# Extract Juzhiyun manual: TOC + master-data / order / schedule chapters
from __future__ import annotations
import glob, os, re, fitz

ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
pdf = glob.glob(os.path.join(ROOT, "demand", "*.pdf"))[0]
out_dir = os.path.join(ROOT, "demand", "_extracted")
os.makedirs(out_dir, exist_ok=True)

doc = fitz.open(pdf)
print(f"pages={doc.page_count} file={os.path.basename(pdf)}")

toc = doc.get_toc()
toc_path = os.path.join(out_dir, "juzhiyun_toc.txt")
with open(toc_path, "w", encoding="utf-8") as f:
    for lv, title, page in toc:
        f.write(f"{'  ' * (lv - 1)}{title}\tp{page}\n")
print(f"toc entries={len(toc)} -> {toc_path}")

# Keywords to pull relevant pages
KEYS = [
    "主数据", "物料", "BOM", "工艺路线", "工艺", "工序", "工厂", "车间", "产线",
    "设备", "班次", "日历", "库存", "订单", "分解", "MRP", "排产", "计划",
    "负荷", "甘特", "齐套", "采购", "委外", "生产订单", "工单",
]

# Build page index from TOC titles matching keys
interesting_pages: set[int] = set()
for lv, title, page in toc:
    if any(k in title for k in KEYS):
        # include this page and next few for section body
        for p in range(max(1, page), min(doc.page_count, page + 4) + 1):
            interesting_pages.add(p)

# Also scan first 30 pages of text for section headers if TOC sparse
if len(interesting_pages) < 20:
    for i in range(min(80, doc.page_count)):
        text = doc[i].get_text("text")
        if any(k in text for k in ("主数据维护", "工艺路线", "物料管理", "订单管理", "生产计划", "排产")):
            interesting_pages.add(i + 1)

print(f"interesting pages={len(interesting_pages)}")

# Extract full text of interesting pages + nearby
chunks = []
for pno in sorted(interesting_pages):
    text = doc[pno - 1].get_text("text")
    # compress blank lines
    text = re.sub(r"\n{3,}", "\n\n", text).strip()
    if text:
        chunks.append(f"\n\n===== PAGE {pno} =====\n{text}")

body_path = os.path.join(out_dir, "juzhiyun_master_schedule.txt")
with open(body_path, "w", encoding="utf-8") as f:
    f.write("".join(chunks))
print(f"extracted chars={sum(len(c) for c in chunks)} -> {body_path}")

# Also dump ALL page texts that look like menus / module names (shorter summary)
summary_lines = []
for i in range(doc.page_count):
    text = doc[i].get_text("text")
    hits = [k for k in KEYS if k in text]
    if hits:
        # first non-empty lines as context
        lines = [ln.strip() for ln in text.splitlines() if ln.strip()][:8]
        summary_lines.append(f"p{i+1} hits={hits[:6]} | {' | '.join(lines[:4])}")

sum_path = os.path.join(out_dir, "juzhiyun_page_hits.txt")
with open(sum_path, "w", encoding="utf-8") as f:
    f.write("\n".join(summary_lines))
print(f"hit pages={len(summary_lines)} -> {sum_path}")