jpeg revision 186675
1 2#------------------------------------------------------------------------------ 3# JPEG images 4# SunOS 5.5.1 had 5# 6# 0 string \377\330\377\340 JPEG file 7# 0 string \377\330\377\356 JPG file 8# 9# both of which turn into "JPEG image data" here. 10# 110 beshort 0xffd8 JPEG image data 12>6 string JFIF \b, JFIF standard 13# The following added by Erik Rossen <rossen@freesurf.ch> 1999-09-06 14# in a vain attempt to add image size reporting for JFIF. Note that these 15# tests are not fool-proof since some perfectly valid JPEGs are currently 16# impossible to specify in magic(4) format. 17# First, a little JFIF version info: 18>>11 byte x \b %d. 19>>12 byte x \b%02d 20# Next, the resolution or aspect ratio of the image: 21#>>13 byte 0 \b, aspect ratio 22#>>13 byte 1 \b, resolution (DPI) 23#>>13 byte 2 \b, resolution (DPCM) 24#>>4 beshort x \b, segment length %d 25# Next, show thumbnail info, if it exists: 26>>18 byte !0 \b, thumbnail %dx 27>>>19 byte x \b%d 28 29# EXIF moved down here to avoid reporting a bogus version number, 30# and EXIF version number printing added. 31# - Patrik R=E5dman <patrik+file-magic@iki.fi> 32>6 string Exif \b, EXIF standard 33# Look for EXIF IFD offset in IFD 0, and then look for EXIF version tag in EXIF IFD. 34# All possible combinations of entries have to be enumerated, since no looping 35# is possible. And both endians are possible... 36# The combinations included below are from real-world JPEGs. 37# Little-endian 38>>12 string II 39# IFD 0 Entry #5: 40>>>70 leshort 0x8769 41# EXIF IFD Entry #1: 42>>>>(78.l+14) leshort 0x9000 43>>>>>(78.l+23) byte x %c 44>>>>>(78.l+24) byte x \b.%c 45>>>>>(78.l+25) byte !0x30 \b%c 46# IFD 0 Entry #9: 47>>>118 leshort 0x8769 48# EXIF IFD Entry #3: 49>>>>(126.l+38) leshort 0x9000 50>>>>>(126.l+47) byte x %c 51>>>>>(126.l+48) byte x \b.%c 52>>>>>(126.l+49) byte !0x30 \b%c 53# IFD 0 Entry #10 54>>>130 leshort 0x8769 55# EXIF IFD Entry #3: 56>>>>(138.l+38) leshort 0x9000 57>>>>>(138.l+47) byte x %c 58>>>>>(138.l+48) byte x \b.%c 59>>>>>(138.l+49) byte !0x30 \b%c 60# EXIF IFD Entry #4: 61>>>>(138.l+50) leshort 0x9000 62>>>>>(138.l+59) byte x %c 63>>>>>(138.l+60) byte x \b.%c 64>>>>>(138.l+61) byte !0x30 \b%c 65# EXIF IFD Entry #5: 66>>>>(138.l+62) leshort 0x9000 67>>>>>(138.l+71) byte x %c 68>>>>>(138.l+72) byte x \b.%c 69>>>>>(138.l+73) byte !0x30 \b%c 70# IFD 0 Entry #11 71>>>142 leshort 0x8769 72# EXIF IFD Entry #3: 73>>>>(150.l+38) leshort 0x9000 74>>>>>(150.l+47) byte x %c 75>>>>>(150.l+48) byte x \b.%c 76>>>>>(150.l+49) byte !0x30 \b%c 77# EXIF IFD Entry #4: 78>>>>(150.l+50) leshort 0x9000 79>>>>>(150.l+59) byte x %c 80>>>>>(150.l+60) byte x \b.%c 81>>>>>(150.l+61) byte !0x30 \b%c 82# EXIF IFD Entry #5: 83>>>>(150.l+62) leshort 0x9000 84>>>>>(150.l+71) byte x %c 85>>>>>(150.l+72) byte x \b.%c 86>>>>>(150.l+73) byte !0x30 \b%c 87# Big-endian 88>>12 string MM 89# IFD 0 Entry #9: 90>>>118 beshort 0x8769 91# EXIF IFD Entry #1: 92>>>>(126.L+14) beshort 0x9000 93>>>>>(126.L+23) byte x %c 94>>>>>(126.L+24) byte x \b.%c 95>>>>>(126.L+25) byte !0x30 \b%c 96# EXIF IFD Entry #3: 97>>>>(126.L+38) beshort 0x9000 98>>>>>(126.L+47) byte x %c 99>>>>>(126.L+48) byte x \b.%c 100>>>>>(126.L+49) byte !0x30 \b%c 101# IFD 0 Entry #10 102>>>130 beshort 0x8769 103# EXIF IFD Entry #3: 104>>>>(138.L+38) beshort 0x9000 105>>>>>(138.L+47) byte x %c 106>>>>>(138.L+48) byte x \b.%c 107>>>>>(138.L+49) byte !0x30 \b%c 108# EXIF IFD Entry #5: 109>>>>(138.L+62) beshort 0x9000 110>>>>>(138.L+71) byte x %c 111>>>>>(138.L+72) byte x \b.%c 112>>>>>(138.L+73) byte !0x30 \b%c 113# IFD 0 Entry #11 114>>>142 beshort 0x8769 115# EXIF IFD Entry #4: 116>>>>(150.L+50) beshort 0x9000 117>>>>>(150.L+59) byte x %c 118>>>>>(150.L+60) byte x \b.%c 119>>>>>(150.L+61) byte !0x30 \b%c 120# Here things get sticky. We can do ONE MORE marker segment with 121# indirect addressing, and that's all. It would be great if we could 122# do pointer arithemetic like in an assembler language. Christos? 123# And if there was some sort of looping construct to do searches, plus a few 124# named accumulators, it would be even more effective... 125# At least we can show a comment if no other segments got inserted before: 126>(4.S+5) byte 0xFE 127>>(4.S+8) string >\0 \b, comment: "%s" 128#>(4.S+5) byte 0xFE \b, comment 129#>>(4.S+6) beshort x \b length=%d 130#>>(4.S+8) string >\0 \b, "%s" 131# Or, we can show the encoding type (I've included only the three most common) 132# and image dimensions if we are lucky and the SOFn (image segment) is here: 133>(4.S+5) byte 0xC0 \b, baseline 134>>(4.S+6) byte x \b, precision %d 135>>(4.S+7) beshort x \b, %dx 136>>(4.S+9) beshort x \b%d 137>(4.S+5) byte 0xC1 \b, extended sequential 138>>(4.S+6) byte x \b, precision %d 139>>(4.S+7) beshort x \b, %dx 140>>(4.S+9) beshort x \b%d 141>(4.S+5) byte 0xC2 \b, progressive 142>>(4.S+6) byte x \b, precision %d 143>>(4.S+7) beshort x \b, %dx 144>>(4.S+9) beshort x \b%d 145# I've commented-out quantisation table reporting. I doubt anyone cares yet. 146#>(4.S+5) byte 0xDB \b, quantisation table 147#>>(4.S+6) beshort x \b length=%d 148#>14 beshort x \b, %d x 149#>16 beshort x \b %d 150 151# HSI is Handmade Software's proprietary JPEG encoding scheme 1520 string hsi1 JPEG image data, HSI proprietary 153 154# From: David Santinoli <david@santinoli.com> 1550 string \x00\x00\x00\x0C\x6A\x50\x20\x20\x0D\x0A\x87\x0A JPEG 2000 image data 156