#!/usr/bin/env python3 from __future__ import annotations import argparse import base64 import json import os from pathlib import Path import subprocess import time import urllib.request def main() -> int: ap = argparse.ArgumentParser(description='Firefox no-JS/semantic smoke for canonical pdf2html output.') ap.add_argument('url') ap.add_argument('output', type=Path) ap.add_argument('--geckodriver', default=os.environ.get('GECKODRIVER', '/tmp/geckodriver')) ap.add_argument('--port', type=int, default=4457) args = ap.parse_args() process = subprocess.Popen([args.geckodriver, '--port', str(args.port), '--log', 'error'], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL) def request(method: str, path: str, payload: dict | None = None, timeout: float = 60) -> dict: data = None if payload is None else json.dumps(payload).encode() req = urllib.request.Request( f'http://127.0.0.1:{args.port}{path}', data=data, method=method, headers={'Content-Type': 'application/json'}, ) with urllib.request.urlopen(req, timeout=timeout) as response: return json.load(response) metadata_url = args.url.rsplit('/', 1)[0] + '/metadata.json' with urllib.request.urlopen(metadata_url, timeout=60) as response: canonical_metadata = json.load(response) expected_figures = sum(len(page.get('figures', [])) for page in canonical_metadata['pages']) session_id = None try: for _ in range(60): try: request('GET', '/status', timeout=1) break except Exception: time.sleep(0.1) created = request('POST', '/session', { 'capabilities': {'alwaysMatch': { 'browserName': 'firefox', 'moz:firefoxOptions': {'args': ['-headless'], 'prefs': { 'browser.cache.disk.enable': False, 'browser.cache.memory.enable': False, 'browser.shell.checkDefaultBrowser': False, }}, }}, })['value'] session_id = created['sessionId'] request('POST', f'/session/{session_id}/window/rect', {'width': 1400, 'height': 1000}) request('POST', f'/session/{session_id}/url', {'url': args.url}, timeout=120) script = r'''return (async()=>{ await document.fonts.ready; const pages=[...document.querySelectorAll('.pf')],first=pages[0],last=pages[pages.length-1]; first.scrollIntoView();await new Promise(resolve=>setTimeout(resolve,200)); last.scrollIntoView();await new Promise(resolve=>setTimeout(resolve,600)); const p4=document.querySelector('[data-page-no="4"]'); const paragraph=[...p4.querySelectorAll('p')].find(node=>node.textContent.includes('We take pride in our role as a center of excellence')); const second=[...p4.querySelectorAll('p')].find(node=>node.textContent.includes('Danziger Kenya is known for its high-quality URC')); const list=[...p4.querySelectorAll('ul')].find(node=>node.textContent.includes('End-to-end production supply chain')); const ids=[...document.querySelectorAll('[id]')].map(node=>node.id),duplicates=[...new Set(ids.filter((id,index,all)=>all.indexOf(id)!==index))]; const lastRect=last.getBoundingClientRect(); const html=document.documentElement.innerHTML; return { pages:pages.length,visualLines:document.querySelectorAll('.vl').length,rendererTextDivs:document.querySelectorAll('div.t').length, paragraphs:document.querySelectorAll('p').length,headings:document.querySelectorAll('h1,h2,h3,h4,h5,h6').length, lists:document.querySelectorAll('ul,ol').length,listItems:document.querySelectorAll('li').length,tables:document.querySelectorAll('table').length, rows:document.querySelectorAll('tr').length,cells:document.querySelectorAll('th,td').length,figures:document.querySelectorAll('.semantic-figure').length, residuals:document.querySelectorAll('.residual-artwork').length,scripts:document.scripts.length,title:document.title,lang:document.documentElement.lang, description:document.querySelector('meta[name="description"]')?.content||'',duplicateIds:duplicates, p4Lines:paragraph?.querySelectorAll('.vl').length||0,p4SecondLines:second?.querySelectorAll('.vl').length||0,p4ListItems:list?.querySelectorAll('li').length||0, p4Order:{paragraph:html.indexOf('We take pride in our role as a center of excellence'),second:html.indexOf('Danziger Kenya is known for its high-quality URC'),list:html.indexOf('End-to-end production supply chain')}, phrase:window.find('flowers. Our State-of-the-Art facilities use cutting-'), lastRect:{w:lastRect.width,h:lastRect.height},contentVisibility:getComputedStyle(first).contentVisibility, remoteResources:performance.getEntriesByType('resource').map(row=>row.name).filter(name=>!name.startsWith(location.origin+'/')), residualPointer:[...document.querySelectorAll('.residual-artwork')].filter(node=>getComputedStyle(node).pointerEvents!=='none').length, loadedLast:[...last.querySelectorAll('img')].every(img=>img.complete&&img.naturalWidth>0) }; })()''' value = request('POST', f'/session/{session_id}/execute/sync', {'script': script, 'args': []}, timeout=90)['value'] page4 = request('POST', f'/session/{session_id}/element', {'using': 'css selector', 'value': '[data-page-no="4"]'})['value'] element_id = page4['element-6066-11e4-a52e-4f735466cecf'] png = base64.b64decode(request('GET', f'/session/{session_id}/element/{element_id}/screenshot', timeout=60)['value']) screenshot_path = args.output.with_suffix('.page4.png') screenshot_path.write_bytes(png) value['page4Screenshot'] = str(screenshot_path) value['page4ScreenshotBytes'] = len(png) printed = base64.b64decode(request('POST', f'/session/{session_id}/print', {'background': True, 'pageRanges': ['4'], 'shrinkToFit': True}, timeout=180)['value']) print_path = args.output.with_suffix('.page4.pdf') print_path.write_bytes(printed) pdfinfo = subprocess.run(['pdfinfo', str(print_path)], check=True, capture_output=True, text=True).stdout print_pages = next((line.split(':', 1)[1].strip() for line in pdfinfo.splitlines() if line.startswith('Pages:')), '') print_size = next((line.split(':', 1)[1].strip() for line in pdfinfo.splitlines() if line.startswith('Page size:')), '') value['page4Print'] = str(print_path) value['page4PrintBytes'] = len(printed) value['page4PrintPages'] = int(print_pages) if print_pages.isdigit() else -1 value['page4PrintSize'] = print_size value['firefox'] = created.get('capabilities', {}).get('browserVersion', '') value['platform'] = created.get('capabilities', {}).get('platformName', '') args.output.write_text(json.dumps(value, indent=2) + '\n', encoding='utf-8') print(json.dumps(value, sort_keys=True)) expected = 'ANNUALS & PERENNIALS CATALOG EU & ROW — 2026-28' ok = ( value['pages'] == 176 and value['visualLines'] == 7128 and value['rendererTextDivs'] == 0 and value['figures'] == expected_figures and value['residuals'] == 176 and value['scripts'] == 0 and value['title'] == expected and value['lang'] == 'en' and value['description'].startswith(expected + '. ') and not value['duplicateIds'] and value['p4Lines'] == 5 and value['p4SecondLines'] > 0 and value['p4ListItems'] == 7 and value['p4Order']['paragraph'] >= 0 and value['p4Order']['second'] > value['p4Order']['paragraph'] and value['p4Order']['list'] > value['p4Order']['second'] and value['phrase'] and value['lastRect'] == {'w': 612, 'h': 792} and value['contentVisibility'] == 'auto' and not value['remoteResources'] and value['residualPointer'] == 0 and value['loadedLast'] and len(png) > 10000 and value['page4PrintPages'] == 1 and value['page4PrintSize'].startswith('612 x 792 pts') and value['page4PrintBytes'] > 100000 ) return 0 if ok else 2 finally: if session_id is not None: try: request('DELETE', f'/session/{session_id}', {}) except Exception: pass process.terminate() try: process.wait(timeout=5) except subprocess.TimeoutExpired: process.kill() if __name__ == '__main__': raise SystemExit(main())