Tim Jenness added a comment - 22/Feb/21 11:14 PM Fred Moolekamp thanks for doing the review. As described in the ticket description there are multiple parts here: Adding the new astrometadata command to replace translate_header.py . Adding JSON sidecar and indexing creation $ astrometadata -h Usage: astrometadata [OPTIONS] COMMAND [ARGS]...   Options: --log-level [CRITICAL|ERROR|WARNING|INFO|DEBUG] Python logging level to use. --traceback / --no-traceback Give detailed trace back when any errors encountered.   -p, --packages TEXT Python packages to import to register additional translators. This is in addition to any packages specified in the METADATA_TRANSLATORS environment variable (colon-separated python module names).   -h, --help Show this message and exit.   Commands: dump Dump data header to standard out in YAML format. translate Translate metadata in supplied files and report. write-index Write JSON index file of ObervationInfo for entire...   write-metadata-index Write JSON index file of original data headers for...   write-metadata-sidecar Write metadata sidecar files with ObservationInfo... write-sidecar Write JSON sidecar files with ObservationInfo... This required that I added a "diff" merging mode to merge_headers function. JSON files can be either raw metadata or translated metadata. The mode is stored in the JSON file as either "obsInfo" or "metadata". The code for reading from a sidecar or index file reads that mode key. Index files share common metadata into a __COMMON__ key which is merged in on read. This allows the index for a directory of a single exposure to only list the detector changes per file. There is some debate over the subcommands. At present I've tried to split them up into distinct commands for writing sidecar files vs index files and for dumping raw or fixed metadata to standard out. The code for writing sidecars is distinct from that writing index files and for writing an index file there is an option to write a standalone index rather than one per directory. It's certainly possible that people would prefer a single index writing command that took an argument to decide if translated or raw content should go in the index file. Similarly when writing sidecar JSON files. I'm trying to avoid combining all the subcommands back into a single one with lots of options that can do sidecar, indexing, and header dumping to standard out.

Fred Moolekamp added a comment - 01/Mar/21 9:06 PM Thanks for making the suggested changes. This looks good to me.