From: Roland Häder <roland@mxchange.org>
Date: Thu, 8 Feb 2024 17:00:57 +0000 (+0100)
Subject: Continued:
X-Git-Url: https://git.mxchange.org/?a=commitdiff_plain;h=618a7b0293a20f9f19608523450ae6178e3c2322;p=fba.git

Continued:
- make --domain parameter for command fetch_instances optional so --software
  can be handled, too
- skip recently crawled domains in same loop
- args.<domain|software> can be both of type None, too
---

diff --git a/fba/boot.py b/fba/boot.py
index ea31d7e..0fb55ec 100644
--- a/fba/boot.py
+++ b/fba/boot.py
@@ -144,7 +144,8 @@ def init_parser():
         help="Fetches instances (aka. \"domains\") from an initial instance. You may want to re-run this command several times (at least 3 with big instances) to have a decent amount of valid instances.",
     )
     parser.set_defaults(command=commands.fetch_instances)
-    parser.add_argument("--domain", required=True, help="Instance name (aka. domain) to fetch further instances from. Start with a large instance, e.g. mastodon.social .")
+    parser.add_argument("--domain", help="Instance name (aka. domain) to fetch further instances from. Start with a large instance, e.g. mastodon.social .")
+    parser.add_argument("--software", help="Name of software, e.g. 'lemmy'")
     parser.add_argument("--single", action="store_true", help="Only fetch given instance.")
 
     ### Fetch blocks from static text file(s) ###
diff --git a/fba/commands.py b/fba/commands.py
index 561815c..6999e2a 100644
--- a/fba/commands.py
+++ b/fba/commands.py
@@ -962,7 +962,7 @@ def fetch_instances(args: argparse.Namespace) -> int:
     rows = list()
 
     # Is domain or software set?
-    if args.domain != "":
+    if args.domain not in [None, ""]:
         logger.debug("args.domain='%s' - checking ...", args.domain)
         if not validators.domain(args.domain):
             logger.warning("args.domain='%s' is not valid.", args.domain)
@@ -978,6 +978,14 @@ def fetch_instances(args: argparse.Namespace) -> int:
         # Fetch record
         database.cursor.execute("SELECT domain, origin, software FROM instances WHERE domain = ? LIMIT 1", [domain])
         rows = database.cursor.fetchall()
+    elif args.software not in [None, ""]:
+        logger.debug("args.software='%s' - BEFORE!", args.software)
+        software = software_helper.alias(args.software)
+        logger.debug("software='%s' - AFTER!", software)
+
+        # Fetch records
+        database.cursor.execute("SELECT domain, origin, software FROM instances WHERE software = ? ORDER BY last_updated ASC", [software])
+        rows = database.cursor.fetchall()
 
     logger.info("Checking %d entries ...", len(rows))
     for row in rows:
@@ -988,6 +996,9 @@ def fetch_instances(args: argparse.Namespace) -> int:
         elif instances.is_registered(row["domain"]) and software_helper.is_relay(row["software"]):
             logger.warning("row[domain]='%s' is of software type '%s' which is not supported by this command. Please invoke fetch_relays instead - SKIPPED!", row["domain"], row["software"])
             continue
+        elif instances.is_recent(row["domain"]):
+            logger.debug("row[domain]='%s' has been recently crawled - SKIPPED!", row["domain"])
+            continue
 
         # Initial fetch
         try: