readability pass on word_count_eval.lua

This commit is contained in:
2026-07-10 17:53:13 -04:00
parent a928d06ac9
commit fa598a41c6
2 changed files with 186 additions and 94 deletions
+71 -7
View File
@@ -1231,10 +1231,20 @@ local function validate(ctx, src)
warnings[#warnings + 1] = { line = f.line, msg = f.msg }
end
end
info[#info + 1] = {
line = 0,
msg = string.format("scanned: %d atom bodies; %d findings", #atoms, #findings),
}
-- Per-source "scanned:" summary line. Includes the source basename
-- for traceability (the old format was just "scanned: N atom
-- bodies; M findings" which is unidentifiable when the module has
-- multiple sources). Sources with 0 atoms (pure-header files like
-- dsl.h, mips.h, etc.) are SKIPPED — the per-module header already
-- lists them in the "Sources:" section, and emitting a noisy
-- "0 atom bodies" line per header is just clutter.
if #atoms > 0 or #findings > 0 then
info[#info + 1] = {
line = 0,
msg = string.format("scanned: %s: %d atom bodies; %d findings",
src.basename, #atoms, #findings),
}
end
-- Phase 3: cycle-budget summary line. Per-path min/max totals.
if #atoms > 0 then
@@ -1476,9 +1486,63 @@ local function emit_module_static_analysis_txt(ctx, dir, dir_sources, atoms, fin
end
add("")
add("── Info ────────────────────────────────────────────────")
for _, i_ in ipairs(info) do
add(string.format(" %s", i_.msg))
add("── Per-source scan summary ──────────────────────────────")
-- One line per source that contributed atoms. The line includes
-- the source basename + per-source atom count + (if path-aware
-- cycle data is present) the min..max cycle range. Sources with
-- 0 atoms are skipped (they're just header files that declared
-- no MipsAtom_ — they're already listed in the module's
-- "Sources:" section above).
--
-- TODO: per-source finding attribution. Currently we can't tell
-- which source a given error/warning came from (errors/warnings
-- only carry atom-name + line, not source-path). The per-atom
-- cycle section already shows which atoms are in which source
-- via the `(file_basename)` suffix. Adding source attribution to
-- error/warning would be a future enhancement.
for _, src in ipairs(dir_sources) do
local src_atoms = {}
for _, a in ipairs(atoms) do
if a.source_path == src.path then
src_atoms[#src_atoms + 1] = a
end
end
if #src_atoms == 0 then
goto continue
end
local atom_count = #src_atoms
local mn, mx = math.huge, -1
for _, a in ipairs(src_atoms) do
local p = a.paths or {}
if (p.cycles_min or 0) < mn then mn = p.cycles_min or 0 end
if (p.cycles_max or 0) > mx then mx = p.cycles_max or 0 end
end
local path_str
if mx > 0 then
path_str = string.format(" cycles=%d..%d", mn, mx)
else
path_str = string.format(" %d cycles", mn)
end
add(string.format(" %-30s %d atom%s%s",
src.basename, atom_count,
atom_count == 1 and "" or "s",
path_str))
::continue::
end
-- Module-level findings summary (across all sources).
local total_errs = #errors
local total_warns = #warnings
add("")
add(string.format("Module findings: %d error(s), %d warning(s)", total_errs, total_warns))
-- Per-source "scanned:" info lines (each line includes the source
-- basename for traceability).
if #info > 0 then
add("")
for _, i_ in ipairs(info) do
add(string.format(" %s", i_.msg))
end
end
write_file(out_path, table.concat(lines, "\n") .. "\n")
+115 -87
View File
@@ -1,39 +1,53 @@
-- word_count_eval.lua
--
-- Word-counting logic for the tape-atom metaprogram pipeline.
-- Used by:
-- - passes/components.lua (compute_component_word_count)
-- - passes/offsets.lua (scan_atom_body)
-- - passes/annotation.lua (TAPE_WORDS <-> WORD_COUNT drift check)
--
-- This module ALSO exposes M.run(ctx) — the "word-counts" pass entry in
-- the PASSES table — which loads metadata.h + scans for existing
-- *.macs.h files into ctx.shared.word_counts.
--
-- Coding standard: tabs (1/level), EmmyLua annotations, no regex,
-- Lua 5.3 compatible.
--- word_count_eval.lua — Word-counting logic for the tape-atom metaprogram
--- pipeline.
---
--- Three responsibilities:
--- 1. **Public utilities** (used by `passes/components.lua`,
--- `passes/offsets.lua`, `passes/annotation.lua`):
--- - `M.count_token_words(token, wc)` — words emitted by one token
--- - `M.scan_dir(dir, suffix)` — glob walk for *.macs.h
--- - `M.count_body_words(body, wc)` — words emitted by an atom body
--- 2. **Pass entry** `M.run(ctx)` — loads metadata.h + *.macs.h into
--- `ctx.shared.word_counts` for downstream passes.
--- 3. **Internal helpers** for the body scanner.
---
--- **Conventions**: tabs (1/level), EmmyLua annotations, no regex,
--- Lua 5.3 compatible. See
--- `C:\projects\Pikuma\ps1-ai\conductor\code_styleguides\lua.md`.
-- ════════════════════════════════════════════════════════════════════════════
-- Module-scope requires + package.path setup
-- ════════════════════════════════════════════════════════════════════════════
-- Resolve `arg[0]` to an absolute-ish script directory so that
-- `require("duffle")` resolves against `scripts/` regardless of CWD.
-- Note: this boilerplate is duplicated in 6 other entry scripts; a
-- Phase-6 extraction target (`duffle.setup_package_path()`).
local script_path = arg and arg[0] or "?"
local last_sep = 0
for i = 1, #script_path do
local c = script_path:sub(i, i)
if c == "/" or c == "\\" then last_sep = i end
for sep_pos = 1, #script_path do
local ch = script_path:sub(sep_pos, sep_pos)
if ch == "/" or ch == "\\" then last_sep = sep_pos end
end
local script_dir = last_sep == 0 and "./" or script_path:sub(1, last_sep)
package.path = script_dir .. "?.lua;" .. script_dir .. "?/init.lua;" .. package.path
package.cpath = "C:\\projects\\Pikuma\\ps1\\toolchain\\luajit-2.1\\lib\\lua\\5.1\\?.dll;" .. package.cpath
local duffle = require("duffle")
local trim = duffle.trim
local read_ident = duffle.read_ident
local skip_ws_and_cmt = duffle.skip_ws_and_cmt
local load_word_counts = duffle.load_word_counts
local split_top_level_commas = duffle.split_top_level_commas
local is_space = duffle.is_space
-- ════════════════════════════════════════════════════════════════════════════
-- Constants
-- ════════════════════════════════════════════════════════════════════════════
-- Windows separator chars — used to convert `dir /b /s` output (which uses
-- `\`) into POSIX paths (which our scripts expect).
local PATH_SEP_BACKSLASH = "\\"
local PATH_SEP_FORWARD = "/"
-- Glob command for Windows directory walk. `dir /b /s` lists all matching
-- files recursively with bare paths (no headers); `2>nul` discards the
-- "file not found" stderr when nothing matches.
local DIR_GLOB_CMD = 'dir /b /s "%s\\%s" 2>nul'
-- ════════════════════════════════════════════════════════════════════════════
-- Type declarations
@@ -42,28 +56,28 @@ local is_space = duffle.is_space
--- @class WordCounts
--- @field [string] integer -- macro name -> word count
--- @class SourceFile
--- @field path string -- absolute path to the source file
--- @field text string -- the full source text
--- @field dir string -- the directory containing the source
--- @field basename string -- filename without extension
--- @class PassCtx
--- @field sources SourceFile[]
--- @field metadata_path string
--- @field shared table
--- @field shared.word_counts WordCounts
--- @field out_root string
--- @field project_root string
--- @field upstream table<string, table>
--- @field flags table
--- @field dry_run boolean
--- @field verbose boolean
--- @field sources SourceFile[] -- all source files in the build
--- @field metadata_path string -- path to word_count.metadata.h
--- @field shared table -- cross-pass shared state
--- @field shared.word_counts WordCounts -- populated by this pass
--- @field out_root string -- output root (e.g. "build/gen")
--- @field project_root string -- project root (e.g. "code/")
--- @field upstream table<string, table> -- per-pass upstream outputs
--- @field flags table -- CLI flags
--- @field dry_run boolean -- if true, compute but don't write
--- @field verbose boolean -- if true, log diagnostic info
--- @class PassResult
--- @field outputs table[]
--- @field errors table[]
--- @field warnings table[]
--- @class SourceFile
--- @field path string
--- @field text string
--- @field dir string
--- @field basename string
--- @field outputs table[] -- {kind=, path=} entries describing emit files
--- @field errors table[] -- {line=, msg=} entries; build-stops
--- @field warnings table[] -- {line=, msg=} entries; build-succeeds
-- ════════════════════════════════════════════════════════════════════════════
-- Module exports
@@ -85,13 +99,13 @@ local M = {}
--- @param wc WordCounts -- the shared word-count table
--- @return integer
function M.count_token_words(token, wc)
local s = trim(token)
local s = duffle.trim(token)
if s == "" then return 0 end
local name, after = read_ident(s, 1)
local name, after = duffle.read_ident(s, 1)
if not name then return 1 end
if wc[name] then return wc[name] end
local j = skip_ws_and_cmt(s, after)
if s:sub(j, j) == "(" then
local paren_pos = duffle.skip_ws_and_cmt(s, after)
if s:sub(paren_pos, paren_pos) == "(" then
io.stderr:write(" warning: unknown macro '" .. name .. "', assuming 1 word\n")
end
return 1
@@ -110,13 +124,13 @@ end
--- @return string[]
function M.scan_dir(dir, suffix)
local results = {}
local p = io.popen('dir /b /s "' .. dir .. '\\' .. suffix .. '" 2>nul')
if not p then return results end
for raw_line in p:lines() do
local path = raw_line:gsub("\\", "/")
local pipe = io.popen(DIR_GLOB_CMD:format(dir, suffix))
if not pipe then return results end
for raw_line in pipe:lines() do
local path = raw_line:gsub(PATH_SEP_BACKSLASH, PATH_SEP_FORWARD)
results[#results + 1] = path
end
p:close()
pipe:close()
return results
end
@@ -134,13 +148,16 @@ end
--- @param wc WordCounts -- the shared word-count table
--- @return integer -- total words
function M.count_body_words(body, wc)
local pos = 0
for _, tok in ipairs(split_top_level_commas(body)) do
local k = 1
local tlen = #tok
while k <= tlen and is_space(tok:sub(k, k)) do k = k + 1 end
local leading_ident = read_ident(tok, k)
if leading_ident == "atom_label" or leading_ident == "atom_offset" then
local total = 0
for _, tok in ipairs(duffle.split_top_level_commas(body)) do
local pos = 1
local tok_len = #tok
while pos <= tok_len and duffle.is_space(tok:sub(pos, pos)) do
pos = pos + 1
end
local leading_ident = duffle.read_ident(tok, pos)
local is_marker = leading_ident == "atom_label" or leading_ident == "atom_offset"
if is_marker then
-- Marker call: record at current pos, do NOT advance pos.
-- But the source pattern may bundle the marker with the next
-- instruction on a new line (no top-level comma between them).
@@ -148,17 +165,16 @@ function M.count_body_words(body, wc)
-- a real instruction that must still be counted.
local marker_end = M.find_marker_call_end(tok)
if marker_end > 0 and marker_end < #tok then
local rest = trim(tok:sub(marker_end + 1))
local rest = duffle.trim(tok:sub(marker_end + 1))
if rest ~= "" then
local rest_words = M.count_token_words(rest, wc)
pos = pos + rest_words
total = total + M.count_token_words(rest, wc)
end
end
else
pos = pos + M.count_token_words(tok, wc)
total = total + M.count_token_words(tok, wc)
end
end
return pos
return total
end
--- Find the end position (just past the closing ')') of the first
@@ -168,34 +184,42 @@ end
--- @param tok string
--- @return integer -- 0 if no marker call found
function M.find_marker_call_end(tok)
local i = 1
local len = #tok
while i <= len do
i = skip_ws_and_cmt(tok, i)
if i > len then break end
local c = tok:sub(i, i)
if is_space(c) then
i = i + 1
elseif c == "/" then
local pos = 1
local tok_len = #tok
while pos <= tok_len do
pos = duffle.skip_ws_and_cmt(tok, pos)
if pos > tok_len then break end
local ch = tok:sub(pos, pos)
if duffle.is_space(ch) then
pos = pos + 1
elseif ch == "/" then
-- comment — skip past it (delegated to duffle.skip_str_or_cmt)
local nx = duffle.skip_str_or_cmt(tok, i)
if nx > i then i = nx else i = i + 1 end
local nx = duffle.skip_str_or_cmt(tok, pos)
pos = (nx > pos) and nx or (pos + 1)
else
local ident, after = read_ident(tok, i)
if ident == "atom_label" or ident == "atom_offset" then
local j = skip_ws_and_cmt(tok, after)
if tok:sub(j, j) == "(" then
local _, end_paren = duffle.read_parens(tok, j)
return end_paren - 1
end
return 0
end
i = after or (i + 1)
local ident, after_ident = duffle.read_ident(tok, pos)
local marker_end = find_marker_end(tok, ident, after_ident)
if marker_end > 0 then return marker_end end
pos = after_ident or (pos + 1)
end
end
return 0
end
-- (internal) If `ident` is `atom_label`/`atom_offset` followed by `(...)`,
-- return the position just past the closing ')'. Otherwise 0.
-- @param tok string
-- @param ident string|nil
-- @param after_ident integer
-- @return integer
local function find_marker_end(tok, ident, after_ident)
if ident ~= "atom_label" and ident ~= "atom_offset" then return 0 end
local open_paren = duffle.skip_ws_and_cmt(tok, after_ident)
if tok:sub(open_paren, open_paren) ~= "(" then return 0 end
local _, end_paren = duffle.read_parens(tok, open_paren)
return end_paren - 1
end
-- ┌────────────────────────────────────────────────────────────────────┐
-- │ Pass entry: M.run(ctx) — "word-counts" pass │
-- └────────────────────────────────────────────────────────────────────┘
@@ -211,14 +235,18 @@ function M.run(ctx)
local wc = {}
-- 1. Load metadata.h (the encoding-macro source of truth).
local meta_counts = load_word_counts(ctx.metadata_path)
local meta_counts = duffle.load_word_counts(ctx.metadata_path)
for name, count in pairs(meta_counts) do wc[name] = count end
-- 2. Scan project_root recursively for *.macs.h files (component-macro source).
local macs_files = M.scan_dir(ctx.project_root, "*.macs.h")
for _, macs_path in ipairs(macs_files) do
local ok, mc = pcall(load_word_counts, macs_path)
if ok and type(mc) == "table" then
local ok, mc = pcall(duffle.load_word_counts, macs_path)
if not ok then
io.stderr:write(string.format("[word_count_eval] parse error in '%s': %s\n", macs_path, tostring(mc)))
elseif type(mc) ~= "table" then
io.stderr:write(string.format("[word_count_eval] '%s' did not return a table (got %s)\n", macs_path, type(mc)))
else
for name, count in pairs(mc) do wc[name] = count end
end
end