def check_memory_usage(buffered_geo_extent, cell_size):
"""Helper to check if analysis is feasible when extents change.
For simplicity, we will do all our calculations in geocrs.
:param buffered_geo_extent: An extent in the for [xmin, ymin, xmax, ymax]
:type buffered_geo_extent: list
:param cell_size: The size of a cell (assumes in the X direction).
:type cell_size: float
:returns: True if it appears we have enough memory (or we can't compute
it), False if it appears we do not have enough.
:rtype: bool
:raises: A Message containing notes about how much memory is needed
for a single raster and if this is likely to result in an error.
:returns: True if it is supposed that there is sufficient memory,
False if it is supposed that too little memory exists.
:rtype: bool
"""
message = m.Message()
check_heading = m.Heading(tr("Checking available memory"), **PROGRESS_UPDATE_STYLE)
message.add(check_heading)
width = buffered_geo_extent[2] - buffered_geo_extent[0]
height = buffered_geo_extent[3] - buffered_geo_extent[1]
try:
# noinspection PyAugmentAssignment
width = width / cell_size
# noinspection PyAugmentAssignment
height = height / cell_size
except TypeError:
# Could have been a vector layer for example
reason = tr("Computed cellsize was None. Memory check currently only works " "for raster input layers.")
message.add(reason)
send_message(message)
return True # assume enough mem since we have no vector check logic
bullet_list = m.BulletedList()
bullet = m.Paragraph(m.ImportantText(tr("Width: ")), str(width))
bullet_list.add(bullet)
bullet = m.Paragraph(m.ImportantText(tr("Height: ")), str(height))
bullet_list.add(bullet)
bullet = m.Paragraph(m.ImportantText(tr("Cell Size: ")), str(cell_size))
bullet_list.add(bullet)
message.add(bullet_list)
# Compute mem requirement in MB (assuming numpy uses 8 bytes by per
# cell) see this link:
# http://stackoverflow.com/questions/11784329/
# python-memory-usage-of-numpy-arrays
# Also note that the on-disk requirement of the clipped tifs is about
# half this since the tifs as in single precision,
# whereas numpy arrays are in double precision.
requirement = (width * height * 8) / 1024 / 1024
try:
free_memory = get_free_memory()
except ValueError:
error_heading = m.Heading(tr("Memory check error"), **WARNING_STYLE)
error_message = tr("Could not determine free memory")
message.add(error_heading)
message.add(error_message)
send_message(message)
LOGGER.exception(message)
return True # still let the user try to run their analysis
# We work on the assumption that if more than 10% of the available
# memory is occupied by a single layer we could run out of memory
# (depending on the impact function). This is because multiple
# in memory copies of the layer are often made during processing.
warning_limit = 10
usage_indicator = (float(requirement) / float(free_memory)) * 100
counts_message = tr("Memory requirement: about %d mb per raster layer (" "%d mb available)") % (
requirement,
free_memory,
)
usage_message = tr("Memory used / available: %d/%d") % (usage_indicator, warning_limit)
message.add(counts_message)
message.add(usage_message)
if warning_limit <= usage_indicator:
warning_heading = m.Heading(tr("Potential memory issue"), **WARNING_STYLE)
warning_message = tr(
"There may not be enough free memory to run this analysis. You can"
" attempt to run the analysis anyway, but note that your computer "
"may become unresponsive during execution, and / or the analysis "
"may fail due to insufficient memory. Proceed at your own risk."
)
suggestion_heading = m.Heading(tr("Suggestion"), **INFO_STYLE)
suggestion = tr(
"Try zooming in to a smaller area or using a raster layer with a "
"coarser resolution to speed up execution and reduce memory "
"requirements. You could also try adding more RAM to your "
"computer."
)
message.add(warning_heading)
message.add(warning_message)
message.add(suggestion_heading)
message.add(suggestion)
send_message(message)
LOGGER.info(message.to_text())
return False
send_message(message)
LOGGER.info(message.to_text())
return True
def check_memory_usage(buffered_geo_extent, cell_size):
"""Helper to check if analysis is feasible when extents change.
For simplicity, we will do all our calculations in geocrs.
:param buffered_geo_extent: An extent in the for [xmin, ymin, xmax, ymax]
:type buffered_geo_extent: list
:param cell_size: The size of a cell (assumes in the X direction).
:type cell_size: float
:returns: True if it appears we have enough memory (or we can't compute
it), False if it appears we do not have enough.
:rtype: bool
:raises: A Message containing notes about how much memory is needed
for a single raster and if this is likely to result in an error.
:returns: True if it is supposed that there is sufficient memory,
False if it is supposed that too little memory exists.
:rtype: bool
"""
message = m.Message()
check_heading = m.Heading(tr('Checking available memory'),
**PROGRESS_UPDATE_STYLE)
message.add(check_heading)
width = buffered_geo_extent[2] - buffered_geo_extent[0]
height = buffered_geo_extent[3] - buffered_geo_extent[1]
try:
# noinspection PyAugmentAssignment
width = width / cell_size
# noinspection PyAugmentAssignment
height = height / cell_size
except TypeError:
# Could have been a vector layer for example
reason = tr(
'Computed cellsize was None. Memory check currently only works '
'for raster input layers.')
message.add(reason)
send_dynamic_message(dispatcher.Anonymous, message)
return True # assume enough mem since we have no vector check logic
bullet_list = m.BulletedList()
bullet = m.Paragraph(m.ImportantText(tr('Width: ')), str(width))
bullet_list.add(bullet)
bullet = m.Paragraph(m.ImportantText(tr('Height: ')), str(height))
bullet_list.add(bullet)
bullet = m.Paragraph(m.ImportantText(tr('Cell Size: ')), str(cell_size))
bullet_list.add(bullet)
message.add(bullet_list)
# Compute mem requirement in MB (assuming numpy uses 8 bytes by per
# cell) see this link:
# http://stackoverflow.com/questions/11784329/
# python-memory-usage-of-numpy-arrays
# Also note that the on-disk requirement of the clipped tifs is about
# half this since the tifs as in single precision,
# whereas numpy arrays are in double precision.
requirement = ((width * height * 8) / 1024 / 1024)
try:
free_memory = get_free_memory()
except ValueError:
error_heading = m.Heading(tr('Memory check error'), **WARNING_STYLE)
error_message = tr('Could not determine free memory')
message.add(error_heading)
message.add(error_message)
send_dynamic_message(dispatcher.Anonymous, message)
LOGGER.exception(message)
return True # still let the user try to run their analysis
# We work on the assumption that if more than 10% of the available
# memory is occupied by a single layer we could run out of memory
# (depending on the impact function). This is because multiple
# in memory copies of the layer are often made during processing.
warning_limit = 10
usage_indicator = (float(requirement) / float(free_memory)) * 100
counts_message = tr('Memory requirement: about %d mb per raster layer ('
'%d mb available)') % (requirement, free_memory)
usage_message = tr('Memory used / available: %d/%d') % (usage_indicator,
warning_limit)
message.add(counts_message)
message.add(usage_message)
if warning_limit <= usage_indicator:
warning_heading = m.Heading(tr('Potential memory issue'),
**WARNING_STYLE)
warning_message = tr(
'There may not be enough free memory to run this analysis. You can'
' attempt to run the analysis anyway, but note that your computer '
'may become unresponsive during execution, and / or the analysis '
'may fail due to insufficient memory. Proceed at your own risk.')
suggestion_heading = m.Heading(tr('Suggestion'), **SUGGESTION_STYLE)
suggestion = tr(
'Try zooming in to a smaller area or using a raster layer with a '
'coarser resolution to speed up execution and reduce memory '
'requirements. You could also try adding more RAM to your '
'computer.')
message.add(warning_heading)
message.add(warning_message)
message.add(suggestion_heading)
message.add(suggestion)
send_dynamic_message(dispatcher.Anonymous, message)
# LOGGER.info(message.to_text())
return False
send_dynamic_message(dispatcher.Anonymous, message)
# LOGGER.info(message.to_text())
return True