Change combine_metadata to average any 'time' fields #473
Conversation
Observation time, which was used previously, varies between bands and can make compositing difficult (no resulting start_time). This uses the scheduled time of the image which should be the same for all bands for a particular region/sector.
|
I just realized that the This makes me think we need to modify the metadata merging functions so that they are semi-aware of |
Codecov Report
@@ Coverage Diff @@
## master #473 +/- ##
==========================================
+ Coverage 73.57% 73.96% +0.38%
==========================================
Files 135 136 +1
Lines 17833 17954 +121
==========================================
+ Hits 13121 13280 +159
+ Misses 4712 4674 -38
Continue to review full report at Codecov.
|
satpy/dataset.py
Outdated
| if sys.version_info < (3, 3): | ||
| # timestamp added in python 3.3 | ||
| import time | ||
| timestamp_func = lambda x: time.mktime(x.timetuple()) |
There was a problem hiding this comment.
E731 do not assign a lambda expression, use a def
satpy/dataset.py
Outdated
| import time | ||
| timestamp_func = lambda x: time.mktime(x.timetuple()) | ||
| else: | ||
| timestamp_func = lambda x: x.timestamp() |
There was a problem hiding this comment.
E731 do not assign a lambda expression, use a def
satpy/dataset.py
Outdated
| return datetime.fromtimestamp(sum(total) / len(total)) | ||
|
|
||
|
|
||
| def combine_metadata(*metadata_objects, average_times=True): |
There was a problem hiding this comment.
E999 SyntaxError: invalid syntax
| return datetime.fromtimestamp(sum(total) / len(total)) | ||
|
|
||
|
|
||
| def combine_metadata(*metadata_objects, **kwargs): |
There was a problem hiding this comment.
why not def combine_metadata(*metadata_objects, average_times=True):?
There was a problem hiding this comment.
Read the comment below. That syntax isn't valid in python 2. I had it that way first and the tests failed. Unless you're saying I should drop python 2 support now... 😜
|
Which time attributes specifically? The yaml_reader always overwrites start and end time with the first and last times of the file handlers. |
Edit: As discussed below and on slack, this PR has been modified from its original purpose. It now does as the title states by modifying
combine_metadatainstead of modifying theahi_hsdreader with a workaround (using scheduled time for start time). The original PR description is below:Observation time, which was used previously, varies between bands and
can make compositing difficult (no resulting start_time). This uses the
scheduled time of the image which should be the same for all bands for
a particular region/sector.
I'm still going to add some documentation to the top of the reader module to mention this time difference.
git diff origin/master -- "*py" | flake8 --diff