import requests, json, sys, os, concurrent.futures, urllib3 urllib3.disable_warnings() TARGETS = { # MILITARY "armada-sigeda": "https://sigeda.armada.mil.co", "armada-dicodarc": "https://dicodarcgisarc.armada.mil.co", "armada-registro": "https://registro.armada.mil.co", "fac-sigsa": "https://sigsa.fac.mil.co", # POLICE GIS (24 subdomains) "policia-gisponal": "https://gisponal.policia.gov.co", "policia-maps-analytics": "https://maps.analytics.ia.policia.gov.co", "policia-srvgis1": "https://srvgis1.policia.gov.co", "policia-srvgis2": "https://srvgis2.policia.gov.co", "policia-srvgis3": "https://srvgis3.policia.gov.co", "policia-srvgis4": "https://srvgis4.policia.gov.co", "policia-srvgis5": "https://srvgis5.policia.gov.co", "policia-srvgis6": "https://srvgis6.policia.gov.co", "policia-srvgis7": "https://srvgis7.policia.gov.co", "policia-srvgis8": "https://srvgis8.policia.gov.co", "policia-portalgis1": "https://srvportalgis1.policia.gov.co", "policia-portalgis2": "https://srvportalgis2.policia.gov.co", "policia-sigadmin": "https://srvsigadmin.policia.gov.co", "policia-sigadmin1": "https://srvsigadmin1.policia.gov.co", "policia-sigaplica": "https://srvsigaplica.policia.gov.co", "policia-sigcarto": "https://srvsigcarto.policia.gov.co", "policia-sigexterno": "https://srvsigexterno.policia.gov.co", "policia-sigponal1": "https://srvsigponal1.policia.gov.co", "policia-sigponal2": "https://srvsigponal2.policia.gov.co", "policia-sigponalapp1": "https://srvsigponalapp1.policia.gov.co", "policia-sigponalapp2": "https://srvsigponalapp2.policia.gov.co", "policia-sigponalext1": "https://srvsigponalext1.policia.gov.co", "policia-sigponalext2": "https://srvsigponalext2.policia.gov.co", # DANE - Statistics "dane-geoportal": "https://geoportal.dane.gov.co", "dane-geoserver": "https://geoserverportal.dane.gov.co", "dane-sige": "https://sige.dane.gov.co", # ANM - Mining "anm-arcgis": "https://arcgisserver.anm.gov.co", # IDEAM - Meteorological "ideam-geoservicio": "https://geoservicio.smartmetospa.ideam.gov.co", "ideam-geoweb": "https://geoweb.smartmetospa.ideam.gov.co", # CAR - Environmental "car-sig": "https://sig.car.gov.co", "car-geoambiental": "https://geoambiental.car.gov.co", "car-geourbana": "https://geourbana.car.gov.co", "car-sigci": "https://sigci.car.gov.co", "car-sigu": "https://sigu.car.gov.co", "car-sigriobogota": "https://sigriobogota.car.gov.co", "car-datosgeograficos": "https://datosgeograficos.car.gov.co", # Transport "mintransporte-sigt": "https://sigt.mintransporte.gov.co", "mintransporte-sigtdev": "https://sigtdev.mintransporte.gov.co", # Geological Survey "sgc-geored": "https://geored.sgc.gov.co", "sgc-geored2": "https://geored2.sgc.gov.co", # Hydrocarbons "anh-geoportal": "https://geoportal.anh.gov.co", "anh-geovisor": "https://geovisor.anh.gov.co", "anh-dataroom": "https://geovisordataroom.anh.gov.co", # Roads "invias-sigpav": "https://sigpav.invias.gov.co", "invias-sigpinspeccion": "https://sigpinspeccion.invias.gov.co", # Agricultural Institute "ica-sigma": "https://sigma.ica.gov.co", "ica-sigeco": "https://sigeco.ica.gov.co", # Land Restitution "restitucion-geoserver": "https://geoserver.restituciondetierras.gov.co", "restitucion-portalmapas": "https://portalmapas.restituciondetierras.gov.co", "restitucion-portalmapasurt": "https://portalmapasurt.restituciondetierras.gov.co", # Land Agency "agenciatierras-geoportal": "https://geoportal.agenciadetierras.gov.co", "agenciatierras-sig": "https://sig.agenciadetierras.gov.co", # Bogota Cadastre "catastrobogota-serviciosgis": "https://serviciosgis.catastrobogota.gov.co", "catastrobogota-sig": "https://sig.catastrobogota.gov.co", # Fiscalia GIS "fiscalia-geoapp": "https://geoappfgn.fiscalia.gov.co", "fiscalia-sigob": "https://sigob-web.fiscalia.gov.co", } PATHS = [ "/arcgis/rest/services?f=json", "/server/rest/services?f=json", "/arcgis/rest/services?f=pjson", "/geoserver/web/", "/arcgis/rest/info?f=json", ] results = {"arcgis_open": [], "geoserver": [], "alive": [], "dead": []} def probe(name, base_url): hits = [] for path in PATHS: url = base_url + path try: r = requests.get(url, timeout=10, verify=False, allow_redirects=True) body = r.text[:2000] if r.status_code == 200: if '"services"' in body or '"folders"' in body or '"currentVersion"' in body: hits.append({"name": name, "url": url, "status": r.status_code, "type": "arcgis", "size": len(body), "snippet": body[:500]}) elif "GeoServer" in body or "geoserver" in body.lower(): hits.append({"name": name, "url": url, "status": r.status_code, "type": "geoserver", "size": len(body)}) elif len(body) < 500 or "blocked" in body.lower() or "denied" in body.lower(): hits.append({"name": name, "url": url, "status": r.status_code, "type": "maybe_waf", "size": len(body)}) else: hits.append({"name": name, "url": url, "status": r.status_code, "type": "other_200", "size": len(body)}) elif r.status_code in [301, 302, 307, 308]: hits.append({"name": name, "url": url, "status": r.status_code, "type": "redirect", "location": r.headers.get("Location", "")}) elif r.status_code == 403: hits.append({"name": name, "url": url, "status": 403, "type": "forbidden"}) elif r.status_code == 401: hits.append({"name": name, "url": url, "status": 401, "type": "auth_required"}) else: hits.append({"name": name, "url": url, "status": r.status_code, "type": f"http_{r.status_code}"}) except requests.exceptions.ConnectTimeout: hits.append({"name": name, "url": url, "type": "timeout"}) except requests.exceptions.ConnectionError as e: hits.append({"name": name, "url": url, "type": "conn_error", "error": str(e)[:100]}) except Exception as e: hits.append({"name": name, "url": url, "type": "error", "error": str(e)[:100]}) return name, hits print(f"Probing {len(TARGETS)} targets across {len(PATHS)} paths ({len(TARGETS)*len(PATHS)} requests)...") sys.stdout.flush() with concurrent.futures.ThreadPoolExecutor(max_workers=20) as executor: futures = {executor.submit(probe, name, url): name for name, url in TARGETS.items()} for future in concurrent.futures.as_completed(futures): name, hits = future.result() arcgis_hit = any(h.get("type") == "arcgis" for h in hits) geo_hit = any(h.get("type") == "geoserver" for h in hits) if arcgis_hit: results["arcgis_open"].append({"name": name, "hits": [h for h in hits if h.get("type") == "arcgis"]}) print(f" [ARCGIS OPEN] {name}") elif geo_hit: results["geoserver"].append({"name": name, "hits": [h for h in hits if h.get("type") == "geoserver"]}) print(f" [GEOSERVER] {name}") elif any(h.get("type") in ["other_200", "maybe_waf", "forbidden", "auth_required"] for h in hits): results["alive"].append({"name": name, "hits": hits}) best = next((h for h in hits if h.get("type") in ["other_200", "maybe_waf", "forbidden", "auth_required"]), hits[0]) print(f" [ALIVE-{best.get('type', '?')}] {name} ({best.get('status', '?')})") else: results["dead"].append({"name": name, "hits": hits}) err = hits[0].get("type", "?") if hits else "?" print(f" [DEAD] {name} ({err})") sys.stdout.flush() print(f"\n{'='*60}") print(f"=== RESULTS SUMMARY ===") print(f"{'='*60}") print(f"ArcGIS Open (dumpable): {len(results['arcgis_open'])}") print(f"GeoServer found: {len(results['geoserver'])}") print(f"Alive (other): {len(results['alive'])}") print(f"Dead/Unreachable: {len(results['dead'])}") print(f"\n=== ARCGIS OPEN (READY TO DUMP) ===") for r in results["arcgis_open"]: for h in r["hits"]: print(f" {r['name']}: {h['url']}") if "snippet" in h: print(f" Preview: {h['snippet'][:200]}") print(f"\n=== GEOSERVER FOUND ===") for r in results["geoserver"]: for h in r["hits"]: print(f" {r['name']}: {h['url']}") print(f"\n=== ALIVE (NEEDS INVESTIGATION) ===") for r in results["alive"]: best = next((h for h in r["hits"] if h.get("type") in ["other_200", "maybe_waf", "forbidden", "auth_required"]), r["hits"][0]) print(f" {r['name']}: type={best.get('type', '')} status={best.get('status', '?')} url={best.get('url', '')}") with open("mass_probe_results.json", "w") as f: json.dump(results, f, indent=2, default=str) print("\nFull results saved to mass_probe_results.json")