Ensures that Unicode tags are correctly written to files even in non-Unicode locales. Thanks to Trevor Davis for the excellent, as usual, issue description and PR.
Improves the error message returned to the user when the call to Exiftool finds no files to act upon. In addition to more gracefully relaying the error message emitted by Exiftool, such calls now return a value of NULL. Thanks to jcblum for reporting this issue.
config_file= and common_args= to exif_call(). These are passed on to exiftool options -config and -common_args respectively, neither of which was supported here before due to their not being allowed in the -@ ARGFILE option used ‘behind the scenes’ by all calls to exif_call().install_exiftool(). Formerly, the executable was installed into the directory returned by system.file("exiftool", package = "exiftoolr"), which will not always be writable (as, e.g., when the package is installed by an admin in the shared library of a multi-user server, and hence is owned by “root”). Now, the executable is by default installed to the directory given by backports::R_user_dir(), which should be more generally writable.Fixes an issue that caused configure_exiftoolr() (and thus essentially all exiftoolr functionality) to fail on Windows machines that do not have Perl installed in a location findable by the exiftoolr:::configure_perl(). Thanks to Tom Yamashita for reporting this issue.
Adds tests (using the tinytest package).
configure_exiftoolr() to fail if the path to the ExifTool executable on a user’s computer contained any spaces. Now configure_exiftoolr() should work even if there are spaces in the path(s) to the user’s installation of Perl and/or their ExifTool executable or library. Thanks to Lafont Rapnouil Tristan for reporting the issue.exif_call() in 0.1.5 from system() to system2(). Thanks to Daniel Baumgartner for bringing this to my attention.pipeline option to exif_read(), which allows users to direct the exif executable to output results in csv rather than json format. This is helpful for use with images whose metadata contains non-UTF-8-encoded characters. As is documented here, ExifTool’s JSON output does not properly handle non-UTF-8 character sets. Setting pipeline="csv" ensures that non-UTF-8 character sets are properly handled, as demonstrated in a new example in ?exif_read.Fixed exif_read() to now allow repeated elements in args=. This can be necessary when (to take one example) a user needs to separately specify the encoding used in the image file names and in the tags respectively. Now, a call like the following works as it should:
exif_read(path = "myimage.jpg",
args = c("-charset", "exiftool=cp850", "-charset", "filename=cp1250"))configure_exiftoolr() to find a local installation of ExifTool now throws an error rather than just a warning, preventing infinite recursion by exif_read() and exif_version().