format and handle quotes

Author: sarah-witt

datadog_checks_base/datadog_checks/base/checks/openmetrics/parser_optimizations.py

@@ -18,17 +18,52 @@
 import prometheus_client.parser as _prom_parser
 
 
+def _is_char_escaped(text, pos):
+    """Return True if the character at pos is preceded by an odd number of backslashes."""
+    num_bslashes = 0
+    while pos > num_bslashes and text[pos - 1 - num_bslashes] == '\\':
+        num_bslashes += 1
+    return num_bslashes % 2 == 1
+
+
 def _next_unquoted_char(text, chs, startidx=0):
-    """Find the next occurrence of any character in chs."""
+    """Find the next unquoted occurrence of any character in chs.
+
+    Uses str.find() to jump to candidate characters, skipping over quoted regions.
+    """
     if chs is None:
         chs = string.whitespace
 
-    best = -1
-    for ch in chs:
-        p = text.find(ch, startidx)
-        if p != -1 and (best == -1 or p < best):
-            best = p
-    return best
+    i = startidx
+    n = len(text)
+
+    while i < n:
+        best = -1
+        for ch in chs:
+            p = text.find(ch, i)
+            if p != -1 and (best == -1 or p < best):
+                best = p
+
+        # Find the next unescaped opening quote
+        q = text.find('"', i)
+        while q != -1 and _is_char_escaped(text, q):
+            q = text.find('"', q + 1)
+
+        # If no quote comes before the best candidate, return it directly
+        if q == -1 or (best != -1 and best < q):
+            return best
+
+        # A quoted region starts before the candidate; skip over it
+        close = text.find('"', q + 1)
+        while close != -1 and _is_char_escaped(text, close):
+            close = text.find('"', close + 1)
+
+        if close == -1:
+            return -1
+
+        i = close + 1
+
+    return -1
 
 
 def apply():

datadog_checks_base/datadog_checks/base/checks/openmetrics/v2/scraper/base_scraper.py

@@ -13,10 +13,9 @@
 from prometheus_client import Metric
 from prometheus_client.openmetrics.parser import text_fd_to_metric_families as parse_openmetrics
 from prometheus_client.parser import text_fd_to_metric_families as parse_prometheus
-
-import datadog_checks.base.checks.openmetrics.parser_optimizations  # noqa: F401
 from requests.exceptions import ConnectionError
 
+import datadog_checks.base.checks.openmetrics.parser_optimizations  # noqa: F401
 from datadog_checks.base.agent import datadog_agent
 from datadog_checks.base.checks.openmetrics.v2.first_scrape_handler import first_scrape_handler
 from datadog_checks.base.checks.openmetrics.v2.labels import LabelAggregator, get_label_normalizer

datadog_checks_base/tests/base/checks/openmetrics/test_parser_optimizations.py

@@ -7,8 +7,6 @@
 original prometheus_client implementation across representative inputs.
 """
 
-import pytest
-
 from datadog_checks.base.checks.openmetrics.parser_optimizations import (
     _next_unquoted_char,
 )
@@ -71,6 +69,20 @@ def test_last_char_is_target(self):
     def test_multiple_occurrences_returns_first(self):
         assert _next_unquoted_char('a{b{c', '{') == 1
 
+    def test_skip_target_char_inside_quotes(self):
+        # comma inside quoted value must not be returned
+        assert _next_unquoted_char('a="apn,gw",b', ',') == 10
+
+    def test_skip_brace_inside_quotes(self):
+        assert _next_unquoted_char('label="val}ue"}', '}') == 14
+
+    def test_skip_equals_inside_quotes(self):
+        assert _next_unquoted_char('label="a=b"} 1', '}') == 11
+
+    def test_escaped_quote_not_treated_as_delimiter(self):
+        # backslash-escaped quote does not close the quoted region
+        assert _next_unquoted_char(r'label="val\"still,inside",next', ',') == 25
+
 
 class TestNextUnquotedCharWithRealMetrics:
     """Tests using real Prometheus metric line patterns."""
@@ -113,11 +125,7 @@ class TestParseFullMetricText:
     def test_parse_simple_metrics(self):
         from prometheus_client.parser import text_string_to_metric_families
 
-        text = (
-            '# HELP test_gauge A test gauge.\n'
-            '# TYPE test_gauge gauge\n'
-            'test_gauge 42\n'
-        )
+        text = '# HELP test_gauge A test gauge.\n# TYPE test_gauge gauge\ntest_gauge 42\n'
         families = list(text_string_to_metric_families(text))
         assert len(families) == 1
         assert families[0].name == 'test_gauge'
@@ -159,11 +167,7 @@ def test_parse_histogram(self):
     def test_parse_escaped_label_value(self):
         from prometheus_client.parser import text_string_to_metric_families
 
-        text = (
-            '# HELP test_metric A test.\n'
-            '# TYPE test_metric gauge\n'
-            'test_metric{label="value with \\"quotes\\""} 1\n'
-        )
+        text = '# HELP test_metric A test.\n# TYPE test_metric gauge\ntest_metric{label="value with \\"quotes\\""} 1\n'
         families = list(text_string_to_metric_families(text))
         assert len(families) == 1
         assert families[0].samples[0].labels == {'label': 'value with "quotes"'}
@@ -187,21 +191,26 @@ def test_parse_multiple_families(self):
     def test_parse_empty_label_value(self):
         from prometheus_client.parser import text_string_to_metric_families
 
-        text = (
-            '# HELP test_metric A test.\n'
-            '# TYPE test_metric gauge\n'
-            'test_metric{label=""} 1\n'
-        )
+        text = '# HELP test_metric A test.\n# TYPE test_metric gauge\ntest_metric{label=""} 1\n'
         families = list(text_string_to_metric_families(text))
         assert families[0].samples[0].labels == {'label': ''}
 
     def test_parse_newline_in_label_value(self):
         from prometheus_client.parser import text_string_to_metric_families
 
+        text = '# HELP test_metric A test.\n# TYPE test_metric gauge\ntest_metric{label="line1\\nline2"} 1\n'
+        families = list(text_string_to_metric_families(text))
+        assert families[0].samples[0].labels == {'label': 'line1\nline2'}
+
+    def test_parse_comma_in_label_value(self):
+        from prometheus_client.parser import text_string_to_metric_families
+
         text = (
-            '# HELP test_metric A test.\n'
-            '# TYPE test_metric gauge\n'
-            'test_metric{label="line1\\nline2"} 1\n'
+            '# HELP apn_active_connections Active connections.\n'
+            '# TYPE apn_active_connections gauge\n'
+            'apn_active_connections{func="apn,gw",proto="tcp"} 8\n'
         )
         families = list(text_string_to_metric_families(text))
-        assert families[0].samples[0].labels == {'label': 'line1\nline2'}
+        assert len(families) == 1
+        assert families[0].samples[0].labels == {'func': 'apn,gw', 'proto': 'tcp'}
+        assert families[0].samples[0].value == 8