fix(internal): allow tagset values to contain equal signs (#3204)

Revision to the original spec to allow equal signs in tag values to
accommodate base64 encoded values.

Author: brettlangdon

ddtrace/internal/_tagset.pyx

@@ -2,8 +2,9 @@
 Tagset eBNF::
 
     tagset = tag, { ",", tag };
-    tag = ( identifier - space ), "=", identifier;
-    identifier = { ? ASCII 32-126 ? - equal or comma };
+    tag = key, "=", value;
+    key = { ? ASCII 32-126 ? - equal or comma or space };
+    value = { ? ASCII 32-126 ? - comma };
     equal or comma = "=" | ",";
     space = " ";
 """
@@ -64,10 +65,9 @@ cdef inline int is_valid_key_char(int c):
 
 # Same as is_valid_key_char except spaces are allowed
 cdef inline int is_valid_value_char(int c):
-    # string.printable - ",="
+    # string.printable - ","
     # 44 = ",""
-    # 61 = "="
-    return c == 32 or is_valid_key_char(c)
+    return c == 32 or is_equal(c) or is_valid_key_char(c)
 
 
 cpdef dict decode_tagset_string(str tagset):

tests/tracer/test_tagset.py

@@ -29,6 +29,8 @@
         ("key=value,", {"key": "value"}),
         # Values can have spaces
         ("key=value can have spaces", {"key": "value can have spaces"}),
+        # Values can have equals
+        ("key=value=value", {"key": "value=value"}),
         # Remove leading/trailing spaces from values
         ("key= value can have spaces ", {"key": "value can have spaces"}),
         # Non-space whitespace characters are allowed
@@ -41,6 +43,13 @@
             "_dd.p.upstream_services=bWNudWx0eS13ZWI|0|1;dHJhY2Utc3RhdHMtcXVlcnk|2|4,_dd.p.hello=world",
             {"_dd.p.upstream_services": "bWNudWx0eS13ZWI|0|1;dHJhY2Utc3RhdHMtcXVlcnk|2|4", "_dd.p.hello": "world"},
         ),
+        (
+            "_dd.p.upstream_services=bWNudWx0eS13ZWI===|0|1;dHJhY2Utc3RhdHMtcXVlcnk===|2|4,_dd.p.hello=world",
+            {
+                "_dd.p.upstream_services": "bWNudWx0eS13ZWI===|0|1;dHJhY2Utc3RhdHMtcXVlcnk===|2|4",
+                "_dd.p.hello": "world",
+            },
+        ),
     ],
 )
 def test_decode_tagset_string(header, expected):
@@ -61,7 +70,6 @@ def test_decode_tagset_string(header, expected):
         "=value",
         # Extra leading comma
         ",key=value",
-        "key=value=value",
         "key=value,=value",
         "key=value,value",
         # Spaces are not allowed in keys