From c387fc16d9a61c347f36d31dea7b155ea009ad50 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Roland=20H=C3=A4der?= <roland@mxchange.org>
Date: Sun, 19 Nov 2023 17:38:45 +0100
Subject: [PATCH] Continued: - added command "fetch_relaylist" to fetch relays
 from relaylist.com

---
 fba/boot.py            |  7 +++++
 fba/commands.py        | 69 ++++++++++++++++++++++++++++++++++++++++++
 fba/http/federation.py |  2 +-
 3 files changed, 77 insertions(+), 1 deletion(-)

diff --git a/fba/boot.py b/fba/boot.py
index 86411b7..dddc80f 100644
--- a/fba/boot.py
+++ b/fba/boot.py
@@ -250,6 +250,13 @@ def init_parser():
     parser.add_argument("--software", help="Name of software, e.g. 'lemmy'")
     parser.add_argument("--force", action="store_true", help="Forces update of data, no matter what.")
 
+    ### Fetches relay list from relaylist.com
+    parser = subparser_command.add_parser(
+        "fetch_relaylist",
+        help="Fetches relay list from relaylist.com",
+    )
+    parser.set_defaults(command=commands.fetch_relaylist)
+
     ### Remove invalid domains ###
     parser = subparser_command.add_parser(
         "remove_invalid",
diff --git a/fba/commands.py b/fba/commands.py
index e10ab32..b000e3d 100644
--- a/fba/commands.py
+++ b/fba/commands.py
@@ -1599,6 +1599,75 @@ def fetch_instances_social(args: argparse.Namespace) -> int:
     logger.debug("Success! - EXIT!")
     return 0
 
+def fetch_relaylist(args: argparse.Namespace) -> int:
+    logger.debug("args[]='%s' - CALLED!", type(args))
+
+    logger.debug("Invoking locking.acquire() ...")
+    locking.acquire()
+
+    source_domain = "api.relaylist.com"
+
+    if sources.is_recent(source_domain):
+        logger.info("API from source_domain='%s' has recently being accessed - EXIT!", source_domain)
+        return 1
+    else:
+        logger.debug("source_domain='%s' has not been recently used, marking ...", source_domain)
+        sources.update(source_domain)
+
+    logger.info("Fetching list from source_domain='%s' ...", source_domain)
+    fetched = network.get_json_api(
+        source_domain,
+        "/relays",
+        {},
+        (config.get("connection_timeout"), config.get("read_timeout"))
+    )
+    logger.debug("fetched[]='%s'", type(fetched))
+
+    if "error_message" in fetched:
+        logger.warning("Error during fetching API result: '%s' - EXIT!", fetched["error_message"])
+        return 2
+    elif "exception" in fetched:
+        logger.warning("Exception '%s' during fetching API result - EXIT!", type(fetched["exception"]))
+        return 3
+    elif "json" not in fetched:
+        logger.warning("fetched has no element 'json' - EXIT!")
+        return 4
+
+    domains = list()
+
+    logger.info("Checking %d row(s) ...", len(fetched["json"]))
+    for row in fetched["json"]:
+        logger.debug("row[]='%s'", type(row))
+        domain = urlparse(row["url"]).netloc.lower().split(":")[0]
+
+        logger.debug("domain='%s' - AFTER!", domain)
+        if domain is None and domain == "":
+            logger.debug("domain='%s' is empty after tidyup.domain() - SKIPPED!", domain)
+            continue
+
+        logger.debug("domain='%s' - BEFORE!", domain)
+        domain = domain.encode("idna").decode("utf-8")
+        logger.debug("domain='%s' - AFTER!", domain)
+
+        if not domain_helper.is_wanted(domain):
+            logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
+            continue
+        elif domain in domains:
+            logger.debug("domain='%s' is already added - SKIPPED!", domain)
+            continue
+        elif instances.is_registered(domain):
+            logger.debug("domain='%s' is already registered - SKIPPED!", domain)
+            continue
+        elif instances.is_recent(domain):
+            logger.debug("domain='%s' has been recently crawled - SKIPPED!", domain)
+            continue
+
+        logger.info("Fetching instances from domain='%s'", domain)
+        federation.fetch_instances(domain, None, None, inspect.currentframe().f_code.co_name)
+
+    logger.debug("Success! - EXIT!")
+    return 0
+
 def fetch_relays(args: argparse.Namespace) -> int:
     logger.debug("args[]='%s' - CALLED!", type(args))
 
diff --git a/fba/http/federation.py b/fba/http/federation.py
index 50f66c3..6f7f033 100644
--- a/fba/http/federation.py
+++ b/fba/http/federation.py
@@ -56,7 +56,7 @@ def fetch_instances(domain: str, origin: str, software: str, command: str, path:
         raise ValueError(f"Parameter command[]='{type(command)}' is not of type 'str'")
     elif command == "":
         raise ValueError("Parameter 'command' is empty")
-    elif command in ["fetch_blocks", "fetch_cs", "fetch_bkali", "fetch_relays", "fetch_fedipact", "fetch_joinmobilizon", "fetch_joinmisskey", "fetch_joinfediverse"] and origin is None:
+    elif command in ["fetch_blocks", "fetch_cs", "fetch_bkali", "fetch_relays", "fetch_fedipact", "fetch_joinmobilizon", "fetch_joinmisskey", "fetch_joinfediverse", "fetch_relaylist"] and origin is None:
         raise ValueError(f"Parameter command='{command}' but origin is None, please fix invoking this function.")
     elif not isinstance(path, str) and path is not None:
         raise ValueError(f"Parameter path[]='{type(path)}' is not of type 'str'")
-- 
2.39.5