feat(master_data): 건설 기계설비 7장 소요량 옮김

- 소요량_건설품셈_기계설비7장.json (본문 표 18 + 계수 표 17)
- 변환 스크립트 build_소요량_건설품셈_기계설비.py · _설비_읽기 · _설비_표 · _설비_계수 · _설비_손 추가
- _토목_계수.lift 에 규칙 함수 인자 추가

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PAdA5ThqmtVSsbzjusk1cJ
This commit is contained in:
2026-09-19 17:45:10 +09:00
co-authored by Claude Sonnet 5
parent 7bc039a835
commit 9f617fdbfa
7 changed files with 1130 additions and 2 deletions
@@ -0,0 +1,67 @@
# -*- coding: utf-8 -*-
"""기계설비 글 속 계산용 수 → 계수 표 줄 (토목 규칙 + 기계설비 규칙)."""
import re
from _md표_읽기 import R
from _토목_계수 import clean
from _토목_계수 import rows_of as base_rows
PCT = r"([\d.]+)%"
def rows_of(line):
"""주 한 줄 → 계수 줄들 [(구분, 값, 단위, 딸린 칸)]."""
s = clean(line)
out = base_rows(re.sub(r"(인력품|주재료비|재료비)의\([^)]*\) ", r"\1의 ", line))
if out:
return out
m = re.search(r"^(.+?)(?:은|는) 본 품에 " + PCT + r"를 (가산|감)", s)
if m:
return [(m.group(1), R(m.group(2)), "%", {"기준": "본 품", "가감": m.group(3)})]
m = re.search(
r"(\d+)본 감할 때마다 " + PCT + r"씩 감하고, (\d+)본 증할 때마다 " + PCT + r"씩 가산", s
)
if m:
return [
(
f"튜브 {m.group(1)}본 감할 때마다",
R(m.group(2)),
"%",
{"기준": "본 품", "가감": ""},
),
(
f"튜브 {m.group(3)}본 증할 때마다",
R(m.group(4)),
"%",
{"기준": "본 품", "가감": "가산"},
),
]
m = re.search(r"^(.+? 3절 초과하는 경우 매 1절 증가마다) " + PCT + r"씩 가산", s)
if m:
return [(m.group(1), R(m.group(2)), "%", {"기준": "본 품", "가감": "가산"})]
m = re.search(r"^(.+?(?:증가마다)) " + PCT + r"씩 가산", s)
if m:
return [(m.group(1), R(m.group(2)), "%", {"기준": "본 품", "가감": "가산"})]
m = re.search(r"^(.+?구조일 경우) " + PCT + r" 가산", s)
if m:
return [(m.group(1), R(m.group(2)), "%", {"기준": "본 품", "가감": "가산"})]
m = re.search(
r"(높이 69m까지는) 품을 "
+ PCT
+ r" 가산하고 (높이 9m를 초과하는 경우 매 3m 증가마다) 품을 "
+ PCT,
s,
)
if m:
return [
(f"비계 사용 {m.group(1)}", R(m.group(2)), "%", {"기준": "본 품", "가감": "가산"}),
(f"비계 사용 {m.group(3)}", R(m.group(4)), "%", {"기준": "본 품", "가감": "가산"}),
]
m = re.search(r"단독주택 (\d+)호당 1조 및 집단아파트 (\d+)호당 1조", s)
if m:
return [
("단독주택 1조당 호수", R(m.group(1)), "", {}),
("집단아파트 1조당 호수", R(m.group(2)), "", {}),
]
return []
@@ -0,0 +1,6 @@
# -*- coding: utf-8 -*-
"""기계설비 표별 지정 — 「항#순번」: convert 옵션(H 머리 줄 수 · L 라벨 칸 수 · M 직종 칸 · -1 = 직종 칸 없음)."""
SPEC = {
"9-5-1#0": {"L": 3},
}
@@ -0,0 +1,126 @@
# -*- coding: utf-8 -*-
"""건설품셈 기계설비 제7~13장 md 읽기 (build_소요량_건설품셈_기계설비.py 가 씀).
항(###) · 목(####) · 표 · 표 머리 깊이(주석) · 표 앞 글 · [주] [비고] [계산예] 글을 읽음.
"""
import re
from _md표_읽기 import unesc
from _토목_읽기 import ROOT
BASE = ROOT / "resources/knowledge/original/원가계산/건설공사_표준품셈/본문/04_기계설비부문"
DIV = "건설품셈 기계설비 "
NOTE0 = re.compile(r"^[①-⑳]")
BLOCK = re.compile(r"^\[(주|비고|계산예|참고|참고자료|별표)\]\s*(.*)$")
class Tab:
def __init__(self, ident, title, mok, base, rows, k, hdr, pre):
self.ident, self.title, self.mok, self.base = ident, title, mok, base
self.rows, self.k, self.hdr, self.pre = rows, k, hdr, pre
self.notes: list[str] = []
def cells(line):
return [c.strip() for c in line.strip().strip("|").split("|")]
def title_of(s):
return re.sub(r"\s*\((?:['‘’]\d\d[^)]*)\)\s*$", "", s).strip()
def clean(ln):
ln = re.sub(r"<!--.*?-->", "", ln).rstrip()
return re.sub(r"^>\s?", "", ln)
def read_chapter(chdir):
"""장 폴더 → 표 목록 (표마다 항 · 목 · 기준 · 머리 깊이 · 앞 글 · 주)."""
tabs = []
for p in sorted(next(BASE.glob(chdir + "*")).glob("*.md")):
L = p.read_text(encoding="utf-8").split("\n")
L = L[[i for i, x in enumerate(L) if x.strip() == "---"][1] + 1 :]
ident = title = mok = base = ""
pre: list[str] = []
k, mode, i = 0, "", 0
while i < len(L):
raw = L[i]
ln = clean(raw)
m = re.match(r"^(#{2,4})\s+(.*)$", ln)
if m:
mode, base, pre = "", "", []
d = len(m.group(1))
if d == 2:
mm = re.match(r"^(\d+-\d+)\s+(.*)$", m.group(2))
ident, title, mok, k = mm.group(1), title_of(mm.group(2)), "", 0
elif d == 3:
mm = re.match(r"^(\d+-\d+-\d+)\s+(.*)$", m.group(2))
ident, title, mok, k = mm.group(1), title_of(mm.group(2)), "", 0
elif d == 4:
mk = re.match(r"^(?:\d+|[가-힣])\.\s*(.*)$", m.group(2))
mok = title_of(mk.group(1) if mk else m.group(2))
i += 1
continue
if not ident:
i += 1
continue
if ln.startswith("|"):
j, rows = i, []
while j < len(L) and L[j].startswith("|"):
c = cells(L[j])
if not all(re.fullmatch(r":?-+:?", x) for x in c):
rows.append(c)
j += 1
hint = ""
while j < len(L) and (L[j].startswith("<!--") or not L[j].strip()):
hint += L[j] if L[j].startswith("<!--") else ""
if not L[j].strip() and not (j + 1 < len(L) and L[j + 1].startswith("<!--")):
break
j += 1
mh = re.search(r"몸 (\d+)행까지 머리", hint)
b = base[1:-1] if base else ""
tabs.append(
Tab(ident, title, mok, b, rows, k, 1 + int(mh.group(1)) if mh else 1, pre)
)
k += 1
base, pre, mode = "", [], ""
i = j
continue
s = ln.strip()
i += 1
if not s or s.startswith("!["):
continue
mb = BLOCK.match(s)
if mb:
kind, rest = mb.groups()
mode = "note" if kind in ("", "계산예", "비고") else ""
if kind in ("참고", "참고자료", "별표"):
pre = [s] if not rest else [f"[{kind}] {rest}"]
mode = ""
continue
first = rest if kind == "" else f"{kind} {rest}".strip()
if first and tabs and tabs[-1].ident == ident:
tabs[-1].notes.append(unesc(first))
mode = "note" if kind in ("", "계산예") else ""
if kind == "계산예":
mode = "calc"
continue
if re.fullmatch(r"\(.*\)", s) and not s.startswith("(주"):
base, mode = s, ""
continue
if mode in ("note", "calc") and tabs and tabs[-1].ident == ident:
t = tabs[-1]
if mode == "calc":
t.notes.append("계산예 " + unesc(s)) if not t.notes or not t.notes[
-1
].startswith("계산예") else t.notes.__setitem__(
-1, t.notes[-1] + " " + unesc(s)
)
elif NOTE0.match(s) or not t.notes:
t.notes.append(unesc(s))
else:
t.notes[-1] += " " + unesc(s)
else:
pre.append(s)
return tabs
@@ -0,0 +1,334 @@
# -*- coding: utf-8 -*-
"""기계설비 표 만들기 — 원문 표 하나를 조건 칸 + 값 칸의 줄로 폄 (build_소요량_건설품셈_기계설비.py 가 씀).
표 모양은 스스로 알아냄(라벨 칸 수 · 머리 층 · 직종 위치). 못 맞추는 표는 SPEC 으로 지정.
"""
import re
from _md표_읽기 import N, R, join
from _토목_표 import bigo_items, conds_out, is_unit, jn
from _토목_읽기 import cell_val, is_num
MEMBER = re.compile(
r"(공|인부|기사|기능사|운전사|운전조|측량사|조공|반장|감독원|기술자|보조원|기능공|공사)$"
)
MEASURE = re.compile(
r"^(수량|시공량|소요량|사용기간|사용시간|시간|소비량|소요전력|용접봉소비량|중량|작업능력|인력)"
)
LABELNAME = re.compile(
r"^(구분|규격|구경|관경|호칭|외경|내경|두께|길이|용량|치수|사이즈|Pipe Size|압력|명칭|품명|장비명|"
r"직종|공종|공정별|작업구분|종류|형식|재질|층|이름|단위|비고|배관구분|작업횟수|기간|처리)"
)
VALUEISH = re.compile(r"수량|시공량|소비량|소요|인력|시간|기간|작업능력")
DASH = ("", "-")
def strip_unit(name):
"""「배관공(인)」 → (배관공, 인)."""
m = re.search(r"\(([^()]{1,8})\)$", name)
if m and is_unit(m.group(1)):
return name[: m.start()].strip(), m.group(1)
return name, ""
def is_member(x):
return bool(MEMBER.search(re.sub(r"\([^)]*\)", "", jn(x)).strip()))
def nm(x):
"""직종 이름 — 「비계 공」 처럼 줄바꿈으로 벌어진 글은 붙임."""
x = jn(x)
return (
x.replace(" ", "") if re.fullmatch(r"[가-힣 ]+", x) and is_member(x.replace(" ", "")) else x
)
def cv(x):
"""값 칸 하나 → (값, 원문). 「a∼b」 는 [아래, 위]."""
x = jn(x)
m = re.fullmatch(r"(\d[\d,]*(?:\.\d+)?)\s*[~]\s*(\d[\d,]*(?:\.\d+)?)", x)
if m:
return [R(m.group(1)), R(m.group(2))], None
return cell_val(x)
def vlike(x):
x = x.strip()
return x in DASH or is_num(x) or bool(re.fullmatch(r"[\d.,]+\s*%?", x)) or x == ""
def halves(rows, H):
"""왼쪽 · 오른쪽 짝 표(머리 이름이 되풀이) → 위아래로 이음."""
hd = rows[0]
n = len(hd)
if n % 2 == 0 and any(hd) and all(r[: n // 2] == r[n // 2 :] for r in rows[:H]):
left = [r[: n // 2] for r in rows]
right = [r[: n // 2] for r in rows[:H]] + [
r[n // 2 :] for r in rows[H:] if any(x not in DASH for x in r[n // 2 :])
]
return left + right[H:]
return rows
def label_cols(rows, H):
"""왼쪽 라벨 칸 수 L."""
ncol, body = len(rows[0]), rows[H:]
def textual(c):
vs = [r[c] for r in body if r[c] not in DASH]
return bool(vs) and sum(not vlike(v) or v == "" for v in vs) * 2 >= len(vs)
L = 1
while L < ncol - 1:
c, top = L, jn(rows[0][L])
if H > 1:
below = [rows[r][c] for r in range(1, H - 1)]
last = rows[H - 1][c]
span = not any(below) and (last == "" or is_unit(jn(last)))
ok = span and not VALUEISH.search(top) and not is_member(top)
ok = ok and (top == "" or textual(c) or bool(LABELNAME.match(top)))
else:
ok = textual(c) or bool(LABELNAME.match(top) and not VALUEISH.search(top))
ok = ok and not (top in ("수량",) or MEASURE.match(top))
ok = ok and any(not textual(k) for k in range(c + 1, ncol))
if not ok:
break
L += 1
return L
def layers(rows, H, start, ncol):
"""머리 층 H 줄 → [층][열] 글(빈 칸은 같은 윗 무리 안에서 왼쪽 값으로 채움)."""
out = []
for r in range(H):
row = [jn(x) for x in rows[r]]
for c in range(start if r == 0 else start + 1, ncol):
if row[c] or c == 0:
continue
same = all(out[q][c] == out[q][c - 1] for q in range(r))
if row[c - 1] and same and (r == 0 or out[r - 1][c]):
row[c] = row[c - 1]
out.append(row)
return out
def measure_hm(entries):
es = {e for e in entries if e}
return bool(es) and sum(is_member(e) for e in es) * 2 >= len(es)
def measure_score(entries):
es = [e for e in entries if e]
if not es:
return 0
return sum(bool(is_member(e) or MEASURE.match(e) or strip_unit(e)[1]) for e in set(es)) / len(
set(es)
)
def convert(t, spec=None):
"""표 → (조건 사전, 값칸 사전, 줄 목록)."""
spec = spec or {}
rows, H = [list(r) for r in t.rows], spec.get("H", t.hdr)
rows = halves(rows, H)
ncol = max(len(r) for r in rows)
rows = [r + [""] * (ncol - len(r)) for r in rows]
bigo = []
for r in rows[H:]:
if jn(r[0]) == "비고":
bigo += bigo_items(" ".join(x for x in r[1:] if x))
rows = rows[:H] + [r for r in rows[H:] if jn(r[0]) != "비고"]
body = rows[H:]
L = spec.get("L") or label_cols(rows, H)
unitc = next(
(c for c in range(L) if jn(rows[H - 1][c]) == "단위" or jn(rows[0][c]) == "단위"), None
)
lab_unit = None # 머리 없는 「〃」 단위 칸 = 앞 라벨 칸의 단위
if unitc is None:
unitc = next(
(
c
for c in range(1, L)
if sum(r[c] == "" for r in body) * 2 >= max(len(body) - 1, 1)
),
None,
)
lab_unit = unitc
if lab_unit is not None and not any(
measure_hm([jn(rows[r][c]) for c in range(L, ncol)]) for r in range(H)
):
unitc = lab_unit = None # 값 머리가 직종이 아니면 그냥 라벨 칸
# 라벨 칸 이름
names = {}
for c in range(L):
top, low = jn(rows[0][c]), jn(rows[H - 1][c]) if H > 1 else ""
names[c] = (
f"{top}({low})" if low and is_unit(low) else (low if c == 0 and low and H > 1 else top)
) or "구분"
if lab_unit is not None and any(
measure_hm([jn(rows[r][c]) for c in range(L, ncol)]) for r in range(H)
):
u0 = next((jn(r[lab_unit]) for r in body if r[lab_unit] not in ("", "")), "")
names[lab_unit - 1] += f"({u0})"
lab_unit = -1
# 직종 칸(줄에 직종이 있는 표)
txt = [c for c in range(L) if c != unitc]
mc = spec.get("M", None)
head_members = any(measure_hm([jn(rows[r][c]) for c in range(L, ncol)]) for r in range(H))
if mc is None and unitc is not None and not head_members:
mc = max(c for c in txt if c < unitc) if any(c < unitc for c in txt) else None
if mc is None and txt:
last = txt[-1]
vs = [r[last] for r in body if r[last] not in DASH]
if vs and sum(is_member(v) for v in vs) * 2 >= len(vs):
mc = last
if mc == -1:
mc = None
if ( # 첫 칸이 값 이름인 옆으로 누운 표
mc is None
and L == 1
and unitc is None
and not head_members
and H == 1
and body
and all(r[0] not in DASH and not vlike(r[0]) and not is_member(r[0]) for r in body)
):
mc = 0
names[0] = jn(rows[0][0]) or "구분"
tposed = True
else:
tposed = False
# 값 열 머리 층
vcols = list(range(L, ncol))
lay = layers(rows, H, L, ncol)
corner = [jn(rows[r][0]) for r in range(H)]
named = H > 1 and all(corner) and len(set(corner)) == H
if named:
names[0] = corner[-1]
lnames = [corner[r] if named else "" for r in range(H)]
if tposed:
lnames = [names[0]]
mlay, ulay = None, set()
for r in range(H):
ents = [lay[r][c] for c in vcols]
if ents and all(is_unit(e) for e in ents if e) and any(ents):
ulay.add(r)
if mc is None:
sc = {r: measure_score([lay[r][c] for c in vcols]) for r in range(H) if r not in ulay}
if sc:
best = max(sc, key=lambda r: (sc[r], r))
if sc[best] >= 0.5:
mlay = best
else: # 직종이 줄에 있음 — 값 이름 층은 「수량」 따위뿐
for r in range(H):
if (
r not in ulay
and lay[r]
and all(MEASURE.match(lay[r][c]) for c in vcols if lay[r][c])
and any(lay[r][c] for c in vcols)
):
mlay = r
only = {
r
for r in range(H)
if any(lay[r][c] for c in vcols)
and all(lay[r][c] in ("", "수량", "시공량", "소요량") for c in vcols)
}
dropped = ulay | only | ({mlay} if mlay is not None else set())
clay = [r for r in range(H) if r not in dropped]
# 몸 줄 읽기
recs, prev, prevu = [], {}, ""
for r in body:
lab = {}
for c in range(L):
if c == unitc or c == mc:
continue
x = jn(r[c])
if x in ("", "") and not any(r[q] for q in range(c)):
x = prev.get(c, "")
lab[c] = x
prev.update(lab)
if not any(x for x in lab.values()) and not any(r[c] for c in range(L)):
continue
unit = ""
if unitc is not None:
u = jn(r[unitc])
prevu = u if u and u != "" else prevu
unit = "" if lab_unit == -1 else prevu
member = nm(r[mc]) if mc is not None else ""
made = 0
for c in vcols:
v = r[c].strip()
if v in DASH:
continue
u2, meas = unit, "수량"
if mlay is not None:
meas, uu = strip_unit(nm(lay[mlay][c]))
u2 = u2 or uu
if member:
meas = member
for q in ulay:
u2 = u2 or lay[q][c]
cond = {}
for c0 in range(L):
if c0 in lab and lab[c0]:
cond[names[c0]] = lab[c0]
path = [lay[q][c] for q in clay if lay[q][c]]
un = [q for q in clay if lnames[q] and lay[q][c]]
if un:
for q in un:
cond[lnames[q]] = lay[q][c]
path = [lay[q][c] for q in clay if lay[q][c] and not lnames[q]]
if path:
cond["구분" if "구분" not in cond else "세구분"] = " ".join(dict.fromkeys(path))
recs.append((cond, meas, v, u2))
made += 1
if not made and mc is not None and member and not is_member(member):
cond = {names[c0]: lab[c0] for c0 in lab if lab[c0]}
recs.append((cond, "수량", member, unit))
return (*pack(recs, spec, t), bigo)
def pack(recs, spec, t):
"""(조건, 값이름, 값, 단위) 목록 → 조건 · 값칸 · 줄."""
unit_of, multi = {}, False
groups = {}
for cond, meas, v, u in recs:
key = tuple(cond.items())
g = groups.setdefault(key, {})
if meas in g:
multi = True
g[meas] = (v, u)
if unit_of.setdefault(meas, u) != u:
multi = True
rows, cols = [], {}
if not multi:
for key, g in groups.items():
d = dict(key)
for meas, (v, u) in g.items():
val, orig = cv(v)
cols.setdefault(meas, unit_of[meas])
d[meas] = val
if orig:
d[meas + "원문"] = orig
cols[meas + "원문"] = ""
rows.append(d)
cn = list(dict.fromkeys(k for r in rows for k in r if k not in cols))
else: # 값 이름이 조건으로 — 값은 「수량」 하나 · 줄마다 단위
for cond, meas, v, u in recs:
d = dict(cond)
d["항목"] = meas
val, orig = cv(v)
d["수량"] = val
if orig:
d["수량원문"] = orig
if u:
d["단위"] = u
rows.append(d)
cols = {"수량": ""}
if any("수량원문" in r for r in rows):
cols["수량원문"] = ""
cn = list(dict.fromkeys(k for r in rows for k in r if k not in cols and k != "단위"))
cond = conds_out(cn, rows)
return cond, cols, rows
@@ -62,13 +62,13 @@ def rows_of(line):
return out
def lift(chdir, text, out):
def lift(chdir, text, out, rows_fn=rows_of):
src_tabs = list(out.tables)
for t in src_tabs:
ident = t["열쇠"].split()[0]
tags = {}
for line in t[""]:
rs = rows_of(line)
rs = rows_fn(line)
if not rs:
continue
tag = "비고" if line.startswith("비고") else "" + re.match(r"[①-⑳]", line).group(0)
@@ -0,0 +1,53 @@
# -*- coding: utf-8 -*-
"""건설품셈 기계설비 제7~13장 본문 md → resources/master_data/소요량_건설품셈_기계설비{7..13}장.json
값은 본문 md 에서만 읽음(손으로 적지 않음). 표 모양은 _설비_표.convert 가 알아내고 못 맞추는 표만 SPEC 으로 지정.
돌리기: ./venv/Scripts/python.exe resources/master_data/scripts/build_소요량_건설품셈_기계설비.py [7 8 …]
"""
import sys
import _토목_읽기 as base
from _설비_계수 import rows_of
from _설비_읽기 import DIV, read_chapter
from _설비_표 import convert
from _설비_손 import SPEC
from _토목_계수 import lift
from _토목_읽기 import ROOT, Out
base.DIV = DIV # 출처 앞말 「건설품셈 기계설비 」
def label_of(t, count):
n = f"{t.title} {t.mok}" if t.mok else t.title
extra = " ".join(x for x in t.pre if not x.startswith("[별표]"))
kinds = [p for p in t.pre if p.startswith("[")]
if kinds:
n += (
" " + kinds[0].strip("[]").split("] ")[-1]
if "]" in kinds[0] and not kinds[0].endswith("]")
else ""
)
return n
def build(ch, out_name):
tabs = read_chapter(f"{ch:02d}")
out, seen = Out(), {}
for t in tabs:
key = f"{t.ident}#{t.k}"
cond, cols, rows, bigo = convert(t, SPEC.get(key))
lab = label_of(t, None)
seen[(t.ident, lab)] = seen.get((t.ident, lab), 0) + 1
if seen[(t.ident, lab)] > 1:
lab += f"{seen[(t.ident, lab)]}"
out.add(t, lab, cond, cols, rows, notes=bigo + t.notes)
lift("", "", out, rows_of)
out.write(ROOT / "resources/master_data" / out_name)
return out
if __name__ == "__main__":
for w in sys.argv[1:] or ["7", "8", "9", "10", "11", "12", "13"]:
o = build(int(w), f"소요량_건설품셈_기계설비{w}장.json")
print(w, len(o.tables), "")