From: Roland Häder Date: Sun, 28 May 2023 09:19:13 +0000 (+0200) Subject: Continued: X-Git-Url: https://git.mxchange.org/?a=commitdiff_plain;h=a11c067f47e261ff7a750ade0356a2f720268d1e;p=fba.git Continued: - don't name your variables after packages, Python seem to be not strict on checking data types while referencing - also scan misskey instances for new (no filter applied) --- diff --git a/fba.py b/fba.py index 2bdf1fb..e5549ce 100644 --- a/fba.py +++ b/fba.py @@ -106,7 +106,7 @@ patterns = [ ] def remove_version(software: str) -> str: - # NOISY-DEBUG: print(f"DEBUG: software='{software}' - CALLED!") + # DEBUG: print(f"DEBUG: software='{software}' - CALLED!") if not "." in software and " " not in software: print(f"WARNING: software='{software}' does not contain a version number.") return software @@ -119,7 +119,7 @@ def remove_version(software: str) -> str: elif " - " in software: temp = software.split(" - ")[0] - # NOISY-DEBUG: print(f"DEBUG: software='{software}'") + # DEBUG: print(f"DEBUG: software='{software}'") version = None if " " in software: version = temp.split(" ")[-1] @@ -128,39 +128,39 @@ def remove_version(software: str) -> str: elif "-" in software: version = temp.split("-")[-1] else: - # NOISY-DEBUG: print(f"DEBUG: Was not able to find common seperator, returning untouched software='{software}'") + # DEBUG: print(f"DEBUG: Was not able to find common seperator, returning untouched software='{software}'") return software matches = None match = None - # NOISY-DEBUG: print(f"DEBUG: Checking {len(patterns)} patterns ...") + # DEBUG: print(f"DEBUG: Checking {len(patterns)} patterns ...") for pattern in patterns: # Run match() match = pattern.match(version) - # NOISY-DEBUG: print(f"DEBUG: match[]={type(match)}") + # DEBUG: print(f"DEBUG: match[]={type(match)}") if type(match) is re.Match: break - # NOISY-DEBUG: print(f"DEBUG: version[{type(version)}]='{version}',match='{match}'") + # DEBUG: print(f"DEBUG: version[{type(version)}]='{version}',match='{match}'") if type(match) is not re.Match: print(f"WARNING: version='{version}' does not match regex, leaving software='{software}' untouched.") return software - # NOISY-DEBUG: print(f"DEBUG: Found valid version number: '{version}', removing it ...") + # DEBUG: print(f"DEBUG: Found valid version number: '{version}', removing it ...") end = len(temp) - len(version) - 1 - # NOISY-DEBUG: print(f"DEBUG: end[{type(end)}]={end}") + # DEBUG: print(f"DEBUG: end[{type(end)}]={end}") software = temp[0:end].strip() if " version" in software: - # NOISY-DEBUG: print(f"DEBUG: software='{software}' contains word ' version'") + # DEBUG: print(f"DEBUG: software='{software}' contains word ' version'") software = strip_until(software, " version") - # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!") + # DEBUG: print(f"DEBUG: software='{software}' - EXIT!") return software def strip_powered_by(software: str) -> str: - # NOISY-DEBUG: print(f"DEBUG: software='{software}' - CALLED!") + # DEBUG: print(f"DEBUG: software='{software}' - CALLED!") if software == "": print(f"ERROR: Bad method call, 'software' is empty") raise Exception("Parameter 'software' is empty") @@ -169,18 +169,18 @@ def strip_powered_by(software: str) -> str: return software start = software.find("powered by ") - # NOISY-DEBUG: print(f"DEBUG: start[{type(start)}]='{start}'") + # DEBUG: print(f"DEBUG: start[{type(start)}]='{start}'") software = software[start + 11:].strip() - # NOISY-DEBUG: print(f"DEBUG: software='{software}'") + # DEBUG: print(f"DEBUG: software='{software}'") software = strip_until(software, " - ") - # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!") + # DEBUG: print(f"DEBUG: software='{software}' - EXIT!") return software def strip_until(software: str, until: str) -> str: - # NOISY-DEBUG: print(f"DEBUG: software='{software}',until='{until}' - CALLED!") + # DEBUG: print(f"DEBUG: software='{software}',until='{until}' - CALLED!") if software == "": print(f"ERROR: Bad method call, 'software' is empty") raise Exception("Parameter 'software' is empty") @@ -194,11 +194,11 @@ def strip_until(software: str, until: str) -> str: # Next, strip until part end = software.find(until) - # NOISY-DEBUG: print(f"DEBUG: end[{type(end)}]='{end}'") + # DEBUG: print(f"DEBUG: end[{type(end)}]='{end}'") if end > 0: software = software[0:end].strip() - # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!") + # DEBUG: print(f"DEBUG: software='{software}' - EXIT!") return software def is_blacklisted(domain: str) -> bool: @@ -221,7 +221,7 @@ def get_hash(domain: str) -> str: return hashlib.sha256(domain.encode("utf-8")).hexdigest() def update_last_blocked(domain: str): - # NOISY-DEBUG: print("DEBUG: Updating last_blocked for domain", domain) + # DEBUG: print("DEBUG: Updating last_blocked for domain", domain) try: cursor.execute("UPDATE instances SET last_blocked = ?, last_updated = ? WHERE domain = ? LIMIT 1", [ time.time(), @@ -236,29 +236,29 @@ def update_last_blocked(domain: str): print("ERROR: failed SQL query:", domain, e) sys.exit(255) - # NOISY-DEBUG: print("DEBUG: EXIT!") + # DEBUG: print("DEBUG: EXIT!") def update_nodeinfos(domain: str): - # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain) + # DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain) sql_string = '' fields = list() for key in nodeinfos: - # NOISY-DEBUG: print("DEBUG: key:", key) + # DEBUG: print("DEBUG: key:", key) if domain in nodeinfos[key]: - # NOISY-DEBUG: print(f"DEBUG: Adding '{nodeinfos[key][domain]}' for key='{key}' ...") + # DEBUG: print(f"DEBUG: Adding '{nodeinfos[key][domain]}' for key='{key}' ...") fields.append(nodeinfos[key][domain]) sql_string += f" {key} = ?," fields.append(domain) - # NOISY-DEBUG: print(f"DEBUG: sql_string='{sql_string}',fields()={len(fields)}") + # DEBUG: print(f"DEBUG: sql_string='{sql_string}',fields()={len(fields)}") sql = "UPDATE instances SET" + sql_string + " last_status_code = NULL, last_error_details = NULL WHERE domain = ? LIMIT 1" - # NOISY-DEBUG: print("DEBUG: sql:", sql) + # DEBUG: print("DEBUG: sql:", sql) try: - # NOISY-DEBUG: print("DEBUG: Executing SQL:", sql) + # DEBUG: print("DEBUG: Executing SQL:", sql) cursor.execute(sql, fields) - # NOISY-DEBUG: print(f"DEBUG: Success! (rowcount={cursor.rowcount })") + # DEBUG: print(f"DEBUG: Success! (rowcount={cursor.rowcount })") if cursor.rowcount == 0: print("WARNING: Did not update any rows:", domain) @@ -267,33 +267,33 @@ def update_nodeinfos(domain: str): print(f"ERROR: failed SQL query: domain='{domain}',sql='{sql}',exception:'{e}'") sys.exit(255) - # NOISY-DEBUG: print("DEBUG: Deleting nodeinfos for domain:", domain) + # DEBUG: print("DEBUG: Deleting nodeinfos for domain:", domain) for key in nodeinfos: try: - # NOISY-DEBUG: print("DEBUG: Deleting key:", key) + # DEBUG: print("DEBUG: Deleting key:", key) del nodeinfos[key][domain] except: pass - # NOISY-DEBUG: print("DEBUG: EXIT!") + # DEBUG: print("DEBUG: EXIT!") def update_last_error(domain: str, res: any): - # NOISY-DEBUG: print("DEBUG: domain,res[]:", domain, type(res)) + # DEBUG: print("DEBUG: domain,res[]:", domain, type(res)) try: - # NOISY-DEBUG: print("DEBUG: BEFORE res[]:", type(res)) + # DEBUG: print("DEBUG: BEFORE res[]:", type(res)) if isinstance(res, BaseException) or isinstance(res, json.JSONDecodeError): res = str(res) - # NOISY-DEBUG: print("DEBUG: AFTER res[]:", type(res)) + # DEBUG: print("DEBUG: AFTER res[]:", type(res)) if type(res) is str: - # NOISY-DEBUG: print(f"DEBUG: Setting last_error_details='{res}'"); + # DEBUG: print(f"DEBUG: Setting last_error_details='{res}'"); cursor.execute("UPDATE instances SET last_status_code = 999, last_error_details = ?, last_updated = ? WHERE domain = ? LIMIT 1", [ res, time.time(), domain ]) else: - # NOISY-DEBUG: print(f"DEBUG: Setting last_error_details='{res.reason}'"); + # DEBUG: print(f"DEBUG: Setting last_error_details='{res.reason}'"); cursor.execute("UPDATE instances SET last_status_code = ?, last_error_details = ?, last_updated = ? WHERE domain = ? LIMIT 1", [ res.status_code, res.reason, @@ -302,17 +302,17 @@ def update_last_error(domain: str, res: any): ]) if cursor.rowcount == 0: - # NOISY-DEBUG: print("DEBUG: Did not update any rows:", domain) + # DEBUG: print("DEBUG: Did not update any rows:", domain) pending_errors[domain] = res except BaseException as e: print("ERROR: failed SQL query:", domain, e) sys.exit(255) - # NOISY-DEBUG: print("DEBUG: EXIT!") + # DEBUG: print("DEBUG: EXIT!") def update_last_nodeinfo(domain: str): - # NOISY-DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain) + # DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain) try: cursor.execute("UPDATE instances SET last_nodeinfo = ?, last_updated = ? WHERE domain = ? LIMIT 1", [ time.time(), @@ -328,60 +328,95 @@ def update_last_nodeinfo(domain: str): sys.exit(255) connection.commit() - # NOISY-DEBUG: print("DEBUG: EXIT!") + # DEBUG: print("DEBUG: EXIT!") def get_peers(domain: str, software: str) -> list: - # NOISY-DEBUG: print("DEBUG: Getting peers for domain:", domain, software) + # DEBUG: print("DEBUG: Getting peers for domain:", domain, software) peers = list() - if software == "lemmy": - # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' is Lemmy, fetching JSON ...") + if software == "misskey": + print(f"DEBUG: domain='{domain}' is misskey, sending API POST request ...") + + counter = 0 + step = 99 + while True: + if counter == 0: + fetched = post_json_api(domain, "/api/federation/instances", json.dumps({ + "sort" : "+caughtAt", + "host" : None, + "limit": step + })) + else: + fetched = post_json_api(domain, "/api/federation/instances", json.dumps({ + "sort" : "+caughtAt", + "host" : None, + "limit" : step, + "offset": counter - 1 + })) + + # DEBUG: print("DEBUG: fetched():", len(fetched)) + if len(fetched) == 0: + # DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain) + break + + # Check records + for row in fetched: + # DEBUG: print(f"DEBUG: row():{len(row)}") + if "host" in row: + # DEBUG: print(f"DEBUG: Adding peer: '{row['host']}'") + peers.append(row["host"]) + + # DEBUG: print("DEBUG: Returning peers[]:", type(peers)) + sys.exit(255) + return peers + elif software == "lemmy": + # DEBUG: print(f"DEBUG: domain='{domain}' is Lemmy, fetching JSON ...") try: res = reqto.get(f"https://{domain}/api/v3/site", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])) - # NOISY-DEBUG: print(f"DEBUG: res.ok={res.ok},res.json[]={type(res.json())}") + # DEBUG: print(f"DEBUG: res.ok={res.ok},res.json[]={type(res.json())}") if res.ok and isinstance(res.json(), dict): - # NOISY-DEBUG: print("DEBUG: Success, res.json():", len(res.json())) - json = res.json() + # DEBUG: print("DEBUG: Success, res.json():", len(res.json())) + data = res.json() - if "federated_instances" in json and "linked" in json["federated_instances"]: - # NOISY-DEBUG: print("DEBUG: Found federated_instances", domain) - peers = json["federated_instances"]["linked"] + json["federated_instances"]["allowed"] + json["federated_instances"]["blocked"] + if "federated_instances" in data and "linked" in data["federated_instances"]: + # DEBUG: print("DEBUG: Found federated_instances", domain) + peers = data["federated_instances"]["linked"] + data["federated_instances"]["allowed"] + data["federated_instances"]["blocked"] except BaseException as e: print("WARNING: Exception during fetching JSON:", domain, e) update_last_nodeinfo(domain) - # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers)) + # DEBUG: print("DEBUG: Returning peers[]:", type(peers)) return peers elif software == "peertube": - # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' is a PeerTube, fetching JSON ...") + # DEBUG: print(f"DEBUG: domain='{domain}' is a PeerTube, fetching JSON ...") start = 0 for mode in ["followers", "following"]: - # NOISY-DEBUG: print(f"DEBUG: domain='{domain}',mode='{mode}'") + # DEBUG: print(f"DEBUG: domain='{domain}',mode='{mode}'") while True: try: res = reqto.get(f"https://{domain}/api/v1/server/{mode}?start={start}&count=100", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])) - # NOISY-DEBUG: print(f"DEBUG: res.ok={res.ok},res.json[]={type(res.json())}") + # DEBUG: print(f"DEBUG: res.ok={res.ok},res.json[]={type(res.json())}") if res.ok and isinstance(res.json(), dict): - # NOISY-DEBUG: print("DEBUG: Success, res.json():", len(res.json())) - json = res.json() + # DEBUG: print("DEBUG: Success, res.json():", len(res.json())) + data = res.json() - if "data" in json: - # NOISY-DEBUG: print(f"DEBUG: Found {len(json['data'])} record(s).") - for record in json["data"]: - # NOISY-DEBUG: print(f"DEBUG: record()={len(record)}") + if "data" in data: + # DEBUG: print(f"DEBUG: Found {len(data['data'])} record(s).") + for record in data["data"]: + # DEBUG: print(f"DEBUG: record()={len(record)}") if mode in record and "host" in record[mode]: - # NOISY-DEBUG: print(f"DEBUG: Found host={record[mode]['host']}, adding ...") + # DEBUG: print(f"DEBUG: Found host={record[mode]['host']}, adding ...") peers.append(record[mode]["host"]) else: print(f"WARNING: record from '{domain}' has no '{mode}' or 'host' record: {record}") - if len(json["data"]) < 100: - # NOISY-DEBUG: print("DEBUG: Reached end of JSON response:", domain) + if len(data["data"]) < 100: + # DEBUG: print("DEBUG: Reached end of JSON response:", domain) break # Continue with next row @@ -392,14 +427,14 @@ def get_peers(domain: str, software: str) -> list: update_last_nodeinfo(domain) - # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers)) + # DEBUG: print("DEBUG: Returning peers[]:", type(peers)) return peers - # NOISY-DEBUG: print(f"DEBUG: Fetching '{get_peers_url}' from '{domain}' ...") + # DEBUG: print(f"DEBUG: Fetching '{get_peers_url}' from '{domain}' ...") try: res = reqto.get(f"https://{domain}{get_peers_url}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])) - # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json())) + # DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json())) if not res.ok or res.status_code >= 400: res = reqto.get(f"https://{domain}/api/v3/site", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])) @@ -407,13 +442,13 @@ def get_peers(domain: str, software: str) -> list: print("WARNING: Could not reach any JSON API:", domain) update_last_error(domain, res) elif "federated_instances" in res.json() and "linked" in res.json()["federated_instances"]: - # NOISY-DEBUG: print("DEBUG: Found federated_instances", domain) + # DEBUG: print("DEBUG: Found federated_instances", domain) peers = res.json()["federated_instances"]["linked"] + res.json()["federated_instances"]["allowed"] + res.json()["federated_instances"]["blocked"] else: print("WARNING: JSON response does not contain 'federated_instances':", domain) update_last_error(domain, res) else: - # NOISY-DEBUG: print("DEBUG:Querying API was successful:", domain, len(res.json())) + # DEBUG: print("DEBUG:Querying API was successful:", domain, len(res.json())) peers = res.json() nodeinfos["get_peers_url"][domain] = get_peers_url @@ -423,37 +458,37 @@ def get_peers(domain: str, software: str) -> list: update_last_nodeinfo(domain) - # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers)) + # DEBUG: print("DEBUG: Returning peers[]:", type(peers)) return peers def post_json_api(domain: str, path: str, data: str) -> list: - # NOISY-DEBUG: print("DEBUG: Sending POST to domain,path,data:", domain, path, data) - json = {} + # DEBUG: print("DEBUG: Sending POST to domain,path,data:", domain, path, data) + data = {} try: res = reqto.post(f"https://{domain}{path}", data=data, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])) - # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json())) + # DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json())) if not res.ok or res.status_code >= 400: print("WARNING: Cannot query JSON API:", domain, path, data, res.status_code) update_last_error(domain, res) else: update_last_nodeinfo(domain) - json = res.json() + data = res.json() except BaseException as e: print("WARNING: Some error during post():", domain, path, data, e) - # NOISY-DEBUG: print("DEBUG: Returning json():", len(json)) - return json + # DEBUG: print("DEBUG: Returning data():", len(data)) + return data def fetch_nodeinfo(domain: str) -> list: - # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from domain:", domain) + # DEBUG: print("DEBUG: Fetching nodeinfo from domain:", domain) nodeinfo = fetch_wellknown_nodeinfo(domain) - # NOISY-DEBUG: print("DEBUG:nodeinfo:", len(nodeinfo)) + # DEBUG: print("DEBUG:nodeinfo:", len(nodeinfo)) if len(nodeinfo) > 0: - # NOISY-DEBUG: print("DEBUG: Returning auto-discovered nodeinfo:", len(nodeinfo)) + # DEBUG: print("DEBUG: Returning auto-discovered nodeinfo:", len(nodeinfo)) return nodeinfo requests = [ @@ -465,16 +500,16 @@ def fetch_nodeinfo(domain: str) -> list: f"https://{domain}/api/v1/instance" ] - json = {} + data = {} for request in requests: try: - # NOISY-DEBUG: print("DEBUG: Fetching request:", request) + # DEBUG: print("DEBUG: Fetching request:", request) res = reqto.get(request, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])) - # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json())) + # DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json())) if res.ok and isinstance(res.json(), dict): - # NOISY-DEBUG: print("DEBUG: Success:", request) - json = res.json() + # DEBUG: print("DEBUG: Success:", request) + data = res.json() nodeinfos["detection_mode"][domain] = "STATIC_CHECK" nodeinfos["nodeinfo_url"][domain] = request break @@ -484,35 +519,35 @@ def fetch_nodeinfo(domain: str) -> list: continue except BaseException as e: - # NOISY-DEBUG: print("DEBUG: Cannot fetch API request:", request) + # DEBUG: print("DEBUG: Cannot fetch API request:", request) update_last_error(domain, e) pass - # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json)) - return json + # DEBUG: print("DEBUG: Returning data[]:", type(data)) + return data def fetch_wellknown_nodeinfo(domain: str) -> list: - # NOISY-DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain) - json = {} + # DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain) + data = {} try: res = reqto.get(f"https://{domain}/.well-known/nodeinfo", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])) - # NOISY-DEBUG: print("DEBUG: domain,res.ok,res.json[]:", domain, res.ok, type(res.json())) + # DEBUG: print("DEBUG: domain,res.ok,res.json[]:", domain, res.ok, type(res.json())) if res.ok and isinstance(res.json(), dict): nodeinfo = res.json() - # NOISY-DEBUG: print("DEBUG: Found entries:", len(nodeinfo), domain) + # DEBUG: print("DEBUG: Found entries:", len(nodeinfo), domain) if "links" in nodeinfo: - # NOISY-DEBUG: print("DEBUG: Found links in nodeinfo():", len(nodeinfo["links"])) + # DEBUG: print("DEBUG: Found links in nodeinfo():", len(nodeinfo["links"])) for link in nodeinfo["links"]: - # NOISY-DEBUG: print("DEBUG: rel,href:", link["rel"], link["href"]) + # DEBUG: print("DEBUG: rel,href:", link["rel"], link["href"]) if link["rel"] in nodeinfo_identifier: - # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from:", link["href"]) + # DEBUG: print("DEBUG: Fetching nodeinfo from:", link["href"]) res = reqto.get(link["href"]) - # NOISY-DEBUG: print("DEBUG: href,res.ok,res.status_code:", link["href"], res.ok, res.status_code) + # DEBUG: print("DEBUG: href,res.ok,res.status_code:", link["href"], res.ok, res.status_code) if res.ok and isinstance(res.json(), dict): - # NOISY-DEBUG: print("DEBUG: Found JSON nodeinfo():", len(res.json())) - json = res.json() + # DEBUG: print("DEBUG: Found JSON nodeinfo():", len(res.json())) + data = res.json() nodeinfos["detection_mode"][domain] = "AUTO_DISCOVERY" nodeinfos["nodeinfo_url"][domain] = link["href"] break @@ -526,95 +561,95 @@ def fetch_wellknown_nodeinfo(domain: str) -> list: update_last_error(domain, e) pass - # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json)) - return json + # DEBUG: print("DEBUG: Returning data[]:", type(data)) + return data def fetch_generator_from_path(domain: str, path: str = "/") -> str: - # NOISY-DEBUG: print(f"DEBUG: domain='{domain}',path='{path}' - CALLED!") + # DEBUG: print(f"DEBUG: domain='{domain}',path='{path}' - CALLED!") software = None try: - # NOISY-DEBUG: print(f"DEBUG: Fetching '{path}' from '{domain}' ...") + # DEBUG: print(f"DEBUG: Fetching '{path}' from '{domain}' ...") res = reqto.get(f"https://{domain}{path}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])) - # NOISY-DEBUG: print("DEBUG: domain,res.ok,res.status_code,res.text[]:", domain, res.ok, res.status_code, type(res.text)) + # DEBUG: print("DEBUG: domain,res.ok,res.status_code,res.text[]:", domain, res.ok, res.status_code, type(res.text)) if res.ok and res.status_code < 300 and len(res.text) > 0: - # NOISY-DEBUG: print("DEBUG: Search for :", domain) + # DEBUG: print("DEBUG: Search for :", domain) doc = bs4.BeautifulSoup(res.text, "html.parser") - # NOISY-DEBUG: print("DEBUG: doc[]:", type(doc)) + # DEBUG: print("DEBUG: doc[]:", type(doc)) tag = doc.find("meta", {"name": "generator"}) - # NOISY-DEBUG: print(f"DEBUG: tag[{type(tag)}: {tag}") + # DEBUG: print(f"DEBUG: tag[{type(tag)}: {tag}") if isinstance(tag, bs4.element.Tag): - # NOISY-DEBUG: print("DEBUG: Found generator meta tag: ", domain) + # DEBUG: print("DEBUG: Found generator meta tag: ", domain) software = tidyup(tag.get("content")) print(f"INFO: domain='{domain}' is generated by '{software}'") nodeinfos["detection_mode"][domain] = "GENERATOR" remove_pending_error(domain) except BaseException as e: - # NOISY-DEBUG: print(f"DEBUG: Cannot fetch / from '{domain}':", e) + # DEBUG: print(f"DEBUG: Cannot fetch / from '{domain}':", e) update_last_error(domain, e) pass - # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}") + # DEBUG: print(f"DEBUG: software[]={type(software)}") if type(software) is str and software == "": - # NOISY-DEBUG: print(f"DEBUG: Corrected empty string to None for software of domain='{domain}'") + # DEBUG: print(f"DEBUG: Corrected empty string to None for software of domain='{domain}'") software = None elif type(software) is str and ("." in software or " " in software): - # NOISY-DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...") + # DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...") software = remove_version(software) - # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}") + # DEBUG: print(f"DEBUG: software[]={type(software)}") if type(software) is str and "powered by" in software: - # NOISY-DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it") + # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it") software = remove_version(strip_powered_by(software)) elif type(software) is str and " by " in software: - # NOISY-DEBUG: print(f"DEBUG: software='{software}' has ' by ' in it") + # DEBUG: print(f"DEBUG: software='{software}' has ' by ' in it") software = strip_until(software, " by ") elif type(software) is str and " see " in software: - # NOISY-DEBUG: print(f"DEBUG: software='{software}' has ' see ' in it") + # DEBUG: print(f"DEBUG: software='{software}' has ' see ' in it") software = strip_until(software, " see ") - # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!") + # DEBUG: print(f"DEBUG: software='{software}' - EXIT!") return software def determine_software(domain: str) -> str: - # NOISY-DEBUG: print("DEBUG: Determining software for domain:", domain) + # DEBUG: print("DEBUG: Determining software for domain:", domain) software = None - # NOISY-DEBUG: print(f"DEBUG: Fetching nodeinfo from '{domain}' ...") - json = fetch_nodeinfo(domain) + # DEBUG: print(f"DEBUG: Fetching nodeinfo from '{domain}' ...") + data = fetch_nodeinfo(domain) - # NOISY-DEBUG: print("DEBUG: json[]:", type(json)) - if not isinstance(json, dict) or len(json) == 0: - # NOISY-DEBUG: print("DEBUG: Could not determine software type:", domain) + # DEBUG: print("DEBUG: data[]:", type(data)) + if not isinstance(data, dict) or len(data) == 0: + # DEBUG: print("DEBUG: Could not determine software type:", domain) return fetch_generator_from_path(domain) - # NOISY-DEBUG: print("DEBUG: json():", len(json), json) - if "status" in json and json["status"] == "error" and "message" in json: - print("WARNING: JSON response is an error:", json["message"]) - update_last_error(domain, json["message"]) + # DEBUG: print("DEBUG: data():", len(data), data) + if "status" in data and data["status"] == "error" and "message" in data: + print("WARNING: JSON response is an error:", data["message"]) + update_last_error(domain, data["message"]) return fetch_generator_from_path(domain) - elif "software" not in json or "name" not in json["software"]: - # NOISY-DEBUG: print(f"DEBUG: JSON response from {domain} does not include [software][name], fetching / ...") + elif "software" not in data or "name" not in data["software"]: + # DEBUG: print(f"DEBUG: JSON response from {domain} does not include [software][name], fetching / ...") software = fetch_generator_from_path(domain) - # NOISY-DEBUG: print(f"DEBUG: Generator for domain='{domain}' is: {software}, EXIT!") + # DEBUG: print(f"DEBUG: Generator for domain='{domain}' is: {software}, EXIT!") return software - software = tidyup(json["software"]["name"]) + software = tidyup(data["software"]["name"]) - # NOISY-DEBUG: print("DEBUG: sofware after tidyup():", software) + # DEBUG: print("DEBUG: sofware after tidyup():", software) if software in ["akkoma", "rebased"]: - # NOISY-DEBUG: print("DEBUG: Setting pleroma:", domain, software) + # DEBUG: print("DEBUG: Setting pleroma:", domain, software) software = "pleroma" elif software in ["hometown", "ecko"]: - # NOISY-DEBUG: print("DEBUG: Setting mastodon:", domain, software) + # DEBUG: print("DEBUG: Setting mastodon:", domain, software) software = "mastodon" elif software in ["calckey", "groundpolis", "foundkey", "cherrypick", "meisskey"]: - # NOISY-DEBUG: print("DEBUG: Setting misskey:", domain, software) + # DEBUG: print("DEBUG: Setting misskey:", domain, software) software = "misskey" elif software.find("/") > 0: print("WARNING: Spliting of slash:", software) @@ -623,38 +658,38 @@ def determine_software(domain: str) -> str: print("WARNING: Spliting of pipe:", software) software = tidyup(software.split("|")[0]); elif "powered by" in software: - # NOISY-DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it") + # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it") software = strip_powered_by(software) elif type(software) is str and " by " in software: - # NOISY-DEBUG: print(f"DEBUG: software='{software}' has ' by ' in it") + # DEBUG: print(f"DEBUG: software='{software}' has ' by ' in it") software = strip_until(software, " by ") elif type(software) is str and " see " in software: - # NOISY-DEBUG: print(f"DEBUG: software='{software}' has ' see ' in it") + # DEBUG: print(f"DEBUG: software='{software}' has ' see ' in it") software = strip_until(software, " see ") - # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}") + # DEBUG: print(f"DEBUG: software[]={type(software)}") if software == "": print("WARNING: tidyup() left no software name behind:", domain) software = None - # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}") + # DEBUG: print(f"DEBUG: software[]={type(software)}") if str(software) == "": - # NOISY-DEBUG: print(f"DEBUG: software for '{domain}' was not detected, trying generator ...") + # DEBUG: print(f"DEBUG: software for '{domain}' was not detected, trying generator ...") software = fetch_generator_from_path(domain) elif len(str(software)) > 0 and ("." in software or " " in software): - # NOISY-DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...") + # DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...") software = remove_version(software) - # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}") + # DEBUG: print(f"DEBUG: software[]={type(software)}") if type(software) is str and "powered by" in software: - # NOISY-DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it") + # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it") software = remove_version(strip_powered_by(software)) - # NOISY-DEBUG: print("DEBUG: Returning domain,software:", domain, software) + # DEBUG: print("DEBUG: Returning domain,software:", domain, software) return software def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str): - # NOISY-DEBUG: print("DEBUG: Updating block reason:", reason, blocker, blocked, block_level) + # DEBUG: print("DEBUG: Updating block reason:", reason, blocker, blocked, block_level) try: cursor.execute( "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''", @@ -667,7 +702,7 @@ def update_block_reason(reason: str, blocker: str, blocked: str, block_level: st ), ) - # NOISY-DEBUG: print(f"DEBUG: cursor.rowcount={cursor.rowcount}") + # DEBUG: print(f"DEBUG: cursor.rowcount={cursor.rowcount}") if cursor.rowcount == 0: print("WARNING: Did not update any rows:", domain) @@ -675,10 +710,10 @@ def update_block_reason(reason: str, blocker: str, blocked: str, block_level: st print("ERROR: failed SQL query:", reason, blocker, blocked, block_level, e) sys.exit(255) - # NOISY-DEBUG: print("DEBUG: EXIT!") + # DEBUG: print("DEBUG: EXIT!") def update_last_seen(blocker: str, blocked: str, block_level: str): - # NOISY-DEBUG: print("DEBUG: Updating last_seen for:", blocker, blocked, block_level) + # DEBUG: print("DEBUG: Updating last_seen for:", blocker, blocked, block_level) try: cursor.execute( "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?", @@ -697,10 +732,10 @@ def update_last_seen(blocker: str, blocked: str, block_level: str): print("ERROR: failed SQL query:", last_seen, blocker, blocked, block_level, e) sys.exit(255) - # NOISY-DEBUG: print("DEBUG: EXIT!") + # DEBUG: print("DEBUG: EXIT!") def block_instance(blocker: str, blocked: str, reason: str, block_level: str): - # NOISY-DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level) + # DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level) if not validators.domain(blocker): print("WARNING: Bad blocker:", blocker) raise @@ -726,10 +761,10 @@ def block_instance(blocker: str, blocked: str, reason: str, block_level: str): print("ERROR: failed SQL query:", blocker, blocked, reason, block_level, first_added, last_seen, e) sys.exit(255) - # NOISY-DEBUG: print("DEBUG: EXIT!") + # DEBUG: print("DEBUG: EXIT!") def is_instance_registered(domain: str) -> bool: - # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' - CALLED!") + # DEBUG: print(f"DEBUG: domain='{domain}' - CALLED!") # Default is not registered registered = False @@ -744,11 +779,11 @@ def is_instance_registered(domain: str) -> bool: print("ERROR: failed SQL query:", blocker, blocked, reason, block_level, first_added, last_seen, e) sys.exit(255) - # NOISY-DEBUG: print("DEBUG: registered='{registered}' - EXIT!") + # DEBUG: print("DEBUG: registered='{registered}' - EXIT!") return registered def add_instance(domain: str, origin: str, originator: str): - # NOISY-DEBUG: print("DEBUG: domain,origin:", domain, origin, originator) + # DEBUG: print("DEBUG: domain,origin:", domain, origin, originator) if not validators.domain(domain): print("WARNING: Bad domain name:", domain) raise @@ -757,7 +792,7 @@ def add_instance(domain: str, origin: str, originator: str): raise software = determine_software(domain) - # NOISY-DEBUG: print("DEBUG: Determined software:", software) + # DEBUG: print("DEBUG: Determined software:", software) print(f"INFO: Adding instance {domain} (origin: {origin})") try: @@ -774,15 +809,15 @@ def add_instance(domain: str, origin: str, originator: str): ) for key in nodeinfos: - # NOISY-DEBUG: pprint(f"DEBUG: key='{key}',domain='{domain}',nodeinfos[key]={nodeinfos[key]}") + # DEBUG: pprint(f"DEBUG: key='{key}',domain='{domain}',nodeinfos[key]={nodeinfos[key]}") if domain in nodeinfos[key]: - # NOISY-DEBUG: pprint(f"DEBUG: domain='{domain}' has pending nodeinfo being updated ...") + # DEBUG: pprint(f"DEBUG: domain='{domain}' has pending nodeinfo being updated ...") update_nodeinfos(domain) remove_pending_error(domain) break if domain in pending_errors: - # NOISY-DEBUG: print("DEBUG: domain has pending error being updated:", domain) + # DEBUG: print("DEBUG: domain has pending error being updated:", domain) update_last_error(domain, pending_errors[domain]) remove_pending_error(domain) @@ -790,10 +825,10 @@ def add_instance(domain: str, origin: str, originator: str): print("ERROR: failed SQL query:", domain, e) sys.exit(255) else: - # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain) + # DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain) update_last_nodeinfo(domain) - # NOISY-DEBUG: print("DEBUG: EXIT!") + # DEBUG: print("DEBUG: EXIT!") def send_bot_post(instance: str, blocks: dict): message = instance + " has blocked the following instances:\n\n" @@ -831,7 +866,7 @@ def send_bot_post(instance: str, blocks: dict): return True def get_mastodon_blocks(domain: str) -> dict: - # NOISY-DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain) + # DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain) blocks = { "Suspended servers": [], "Filtered media" : [], @@ -853,7 +888,7 @@ def get_mastodon_blocks(domain: str) -> dict: header_text = tidyup(header.text) if header_text in language_mapping: - # NOISY-DEBUG: print(f"DEBUG: header_text='{header_text}'") + # DEBUG: print(f"DEBUG: header_text='{header_text}'") header_text = language_mapping[header_text] if header_text in blocks or header_text.lower() in blocks: @@ -867,7 +902,7 @@ def get_mastodon_blocks(domain: str) -> dict: } ) - # NOISY-DEBUG: print("DEBUG: Returning blocks for domain:", domain) + # DEBUG: print("DEBUG: Returning blocks for domain:", domain) return { "reject" : blocks["Suspended servers"], "media_removal" : blocks["Filtered media"], @@ -875,7 +910,7 @@ def get_mastodon_blocks(domain: str) -> dict: } def get_friendica_blocks(domain: str) -> dict: - # NOISY-DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain) + # DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain) blocks = [] try: @@ -892,7 +927,7 @@ def get_friendica_blocks(domain: str) -> dict: # Prevents exceptions: if blocklist is None: - # NOISY-DEBUG: print("DEBUG:Instance has no block list:", domain) + # DEBUG: print("DEBUG:Instance has no block list:", domain) return {} for line in blocklist.find("table").find_all("tr")[1:]: @@ -901,13 +936,13 @@ def get_friendica_blocks(domain: str) -> dict: "reason": tidyup(line.find_all("td")[1].text) }) - # NOISY-DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks)) + # DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks)) return { "reject": blocks } def get_misskey_blocks(domain: str) -> dict: - # NOISY-DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain) + # DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain) blocks = { "suspended": [], "blocked" : [] @@ -921,7 +956,7 @@ def get_misskey_blocks(domain: str) -> dict: # sending them all at once try: if counter == 0: - # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter) + # DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter) doc = post_json_api(domain, "/api/federation/instances/", json.dumps({ "sort" : "+caughtAt", "host" : None, @@ -929,7 +964,7 @@ def get_misskey_blocks(domain: str) -> dict: "limit" : step })) else: - # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter) + # DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter) doc = post_json_api(domain, "/api/federation/instances/", json.dumps({ "sort" : "+caughtAt", "host" : None, @@ -938,9 +973,9 @@ def get_misskey_blocks(domain: str) -> dict: "offset" : counter-1 })) - # NOISY-DEBUG: print("DEBUG: doc():", len(doc)) + # DEBUG: print("DEBUG: doc():", len(doc)) if len(doc) == 0: - # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain) + # DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain) break for instance in doc: @@ -955,10 +990,10 @@ def get_misskey_blocks(domain: str) -> dict: ) if len(doc) < step: - # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step) + # DEBUG: print("DEBUG: End of request:", len(doc), step) break - # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step) + # DEBUG: print("DEBUG: Raising counter by step:", step) counter = counter + step except BaseException as e: @@ -971,7 +1006,7 @@ def get_misskey_blocks(domain: str) -> dict: # same shit, different asshole ("blocked" aka full suspend) try: if counter == 0: - # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter) + # DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter) doc = post_json_api(domain,"/api/federation/instances", json.dumps({ "sort" : "+caughtAt", "host" : None, @@ -979,7 +1014,7 @@ def get_misskey_blocks(domain: str) -> dict: "limit" : step })) else: - # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter) + # DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter) doc = post_json_api(domain,"/api/federation/instances", json.dumps({ "sort" : "+caughtAt", "host" : None, @@ -988,9 +1023,9 @@ def get_misskey_blocks(domain: str) -> dict: "offset" : counter-1 })) - # NOISY-DEBUG: print("DEBUG: doc():", len(doc)) + # DEBUG: print("DEBUG: doc():", len(doc)) if len(doc) == 0: - # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain) + # DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain) break for instance in doc: @@ -1001,10 +1036,10 @@ def get_misskey_blocks(domain: str) -> dict: }) if len(doc) < step: - # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step) + # DEBUG: print("DEBUG: End of request:", len(doc), step) break - # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step) + # DEBUG: print("DEBUG: Raising counter by step:", step) counter = counter + step except BaseException as e: @@ -1013,7 +1048,7 @@ def get_misskey_blocks(domain: str) -> dict: counter = 0 break - # NOISY-DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"])) + # DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"])) return { "reject" : blocks["blocked"], "followers_only": blocks["suspended"]