|
| 1 | +# -*- Mode: Python -*- |
| 2 | + |
| 3 | +# Find Exif data in an HEIC file. |
| 4 | + |
| 5 | +# As of 2019, the latest standard seems to be "ISO/IEC 14496-12:2015" |
| 6 | +# There are many different related standards. (quicktime, mov, mp4, etc...) |
| 7 | +# See https://en.wikipedia.org/wiki/ISO_base_media_file_format for more details. |
| 8 | + |
| 9 | +# We parse just enough of the iso format to locate the Exif data in the file. |
| 10 | +# Inside the 'meta' box are two directories we need: |
| 11 | +# 1) the 'iinf' box contains 'infe' records, we look for the item_ID for 'Exif'. |
| 12 | +# 2) once we have the item_ID, we find a matching entry in the 'iloc' box, which |
| 13 | +# gives us position and size information. |
| 14 | + |
| 15 | +import struct |
| 16 | + |
| 17 | +from .exif_log import get_logger |
| 18 | + |
| 19 | +logger = get_logger() |
| 20 | + |
| 21 | +class WrongBox (Exception): |
| 22 | + pass |
| 23 | +class NoParser (Exception): |
| 24 | + pass |
| 25 | +class BoxVersion (Exception): |
| 26 | + pass |
| 27 | +class BadSize (Exception): |
| 28 | + pass |
| 29 | + |
| 30 | +class Box: |
| 31 | + def __init__ (self, name): |
| 32 | + self.name = name |
| 33 | + |
| 34 | + def __repr__ (self): |
| 35 | + return "<box '%s'>" % (self.name,) |
| 36 | + |
| 37 | +class HEICExifFinder: |
| 38 | + |
| 39 | + def __init__ (self, file): |
| 40 | + self.file = file |
| 41 | + |
| 42 | + def get (self, nbytes): |
| 43 | + r = self.file.read (nbytes) |
| 44 | + if not r: |
| 45 | + raise EOFError |
| 46 | + else: |
| 47 | + return r |
| 48 | + |
| 49 | + def get16 (self): |
| 50 | + return struct.unpack ('>H', self.get (2))[0] |
| 51 | + |
| 52 | + def get32 (self): |
| 53 | + return struct.unpack ('>L', self.get (4))[0] |
| 54 | + |
| 55 | + def get64 (self): |
| 56 | + return struct.unpack ('>Q', self.get (8))[0] |
| 57 | + |
| 58 | + def get_int4x2 (self): |
| 59 | + n = struct.unpack ('>B', self.get(1))[0] |
| 60 | + n0 = n >> 4 |
| 61 | + n1 = n & 0xf |
| 62 | + return n0, n1 |
| 63 | + |
| 64 | + # some fields have variant-sized data. |
| 65 | + def get_int (self, size): |
| 66 | + if size == 2: |
| 67 | + return self.get16() |
| 68 | + elif size == 4: |
| 69 | + return self.get32() |
| 70 | + elif size == 8: |
| 71 | + return self.get64() |
| 72 | + elif size == 0: |
| 73 | + return 0 |
| 74 | + else: |
| 75 | + raise BadSize (size) |
| 76 | + |
| 77 | + def get_string (self): |
| 78 | + r = [] |
| 79 | + while 1: |
| 80 | + ch = self.get (1) |
| 81 | + if ch == b'\x00': |
| 82 | + break |
| 83 | + else: |
| 84 | + r.append (ch) |
| 85 | + return b''.join (r) |
| 86 | + |
| 87 | + def next_box (self, depth=0): |
| 88 | + pos = self.file.tell() |
| 89 | + size = self.get32() |
| 90 | + kind = self.get(4).decode('ascii') |
| 91 | + b = Box (kind) |
| 92 | + if size == 0: |
| 93 | + # signifies 'to the end of the file', we shouldn't see this. |
| 94 | + raise NotImplementedError |
| 95 | + elif size == 1: |
| 96 | + # 64-bit size follows type. |
| 97 | + size = self.get64() |
| 98 | + b.size = size - 16 |
| 99 | + b.after = pos + size |
| 100 | + else: |
| 101 | + b.size = size - 8 |
| 102 | + b.after = pos + size |
| 103 | + b.pos = self.file.tell() |
| 104 | + return b |
| 105 | + |
| 106 | + def get_full (self, box): |
| 107 | + # iso boxes come in 'old' and 'full' variants. the 'full' variant |
| 108 | + # contains version and flags information. |
| 109 | + vflags = self.get32() |
| 110 | + box.version = vflags >> 24 |
| 111 | + box.flags = vflags & 0x00ffffff |
| 112 | + |
| 113 | + def skip (self, box): |
| 114 | + self.file.seek (box.after) |
| 115 | + |
| 116 | + def expect_parse (self, name): |
| 117 | + b = self.next_box() |
| 118 | + if b.name == name: |
| 119 | + return self.parse_box (b) |
| 120 | + else: |
| 121 | + raise WrongBox (name, b.name) |
| 122 | + |
| 123 | + def get_parser (self, box): |
| 124 | + method = 'parse_%s' % (box.name,) |
| 125 | + return getattr (self, method, None) |
| 126 | + |
| 127 | + def parse_box (self, b): |
| 128 | + probe = self.get_parser (b) |
| 129 | + if probe is None: |
| 130 | + raise NoParser (b.name) |
| 131 | + else: |
| 132 | + probe (b) |
| 133 | + # in case anything is left unread |
| 134 | + self.file.seek (b.after) |
| 135 | + return b |
| 136 | + |
| 137 | + def parse_ftyp (self, box): |
| 138 | + box.major_brand = self.get(4) |
| 139 | + box.minor_version = self.get32() |
| 140 | + box.compat = [] |
| 141 | + size = box.size - 8 |
| 142 | + while size > 0: |
| 143 | + box.compat.append (self.get (4)) |
| 144 | + size -= 4 |
| 145 | + |
| 146 | + def parse_meta (self, meta): |
| 147 | + self.get_full (meta) |
| 148 | + # this is full of boxes, but not in a predictable order. |
| 149 | + meta.subs = {} |
| 150 | + while self.file.tell() < meta.after: |
| 151 | + box = self.next_box() |
| 152 | + psub = self.get_parser (box) |
| 153 | + if psub is not None: |
| 154 | + psub (box) |
| 155 | + meta.subs[box.name] = box |
| 156 | + else: |
| 157 | + logger.debug("HEIC: skipping %r" % (box,)) |
| 158 | + # skip any unparsed data |
| 159 | + self.skip (box) |
| 160 | + |
| 161 | + def parse_infe (self, box): |
| 162 | + self.get_full (box) |
| 163 | + if box.version >= 2: |
| 164 | + if box.version == 2: |
| 165 | + box.item_ID = self.get16() |
| 166 | + elif box.version == 3: |
| 167 | + box.item_ID = self.get32() |
| 168 | + box.item_protection_index = self.get16() |
| 169 | + box.item_type = self.get(4) |
| 170 | + box.item_name = self.get_string() |
| 171 | + # ignore the rest |
| 172 | + else: |
| 173 | + box.item_type = '' |
| 174 | + |
| 175 | + def parse_iinf (self, box): |
| 176 | + self.get_full (box) |
| 177 | + count = self.get16() |
| 178 | + box.exif_infe = None |
| 179 | + for _ in range (count): |
| 180 | + infe = self.expect_parse ('infe') |
| 181 | + if infe.item_type == b'Exif': |
| 182 | + logger.debug("HEIC: found Exif 'infe' box") |
| 183 | + box.exif_infe = infe |
| 184 | + break |
| 185 | + |
| 186 | + def parse_iloc (self, box): |
| 187 | + self.get_full (box) |
| 188 | + s0, s1 = self.get_int4x2() |
| 189 | + s2, s3 = self.get_int4x2() |
| 190 | + box.offset_size = s0 |
| 191 | + box.length_size = s1 |
| 192 | + box.base_offset_size = s2 |
| 193 | + box.index_size = s3 |
| 194 | + if box.version < 2: |
| 195 | + box.item_count = self.get16() |
| 196 | + elif box.version == 2: |
| 197 | + box.item_count = self.get32() |
| 198 | + else: |
| 199 | + raise BoxVersion (2, box.version) |
| 200 | + box.locs = {} |
| 201 | + logger.debug("HEIC: %d iloc items" % (box.item_count,)) |
| 202 | + for i in range (box.item_count): |
| 203 | + if box.version < 2: |
| 204 | + item_ID = self.get16() |
| 205 | + elif box.version == 2: |
| 206 | + item_ID = self.get32() |
| 207 | + else: |
| 208 | + # notreached |
| 209 | + raise BoxVersion (2, box.version) |
| 210 | + if box.version in (1, 2): |
| 211 | + # ignore construction_method |
| 212 | + _ = self.get16() |
| 213 | + data_reference_index = self.get16() |
| 214 | + box.base_offset = self.get_int (box.base_offset_size) |
| 215 | + extent_count = self.get16() |
| 216 | + extents = [] |
| 217 | + for _ in range (extent_count): |
| 218 | + if box.version in (1, 2) and box.index_size > 0: |
| 219 | + extent_index = self.get_int (box.index_size) |
| 220 | + else: |
| 221 | + extent_index = -1 |
| 222 | + extent_offset = self.get_int (box.offset_size) |
| 223 | + extent_length = self.get_int (box.length_size) |
| 224 | + extents.append ((extent_offset, extent_length)) |
| 225 | + box.locs[item_ID] = extents |
| 226 | + |
| 227 | + def find_exif (self): |
| 228 | + ftyp = self.expect_parse ('ftyp') |
| 229 | + assert ftyp.major_brand == b'heic' |
| 230 | + assert ftyp.minor_version == 0 |
| 231 | + meta = self.expect_parse ('meta') |
| 232 | + item_ID = meta.subs['iinf'].exif_infe.item_ID |
| 233 | + extents = meta.subs['iloc'].locs[item_ID] |
| 234 | + logger.debug("HEIC: found Exif location.") |
| 235 | + # we expect the Exif data to be in one piece. |
| 236 | + assert len(extents) == 1 |
| 237 | + pos, size = extents[0] |
| 238 | + # looks like there's a kind of pseudo-box here. |
| 239 | + self.file.seek (pos) |
| 240 | + size1 = self.get32() |
| 241 | + assert self.get(size1)[0:4] == b'Exif' |
| 242 | + offset = self.file.tell() |
| 243 | + endian = self.file.read(1) |
| 244 | + return offset, endian |
0 commit comments