allaun/Research-Stack
1
0
Fork 0
mirror of https://github.com/allaunthefox/Research-Stack.git synced 2026-08-15 14:40:35 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
618 commits 18 branches 0 tags 425 MiB
Commit graph

4 commits

Author SHA1 Message Date
Devin AI
f6f9122d24 fix(infra): write_jsonl mkdir+sort_keys, q16_div raise on zero 
- write_jsonl: add path.parent.mkdir(parents=True, exist_ok=True) to
  prevent FileNotFoundError when output dir doesn't exist
- write_jsonl: add sort_keys=True to match original callers' deterministic
  output behavior (AGENTS.md: identical SHA256 on consecutive runs)
- q16_div: raise ZeroDivisionError instead of silently returning 0,
  matching the original fractal_dimension.py semantics

Co-Authored-By: Allaun Silverfox <bigdataiscoming+9i37y6j2@protonmail.com>
 2026-06-16 01:20:56 +00:00
Devin AI
f9dcec7369 fix(infra): restore canonical_json_bytes non-ASCII behavior and dataset_ingest list safety 
- canonical_json_bytes now delegates to canonical_json (ensure_ascii=False)
  matching the original per-file implementations in emitter and loopback
- dataset_ingest_rds.py restores isinstance(data, list) safety check from
  original load_jsonl, wrapping lib.jsonl.load_json

Co-Authored-By: Allaun Silverfox <bigdataiscoming+9i37y6j2@protonmail.com>
 2026-06-16 01:11:35 +00:00
Devin AI
38b48e30d2 fix: preserve semantic differences in refactored shared utilities 
- q16_lut_generator.py: revert to local unsigned-wrapping Q16 functions
  (hardware LUT uses 32-bit wrapping + 0xFFFFFFFF div-by-zero sentinel)
- dataset_ingest_rds.py: import load_json (not load_jsonl) — files are
  standard JSON arrays, not JSONL
- tiddlywiki_ene_bridge.py: use canonical_json (ensure_ascii=False) to
  preserve non-ASCII hash stability
- ollama_deepseek_review_emitter.py: restore "sha256:" prefix wrapper
- md_to_jsonl_converter.py: restore "sha256:" prefix via compute_sha256
- text_container_to_jsonl.py: restore "sha256:" prefix via compute_sha256
- kotc_sim.py: restore "sha256:" prefix on sha256_text
- gccl_waveprobe.py: restore Q16_MIN/Q16_MAX clamping on q16_add/sub/mul
- vcn_famm_transport.py: restore Q16_MIN/Q16_MAX clamping on q16_add/sub/mul
- lib/jsonl.py: add canonical_json() with ensure_ascii=False

Build: py_compile verified on all 71 modified files
Co-Authored-By: Allaun Silverfox <bigdataiscoming+9i37y6j2@protonmail.com>
 2026-06-15 02:26:45 +00:00
Devin AI
e5f04ee6c3 refactor(infra): extract shared utilities from duplicated code patterns 
Create 4-Infrastructure/lib/ with canonical implementations of:
- hashing.py: sha256_bytes, sha256_text, sha256_file
- q16.py: Q16_16 fixed-point constants and arithmetic
- jsonl.py: load_json, load_jsonl, write_jsonl, stable_json, canonical_json_bytes
- fraction_utils.py: Fraction serialization helpers for hardware probes

Refactor 66 files across 4-Infrastructure/{hardware,shim,infra} and
5-Applications/{scripts,tools-scripts,hutter_prize,text-to-cad} to import
from the shared library instead of maintaining local copies.

4-Infrastructure/auto/lib/q16.py now re-exports from lib.q16.

Net: -743 lines (490 added, 1233 removed)

Build: not applicable (Python-only change, py_compile verified on all 71 files)
Co-Authored-By: Allaun Silverfox <bigdataiscoming+9i37y6j2@protonmail.com>
 2026-06-15 02:26:45 +00:00