[DicksonWu654]

Every header that is already in the static or dynamic table is encoded by Encoder._encode_indexed(), which calls encode_integer() (building a bytearray), sets the top bit and copies the result to bytes. Any index below 127 fits in the 7-bit prefix and always encodes to the single byte 0x80 | index, so this looks those up in a 127-entry tuple of precomputed bytes instead.

• Other values and types (127 and above, bool, negative numbers, non-ints) still go through encode_integer(), with the same output and errors.
• When the hpack logger has DEBUG enabled, the existing path is used as well, so the Encoding %d with %d bits records (including their funcName) are unchanged.

The table adds 127 one-byte bytes objects at import time.

Benchmarks

Each workload runs header blocks through Encoder.encode() and, for *_roundtrip, Decoder.decode(). browser_mixed alternates 16 browser-like requests and 16 responses; repeated_requests sends one request 16 times, so after the first block every field is indexed; long_cookie is 16 requests whose cookie is about 1.2 KB; no_huffman_encode is browser_mixed with huffman=False; fixture_stories_roundtrip replays the first four stories of the hpack-test-case fixtures in tests/.

master and this branch were imported side by side in the same interpreter and timed in alternating ~20 ms CPU-time blocks (40 rounds in each of 6 processes, pinned to one core). The table shows the median per-round speedup with a 95% bootstrap interval. The machine was shared with other work, so the absolute times are only indicative.

Workload	CPython 3.12.3 master → branch (µs)	CPython 3.12.3 speedup	CPython 3.14.8 speedup
browser_mixed_encode	1,794.3 → 1,695.9	1.06x [1.03, 1.08]	1.06x [1.05, 1.08]
browser_mixed_roundtrip	3,938.6 → 3,755.2	1.04x [1.02, 1.06]	1.02x [1.01, 1.04]
repeated_requests_encode	414.7 → 337.5	1.24x [1.22, 1.26]	1.23x [1.22, 1.25]
repeated_requests_roundtrip	953.5 → 872.3	1.10x [1.07, 1.13]	1.11x [1.09, 1.12]
responses_decode	669.4 → 666.5	1.01x [0.99, 1.03]	1.00x [0.99, 1.02]
long_cookie_encode	7,183.9 → 7,029.2	1.00x [0.99, 1.02]	1.01x [1.00, 1.02]
long_cookie_roundtrip	12,117.6 → 11,750.4	1.02x [1.00, 1.03]	1.00x [0.98, 1.01]
no_huffman_encode	1,028.6 → 944.7	1.10x [1.09, 1.12]	1.09x [1.07, 1.10]
fixture_stories_roundtrip	2,086.2 → 1,987.1	1.06x [1.04, 1.08]	1.07x [1.05, 1.08]

Behaviour

• New tests: the indexed encoding at the 7-bit prefix boundary (0, 1, 126, 127, 128, 255, True), invalid values, and a DEBUG-logging round trip that checks the redaction of sensitive values and that the integer-encoding records still come from encode_integer.
• pytest passes with 100% statement and branch coverage on CPython 3.11, 3.12, 3.13 and 3.14; ruff check src/ and mypy src/ (the lint tox env) pass.
• I also compared master and this branch on 1,500 random encoder/decoder sessions per interpreter (CPython 3.11 and 3.14), including repeated headers so the dynamic table is hit, table size changes, sensitive headers and direct _encode_indexed() calls with indices from 0 to 200. The encoded bytes, decoded headers, exceptions and, with DEBUG logging enabled, every log record were identical.

🤖 Generated with Claude Code