Files
sehistory/influx.py

683 lines
26 KiB
Python

# *****************************************************************************
# Copyright (c) 2024 ff by the module authors
#
# This program is free software; you can redistribute it and/or modify it under
# the terms of the GNU General Public License as published by the Free Software
# Foundation; either version 2 of the License, or (at your option) any later
# version.
#
# This program is distributed in the hope that it will be useful, but WITHOUT
# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
# FOR A PARTICULAR PURPOSE. See the GNU General Public License for more
# details.
#
# You should have received a copy of the GNU General Public License along with
# this program; if not, write to the Free Software Foundation, Inc.,
# 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
#
# Module authors:
# Markus Zolliker <markus.zolliker@psi.ch>
#
# *****************************************************************************
import re
import time
from pathlib import Path
from configparser import ConfigParser
from datetime import datetime, timezone
from math import floor, ceil
from influxdb_client import InfluxDBClient, BucketRetentionRules, Point
from influxdb_client.client.write_api import SYNCHRONOUS
DAY = 24 * 3600
YEAR = 366 * DAY
# write_precision from digits after decimal point
TIME_PRECISION = ['s'] + ['ms'] * 3 + ['us'] * 3 + ['ns'] * 3
UNDEF = '<undef>'
try:
parse_time = datetime.fromisoformat
except AttributeError:
from dateutil.parser import parse as parse_time
def to_time(v):
return parse_time(v).timestamp()
def to_iso(t):
return datetime.fromtimestamp(t, timezone.utc).isoformat().replace('+00:00', 'Z')
class PrettyFloat(float):
"""saves bandwidth when converting to JSON
a lot of numbers originally have a fixed (low) number of decimal digits.
as the binary representation is not exact, it might happen, that a
lot of superfluous digits are transmitted:
str(1/10*3) == '0.30000000000000004'
str(PrettyFloat(1/10*3)) == '0.3'
"""
def __new__(cls, value):
return None if value == '-0' else super().__new__(cls, value)
def __repr__(self):
return '%.15g' % self
class Converters(dict):
def __init__(self, datatypes):
super().__init__((i, getattr(self, f"cvt_{d.split(':')[0]}"))
for i, d in enumerate(datatypes) if i > 2)
def as_tuple(self, row):
"""get selected columns as tuple"""
return tuple(f(row[i]) for i, f in self.items())
cvt_double = staticmethod(PrettyFloat)
@staticmethod
def cvt_string(value):
return value
@staticmethod
def cvt_long(value):
return int(value)
@staticmethod
def cvt_dateTime(value):
return to_time(value)
@staticmethod
def cvt_boolean(value):
return value == 'true'
cvt_unsigned_long = cvt_duration = cvt_long
class Table(list):
"""a list of tuples with meta info"""
def __init__(self, tags={}, key_names=(), column_names=(), rows=None):
super().__init__()
self.tags = tags
self.key_names = key_names
self.column_names = column_names
if rows:
self[:] = rows
def to_csv_rows(self, timeoffset=0, sep='\t', none='none', float_format='%.15g'):
for row in self:
result = ['%.15g' % (row[0] - timeoffset)]
for value in row[1:]:
try:
result.append(float_format % value)
except TypeError:
if value is None:
result.append(none)
else:
result.append(str(value).replace(sep, ' '))
yield sep.join(result)
class Single(Table):
"""a single row of a table, as a list with meta info"""
def __init__(self, tags={}, key_names=(), column_names=(), rows=None):
super().__init__(tags, key_names, column_names)
if rows:
single_row, = rows
self[:] = single_row
def summarize_tags(curves, remove_multiple=False):
"""summarize tags
:param curves: list of curves (type Table)
:param remove_multiple: True: remove non-unique values
:return: dict <key> of comma separated values
"""
result = {}
for curve in curves:
for k, v in curve.tags.items():
result.setdefault(k, set()).add(str(v))
if remove_multiple:
return {k: ','.join(v) for k, v in result.items() if len(v) == 1}
return {k: ','.join(v) for k, v in result.items()}
class RegExp(str):
"""indicates, tht this string should be treated as regexp
Usage: RegExp(<pattern>)
when used in InfluxDBWrapper.query, uses Go regexp syntax!
"""
def append_wildcard_filter(msg, key, names):
patterns = []
for pattern in names:
if isinstance(pattern, RegExp):
patterns.append(pattern)
else:
# patterns.append('[^.]*'.join(re.escape(v) for v in pattern.split('*')))
patterns.append('.*'.join(re.escape(v) for v in pattern.split('*')))
if patterns:
pattern = '|'.join(patterns)
msg.append(f'|> filter(fn:(r) => r.{key} =~ /^({pattern})$/)')
class CurveDict(dict):
def __missing__(self, key):
return []
def abs_range(start=None, stop=None):
now = time.time()
if start is None: # since ever
start = 0
elif start < YEAR:
start = int(now + start)
if stop is None:
stop = int(now + YEAR)
elif stop < YEAR:
stop = ceil(now + stop)
return start, stop
def round_range(start, stop, interval=None):
interval = max(1, interval or 0)
start = floor(floor(start) // interval * interval)
stop = ceil(ceil(stop // interval) * interval)
return start, stop
class InfluxDBWrapper:
"""Wrapper for InfluxDB API 2.0.
based on nicos.services.cache.database.influxdb
(work of Konstantin Kholostov <k.kholostov@fz-juelich.de>)
"""
_update_queue = None
_write_api_write = None
def __init__(self, uri=None, token=None, org=None, bucket=None, access='readonly'):
"""initialize
:param uri: the uri for the influx DB or a name to look up in ~/.sehistory
:param token: the token
:param org: the organisation
:param bucket: the bucket name
:param access: 'readonly', 'write' (RW) or 'create' (incl. RW)
"""
if ':' in uri:
args = uri, token, org, bucket
else:
parser = ConfigParser()
parser.optionxform = str
parser.read([Path('~').expanduser() / '.sehistory'])
section = parser[uri]
args = [section[k] for k in ('uri', 'token', 'org', 'bucket')]
self._url, self._token, self._org, self._bucket =args
self._client = InfluxDBClient(url=self._url, token=self._token,
org=self._org)
if access != 'readonly':
self.enable_write_access()
self._deadline = 0
self.set_time_precision(3)
self.add_new_bucket(self._bucket, access == 'create')
self._write_buffer = []
self._alias = {}
print('InfluxDBWrapper', self._url, self._org, self._bucket)
def enable_write_access(self):
self._write_api_write = self._client.write_api(write_options=SYNCHRONOUS).write
def set_time_precision(self, digits):
self.timedig = max(0, min(digits, 9))
self._write_precision = TIME_PRECISION[self.timedig]
def disconnect(self):
self.flush()
self._client.close()
def get_bucket_names(self):
bucket_names = []
buckets = self._client.buckets_api().find_buckets().buckets
for bucket in buckets:
bucket_names.append(bucket.name)
return bucket_names
def add_new_bucket(self, bucket_name, create):
if bucket_name not in self.get_bucket_names():
if not create:
raise ValueError(f'unknown bucket {bucket_name}')
retention_rules = BucketRetentionRules(type='expire',
every_seconds=0)
self._client.buckets_api().create_bucket(
bucket_name=bucket_name,
retention_rules=retention_rules, org=self._org)
def get_measurements(self):
return [r['_value'] for t in self._client.query_api().query(f"""
import "influxdata/influxdb/schema"
schema.measurements(bucket: "{self._bucket}")""") for r in t]
def delete_measurement(self, measurement, start=None, stop=None):
delete_api = self._client.delete_api()
start, stop = abs_range(start, stop)
if stop is None:
stop = time.time() + DAY
delete_api.delete(to_iso(start), to_iso(stop), f'_measurement="{measurement}"',
bucket=self._bucket, org=self._org)
def delete_all_measurements(self, measurements=None, start=0, stop=None):
if measurements is None:
measurements = self.get_measurements()
for meas in measurements:
self.delete_measurement(meas, start, stop)
def _get_rows(self, reader, as_tuple, first_row):
row = first_row
tableno = row[2]
try:
while 1:
if row[0]:
first_row[:] = row
return
if row[2] != tableno:
# table id changed: new table, store first row for next call
first_row[:] = row
return
yield as_tuple(row)
row = next(reader)
if not row:
raise ValueError('EMPTY')
except StopIteration:
first_row.clear() # indicate end of data
# query the database
def query(self, start=None, stop=None, interval=None, single=None, columns=None, **tags):
"""Returns queried data as InfluxDB tables
:param start: start time (default: since ever)
:param stop: end time (default: eternity = 1 year in the future)
:param interval: if set an aggregation filter will be applied. This will
return only the latest values per time interval in seconds.
:param single: when True (or 1), only the last value within the interval is returned
(for any existing combinations of tags!)
single=-1: return the first value instead
:param columns: if given, return only these columns (in addition to '_time' and '_value')
:param tags: selection criteria:
<tag>=None
return records independent of this tag.
the obtained value will be contained in the result dicts key
<tag>=[<value1>, <value2>, ...]
return records where tag is one from the list.
the obtained value is contained in the result dicts key
<tag>=<value>:
return only records with given tag matching <value>.
the obtained value is contained in the result dicts key only
if the value is an instance of RegExp or when it contains an asterisk ('*')
:return: a dict <tuple of key values> of <Table instance>
Table is an extension of list, with some meta info
"""
result = {}
for rows, key, props in self.query_gen(start, stop, interval, single, columns, **tags):
if single:
result[key] = Single(*props, rows=rows)
else:
table = Table(*props, rows=rows)
table.sort()
result[key] = table
return result
def query_gen(self, start=None, stop=None, interval=None, single=None, columns=None, **tags):
"""Returns queried data as InfluxDB as a generator
argument description: see query methods
:return: an iterator of (rows, key, (tags, key_names, column_names))
remark: rows not consumed in between iteration steps are lost
using this generator version does reduce memory usage
"""
self.flush()
start, stop = round_range(*abs_range(start, stop))
msg = [f'from(bucket:"{self._bucket}")',
f'|> range(start: {start}, stop: {stop})']
keylist = []
dropcols = ['_start', '_stop']
fixed_tags = {}
for key, crit in tags.items():
if crit is None:
keylist.append(key)
continue
if isinstance(crit, str):
if isinstance(crit, RegExp) or '*' in crit:
keylist.append(key)
append_wildcard_filter(msg, key, [crit])
continue
fixed_tags[key] = crit
dropcols.append(key)
crit = f'"{crit}"'
elif isinstance(crit, (int, float)):
fixed_tags[key] = crit
dropcols.append(key)
else:
try:
keylist.append(key)
append_wildcard_filter(msg, key, crit)
continue
except Exception:
raise ValueError(f'illegal value for {key}: {crit}')
msg.append(f'|> filter(fn:(r) => r.{key} == {crit})')
if single:
if single < 0:
msg.append('|> first(column: "_time")')
else:
msg.append('|> last(column: "_time")')
if interval:
msg.append(f'|> aggregateWindow(every: {interval:g}s, fn: last, createEmpty: false)')
if columns is None:
msg.append(f'''|> drop(columns:["{'","'.join(dropcols)}"])''')
else:
columns = ['_time', '_value'] + list(columns)
msg.append(f'''|> keep(columns:["{'","'.join(columns + keylist)}"])''')
msg = '\n'.join(msg)
# print(msg)
self.msg = msg
try:
reader = self._client.query_api().query_csv(msg)
except Exception:
print(msg)
raise
try:
row = next(reader)
except StopIteration:
return
converters = key_dict = table_properties = None # make IDE happy
while 1:
header = {}
if row[0]: # this is a header
header[row[0]] = row
for row in reader:
if row:
if not row[0]:
break
header[row[0]] = row
else:
return # this should not happen
# we are now at the row with the column names
column_names = row
converters = Converters(header['#datatype'])
group = header['#group']
keys = {k: None for k in keylist}
for col, (name, grp) in enumerate(zip(column_names, group)):
if grp != 'true':
continue
if columns is None or name in keys:
keys[name] = col, converters.pop(col)
none_keys = [k for k, v in keys.items() if v is None]
if none_keys:
for k in none_keys:
keys.pop(k)
# break
row = next(reader)
# we are at the first data row
key_dict = {n: f(row[i]) for n, (i, f) in keys.items()}
column_keys = tuple(column_names[i] for i in converters)
table_properties = {**fixed_tags, **key_dict}, tuple(keys), column_keys
key = tuple(key_dict.values())
row = list(row) # copy row, as it will be modified
rows = self._get_rows(reader, converters.as_tuple, row)
yield rows, key, table_properties
# consume unused rows
consumed = sum(1 for _ in rows)
if consumed:
print('skip', consumed, 'rows')
if not row: # reader is at end
return
def curves(self, start=None, stop=None, measurement=('*.value', '*.target'), field='float',
interval=None, add_prev=3600, add_end=False, merge=None, pivot=False, **kwds):
"""get curves
:param start: start time (default: since ever)
:param stop: end time (default: eternity = 1 year in the future)
:param measurement: '<module>.<parameter>' (default: ['*.value', '*.target'])
:param field: default 'float' (only numeric curves)
:param interval: if given, the result is binned
:param add_prev: amount of time to look back for the last previous point (default: 1 hr)
:param add_end: whether to add endpoint at stop time (default: False)
:param merge: None: no merge happens, else curves with the same final key are merged. 2 cases:
one key (str): the name of final key. result will be a dict of <str> of <Table>
a tuple of keys: the names of the final key elements. result: dict of <tuple> of Table
:param pivot: sort values in to columns of one big table
:param kwds: further selection criteria
:return: a dict <key or tuple of key values> of <Table> or <Single>
where <Table> is a list of tuples with some meta info (table.tags, table.column_names)
and <Single> is a list (a single row of a table) with the same meta info
when _field='float' (the default), the returned values are either a floats or None
"""
tags = {k: v for k, v in (('_measurement', measurement), ('_field', field)) if v is not None}
tags.update(kwds)
start, stop = abs_range(start, stop)
rstart, rstop = round_range(start, stop, interval)
if rstart < rstop:
result = self.query(rstart, rstop, interval, columns=None, **tags)
# result = self.query(rstart, rstop, interval, columns=['stream', 'device'], **tags)
else:
result = {}
start_row = {}
if add_prev:
prev_data = self.query(rstart - add_prev, rstart, single=1, **tags)
for key, first in prev_data.items():
curve = result.get(key)
if first[1] is not None:
if curve:
if first[0] < curve[0][0]:
if pivot:
curve.insert(0, (rstart,) + tuple(first[1:]))
# start_row.setdefault(key[1:], {})[key[0]] = first[1]
else:
curve.insert(0, tuple(first))
else:
result[key] = table = Table(first.tags, first.key_names, first.column_names)
table.append(tuple(first))
if add_end:
self.complete(result, stop)
if merge:
single_merge = isinstance(merge, str)
if single_merge:
merge = [merge]
rows = []
common_tags = summarize_tags(result.values(), True)
col_keys = {}
col_info = {}
for key, curve in result.items():
merge_tags = {k: curve.tags.get(k, '') for k in merge}
for k, v in zip(curve.key_names, key):
merge_tags.setdefault(k, v)
merge_key = tuple(zip(*merge_tags.items())) # (<keys tuple>, <values tuple>)
info = col_info.get(merge_key)
if info is None:
col_idx = len(col_keys) + 1
col_keys[col_idx] = merge_key
col_info[merge_key] = info = [col_idx, []]
else:
col_idx = info[0]
info[1].append(curve)
assert curve.column_names[1] == '_value'
for row in curve:
rows.append((row[0], row[1], col_idx))
# merged.append((merge_key, curve))
rows.sort(key=lambda x: x[0])
if pivot:
header = []
for merge_key, (col_idx, curves) in col_info.items():
tags = summarize_tags(curves)
primary = tags.pop(merge[0])
header.append(' '.join([primary] + [f"{k}={v}" for k, v in tags.items() if k not in common_tags]))
result = Table(common_tags, (), ('_time',) + tuple(header))
values = [0] + [None] * len(col_keys)
for row in rows:
col_nr = row[2]
values[col_nr] = row[1]
if row[0] > values[0]:
values[0] = row[0]
result.append(tuple(values))
elif row[0] < values[0]:
raise ValueError(f'{rstart} {values[0]} {row[0]}')
else:
result = {}
by_idx = {}
for merge_key, (col_idx, curves) in col_info.items():
tags = summarize_tags(curves)
primary = tags[merge[0]]
table = Table(tags, merge_key[0], ('_time', primary))
result[primary if single_merge else merge_key[1][:len(merge)]] = table
by_idx[col_idx] = table
for row in rows:
by_idx[row[2]].append((row[0], row[1]))
return result
@staticmethod
def complete(curve_dict, end_time=0, tag='stream'):
"""complete to end_time
if end_time is not given, is is the max timestamp within the same stream
"""
end_time_dict = {}
if not end_time:
for curve in curve_dict.values():
key = curve.tags.get(tag)
end_time_dict[key] = max(end_time_dict.get(key, 0), curve[-1][0])
for curve in curve_dict.values():
if len(curve):
tlast, value = curve[-1]
etime = end_time_dict.get(curve.tags.get(tag), end_time)
if value is not None and tlast < etime:
curve.append((etime, value))
def export(self, start, stop, measurement=('*.value', '*.target'), field='float',
interval=None, add_prev=3600, add_end=False, timeoffset=None, none='', **tags):
result = self.curves(start, stop, measurement, field, interval, add_prev, add_end,
merge=('_measurement', 'device', 'stream'), pivot=True, **tags)
if timeoffset is None:
timeoffset = int(start)
result.tags.pop('_field', None)
rows = [f"# {' '.join(f'{k}={v}' for k, v in result.tags.items())}"]
rows.extend(f'# col {i} {k}' for i, k in enumerate(result.column_names))
rows.extend(result.to_csv_rows(timeoffset, none=none))
rows.append('')
return '\n'.join(rows)
# write to the database
def _add_point(self, measurement, field, value, ts, tags):
point = Point(measurement).field(f'{field}', value)
if ts:
point.time(datetime.utcfromtimestamp(ts), write_precision=self._write_precision)
for key, val in tags.items():
point.tag(key, val)
self._write_buffer.append(point)
def write(self, measurement, field, value, ts, **tags):
"""add point and flush"""
self._add_point(measurement, field, value, ts, tags)
self.flush()
def flush(self):
"""flush write buffer"""
points = self._write_buffer
self._write_buffer = []
if points:
try:
self._write_api_write(bucket=self._bucket, record=points)
except TypeError as e:
if self._write_api_write is None:
raise PermissionError('no write access - need access="write"') from None
raise
# TODO: move these sehistory related methods to a subclass
def add_float(self, value, key, tags, ts):
self._add_point('.'.join(key), 'float', -0.0 if value is None else float(value), ts, tags)
def add_error(self, value, key, tags, ts):
self._add_point('.'.join(key), 'error', '' if value is None else str(value), ts, tags)
def get_instrument(self, stream, ts=None, **tags):
if ts is None:
ts = int(time.time())
reply = self.query(None, int(ts) + 1, _measurement='_stream_', _field='on',
stream=stream, single=1, **tags)
if reply:
entry = sorted(reply.values(), key=lambda r: r[0])[-1]
return entry.tags.get('instrument'), entry[0]
return None, None
def get_streams(self, instrument=None, ts=None, **tags):
"""get streams for one or all instruments
:param instrument: None when looking for all instruments
:param ts: the time or None when now
:return: dict <stream> of <instrument> or '0' when instrument is not known
"""
if ts is None:
ts = int(time.time())
reply = self.query(None, int(ts) + 1, _measurement='_stream_', _field='on',
single=1, instrument=instrument, **tags)
all_entries = {}
for entry in reply.values():
all_entries.setdefault(entry.tags.get('stream'), []).append(entry)
result = {}
for stream, entries in all_entries.items():
entry = sorted(entries, key=lambda r: r[0])[-1]
if entry[1]: # on=True
result[stream] = entry.tags.get('instrument', '0')
return result
def set_instrument(self, stream, value, ts=None, guess=True, **tags):
"""set stream and instrument on or off
:param stream: the uri of the stream
:param value: instrument, "0" when unknown or None when switching to off
:param ts: the time or None when now
:param guess: when instrument is undefined, take from previous
"""
prev, prevts = self.get_instrument(stream, ts, **tags)
if prevts is not None:
if prev in (None, '0'):
ts = prevts + 0.001
else:
if value == '0' and guess:
value = prev
if ts < prevts:
ts = prevts + 0.001
tags['stream'] = stream
if value:
tags['instrument'] = value
flag = True
else:
tags['instrument'] = prev or '0'
flag = False
self._add_point('_stream_', 'on', flag, ts, tags)
def add_stream(self, value, tags, key, ts):
self.set_instrument(key, value, ts, **tags)
def testdb():
token = "zqDbTcMv9UizfdTj15Fx_6vBetkM5mXN56EE9CiDaFsh7O2FFWZ2X4VwAAmdyqZr3HbpIr5ixRju07-oQmxpXw=="
return InfluxDBWrapper('http://pc16392:8086', token, 'linse', 'curve-test')