Google Maps business leads

This Python runs in your browser session when you use the workflow. Review it before running.

  1import re  2from playwright.sync_api import expect, TimeoutError as PlaywrightTimeoutError  3   4QUERY = "dentists in Austin"  5SEARCH_URL = "https://www.google.com/maps/search/dentists+in+Austin"  6FEED = 'div[role="feed"]'  7LINKS = 'div[role="feed"] a.hfpxzc'  8   9PANEL_JS = """()=>{ 10  const m=document.querySelector('div[role=main][aria-label]'); 11  const r=m||document; 12  const ph=r.querySelector('button[data-item-id^="phone:tel:"]'); 13  const w=r.querySelector('a[data-item-id="authority"]'); 14  const rt=r.querySelector('div.F7nice span[aria-hidden="true"]'); 15  return {main:m&&m.getAttribute('aria-label'), 16          phone:ph?(ph.getAttribute('aria-label')||'').replace('Phone: ','').trim():null, 17          website:w?w.href:null, 18          rating:rt?rt.textContent:null}}""" 19  20READY_JS = """(name)=>{ 21  const m=document.querySelector('div[role=main][aria-label]'); 22  if(!m || m.getAttribute('aria-label')!==name) return false; 23  const ph=m.querySelector('button[data-item-id^="phone:tel:"]'); 24  const w=m.querySelector('a[data-item-id="authority"]'); 25  return !!(ph||w); 26}""" 27  28  29def run(page, context, browser): 30    with browser.scope("open-maps-search", goal="Open Google Maps search results for the query straight from the URL", returns=None) as s_open: 31        page.goto(SEARCH_URL, wait_until="domcontentloaded") 32        page.wait_for_load_state("domcontentloaded") 33        try: 34            expect(page.locator(FEED)).to_be_visible(timeout=15000) 35        except AssertionError: 36            s_open.deopt(reason="results feed did not appear", evidence={"url": page.url}) 37        s_open.value = None 38  39    with browser.scope("load-results-feed", goal="Scroll the results feed to the bottom several times so more listings load, then count the listing links", 40                       returns={"type": "object", "properties": {"count": {"type": "integer"}}, "required": ["count"]}) as s_load: 41        feed = page.locator(FEED) 42        links = page.locator(LINKS) 43        try: 44            expect(feed).to_be_visible(timeout=10000) 45        except AssertionError: 46            s_load.deopt(reason="results feed not visible", evidence={"url": page.url}) 47        for _ in range(6): 48            before = links.count() 49            feed.evaluate("e=>e.scrollTo(0,e.scrollHeight)") 50            try: 51                expect(links).not_to_have_count(before, timeout=3000) 52            except AssertionError: 53                pass 54        try: 55            expect(links).not_to_have_count(0, timeout=10000) 56        except AssertionError: 57            s_load.deopt(reason="the results feed has no listing links", evidence={"url": page.url}) 58        s_load.value = {"count": links.count()} 59  60    with browser.scope("list-result-links", goal="Read every listing link's name, href and visibility so hidden or duplicate entries can be skipped", 61                       returns={"type": "object", "properties": {"items": {"type": "array", "items": { 62                           "type": "object", 63                           "properties": {"index": {"type": "integer"}, "name": {"type": ["string", "null"]}, 64                                          "href": {"type": ["string", "null"]}, "vis": {"type": "boolean"}}, 65                           "required": ["index", "name", "vis"]}}}, "required": ["items"]}) as s_list: 66        links = page.locator(LINKS) 67        raw = links.evaluate_all("els=>els.map(e=>({name:e.getAttribute('aria-label'),href:e.href,vis:e.offsetParent!==null}))") 68        if not raw: 69            s_list.deopt(reason="no listing links read from feed", evidence={"url": page.url}) 70        items = [{"index": i, "name": it.get("name"), "href": it.get("href"), "vis": bool(it.get("vis"))} for i, it in enumerate(raw)] 71        s_list.value = {"items": items} 72  73    results = [] 74    seen = set() 75    for it in s_list.value["items"]: 76        name = it["name"] 77        if not it["vis"] or not name or name in seen: 78            continue 79        seen.add(name) 80        with browser.scope("extract-place-details", goal="Click the listing and read phone, website and rating from its detail panel", 81                           returns={"type": "object", "properties": { 82                               "name": {"type": "string"}, "phone": {"type": ["string", "null"]}, 83                               "website": {"type": ["string", "null"]}, "rating": {"type": ["string", "null"]}}, 84                               "required": ["name", "phone", "website", "rating"]}) as s_item: 85            link = page.locator(LINKS).nth(it["index"]) 86            try: 87                expect(link).to_have_attribute("aria-label", name, timeout=10000) 88            except AssertionError: 89                s_item.deopt(reason="listing link at index no longer matches the name read", evidence={"index": it["index"], "name": name}) 90            link.scroll_into_view_if_needed(timeout=5000) 91            link.click(timeout=5000) 92            try: 93                page.wait_for_function(READY_JS, arg=name, timeout=8000) 94            except PlaywrightTimeoutError: 95                pass 96            d = page.evaluate(PANEL_JS) 97            if d.get("main") != name: 98                s_item.deopt(reason="detail panel did not show the clicked listing", evidence={"expected": name, "panel": d.get("main")}) 99            s_item.value = {"name": name, "phone": d.get("phone") or None,100                            "website": d.get("website") or None, "rating": d.get("rating") or None}101        if s_item.value:102            results.append(s_item.value)103 104    browser.done(result={"query": QUERY, "source": "Google Maps", "count": len(results), "results": results})