-
-
Notifications
You must be signed in to change notification settings - Fork 1k
/
tracking.py
238 lines (200 loc) · 9.54 KB
/
tracking.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
import contextlib
import contextvars
import datetime
import json
from time import perf_counter
import django.test.testcases
from django.utils.encoding import force_str
from debug_toolbar.utils import get_stack_trace, get_template_info
try:
import psycopg
PostgresJson = psycopg.types.json.Jsonb
STATUS_IN_TRANSACTION = psycopg.pq.TransactionStatus.INTRANS
except ImportError:
try:
from psycopg2._json import Json as PostgresJson
from psycopg2.extensions import STATUS_IN_TRANSACTION
except ImportError:
PostgresJson = None
STATUS_IN_TRANSACTION = None
# Prevents SQL queries from being sent to the DB. It's used
# by the TemplatePanel to prevent the toolbar from issuing
# additional queries.
allow_sql = contextvars.ContextVar("debug-toolbar-allow-sql", default=True)
class SQLQueryTriggered(Exception):
"""Thrown when template panel triggers a query"""
def wrap_cursor(connection):
# When running a SimpleTestCase, Django monkey patches some DatabaseWrapper
# methods, including .cursor() and .chunked_cursor(), to raise an exception
# if the test code tries to access the database, and then undoes the monkey
# patching when the test case is finished. If we monkey patch those methods
# also, Django's process of undoing those monkey patches will fail. To
# avoid this failure, and because database access is not allowed during a
# SimpleTestCase anyway, skip applying our instrumentation monkey patches if
# we detect that Django has already monkey patched DatabaseWrapper.cursor().
if isinstance(connection.cursor, django.test.testcases._DatabaseFailure):
return
if not hasattr(connection, "_djdt_cursor"):
connection._djdt_cursor = connection.cursor
connection._djdt_chunked_cursor = connection.chunked_cursor
connection._djdt_logger = None
def cursor(*args, **kwargs):
# Per the DB API cursor() does not accept any arguments. There's
# some code in the wild which does not follow that convention,
# so we pass on the arguments even though it's not clean.
# See:
# https://github.com/jazzband/django-debug-toolbar/pull/615
# https://github.com/jazzband/django-debug-toolbar/pull/896
logger = connection._djdt_logger
cursor = connection._djdt_cursor(*args, **kwargs)
if logger is None:
return cursor
mixin = NormalCursorMixin if allow_sql.get() else ExceptionCursorMixin
return patch_cursor_wrapper_with_mixin(cursor.__class__, mixin)(
cursor.cursor, connection, logger
)
def chunked_cursor(*args, **kwargs):
# prevent double wrapping
# solves https://github.com/jazzband/django-debug-toolbar/issues/1239
logger = connection._djdt_logger
cursor = connection._djdt_chunked_cursor(*args, **kwargs)
if logger is not None and not isinstance(cursor, DjDTCursorWrapperMixin):
mixin = NormalCursorMixin if allow_sql.get() else ExceptionCursorMixin
return patch_cursor_wrapper_with_mixin(cursor.__class__, mixin)(
cursor.cursor, connection, logger
)
return cursor
connection.cursor = cursor
connection.chunked_cursor = chunked_cursor
def patch_cursor_wrapper_with_mixin(base_wrapper, mixin):
class DjDTCursorWrapper(mixin, base_wrapper):
pass
return DjDTCursorWrapper
class DjDTCursorWrapperMixin:
def __init__(self, cursor, db, logger):
super().__init__(cursor, db)
# logger must implement a ``record`` method
self.logger = logger
class ExceptionCursorMixin(DjDTCursorWrapperMixin):
"""
Wraps a cursor and raises an exception on any operation.
Used in Templates panel.
"""
def __getattr__(self, attr):
raise SQLQueryTriggered()
class NormalCursorMixin(DjDTCursorWrapperMixin):
"""
Wraps a cursor and logs queries.
"""
def _decode(self, param):
if PostgresJson and isinstance(param, PostgresJson):
# psycopg3
if hasattr(param, "obj"):
return param.dumps(param.obj)
# psycopg2
if hasattr(param, "adapted"):
return param.dumps(param.adapted)
# If a sequence type, decode each element separately
if isinstance(param, (tuple, list)):
return [self._decode(element) for element in param]
# If a dictionary type, decode each value separately
if isinstance(param, dict):
return {key: self._decode(value) for key, value in param.items()}
# make sure datetime, date and time are converted to string by force_str
CONVERT_TYPES = (datetime.datetime, datetime.date, datetime.time)
try:
return force_str(param, strings_only=not isinstance(param, CONVERT_TYPES))
except UnicodeDecodeError:
return "(encoded string)"
def _last_executed_query(self, sql, params):
"""Get the last executed query from the connection."""
# Django's psycopg3 backend creates a new cursor in its implementation of the
# .last_executed_query() method. To avoid wrapping that cursor, temporarily set
# the DatabaseWrapper's ._djdt_logger attribute to None. This will cause the
# monkey-patched .cursor() and .chunked_cursor() methods to skip the wrapping
# process during the .last_executed_query() call.
self.db._djdt_logger = None
try:
return self.db.ops.last_executed_query(self.cursor, sql, params)
finally:
self.db._djdt_logger = self.logger
def _record(self, method, sql, params):
alias = self.db.alias
vendor = self.db.vendor
if vendor == "postgresql":
# The underlying DB connection (as opposed to Django's wrapper)
conn = self.db.connection
initial_conn_status = conn.info.transaction_status
start_time = perf_counter()
try:
return method(sql, params)
finally:
stop_time = perf_counter()
duration = (stop_time - start_time) * 1000
_params = ""
with contextlib.suppress(TypeError):
# object JSON serializable?
_params = json.dumps(self._decode(params))
template_info = get_template_info()
# Sql might be an object (such as psycopg Composed).
# For logging purposes, make sure it's str.
if vendor == "postgresql" and not isinstance(sql, str):
if isinstance(sql, bytes):
sql = sql.decode("utf-8")
else:
sql = sql.as_string(conn)
else:
sql = str(sql)
kwargs = {
"vendor": vendor,
"alias": alias,
"sql": self._last_executed_query(sql, params),
"duration": duration,
"raw_sql": sql,
"params": _params,
"raw_params": params,
"stacktrace": get_stack_trace(skip=2),
"template_info": template_info,
}
if vendor == "postgresql":
# If an erroneous query was ran on the connection, it might
# be in a state where checking isolation_level raises an
# exception.
try:
iso_level = conn.isolation_level
except conn.InternalError:
iso_level = "unknown"
# PostgreSQL does not expose any sort of transaction ID, so it is
# necessary to generate synthetic transaction IDs here. If the
# connection was not in a transaction when the query started, and was
# after the query finished, a new transaction definitely started, so get
# a new transaction ID from logger.new_transaction_id(). If the query
# was in a transaction both before and after executing, make the
# assumption that it is the same transaction and get the current
# transaction ID from logger.current_transaction_id(). There is an edge
# case where Django can start a transaction before the first query
# executes, so in that case logger.current_transaction_id() will
# generate a new transaction ID since one does not already exist.
final_conn_status = conn.info.transaction_status
if final_conn_status == STATUS_IN_TRANSACTION:
if initial_conn_status == STATUS_IN_TRANSACTION:
trans_id = self.logger.current_transaction_id(alias)
else:
trans_id = self.logger.new_transaction_id(alias)
else:
trans_id = None
kwargs.update(
{
"trans_id": trans_id,
"trans_status": conn.info.transaction_status,
"iso_level": iso_level,
}
)
# We keep `sql` to maintain backwards compatibility
self.logger.record(**kwargs)
def callproc(self, procname, params=None):
return self._record(super().callproc, procname, params)
def execute(self, sql, params=None):
return self._record(super().execute, sql, params)
def executemany(self, sql, param_list):
return self._record(super().executemany, sql, param_list)