = field_size * count - is_local = total_size <= 4 + fmt_size = 8 if self._bigtiff else 4 + is_local = total_size <= fmt_size if not is_local: - offset = self.readLong() + self.offsetOfNewPage - self.rewriteLastLong(offset) + offset = self._read(fmt_size) + self.offsetOfNewPage + self._rewriteLast(offset, fmt_size) if tag in self.Tags: cur_pos = self.f.tell() if is_local: self._fixOffsets(count, field_size) - self.f.seek(cur_pos + 4) + self.f.seek(cur_pos + fmt_size) else: self.f.seek(offset) self._fixOffsets(count, field_size) @@ -2224,7 +2239,7 @@ def fixIFD(self) -> None: elif is_local: # skip the locally stored value that is not an offset - self.f.seek(4, os.SEEK_CUR) + self.f.seek(fmt_size, os.SEEK_CUR) def _fixOffsets(self, count: int, field_size: int) -> None: for i in range(count):