"""Portable one-outfit inference with the dataset's neutral reference in memory."""
import argparse
import hashlib
import io
import json
from pathlib import Path
import tarfile
import time
from urllib.request import urlopen,Request
from PIL import Image
import torch
from background_production import add_member
from character_pilot import metadata
from native_flux_inference import create_native_pipeline,generate_native_image,encode_text

REPO='laion/Vivarium-Greenscreen-Characters-FLUX-9B'

def main():
    p=argparse.ArgumentParser();p.add_argument('--id',default='person-0281-job-001');p.add_argument('--plan',type=Path)
    p.add_argument('--revision',default='main');p.add_argument('--reference',type=Path);p.add_argument('--base-path',type=Path)
    p.add_argument('--output',type=Path,default=Path('job-outfit.tar'));p.add_argument('--no-compile',action='store_true');a=p.parse_args()
    root=f'https://huggingface.co/datasets/{REPO}/resolve/{a.revision}/'
    plan=json.loads(a.plan.read_text()) if a.plan else json.load(urlopen(root+'metadata/job-outfits/image-plan.json',timeout=90))
    r=next(r for r in plan if r['id']==a.id)
    if a.reference:data=a.reference.read_bytes()
    else:
        index_member='metadata/sample-index.json' if r['reference_source_phase']=='character-production' else 'metadata/extension/sample-index.json'
        entries=json.load(urlopen(root+index_member,timeout=90));entry=next(x for x in entries if x['id']==r['reference_id'])
        start=entry['image_offset'];end=start+entry['image_bytes']-1;req=Request(root+entry['repo_path'],headers={'Range':f'bytes={start}-{end}'})
        with urlopen(req,timeout=90) as response:
            raw=response.read();data=raw if response.status==206 else raw[start:end+1]
        assert hashlib.sha256(data).hexdigest()==entry['image_sha256']
    digest=hashlib.sha256(data).hexdigest()
    with Image.open(io.BytesIO(data)) as im:im.load();reference=im.convert('RGB').resize((416,624),Image.Resampling.BICUBIC)
    pipe=create_native_pipeline('characters',base_path=a.base_path,compile_mode=None if a.no_compile else 'default',adapter_scale=r['lora_scale'])
    emb=encode_text(pipe,r['prompt']);torch.cuda.synchronize();t=time.perf_counter()
    im=generate_native_image(pipe,prompt_embeds=emb,seed=r['seed'],steps=50,guidance=4.,negative_prompt=r['negative_prompt'],image=reference,width=832,height=1248)[0]
    torch.cuda.synchronize();seconds=time.perf_counter()-t;b=io.BytesIO();im.save(b,format='PNG',compress_level=4);pixels=b.getvalue()
    m=metadata(r,dict(id=r['id'],seed=r['seed']),pixels,digest,seconds,0.,a.output.name,'portable-job-inference')
    m['inference']['compile_mode']=None if a.no_compile else 'default';m['runtime']['includes_compile_warmup']=not a.no_compile
    a.output.parent.mkdir(parents=True,exist_ok=True);temporary=a.output.with_name(a.output.name+'.partial')
    with tarfile.open(temporary,'w',format=tarfile.USTAR_FORMAT) as tar:
        add_member(tar,r['id']+'.png',pixels);add_member(tar,r['id']+'.json',json.dumps(m,indent=2).encode());add_member(tar,r['id']+'.txt',r['prompt'].encode())
    temporary.replace(a.output);print(r['id'],round(seconds,2),'seconds')

if __name__=='__main__':main()