Coverage for /Users/buh/.pyenv/versions/3.12.2/envs/es-testbed/lib/python3.12/site-packages/es_testbed/helpers/utils.py: 64%

103 statements  

« prev     ^ index     » next       coverage.py v7.4.4, created at 2024-08-21 12:05 -0600

1"""Utility helper functions""" 

2 

3import sys 

4import typing as t 

5import random 

6import string 

7import logging 

8from datetime import datetime, timezone 

9from pathlib import Path 

10from pprint import pformat 

11from shutil import rmtree 

12from tempfile import mkdtemp 

13from git import Repo 

14from ..defaults import ilm_force_merge, ilm_phase, TIER 

15from ..exceptions import TestbedMisconfig 

16 

17logger = logging.getLogger(__name__) 

18 

19 

20def build_ilm_phase( 

21 phase: str, 

22 actions: t.Union[t.Dict, None] = None, 

23 repo: t.Union[str, None] = None, 

24 fm: bool = False, 

25) -> t.Dict: 

26 """Build a single ILM policy step based on phase""" 

27 retval = ilm_phase(phase) 

28 if phase in ['cold', 'frozen']: 

29 if repo: 

30 retval[phase]['actions']['searchable_snapshot'] = { 

31 'snapshot_repository': repo, 

32 'force_merge_index': fm, 

33 } 

34 else: 

35 msg = ( 

36 f'Unable to build {phase} ILM phase. Value for repository not ' 

37 f'provided' 

38 ) 

39 raise TestbedMisconfig(msg) 

40 if actions: 

41 retval[phase]['actions'].update(actions) 

42 return retval 

43 

44 

45def build_ilm_policy( 

46 phases: list = None, 

47 forcemerge: bool = False, 

48 max_num_segments: int = 1, 

49 readonly: t.Union[str, None] = None, 

50 repository: t.Union[str, None] = None, 

51) -> t.Dict: 

52 """ 

53 Build a full ILM policy based on the provided phases. 

54 Put forcemerge in the last phase before cold or frozen (whichever comes first) 

55 """ 

56 if not phases: 

57 phases = ['hot', 'delete'] 

58 retval = {} 

59 if ('cold' in phases or 'frozen' in phases) and not repository: 

60 raise TestbedMisconfig('Cannot build cold or frozen phase without repository') 

61 for phase in phases: 

62 actions = None 

63 if readonly == phase: 

64 actions = {"readonly": {}} 

65 phase = build_ilm_phase(phase, repo=repository, fm=forcemerge, actions=actions) 

66 retval.update(phase) 

67 if forcemerge: 

68 retval['hot']['actions'].update( 

69 ilm_force_merge(max_num_segments=max_num_segments) 

70 ) 

71 return {'phases': retval} 

72 

73 

74# def doc_gen( 

75# count: int = 10, start_at: int = 0, match: bool = True 

76# ) -> t.Generator[t.Dict, None, None]: 

77# """Create this doc for each count""" 

78# keys = ['message', 'nested', 'deep'] 

79# # Start with an empty map 

80# matchmap = {} 

81# # Iterate over each key 

82# for key in keys: 

83# # If match is True 

84# if match: 

85# # Set matchmap[key] to key 

86# matchmap[key] = key 

87# else: 

88# # Otherwise matchmap[key] will have a random string value 

89# matchmap[key] = randomstr() 

90 

91# # This is where count and start_at matter 

92# for num in range(start_at, start_at + count): 

93# yield { 

94# '@timestamp': iso8601_now(), 

95# 'message': f'{matchmap["message"]}{num}', # message# or randomstr# 

96# 'number': ( 

97# num if match else random.randint(1001, 32767) 

98# ), # value of num or random int 

99# 'nested': {'key': f'{matchmap["nested"]}{num}'}, # nested# 

100# 'deep': {'l1': {'l2': {'l3': f'{matchmap["deep"]}{num}'}}}, # deep# 

101# } 

102 

103 

104# def getlogger(name: str) -> logging.getLogger: 

105# """Return a named logger""" 

106# return logging.getLogger(name) 

107 

108 

109def get_routing(tier='hot') -> t.Dict: 

110 """Return the routing allocation tier preference""" 

111 try: 

112 pref = TIER[tier]['pref'] 

113 except KeyError: 

114 # Fallback value 

115 pref = 'data_content' 

116 return {'index.routing.allocation.include._tier_preference': pref} 

117 

118 

119def iso8601_now() -> str: 

120 """ 

121 :returns: An ISO8601 timestamp based on now 

122 :rtype: str 

123 """ 

124 # Because Python 3.12 now requires non-naive timezone declarations, we must change. 

125 # 

126 # ## Example: 

127 # ## The new way: 

128 # ## datetime.now(timezone.utc).isoformat() 

129 # ## Result: 2024-04-16T16:00:00+00:00 

130 # ## End Example 

131 # 

132 # Note that the +00:00 is appended now where we affirmatively declare the 

133 # UTC timezone 

134 # 

135 # As a result, we will use this function to prune away the timezone if it is 

136 # +00:00 and replace it with Z, which is shorter Zulu notation for UTC (which 

137 # Elasticsearch uses) 

138 # 

139 # We are MANUALLY, FORCEFULLY declaring timezone.utc, so it should ALWAYS be 

140 # +00:00, but could in theory sometime show up as a Z, so we test for that. 

141 

142 parts = datetime.now(timezone.utc).isoformat().split('+') 

143 if len(parts) == 1: 

144 if parts[0][-1] == 'Z': 

145 return parts[0] # Our ISO8601 already ends with a Z for Zulu/UTC time 

146 return f'{parts[0]}Z' # It doesn't end with a Z so we put one there 

147 if parts[1] == '00:00': 

148 return f'{parts[0]}Z' # It doesn't end with a Z so we put one there 

149 return f'{parts[0]}+{parts[1]}' # Fallback publishes the +TZ, whatever that was 

150 

151 

152def mounted_name(index: str, tier: str): 

153 """Return a value for renamed_index for mounting a searchable snapshot index""" 

154 return f'{TIER[tier]["prefix"]}-{index}' 

155 

156 

157def prettystr(*args, **kwargs) -> str: 

158 """ 

159 A (nearly) straight up wrapper for pprint.pformat, except that we provide our own 

160 default values for 'indent' (2) and 'sort_dicts' (False). Primarily for debug 

161 logging and showing more readable dictionaries. 

162 

163 'Return the formatted representation of object as a string. indent, width, depth, 

164 compact, sort_dicts and underscore_numbers are passed to the PrettyPrinter 

165 constructor as formatting parameters' (from pprint documentation). 

166 """ 

167 defaults = [ 

168 ('indent', 2), 

169 ('width', 80), 

170 ('depth', None), 

171 ('compact', False), 

172 ('sort_dicts', False), 

173 ('underscore_numbers', False), 

174 ] 

175 kw = {} 

176 for tup in defaults: 

177 key, default = tup 

178 kw[key] = kwargs[key] if key in kwargs else default 

179 

180 return f"\n{pformat(*args, **kw)}" # newline in front so it always starts clean 

181 

182 

183def process_preset( 

184 builtin: t.Union[str, None], 

185 path: t.Union[str, None], 

186 ref: t.Union[str, None], 

187 url: t.Union[str, None], 

188) -> t.Tuple: 

189 """Process the preset settings 

190 :param preset: One of `builtin`, `git`, or `path` 

191 :param builtin: The name of a builtin preset 

192 :param path: A relative or absolute file path. Used by presets `git` and `path` 

193 :param ref: A Git ref (e.g. 'main'). Only used by preset `git` 

194 :param url: A Git repository URL. Only used by preset `git` 

195 """ 

196 modpath = None 

197 tmpdir = None 

198 if builtin: # Overrides any other options 

199 modpath = f'es_testbed.presets.{builtin}' 

200 else: 

201 trygit = False 

202 try: 

203 kw = {'path': path, 'ref': ref, 'url': url} 

204 raise_on_none(**kw) 

205 trygit = True # We have all 3 kwargs necessary for git 

206 except ValueError as resp: # Not able to do a git preset 

207 logger.debug('Unable to import a git-based preset: %s', resp) 

208 if trygit: # Trying a git import 

209 tmpdir = mkdtemp() 

210 try: 

211 _ = Repo.clone_from(url, tmpdir, branch=ref, depth=1) 

212 filepath = Path(tmpdir) / path 

213 except Exception as err: 

214 logger.error('Git clone failed: %s', err) 

215 rmtree(tmpdir) # Clean up after failed attempt 

216 raise err 

217 if path: 

218 filepath = Path(path) 

219 if not filepath.resolve().is_dir(): 

220 raise ValueError(f'The provided path "{path}" is not a directory') 

221 modpath = filepath.resolve().name # The final dirname 

222 parent = filepath.parent.resolve() # Up one level 

223 # We now make the parent path part of the sys.path. 

224 sys.path.insert(0, parent) # This should persist beyond this module 

225 return modpath, tmpdir 

226 

227 

228def raise_on_none(**kwargs): 

229 """Raise if any kwargs have a None value""" 

230 for key, value in kwargs.items(): 

231 if value is None: 

232 raise ValueError(f'kwarg "{key}" cannot have a None value') 

233 

234 

235def randomstr(length: int = 16, lowercase: bool = False) -> str: 

236 """Generate a random string""" 

237 letters = string.ascii_uppercase 

238 if lowercase: 

239 letters = string.ascii_lowercase 

240 return str(''.join(random.choices(letters + string.digits, k=length))) 

241 

242 

243def storage_type(tier: str) -> t.Dict: 

244 """Return the storage type of a searchable snapshot by tier""" 

245 return TIER[tier]["storage"]