#!/usr/bin/env python3
"""Mirror the app corpus's photographs into both concepts.

Reads afrix-web/lib/api/fixtures.json (the wired corpus) and img/credits-scenes.json,
then for every post carrying a photograph:

  build.py             the matching main-feed entry becomes k="image" with the
                       real tone and byte count (the canvas draws the tone well)
  concept-web/afrix-web.html
                       the matching post gets media {ar, tone, bytes, art} with
                       the 25 KB thumb embedded, behind its tap-to-load chip

Posts are matched by the first 40 characters of their text. Run build.py and
reseed.py afterwards for the canvas; the web concept is written directly.

    python3 mirror_photos.py
"""
import base64, io, json, os, re

HERE = os.path.dirname(os.path.abspath(__file__))
ROOT = os.path.dirname(HERE)
FIX = os.path.join(ROOT, "afrix-web", "lib", "api", "fixtures.json")
MEDIA = os.path.join(ROOT, "afrix-web", "public", "media")
WEB = os.path.join(ROOT, "concept-web", "afrix-web.html")
BUILD = os.path.join(HERE, "build.py")

def key(text: str) -> str:
    """Letters and digits only: "#TemaPort at 06:40" and "Tema port at 06:40" are one post."""
    return re.sub(r"[^a-z0-9]", "", (text or "").lower())[:32]

def main() -> None:
    fx = json.load(io.open(FIX, encoding="utf-8"))
    photos = {}
    for p in fx["posts"]:
        m = (p.get("media") or [None])[0]
        if m and m.get("src") and os.path.exists(os.path.join(MEDIA, os.path.basename(m["src"]))):
            photos[key(p["body"])] = (p["id"], m)
    kb = lambda m: f"{round(m['thumb_bytes'] / 1024)} KB"

    # ---- canvas source ----
    src = io.open(BUILD, encoding="utf-8").read()
    hit = 0
    def fix_entry(mm):
        nonlocal hit
        body = mm.group(0)
        text = re.search(r'x="([^"]*)"', body).group(1)
        k = key(text.encode().decode("unicode_escape") if "\\u" in text else text)
        if k not in photos:
            return body
        pid, m = photos[k]
        hit += 1
        body = re.sub(r'k="[a-z]+"', 'k="image"', body, count=1)
        body = re.sub(r',\s*tone="#[0-9A-Fa-f]{6}",\s*bytes="[^"]*"', "", body)
        return body.replace(f'x="{text}",', f'x="{text}",\n   tone="{m["tone"]}",bytes="{kb(m)}",', 1)
    src = re.sub(r'dict\(i="[^"]*",n="[^"]*",s="[^"]*",h="[^"]*",t="[^"]*",k="[a-z]+",\s*\n\s*x="[^"]*",(?:\s*\n\s*tone="#[0-9A-Fa-f]{6}",bytes="[^"]*",)?', fix_entry, src)
    io.open(BUILD, "w", encoding="utf-8").write(src)
    print(f"build.py: {hit} image entries")

    # ---- web concept ----
    html = io.open(WEB, encoding="utf-8").read()
    mm = re.search(r"var DATA = (\{.*?\});\n", html, re.S)
    D = json.loads(mm.group(1))
    credits = {c["post"]: c for c in json.load(open(os.path.join(HERE, "img", "credits-scenes.json")))}
    n = 0
    for posts in D["posts"].values():
        for po in posts:
            k = key(po.get("text", ""))
            if k in photos:
                pid, m = photos[k]
                b64 = base64.b64encode(open(os.path.join(MEDIA, os.path.basename(m["src"])), "rb").read()).decode()
                alt = credits.get(pid, {}).get("alt") or m.get("alt") or ""
                po["media"] = {"ar": f"{m['width']} / {m['height']}", "tone": m["tone"], "bytes": kb(m),
                               "art": f'<img src="data:image/jpeg;base64,{b64}" alt="{alt}" width="{m["width"]}" height="{m["height"]}">'}
                n += 1
    html = html[:mm.start(1)] + json.dumps(D, ensure_ascii=False, separators=(",", ":")) + html[mm.end(1):]
    io.open(WEB, "w", encoding="utf-8").write(html)
    print(f"web concept: {n} posts with photographs")

if __name__ == "__main__":
    main()