feat(metrics): New normalization of keys, values, units

sentry_sdk/metrics.py

@@ -11,7 +11,7 @@
 from functools import wraps, partial
 
 import sentry_sdk
-from sentry_sdk._compat import text_type, utc_from_timestamp, iteritems
+from sentry_sdk._compat import PY2, text_type, utc_from_timestamp, iteritems
 from sentry_sdk.utils import (
     ContextVar,
     now,
@@ -53,9 +53,15 @@
     from sentry_sdk._types import MetricValue
 
 
+if PY2:
+    import string
+
+    maketrans = string.maketrans
+else:
+    maketrans = str.maketrans
+
+
 _in_metrics = ContextVar("in_metrics", default=False)
-_sanitize_key = partial(re.compile(r"[^a-zA-Z0-9_/.-]+").sub, "_")
-_sanitize_value = partial(re.compile(r"[^\w\d\s_:/@\.{}\[\]$-]+", re.UNICODE).sub, "")
 _set = set  # set is shadowed below
 
 GOOD_TRANSACTION_SOURCES = frozenset(
@@ -67,6 +73,25 @@
     ]
 )
 
+_sanitize_unit = partial(re.compile(r"[^\w]+").sub, "")
+_sanitize_metric_key = partial(re.compile(r"[^\w\-.]+").sub, "_")
+_sanitize_tag_key = partial(re.compile(r"[^\w\-.\/]+", re.UNICODE).sub, "")
+
+
+def _sanitize_tag_value(value):
+    # type: (str) -> str
+    table = maketrans(
+        {
+            "\n": "\\n",
+            "\r": "\\r",
+            "\t": "\\t",
+            "\\": "\\\\",
+            "|": "\\u{7c}",
+            ",": "\\u{2c}",
+        }
+    )
+    return value.translate(table)
+
 
 def get_code_location(stacklevel):
     # type: (int) -> Optional[Dict[str, Any]]
@@ -269,7 +294,8 @@ def _encode_metrics(flushable_buckets):
     for timestamp, buckets in flushable_buckets:
         for bucket_key, metric in iteritems(buckets):
             metric_type, metric_name, metric_unit, metric_tags = bucket_key
-            metric_name = _sanitize_key(metric_name)
+            metric_name = _sanitize_metric_key(metric_name)
+            metric_unit = _sanitize_unit(metric_unit)
             _write(metric_name.encode("utf-8"))
             _write(b"@")
             _write(metric_unit.encode("utf-8"))
@@ -285,7 +311,7 @@ def _encode_metrics(flushable_buckets):
                 _write(b"|#")
                 first = True
                 for tag_key, tag_value in metric_tags:
-                    tag_key = _sanitize_key(tag_key)
+                    tag_key = _sanitize_tag_key(tag_key)
                     if not tag_key:
                         continue
                     if first:
@@ -294,7 +320,7 @@ def _encode_metrics(flushable_buckets):
                         _write(b",")
                     _write(tag_key.encode("utf-8"))
                     _write(b":")
-                    _write(_sanitize_value(tag_value).encode("utf-8"))
+                    _write(_sanitize_tag_value(tag_value).encode("utf-8"))
 
             _write(b"|T")
             _write(str(timestamp).encode("ascii"))
@@ -309,7 +335,9 @@ def _encode_locations(timestamp, code_locations):
 
     for key, loc in code_locations:
         metric_type, name, unit = key
-        mri = "{}:{}@{}".format(metric_type, _sanitize_key(name), unit)
+        mri = "{}:{}@{}".format(
+            metric_type, _sanitize_metric_key(name), _sanitize_unit(unit)
+        )
 
         loc["type"] = "location"
         mapping.setdefault(mri, []).append(loc)

tests/test_metrics.py

@@ -677,56 +677,96 @@ def test_metric_summaries(
 
 @minimum_python_37_with_gevent
 @pytest.mark.forked
-def test_tag_normalization(
-    sentry_init, capture_envelopes, maybe_monkeypatched_threading
+@pytest.mark.parametrize(
+    "metric_name,metric_unit,expected_name",
+    [
+        ("first-metric", "nano-second", "first-metric@nanosecond"),
+        ("another_metric?", "nano second", "another_metric_@nanosecond"),
+        (
+            "metric",
+            "nanosecond",
+            "metric@nanosecond",
+        ),
+        (
+            "my.amaze.metric I guess",
+            "nano|\nsecond",
+            "my.amaze.metric_I_guess@nanosecond",
+        ),
+    ],
+)
+def test_metric_name_normalization(
+    sentry_init,
+    capture_envelopes,
+    metric_name,
+    metric_unit,
+    expected_name,
+    maybe_monkeypatched_threading,
 ):
     sentry_init(
-        release="fun-release@1.0.0",
-        environment="not-fun-env",
         _experiments={"enable_metrics": True, "metric_code_locations": False},
     )
-    ts = time.time()
     envelopes = capture_envelopes()
 
-    # fmt: off
-    metrics.distribution("a", 1.0, tags={"foo-bar": "%$foo"}, timestamp=ts)
-    metrics.distribution("b", 1.0, tags={"foo$$$bar": "blah{}"}, timestamp=ts)
-    metrics.distribution("c", 1.0, tags={u"foö-bar": u"snöwmän"}, timestamp=ts)
-    metrics.distribution("d", 1.0, tags={"route": "GET /foo"}, timestamp=ts)
-    # fmt: on
+    metrics.distribution(metric_name, 1.0, unit=metric_unit)
+
     Hub.current.flush()
 
     (envelope,) = envelopes
 
     assert len(envelope.items) == 1
     assert envelope.items[0].headers["type"] == "statsd"
-    m = parse_metrics(envelope.items[0].payload.get_bytes())
 
-    assert len(m) == 4
-    assert m[0][4] == {
-        "foo-bar": "$foo",
-        "release": "fun-release@1.0.0",
-        "environment": "not-fun-env",
-    }
+    parsed_metrics = parse_metrics(envelope.items[0].payload.get_bytes())
+    assert len(parsed_metrics) == 1
 
-    assert m[1][4] == {
-        "foo_bar": "blah{}",
-        "release": "fun-release@1.0.0",
-        "environment": "not-fun-env",
-    }
+    name = parsed_metrics[0][1]
+    assert name == expected_name
 
-    # fmt: off
-    assert m[2][4] == {
-        "fo_-bar": u"snöwmän",
-        "release": "fun-release@1.0.0",
-        "environment": "not-fun-env",
-    }
-    assert m[3][4] == {
-        "release": "fun-release@1.0.0",
-        "environment": "not-fun-env",
-        "route": "GET /foo",
-    }
-    # fmt: on
+
+@minimum_python_37_with_gevent
+@pytest.mark.forked
+@pytest.mark.parametrize(
+    "metric_tag,expected_tag",
+    [
+        ({"f-oo|bar": "%$foo/"}, {"f-oobar": "%$foo/"}),
+        ({"foo$.$.$bar": "blah{}"}, {"foo..bar": "blah{}"}),
+        # fmt: off
+        ({u"foö-bar": u"snöwmän"}, {"foö-bar": u"snöwmän"},),
+        # fmt: on
+        ({"route": "GET /foo"}, {"route": "GET /foo"}),
+        ({"__bar__": "this | or , that"}, {"__bar__": "this \\u{7c} or \\u{2c} that"}),
+        ({"foo/": "hello!\n\r\t\\"}, {"foo/": "hello!\\n\\r\\t\\\\"}),
+    ],
+)
+def test_metric_tag_normalization(
+    sentry_init,
+    capture_envelopes,
+    metric_tag,
+    expected_tag,
+    maybe_monkeypatched_threading,
+):
+    sentry_init(
+        _experiments={"enable_metrics": True, "metric_code_locations": False},
+    )
+    envelopes = capture_envelopes()
+
+    metrics.distribution("a", 1.0, tags=metric_tag)
+
+    Hub.current.flush()
+
+    (envelope,) = envelopes
+
+    assert len(envelope.items) == 1
+    assert envelope.items[0].headers["type"] == "statsd"
+
+    parsed_metrics = parse_metrics(envelope.items[0].payload.get_bytes())
+    assert len(parsed_metrics) == 1
+
+    tags = parsed_metrics[0][4]
+
+    expected_tag_key, expected_tag_value = expected_tag.popitem()
+    assert expected_tag_key in tags
+    assert tags[expected_tag_key] == expected_tag_value
 
 
 @minimum_python_37_with_gevent