Update canon tags #1628
Update canon tags #1628
Conversation
|
@alexvanderberkel Sure, but that's easy as pie (if there is no additional munging of data required), just add your dictionary (e.g. |
|
Ah, sorry, I just now got |
| @@ -1046,13 +1190,24 @@ namespace Exiv2 { | |||
| { Tag::root, fujiId, exifId, 0x927c }, | |||
| { Tag::root, canonId, exifId, 0x927c }, | |||
| { Tag::root, canonCsId, canonId, 0x0001 }, | |||
| { Tag::root, canonSiId, canonId, 0x0004 }, | |||
| { Tag::root, canonSiId, canonId, 0x0004 }, | |||
| { Tag::root, canonAf2Id, canonId, 0x0026 }, | |||
There was a problem hiding this comment.
I don't think that this works, because the AF field of canon maker notes is handled in a special way.
See line 1816 where we define a special TiffDecoder for this element.
That's why I think that we can actually remove most (maybe all) the AF stuff from canonmn_int.{hpp,cpp}
There was a problem hiding this comment.
Actually I see the following:
Exif.CanonAf2Id.AFInfoSize SLong 1 1216
Exif.CanonAf2Id.AFAreaMode SLong 1 (65679)
Exif.CanonAf2Id.AFNumPoints SLong 1 293608000
Exif.CanonAf2Id.AFValidPoints SLong 1 293608000
Exif.CanonAf2Id.AFCanonImageWidth SLong 1 468
Exif.CanonAf2Id.AFCanonImageHeight SLong 1 0
Exif.CanonAf2Id.AFImageWidth SLong 1 0
Exif.CanonAf2Id.AFImageHeight SLong 1 0
Exif.CanonAf2Id.AFAreaWidths SLong 1 0
Exif.CanonAf2Id.AFAreaHeights SLong 1 0
Exif.CanonAf2Id.AFXPositions SLong 1 0
Exif.CanonAf2Id.AFYPositions SLong 1 0
Exif.CanonAf2Id.AFPointsInFocus SLong 1 0
Exif.CanonAf2Id.AFPointsSelected SLong 1 0
Exif.CanonAf2Id.AFPointsUnusable SLong 1 0
Exif.CanonAf2Id.AFFineRotation SLong 1 0
And:
Exif.CanonAfC.AFConfigTool SLong 1 0
Exif.CanonAfC.AFTrackingSensitivity SLong 1 2
Exif.CanonAfC.AFAccelDecelTracking SLong 1 2
Exif.CanonAfC.AFPointSwitching SLong 1 0
Exif.CanonAfC.AIServoFirstImage SLong 1 Equal Priority
Exif.CanonAfC.AIServoSecondImage SLong 1 Equal Priority
Exif.CanonAfC.USMLensElectronicMF SLong 1 Disable in AF Mode
Exif.CanonAfC.AFAssistBeam SLong 1 Disable
Exif.CanonAfC.OneShotAFRelease SLong 1 Release Priortiy
Exif.CanonAfC.AutoAFPointSelEOSiTRAF SLong 1 Enable
Exif.CanonAfC.LensDriveWhenAFImpossible SLong 1 Continue Focus Search
Exif.CanonAfC.SelectAFAreaSelectionMode SLong 1 (22359)
Exif.CanonAfC.AFAreaSelectionMethod SLong 1 M-Fn Button
Exif.CanonAfC.OrientationLinkedAF SLong 1 Same for Vert/Horiz Points
Exif.CanonAfC.ManualAFPointSelPattern SLong 1 Stops at AF Area Edges
Exif.CanonAfC.AFPointDisplayDuringFocus SLong 1 Selected (constant)
Exif.CanonAfC.VFDisplayIllumination SLong 1 Auto
Exif.CanonAfC.AFStatusViewfinder SLong 1 Auto
Exif.CanonAfC.InitialAFPointInServo SLong 1 Initial AF Point Selected
There was a problem hiding this comment.
Exif.CanonAfC.SelectAFAreaSelectionMode SLong 1 (22359) --> This tag needs some more work though
There was a problem hiding this comment.
Exif.CanonAf2Id.AFAreaMode SLong 1 (65679) --> This could also need some digging
…l?id=37889 Avoid reading 1 byte off the end when the string does not contain a '\0' byte.
Avoid reading 1 byte off the end when the string does not contain a '\0' byte
Update SECURITY.md
Standardize on conan 1.39.0
Add doc to release workflow
Make fields of DataBuf private
Call Metadatum::print() to increase fuzzing coverage.
Update the `COMMANDS` and `EXIV2 GROUPS, TYPES AND VALUES` sections, including any related sections. Changes: + Add new examples and update text + Fix spelling errors
Original manpage file text is replaced with a note pointing to the Exiv2 website.
Changes: + Update BMFF types in FILE TYPES section + Add text in FILE TYPES explaining formats + Change IPTC datasets and XMP properties to IPTC and XMP tags + Fix text and formatting
|
Hi @alexvanderberkel , there seems to be something really wrong with the status of your branch (it says it has 650 commits 🤯 ) . I would recommend you to either rebase your commits in top of |
did a rebase lately. I think it based on the fact that there was quite some work by a lot of other people over the month. I had to catch up with a lot things. |
|
You have two errors being detected by the test suite in github/981 and redmine/445. github/981 This is easy. You are now identifying three unknown tags: OpticalZoomCode (0x10), AEBBracketValue (0x11) and ControlMode (0x12). Fix is easy and in the patch below. redmine/445 Good and Horrible News Good: You have inflicted damage on the the test file. It runs exiv2 6 times. Somehow, you've lost 5 of the outputs. I've restored the "old" test and now it runs. Horrible: I can't remember what redmine/445 is about, but it's got my finger prints all over it. It concerns Canon AFInfo and should report metadata with the keys such as Exif.Canon.AFInfoSize. All of that metadata has disappeared. I'll have to investigate what you've changed to cause this. I recommend that you submit the patch as a step in the right direction. The errors should drop to 1 (plus the bitch that bugfixTests has failed. Patch: |
|
@piponazo Can I ask you to use your awesome git skills to rebase this project and update this PR. GitHub thinks 368 files have been changed. That's not right. 360+ other files have changes in the code-base since this PR was created last May. I'd like to understand what has been changed by Alex, and then fix the disappearance of all Exif.Canon.AFInfo tags. The Canon.AFInfo decoder is { "*", 0x0026, canonId, &TiffDecoder::decodeCanonAFInfo, 0 /* Exiv2.Canon.AFInfo is read-only */ },`Christoph has added something to the state table: I think that's is involved in causing all the Exif.Canon.AFInfo metadata to disappear. A clean and simple diff be of great value. |
Sure, I will give it a try. I'll create a second branch |
|
@piponazo You are a splendid fellow, @piponazo. Thank You. I suspect we'll open a new PR to deal with the issue that Alex is trying to solve. @hassec has made a very clever observation about modifying the state table to deal with 0x0026. It's possible that my code in #981 can be eliminated. However that's different from the issue that Alex set out to solve. |
|
@clanmills @alexvanderberkel I could successfully move all the commits from @alexvanderberkel to the branch You can see that the latest 3 commits are from my side to fix some compilation issues and tests output expectations: Compilation seems to be fine, but there are some tests failing: |
|
@piponazo Thanks for dealing with this, Luis. I'll look at it tomorrow (Thursday). |
|
@alexvanderberkel Even with Luis' great work, I have a problem building both your branch and Luis new branch. The problem is caused on macOS by libiconv. This was fixed in May last year I can't remember how. There's a saying in English "When you are in a hole, stop digging". We need to back up and open a bug report which precisely explains the issue. I realise it has something to do with Canon tags and ExifTool compatibility. Can you document the issue? Let's ask @kmilos and/or @hassec to deal with it. Fixing this is is not urgent as the issue always has been in the code base. |
|
@piponazo thanks. |
|
The linking issue on the MacMini is caused by brew or macports having installed libiconv in /opt/local. cmake correctly generated the build to use libiconv in the macOS/SDK. The linker finds the wrong library and fails as follows: My fix was to remove the library from /opt/local. It's likely that there's poison in my environment that causes this. env | grep -i lib
PERL5LIB=/Users/rmills/bin
PATH=~/Library/Python/3.8/bin:/usr/local/opt/gettext/bin:.:/Users/rmills/bin:/Users/rmills/bin/macosx:/usr/local/bin:/usr/bin:/usr/sbin:/bin:/sbin:/usr/X11R6/bin:~/Library/Python/3.8/bi:/Library/Developer/Toolchains/swift-latest.xctoolchain/usr/bin:/opt/local/bin:/opt/local/sbin:/opt/pkgconfig/bin
DYLD_LIBRARY_PATH=/Users/rmills/gnu/boost/boost_1_66_0
BOOST_LIBRARYDIR=/Users/rmills/gnu/boost/boost_1_66_0/libs
PYTHONPATH=/System/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages:
PKG_CONFIG_PATH=/usr/local/lib/pkgconfig:/usr/lib/pkgconfig:/opt/local/lib/pkgconfig
620 rmills@rmillsmm-local:~/gnu/github/exiv2/update_canon_tags_2/build $ |
|
@alexvanderberkel. Luis has created a new branch |
|
I'm going to close this (without deleting the branch at the moment). Conversation continues in #2087. @piponazo Thank You very much for your help with this. A little easier to understand the changes in 24 files. I'll update README.md and CONTRIBITING.md at the weekend. I'll make sure the documented test procedure is correct and I'll remove all mention of the |
|
I'm going to delete this branch. We don't need it. #2087 deals with this. |
I think we still need to work a bit on the new Canon tags. For instance, we have these tags identified
However, there not nicely being displayed at the moment if you compare it with information from the exiftool / page.
@kmilos I think we need to add somewhere a prettier printer, isn't that right?