From: Roland Häder <roland@mxchange.org>
Date: Sun, 25 Jun 2023 10:32:07 +0000 (+0200)
Subject: Continued:
X-Git-Url: https://git.mxchange.org/?a=commitdiff_plain;h=2d08bbdd47e0ce0c069f20d3773c88f2bead75ee;p=fba.git

Continued:
- if lemmy's API doesn't work or returns zero rows, try /instances instead
---

diff --git a/fba/http/federation.py b/fba/http/federation.py
index ff7cf5e..7d5d97c 100644
--- a/fba/http/federation.py
+++ b/fba/http/federation.py
@@ -82,8 +82,8 @@ def fetch_instances(domain: str, origin: str, software: str, command: str, path:
     logger.debug("Updating last_instance_fetch for domain='%s' ...", domain)
     instances.set_last_instance_fetch(domain)
 
-    logger.debug("Fetching instances for domain='%s',software='%s'", domain, software)
-    peerlist = fetch_peers(domain, software)
+    logger.debug("Fetching instances for domain='%s',software='%s',origin='%s'", domain, software, origin)
+    peerlist = fetch_peers(domain, software, origin)
 
     logger.debug("peerlist[]='%s'", type(peerlist))
     if isinstance(peerlist, list):
@@ -137,8 +137,8 @@ def fetch_instances(domain: str, origin: str, software: str, command: str, path:
 
     logger.debug("EXIT!")
 
-def fetch_peers(domain: str, software: str) -> list:
-    logger.debug("domain='%s',software='%s' - CALLED!", domain, software)
+def fetch_peers(domain: str, software: str, origin: str) -> list:
+    logger.debug("domain='%s',software='%s',origin='%s' - CALLED!", domain, software, origin)
     domain_helper.raise_on(domain)
 
     if not isinstance(software, str) and software is not None:
@@ -148,8 +148,8 @@ def fetch_peers(domain: str, software: str) -> list:
         logger.debug("Invoking misskey.fetch_peers(%s) ...", domain)
         return misskey.fetch_peers(domain)
     elif software == "lemmy":
-        logger.debug("Invoking lemmy.fetch_peers(%s) ...", domain)
-        return lemmy.fetch_peers(domain)
+        logger.debug("Invoking lemmy.fetch_peers(%s,%s) ...", domain, origin)
+        return lemmy.fetch_peers(domain, origin)
     elif software == "peertube":
         logger.debug("Invoking peertube.fetch_peers(%s) ...", domain)
         return peertube.fetch_peers(domain)
diff --git a/fba/networks/lemmy.py b/fba/networks/lemmy.py
index c8d0bad..288280b 100644
--- a/fba/networks/lemmy.py
+++ b/fba/networks/lemmy.py
@@ -32,9 +32,10 @@ from fba.models import instances
 
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
+#logger.setLevel(logging.DEBUG)
 
-def fetch_peers(domain: str) -> list:
-    logger.debug("domain='%s' - CALLED!", domain)
+def fetch_peers(domain: str, origin: str) -> list:
+    logger.debug("domain='%s',origin='%s' - CALLED!", domain, origin)
     domain_helper.raise_on(domain)
 
     peers = list()
@@ -68,8 +69,8 @@ def fetch_peers(domain: str) -> list:
             peers = peers + federation.add_peers(data["json"]["federated_instances"])
             instances.set_success(domain)
         else:
-            logger.warning("JSON response does not contain 'federated_instances', domain='%s'", domain)
-            instances.set_last_error(domain, data)
+            logger.warning("JSON response does not contain 'federated_instances', domain='%s' - trying /instances ...", domain)
+            peers = fetch_instances(domain, origin)
 
     except network.exceptions as exception:
         logger.warning("Exception during fetching JSON: domain='%s',exception[%s]:'%s'", domain, type(exception), str(exception))
@@ -185,3 +186,58 @@ def fetch_blocks(domain: str, nodeinfo_url: str) -> list:
 
     logger.debug("blocklist()=%d - EXIT!", len(blocklist))
     return blocklist
+
+def fetch_instances(domain: str, origin: str) -> list:
+    logger.debug("domain='%s',origin='%s' - CALLED!", domain, origin)
+    domain_helper.raise_on(domain)
+
+    peers = list()
+
+    try:
+        # json endpoint for newer mastodongs
+        logger.debug("Fetching /instances from domain='%s'", domain)
+        response = network.fetch_response(
+            domain,
+            "/instances",
+            network.web_headers,
+            (config.get("connection_timeout"), config.get("read_timeout"))
+        )
+
+        logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
+        if response.ok and response.status_code < 300 and response.text != "":
+            logger.debug("Parsing %s Bytes ...", len(response.text))
+
+            doc = bs4.BeautifulSoup(response.text, "html.parser")
+            logger.debug("doc[]='%s'", type(doc))
+
+            headers = doc.findAll("h5")
+            logger.debug("Checking %d headers ...", len(headers))
+            for header in headers:
+                logger.debug("header[%s]='%s'", type(header), header)
+
+                rows = header.find_next(["ul","table"]).findAll("a")
+                logger.debug("Found %d blocked instance(s) ...", len(rows))
+                for tag in rows:
+                    logger.debug("tag[]='%s'", type(tag))
+                    peer = tidyup.domain(tag.contents[0])
+                    logger.debug("peer='%s'", peer)
+
+                    if peer == "":
+                        logger.debug("peer is empty - SKIPPED!")
+                        continue
+                    elif not utils.is_domain_wanted(peer):
+                        logger.debug("peer='%s' is not wanted - SKIPPED!", peer)
+                        continue
+                    elif peer in peers:
+                        logger.debug("peer='%s' already added - SKIPPED!", peer)
+                        continue
+
+                    logger.debug("Appending peer='%s' ...", peer)
+                    peers.append(peer)
+
+    except network.exceptions as exception:
+        logger.warning("domain='%s',exception[%s]:'%s'", domain, type(exception), str(exception))
+        instances.set_last_error(domain, exception)
+
+    logger.debug("peers()=%d - EXIT!", len(peers))
+    return peers