#!/usr/bin/env python3
import json, re, shutil, subprocess, tempfile
from pathlib import Path
ROOT=Path(__file__).resolve().parent
MD,FIG,SITE=ROOT/'markdown',ROOT/'figures',ROOT/'site'
def main():
 FIG.mkdir(exist_ok=True); sources=sorted(MD.glob('NaturalHazards.Chapter[0-9]*.md'),key=lambda p:int(re.search(r'Chapter(\d+)',p.name).group(1))); mapping={}
 for source in sources:
  ch=int(re.search(r'Chapter(\d+)',source.name).group(1)); pdf=ROOT/f'NaturalHazards.Chapter{ch}.pdf'
  if not pdf.exists(): continue
  text=source.read_text(encoding='utf-8'); text=re.sub(r'\n*!\[Source PDF page \d+, containing Figure [^\]]+\]\([^\n]+\)\n*', '\n', text); pages=subprocess.run(['pdftotext','-layout',str(pdf),'-'],check=True,capture_output=True,text=True).stdout.split('\f')
  raw_ids=re.findall(rf'Figure\s+{ch}\.\d+(?:\.[A-Z])?[A-Z]?(?![\d.])',text,re.I); ids=sorted({x.casefold():re.sub(r'^figure','Figure',x,flags=re.I) for x in raw_ids}.values(),key=lambda s:(int(re.search(r'\.(\d+)',s).group(1)),s)); found={}
  for fig in ids:
   cap=re.escape(fig)+r"(?![\d.])(?:\s*:\s*|\s+)(?=(?-i:[A-Z]))"
   md_caps=list(re.finditer(cap,text,re.I)); candidates=[]
   for i,page in enumerate(pages):
    for pm in re.finditer(cap,page,re.I):
     pdf_words=set(re.findall(r"[a-zA-Z]{4,}",page[pm.end():pm.end()+220].lower()))
     for mm in md_caps:
      md_words=set(re.findall(r"[a-zA-Z]{4,}",text[mm.end():mm.end()+220].lower()))
      score=len(pdf_words & md_words)
      candidates.append((score,i+1,mm.start()))
   if candidates:
    score,page,pos=max(candidates); found[fig]=page; found[fig+'|offset']=pos
  inserts=[]
  for key,page in list(found.items()):
   if '|' in key: continue
   pos=found.get(key+'|offset')
   if pos is not None: inserts.append((pos,f'\n\n![Source PDF page {page}, containing {key}](../figures/chapter-{ch}-page-{page:02}.jpg)\n\n'))
  found={k:v for k,v in found.items() if '|' not in k}; mapping[source.name]=found
  for pos,value in sorted(inserts,reverse=True): text=text[:pos]+value+text[pos:]
  source.write_text(text,encoding='utf-8')
  if found:
   with tempfile.TemporaryDirectory(prefix='natural-hazards-') as temp:
    for page in sorted(set(found.values())):
     target=FIG/f'chapter-{ch}-page-{page:02}.jpg'
     if target.exists(): continue
     prefix=str(Path(temp)/'page')
     subprocess.run(['pdftoppm','-f',str(page),'-l',str(page),'-jpeg','-scale-to','1400','-jpegopt','quality=78',str(pdf),prefix],check=True,stdout=subprocess.DEVNULL,stderr=subprocess.DEVNULL)
     shutil.copy2(f'{prefix}-{page:02}.jpg',target)
 # Appendices use lettered figure numbers and a separate source PDF.
 source=MD/'NaturalHazards.Appendices.md'; pdf=ROOT/'NaturalHazards.Appendices.pdf'
 text=source.read_text(encoding='utf-8'); text=re.sub(r'\n*!\[Source PDF page \d+, containing Figure [^\]]+\]\([^\n]+\)\n*','\n',text)
 pages=subprocess.run(['pdftotext','-layout',str(pdf),'-'],check=True,capture_output=True,text=True).stdout.split('\f')
 raw_ids=re.findall(r'Figure\s+[A-Z]\.\d+[A-Z]?(?![\d.])',text,re.I); ids=sorted({x.casefold():re.sub(r'^figure','Figure',x,flags=re.I) for x in raw_ids}.values()); found={}
 for fig in ids:
  cap=re.escape(fig)+r"(?![\d.])(?:\s*:\s*|\s+)(?=(?-i:[A-Z]))"; md_caps=list(re.finditer(cap,text,re.I)); candidates=[]
  for i,page in enumerate(pages):
   for pm in re.finditer(cap,page,re.I):
    pdf_words=set(re.findall(r'[a-zA-Z]{4,}',page[pm.end():pm.end()+220].lower()))
    for mm in md_caps:
     md_words=set(re.findall(r'[a-zA-Z]{4,}',text[mm.end():mm.end()+220].lower())); candidates.append((len(pdf_words & md_words),i+1,mm.start()))
  if candidates:
   score,page,pos=max(candidates); found[fig]=page; found[fig+'|offset']=pos
 inserts=[]
 for key,page in list(found.items()):
  if '|' not in key: inserts.append((found[key+'|offset'],f'\n\n![Source PDF page {page}, containing {key}](../figures/appendices-page-{page:02}.jpg)\n\n'))
 found={k:v for k,v in found.items() if '|' not in k}; mapping[source.name]=found
 for pos,value in sorted(inserts,reverse=True): text=text[:pos]+value+text[pos:]
 source.write_text(text,encoding='utf-8')
 with tempfile.TemporaryDirectory(prefix='natural-hazards-appendices-') as temp:
  for page in sorted(set(found.values())):
   target=FIG/f'appendices-page-{page:02}.jpg'
   if target.exists(): continue
   prefix=str(Path(temp)/'page'); subprocess.run(['pdftoppm','-f',str(page),'-l',str(page),'-jpeg','-scale-to','1400','-jpegopt','quality=78',str(pdf),prefix],check=True,stdout=subprocess.DEVNULL,stderr=subprocess.DEVNULL); shutil.copy2(f'{prefix}-{page:02}.jpg',target)
 # The complete-book Markdown also gets links to the same chapter and appendix images.
 full=MD/"Natural hazards_ earth's processes as hazards, disasters, and catastrophes.md"
 if full.exists():
  full_text=full.read_text(encoding='utf-8'); full_text=re.sub(r'\n*!\[Source PDF page \d+, containing Figure [^\]]+\]\([^\n]+\)\n*','\n',full_text); assets={}
  for name,items in mapping.items():
   if name==full.name: continue
   for fig,page in items.items():
    if name=='NaturalHazards.Appendices.md': rel=f'../figures/appendices-page-{page:02}.jpg'
    else:
     chapter=re.search(r'Chapter(\d+)',name).group(1); rel=f'../figures/chapter-{chapter}-page-{page:02}.jpg'
    assets[fig]=(page,rel)
  inserts=[]
  for fig,(page,rel) in assets.items():
   cap=re.escape(fig)+r"(?![\d.])(?:\s*:\s*|\s+)(?=(?-i:[A-Z]))"; match=re.search(cap,full_text,re.I)
   if match: inserts.append((match.start(),f'\n\n![Source PDF page {page}, containing {fig}]({rel})\n\n'))
  for pos,value in sorted(inserts,reverse=True): full_text=full_text[:pos]+value+full_text[pos:]
  full.write_text(full_text,encoding='utf-8'); mapping[full.name]={fig:rel for fig,(_,rel) in assets.items() if re.search(re.escape(fig)+r"(?![\d.])(?:\s*:\s*|\s+)(?=(?-i:[A-Z]))",full_text,re.I)}
 (ROOT/'figure-pages.json').write_text(json.dumps(mapping,indent=2)+'\n'); make_site(sources)
def make_site(chapters):
 SITE.mkdir(exist_ok=True); files=list(chapters)+[MD/'NaturalHazards.Appendices.md']; known={p.name for p in files}; files += [p for p in sorted(MD.glob('*.md')) if p.name not in known]
 books=[]
 for p in files:
  if not p.exists(): continue
  title=p.stem.replace('NaturalHazards.','').replace('_',' ')
  if title.startswith('Natural hazards'): title='Full Book — Natural Hazards'
  title=re.sub(r'Chapter(\d+)',r'Chapter \1',title)
  title=re.sub(r'(?<=[a-z])(?=[A-Z])',' ',title).replace('.',' — ')
  books.append({'title':title,'path':'../'+p.relative_to(ROOT).as_posix()})
 (SITE/'books.json').write_text(json.dumps(books,indent=2)+'\n'); (SITE/'index.html').write_text(HTML); (SITE/'app.js').write_text(APP); (SITE/'style.css').write_text(CSS); (SITE/'README.md').write_text('Serve the book directory over HTTP, then open `http://localhost:8000/site/`:\n\n```sh\npython3 -m http.server 8000\n```\n')
HTML='''<!doctype html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width,initial-scale=1"><title>Natural Hazards — Book Browser</title><link rel="stylesheet" href="style.css"></head><body><aside><a class="brand" href="#">Natural Hazards</a><input id="filter" type="search" placeholder="Filter pages…"><nav id="nav"></nav></aside><main><header><button id="menu">☰</button><span id="where">Select a page</span></header><article id="content"><h1>Natural Hazards</h1><p>Choose a Markdown page from the list.</p></article></main><script src="app.js" defer></script></body></html>'''
APP=r'''const nav=document.querySelector('#nav'),content=document.querySelector('#content'),filter=document.querySelector('#filter');let books=[];
fetch('books.json').then(r=>r.json()).then(x=>{books=x;draw()});
function draw(){nav.innerHTML='';books.filter(b=>b.title.toLowerCase().includes(filter.value.toLowerCase())).forEach(b=>{const a=document.createElement('a');a.href='#';a.textContent=b.title;a.onclick=e=>{e.preventDefault();load(b.path)};nav.append(a)})}
filter.oninput=draw;document.querySelector('#menu').onclick=()=>document.body.classList.toggle('show-nav');
async function load(path){try{const r=await fetch(path);if(!r.ok)throw Error();const md=await r.text();document.querySelector('#where').textContent=books.find(b=>b.path===path)?.title||path;content.innerHTML=render(md,new URL(path,location.href));document.body.classList.remove('show-nav')}catch(e){content.innerHTML='<p>Could not load this page. Run <code>python3 -m http.server</code> from the book directory.</p>'}}
function inl(s,b){s=s.replace(/&/g,'&amp;').replace(/</g,'&lt;').replace(/>/g,'&gt;');return s.replace(/!\[([^\]]*)\]\(([^)]+)\)/g,(_,t,u)=>'<img loading="lazy" alt="'+t+'" src="'+new URL(u,b).href+'">').replace(/\[([^\]]+)\]\(([^)]+)\)/g,(_,t,u)=>'<a href="'+new URL(u,b).href+'">'+t+'</a>').replace(/\*\*(.+?)\*\*/g,'<strong>$1</strong>').replace(/\*(.+?)\*/g,'<em>$1</em>')}
function render(md,b){let out=[],p=[],list=false,table=[];function flush(){if(p.length){out.push('<p>'+p.map(x=>inl(x,b)).join(' ')+'</p>');p=[]}if(list){out.push('</ul>');list=false}if(table.length){let rs=table.filter(x=>!/^\|?\s*:?-{3,}/.test(x));out.push('<table>'+rs.map((r,i)=>'<tr>'+r.trim().replace(/^\||\|$/g,'').split('|').map(c=>'<'+(i?'td':'th')+'>'+inl(c.trim(),b)+'</'+(i?'td':'th')+'>').join('')+'</tr>').join('')+'</table>');table=[]}}
for(const l of md.split(/\r?\n/)){if(/^\s*\|.*\|\s*$/.test(l)){if(p.length||list)flush();table.push(l);continue}if(table.length)flush();let m=l.match(/^(#{1,6})\s+(.*)$/);if(m){flush();out.push('<h'+m[1].length+'>'+inl(m[2],b)+'</h'+m[1].length+'>');continue}m=l.match(/^\s*[-*+]\s+(.*)$/);if(m){if(p.length)flush();if(!list){out.push('<ul>');list=true}out.push('<li>'+inl(m[1],b)+'</li>');continue}if(!l.trim()){flush();continue}p.push(l.trim())}flush();return out.join('\n')}'''
CSS='''*{box-sizing:border-box}body{margin:0;color:#202523;background:#f4f4ef;font:16px/1.65 Georgia,serif;display:grid;grid-template-columns:290px 1fr;min-height:100vh}aside{background:#e7e8e0;border-right:1px solid #d2d5cc;padding:24px 18px;position:sticky;top:0;height:100vh;overflow:auto;font:14px/1.4 system-ui,sans-serif}.brand{font:700 20px system-ui;color:#263d35;text-decoration:none;display:block;margin:0 0 18px}#filter{width:100%;padding:9px 10px;border:1px solid #c8cdc3;border-radius:5px;margin-bottom:14px}nav a{display:block;padding:6px 8px;text-decoration:none;color:#3b4943;border-radius:4px}nav a:hover{background:#d9ddd3}main{min-width:0}header{position:sticky;top:0;background:#f4f4efdd;backdrop-filter:blur(8px);padding:14px max(5vw,24px);border-bottom:1px solid #e1e1d9;font:14px system-ui;z-index:1}#menu{display:none}article{max-width:900px;padding:28px max(5vw,24px) 80px;margin:auto}h1,h2,h3,h4{font-family:system-ui,sans-serif;line-height:1.25;color:#263d35}h1{font-size:2.15rem}p{margin:0 0 1.1em}img{display:block;max-width:100%;height:auto;margin:16px auto 24px;box-shadow:0 2px 12px #0002;border-radius:3px}table{border-collapse:collapse;width:100%;margin:1em 0}td,th{border:1px solid #cfd3ca;padding:7px 10px;text-align:left}th{background:#e8ebe3}li{margin:.3em 0}@media(max-width:760px){body{grid-template-columns:1fr}aside{display:none;position:fixed;z-index:4;width:min(85vw,320px);box-shadow:0 0 0 100vw #0005}.show-nav aside{display:block}#menu{display:inline-block;margin-right:10px}article{padding:22px 20px 60px}}'''
if __name__=='__main__': main()