Multiaddr.decapsulate() converts both multiaddrs to strings and uses str.rindex() to find the decapsulation point. This produces incorrect results when the same protocol appears multiple times in the address, or when the decapsulation target is a substring of a protocol value.
Problem
In multiaddr/multiaddr.py:
def decapsulate(self, addr):
addr_str = str(addr)
s = str(self)
i = s.rindex(addr_str) # ← String search, not component-level
if i < 0:
raise ValueError(...)
return Multiaddr(s[:i])Case 1 — Repeated protocol:
ma = Multiaddr("/ip4/1.2.3.4/tcp/80/ip4/5.6.7.8/tcp/443")
result = ma.decapsulate("/ip4/5.6.7.8")
# Expected: /ip4/1.2.3.4/tcp/80
# Actual: /ip4/1.2.3.4/tcp/80 ✅ (happens to work because rindex finds last)Case 2 — Substring collision:
ma = Multiaddr("/dns4/example.com/tcp/80")
result = ma.decapsulate("/tcp/8")
# rindex finds "tcp/8" inside "tcp/80" → incorrect split pointCase 3 — Value contains protocol name:
ma = Multiaddr("/dns4/tcp.example.com/tcp/80")
result = ma.decapsulate("/tcp/80")
# rindex might find "tcp" inside "tcp.example.com" → incorrectGo's implementation does component-by-component comparison:
func (m Multiaddr) Decapsulate(rightPartsAny Multiaddrer) Multiaddr {
rightParts := rightPartsAny.Multiaddr()
leftParts := m
lastIndex := -1
for i := range leftParts {
foundMatch := false
for j, rightC := range rightParts {
if len(leftParts) <= i+j { break }
foundMatch = rightC.Equal(&leftParts[i+j])
if !foundMatch { break }
}
if foundMatch { lastIndex = i }
}
// ...
}Proposed Solution
Rewrite decapsulate() to compare at the component level using bytes_iter():
def decapsulate(self, addr):
other = Multiaddr(addr) if not isinstance(addr, Multiaddr) else addr
other_components = list(bytes_iter(other.to_bytes()))
self_components = list(bytes_iter(self._bytes))
last_match_end = -1
for i in range(len(self_components)):
match = True
for j, (offset, proto, codec, value) in enumerate(other_components):
if i + j >= len(self_components):
match = False
break
s_offset, s_proto, s_codec, s_value = self_components[i + j]
if s_proto != proto or s_value != value:
match = False
break
if match:
last_match_end = self_components[i][0] # byte offset
if last_match_end < 0:
raise ValueError(f"Address {self} does not contain subaddress: {addr}")
if last_match_end == 0:
return Multiaddr("")
return Multiaddr(self._bytes[:last_match_end])Related
Multiaddr.decapsulate()converts both multiaddrs to strings and usesstr.rindex()to find the decapsulation point. This produces incorrect results when the same protocol appears multiple times in the address, or when the decapsulation target is a substring of a protocol value.Problem
In
multiaddr/multiaddr.py:Case 1 — Repeated protocol:
Case 2 — Substring collision:
Case 3 — Value contains protocol name:
Go's implementation does component-by-component comparison:
Proposed Solution
Rewrite
decapsulate()to compare at the component level usingbytes_iter():Related
multiaddr.goDecapsulate()multiaddr/multiaddr.py