Blame for scripts/dump_blocks.py

import requests, re, json, os
H = "https://huggingface.co"
from pymupdf import Document as MuDoc

def norm(s): return re.sub(r'[^0-9]','',str(s))

pdf=requests.get(H+"/datasets/bep40/grob-products-updated/resolve/main/pdfs/imudex_2026.pdf", timeout=600).content
open("/tmp/imudex.pdf","wb").write(pdf)
doc=MuDoc("/tmp/imudex.pdf")
code_pat=re.compile(r'^[\d\s.]{6,}$')

def parse_page(pgn):
    pg=doc[pgn]; d=pg.get_text("dict")
    lines=[]
    for b in d["blocks"]:
        if b["type"]!=0: continue
        blktext="".join(s["text"] for l in b["lines"] for s in l["spans"])
        for l in b["lines"]:
            txt="".join(s["text"] for s in l["spans"]).strip()
            if txt:
                x0,y0,x1,y1=l["bbox"]
                lines.append({"x0":round(x0,1),"y0":round(y0,1),"x1":round(x1,1),"y1":round(y1,1),"t":txt,"block":blktext.strip()})
    imgs=[]
    for ip in pg.get_image_info():
        bb=ip["bbox"]; w=ip.get("width") or 0; h=ip.get("height") or 0
        if w>=160 and h>=110:
            imgs.append({"x0":bb[0],"y0":bb[1],"x1":bb[2],"y1":bb[3],"w":w,"h":h})
    return lines,imgs

def side_of_x(x): return 'L' if x<567 else 'R'

def nearest_img_col(side, code_y, imgs):
    cand=[im for im in imgs if side_of_x((im["x0"]+im["x1"])/2)==side]
    if not cand: return None
    bd=1e15; best=None
    for im in cand:
        icy=(im["y0"]+im["y1"])/2
        dd=abs(icy-code_y)
        if dd<bd: bd=dd; best=im
    return best

def is_footnote_code(line):
    return len(line["block"]) > len(line["t"])+2

pages_cached={pgn:parse_page(pgn) for pgn in range(len(doc))}

def page_code_blocks(pgn):
    lines=pages_cached[pgn][0]
    codes=[]
    for l in lines:
        t=l["t"].strip()
        if code_pat.match(t) and len(t)>=6 and not is_footnote_code(l):
            codes.append(l)
    codes.sort(key=lambda l:(l["x0"],l["y0"]))
    blocks=[]
    for c in codes:
        side=side_of_x(c["x0"])
        cur=None
        for b in blocks:
            if b["side"]==side and abs(c["y0"]-b["codes"][-1]["y0"])<22:
                cur=b; break
        if cur is None:
            cur={"side":side,"codes":[c]}; blocks.append(cur)
        else:
            cur["codes"].append(c)
    for b in blocks:
        first=b["codes"][0]
        b["img"]=nearest_img_col(b["side"], first["y0"], pages_cached[pgn][1])
    return blocks

for pgn in [80, 119, 118]:
    blocks=page_code_blocks(pgn)
    print(f"=== PAGE {pgn+1} (pdf idx {pgn}) ===")
    for b in blocks:
        im=b["img"]
        print(f"side={b['side']} codes={[c['t'] for c in b['codes']]} first_y={b['codes'][0]['y0']} img={[round(im['x0']),round(im['y0']),round(im['x1']),round(im['y1'])] if im else None}")

print("\n=== page 79 'Tên hàng' lines ===")
for l in pages_cached[78][0]:
    if l["t"].strip()=="Tên hàng":
        print(l["x0"], l["y0"])
print("DONE")