Commit b65312da authored by Ophélie Gagnard's avatar Ophélie Gagnard

slapos_metadata_transform_agent: fix data conversion in Stream to Array so...

slapos_metadata_transform_agent: fix data conversion in Stream to Array so json conversion don't fail on special characters
parent 9ba3e103
...@@ -4,7 +4,7 @@ ...@@ -4,7 +4,7 @@
""" """
context.log(reference)
if not reference: if not reference:
raise ValueError('reference is None') raise ValueError('reference is None')
......
# TODO
# + review the code
# + put a slash ('/') at the end of the directories' name in MCA
import numpy as np import numpy as np
import re import re
import json import json
import os.path import os.path
def get_end_and_json_list(start, in_data_stream, max_scan_size = 1024 ** 3): def get_end_and_json_list(start, in_data_stream, chunk_size = 4 * 1024 * 1024):
""" """
Determine the end index of a scan (i.e. the complete scan of a filesystem). Determine the end index of a scan (i.e. the complete scan of a filesystem).
Return the end index and a list of strings assumed to be valid json strings. Return the end index and a list of strings assumed to be valid json strings.
(This assumes the scanning of the file system produces correct strings.) (This assumes the scanning of the file system produces correct strings.)
""" """
chunk_size = 1024 # * 1024 # assume max path size is 4096 and add 4096 for the rest
while chunk_size < max_scan_size: max_remaining_for_eol = 8192
end = min(start+chunk_size, in_data_stream.getSize()) end = min(start + chunk_size + max_remaining_for_eol, in_data_stream.getSize())
unpacked, end = in_data_stream.readMsgpackChunkList(start, end) #unpacked, get_end = in_data_stream.readMsgpackChunkList(start, end) # LEGACY
unpacked = in_data_stream.readChunkList(start, end) # CURRENT
unpacked_string = "".join(unpacked) # CURRENT
# extending the current chunk until the next end of line,
# so json remains valid
if end < in_data_stream.getSize():
new_end_index = chunk_size
while unpacked_string[new_end_index] != '\n': # CURRENT
#while unpacked[new_end_index] != ord('\n'): # LEGACY
new_end_index += 1
end = start + new_end_index + 1
""" not useful anymore? # LEGACY / TOCLEAN
unpacked = unpacked[:end]
try: # efficient but does not prevent errors
raw_data_string = ''.join(map(chr, unpacked)) raw_data_string = ''.join(map(chr, unpacked))
end_scan_regexp = re.compile('.*?\[fluentbit_end\]\n', re.DOTALL) except: # not efficient but does prevent value errors and type errors
complete_scan = end_scan_regexp.match(raw_data_string) cleaned_unpacked = []
if not complete_scan: for i in unpacked:
chunk_size *= 2 if (type(i) == type(1) # only ints
continue and 0 <= i and i < 256): # i in range(256)
chunk_size = len(complete_scan.group()) + 1 cleaned_unpacked.append(i)
raw_data_string = ''.join(map(chr, cleaned_unpacked))
line_list = complete_scan.group().splitlines() #"""
timestamp_json_regexp = re.compile('.*?:(.*?)\[(.*)\]') raw_data_string = unpacked_string[:end] # CURRENT
json_string_list = [timestamp_json_regexp.match(line).group(2) end_scan_regexp = re.compile('.*?\[fluentbit_end\]\n', re.DOTALL)
for line in line_list scan_end = end_scan_regexp.match(raw_data_string)
if len(timestamp_json_regexp.match(line).groups()) == 2] if not scan_end:
is_end_of_scan = False
return start + chunk_size, json_string_list else:
is_end_of_scan = True
raise NotImplementedError( end = start + len(scan_end.group()) + 1
'Scan too big (<= ' raw_data_string = raw_data_string[:len(scan_end.group())]
+ str(chunk_size / 1024 ** 2)
+ ' Mo) or no "fluentbit_end" in the Data Stream.') context.log("DEBUG 020: len(raw_data_string) =", len(raw_data_string))
context.log("DEBUG 020: type(raw_data_string) =", type(raw_data_string))
def get_triplet_list(json_string_list): context.log("DEBUG 020: type(raw_data_string[0]) =", type(raw_data_string[0]))
line_list = raw_data_string.splitlines()
context.log("DEBUG 020: len(line_list) =", len(line_list))
timestamp_json_regexp = re.compile(r'.*?:(.*?)\[(.*)\]')
json_string_list = [timestamp_json_regexp.match(line).group(2)
for line in line_list
if (timestamp_json_regexp.match(line) and (len(timestamp_json_regexp.match(line).groups()) == 2))]
return end, json_string_list, is_end_of_scan
def get_triplet_list(json_string_list, is_end_of_scan):
""" """
Parse unpacked and return a triplet list: (path, slice1, slice2). Parse unpacked and return a triplet list: (path, slice1, slice2).
path is the path of the processed file, slice1 and slice2 are two parts path is the path of the processed file, slice1 and slice2 are two parts
...@@ -50,10 +71,26 @@ def get_triplet_list(json_string_list): ...@@ -50,10 +71,26 @@ def get_triplet_list(json_string_list):
NOTE: timestamps are parsed in case they are needed for future improvement NOTE: timestamps are parsed in case they are needed for future improvement
but they are not used at the moment. but they are not used at the moment.
""" """
json_string_list = json_string_list[:-1] if is_end_of_scan:
tmp_data_list = [json.loads(json_string) for json_string in json_string_list] # this lign deletes the "fluentbit_end" at the end of a scan
data_list = [] # because it is not valid json
json_string_list = json_string_list[:-1]
tmp_data_list = []
fail_counter = 0
for json_string in json_string_list:
tmp_data_list.append(json.loads(json_string))
""" CURRENT
try:
tmp_data_list.append(json.loads(json_string))
except:
context.log('FAILING json_string:', json_string) # DEBUG
fail_counter += 1
pass
if fail_counter > 0:
context.log('FAILED json_string:', fail_counter) # DEBUG
#"""
data_list = []
for data in tmp_data_list: for data in tmp_data_list:
in_list = False in_list = False
if ('path' in data) and exclude_file_list: if ('path' in data) and exclude_file_list:
...@@ -103,32 +140,33 @@ out_data_array.setPublicationSectionList(in_data_stream.getPublicationSectionLis ...@@ -103,32 +140,33 @@ out_data_array.setPublicationSectionList(in_data_stream.getPublicationSectionLis
if 'file_system_image/reference_image' in in_data_stream.getPublicationSectionList(): if 'file_system_image/reference_image' in in_data_stream.getPublicationSectionList():
if out_data_array.getValidationState() == 'draft': if out_data_array.getValidationState() == 'draft':
out_data_array.validate() out_data_array.validate()
if not out_data_array.getCausality(): if not out_data_array.getCausality():
ingestion_line = in_data_stream.getAggregateRelatedValue(portal_type='Data Ingestion Line') ingestion_line = in_data_stream.getAggregateRelatedValue(portal_type='Data Ingestion Line')
resource = ingestion_line.getResource() ressource = ingestion_line.getRessource()
exclude_file_list = ingestion_line.getResourceValue().DataProduct_getExcludeFileList() exclude_file_list = ingestion_line.getRessourceValue().DataProduct_getExcludeFileList()
out_data_array.edit(causality=resource) out_data_array.edit(causality=ressource)
# DEBUG advice: running the script on the whole Data Stream
# (i.e. restarting from the beginning):
#progress_indicator.setIntOffsetIndex(0)
start = progress_indicator.getIntOffsetIndex() start = progress_indicator.getIntOffsetIndex()
end = in_data_stream.getSize() end = in_data_stream.getSize()
if start >= end: if start >= end:
return return
end, json_string_list = get_end_and_json_list(start, in_data_stream) end, json_string_list, is_end_of_scan = get_end_and_json_list(start, in_data_stream)
triplet_list = get_triplet_list(json_string_list) context.log("DEBUG 020: len(json_string_list) =", len(json_string_list))
triplet_list = get_triplet_list(json_string_list, is_end_of_scan)
context.log("DEBUG 020: len(triplet_list) =", len(triplet_list))
uid_list = get_uid_list(triplet_list, in_data_stream) uid_list = get_uid_list(triplet_list, in_data_stream)
context.log("DEBUG 020: len(uid_list) =", len(uid_list))
uid_ndarray = create_ndarray(uid_list) uid_ndarray = create_ndarray(uid_list)
context.log("len(uid_ndarray) =", len(uid_ndarray))
context.log("start =", start)
context.log("end =", end)
if start == 0: if start == 0:
zbigarray = None zbigarray = None
else: else:
zbigarray = out_data_array.getArray() zbigarray = out_data_array.getArray()
# DEBUG advice: reset the Data Array:
# NOTE: currently, the Data Array is reset each time a scan is processed
zbigarray = None
if zbigarray is None: if zbigarray is None:
zbigarray = out_data_array.initArray(shape=(0,), dtype='int64') zbigarray = out_data_array.initArray(shape=(0,), dtype='int64')
......
...@@ -149,7 +149,7 @@ ...@@ -149,7 +149,7 @@
</item> </item>
<item> <item>
<key> <string>serial</string> </key> <key> <string>serial</string> </key>
<value> <string>995.60665.3511.20872</string> </value> <value> <string>997.41827.9954.54391</string> </value>
</item> </item>
<item> <item>
<key> <string>state</string> </key> <key> <string>state</string> </key>
...@@ -167,7 +167,7 @@ ...@@ -167,7 +167,7 @@
</tuple> </tuple>
<state> <state>
<tuple> <tuple>
<float>1636043117.96</float> <float>1642690808.92</float>
<string>UTC</string> <string>UTC</string>
</tuple> </tuple>
</state> </state>
......
return [ return [
u'/dracut-state.sh', u'/dracut-state.sh',
u'/tmp/dhclient', u'/tmp/dhclient',
u'/tmp/net',
u'/etc/machine-id', u'/etc/machine-id',
u'/etc/flb.conf' u'/etc/flb.conf'
] ]
...@@ -31,7 +31,7 @@ for movement in portal_catalog(query = query): ...@@ -31,7 +31,7 @@ for movement in portal_catalog(query = query):
continue continue
if movement.DataIngestionLine_hasMissingRequiredItem(): if movement.DataIngestionLine_hasMissingRequiredItem():
raise ValueError("Transformation requires movement to have " + raise ValueError("Transformation requires movement to have " +
"aggregated data ingestion batch") "aggregated data ingestion batch %s" % movement.getRelativeUrl())
delivery = movement.getParentValue() delivery = movement.getParentValue()
data_supply = delivery.getSpecialiseValue(portal_type="Data Supply") data_supply = delivery.getSpecialiseValue(portal_type="Data Supply")
data_supply_list = delivery.getSpecialiseValueList(portal_type="Data Supply") data_supply_list = delivery.getSpecialiseValueList(portal_type="Data Supply")
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment