{esc(standfirst)}
" if standfirst else "" raw = "" if raw_name: raw = ( f'█ ' f"raw flat file — {esc(raw_name)}
" ) meta = f" · {rows}" if rows else "" return f"""krisrockwell.io / neuromancer-semiotics / data / {esc(raw_name or title)}
██╹{esc(kicker)}{meta}
{count} {raw}Rendered from {esc(raw_name or '')} by scripts/build_data_pages.py. The flat file is canonical; this page is a view of it. · back to the study
| document | id | ' f"{head} |
|---|
| {esc(left)} | {esc(right)} | ' "
|---|
{esc(seg["DEF"])}
') for k in ("INCLUDE", "EXCLUDE"): if seg.get(k): bits.append( f'{k.lower()}{esc(seg[k])}
' ) items.append(f', so column alignment survives untouched."""
raw = read(name).splitlines()
blocks, buf, i = [], [], 0
rule = re.compile(r"^[=\-]{8,}\s*$")
# some reports title their sections with a bare uppercase line instead of
# a fenced banner; the trailing group allows an em-dash gloss
bare = re.compile(r"^[A-Z][A-Z0-9 ]{5,}(\s*[—-].*)?$")
def flush():
while buf and not buf[-1].strip():
buf.pop()
while buf and not buf[0].strip():
buf.pop(0)
if buf:
blocks.append(("pre", "\n".join(buf)))
buf.clear()
while i < len(raw):
line = raw[i]
# a title fenced by rule lines above and/or below
if rule.match(line):
nxt = raw[i + 1] if i + 1 < len(raw) else ""
after = raw[i + 2] if i + 2 < len(raw) else ""
if nxt.strip() and not rule.match(nxt) and rule.match(after):
flush()
blocks.append(("h2", nxt.strip()))
i += 3
continue
if nxt.strip() and not rule.match(nxt) and not raw[i - 1:i]:
pass
i += 1
continue
if bare.match(line):
flush()
blocks.append(("h2", line.strip()))
i += 1
continue
buf.append(line)
i += 1
flush()
out = []
for kind, text in blocks:
if kind == "h2":
out.append(f"{esc(text)}
")
else:
out.append(f'{esc(text)}')
return page(title, kicker, standfirst, name, "".join(out))
def render_passages(name, title, kicker, standfirst):
"""[P1] ... [P2] ... prose passages."""
text = read(name)
parts = re.split(r"\[(P\d+)\]", text)
out = []
for i in range(1, len(parts) - 1, 2):
tag, body = parts[i], parts[i + 1].strip()
out.append(
f'{esc(tag)}'
f"{esc(body)}
"
)
return page(title, kicker, standfirst, name, "".join(out), f"{len(out)} passages")
# ── file registry ────────────────────────────────────────────────────────────
FILES = [
("codes_matrix_pass2.txt", "The Matrix", "second pass · every code applied",
"Every coded passage after the second pass: the chapter document, the passage id, "
"and each code applied to it. Every number on the study page is computed from this "
"file. MODE codes carry the validated triad; all other families are set in neutral ink.",
lambda: render_coded("codes_matrix_pass2.txt", "The Matrix",
"second pass · every code applied",
"Every coded passage after the second pass: the chapter document, "
"the passage id, and each code applied to it. Every number on the "
"study page is computed from this file.")),
("pass2_tables.txt", "Co-occurrence", "all crosstabs · second pass",
"Every co-occurrence matrix the study reports: cue against mode, target against mode, "
"target against verdict, target against cue, verdict against inference.",
lambda: render_report("pass2_tables.txt", "Co-occurrence",
"all crosstabs · second pass",
"Every co-occurrence matrix the study reports: cue against mode, "
"target against mode, target against verdict, target against cue, "
"verdict against inference. Column alignment is preserved exactly.")),
("codebook.txt", "The Instrument", "69 codes · fourteen families",
"The codebook itself: every code with its definition, what it includes and what it "
"excludes. This is the instrument that was applied — and the only thing the blind "
"coder was given besides the passages.",
lambda: render_codebook("codebook.txt", "The Instrument",
"69 codes · fourteen families",
"Every code with its definition, inclusions and exclusions. This is "
"the instrument that was applied, and the only thing the blind coder "
"was given besides the passages.")),
("blind_input.txt", "Blind Input", "exactly what the second coder saw",
"The 36 passages handed to an independent coder — chapter, passage id, and the text. "
"No codes, no context, no argument.",
lambda: render_coded("blind_input.txt", "Blind Input",
"exactly what the second coder saw",
"The 36 passages handed to an independent coder — chapter, passage "
"id, and the text. No codes, no context, no argument.",
quote_col=True)),
("blind_output.txt", "Blind Output", "exactly what they returned",
"What the independent coder returned, against passage id. Compare with the matrix to "
"see where the two readings part.",
lambda: render_pairs("blind_output.txt", "Blind Output",
"exactly what they returned",
"What the independent coder returned, against passage id. Compare with "
"the matrix to see where the two readings part.",
"passage id", "codes returned", True)),
("blind_test_results.txt", "Agreement", "where the two coders diverged",
"Agreement statistics by family and every passage-level disagreement. The verdict "
"family — which carries every headline — agreed 33% of the time.",
lambda: render_report("blind_test_results.txt", "Agreement",
"where the two coders diverged",
"Agreement statistics by family and every passage-level disagreement. "
"The verdict family, which carries every headline, agreed 33% of the time.")),
("selection_test_results.txt", "Selection", "what the corpus missed",
"A blind selection test on three chapters, spans only. The independent reader marked "
"passages this study never saw: 38% coverage overall.",
lambda: render_report("selection_test_results.txt", "Selection",
"what the corpus missed",
"A blind selection test on three chapters, spans only. The independent "
"reader marked passages this study never saw: 38% coverage overall.")),
("blind_spans.txt", "Blind Spans", "every span the independent reader marked",
"The raw spans an independent reader selected across three chapters, before any codes "
"were applied. The basis of the selection test.",
lambda: render_pairs("blind_spans.txt", "Blind Spans",
"every span the independent reader marked",
"The raw spans an independent reader selected across three chapters, "
"before any codes were applied. The basis of the selection test.",
"chapter", "span", False)),
("loopA_test.txt", "Loop A", "instrument shakedown · eleven passages",
"An early shakedown pass: eleven passages pulled to test whether the codebook could be "
"applied at all before the full corpus was attempted.",
lambda: render_passages("loopA_test.txt", "Loop A",
"instrument shakedown · eleven passages",
"An early shakedown pass: eleven passages pulled to test whether the "
"codebook could be applied at all before the full corpus was attempted.")),
]
def build_directory():
rows = []
for fname, title, kicker, blurb, _ in FILES:
stem = fname[:-4]
size = (DATA / fname).stat().st_size
n = len([l for l in (DATA / fname).read_text(encoding="utf-8").splitlines() if l.strip()])
rows.append(
f'{esc(title)}'
f'{esc(blurb)}'
f"{esc(fname)} · {n} lines · {size:,} bytes · "
f'raw'
)
body = (
"Nine flat files. Each is rendered here as a readable view; each view links "
"back to the raw file it was built from. The flat files are canonical — the "
"whole matrix can be recomputed from them at any point, which is what makes the "
"audit possible.
"
'' + "".join(rows) + ""
)
return page("Data", "the coded corpus · nine flat files",
"Everything the study was computed from: the coded matrix, the instrument, "
"the blind re-code, and the tests that refuted it.",
None, body)
def main():
written = []
for fname, _t, _k, _b, fn in FILES:
out = DATA / (fname[:-4] + ".html")
out.write_text(fn(), encoding="utf-8")
written.append((out.name, out.stat().st_size))
idx = DATA / "index.html"
idx.write_text(build_directory(), encoding="utf-8")
written.append((idx.name, idx.stat().st_size))
for name, size in written:
print(f" wrote data/{name} {size:,} bytes")
print(f"{len(written)} pages")
if __name__ == "__main__":
main()