scan timeline, formatting

3 files changed, 81 insertions, 26 deletions

diff --git a/config.py b/config.py
index 0469806..c1e4e7e 100644
--- a/config.py
+++ b/config.py
@@ -8,6 +8,9 @@ TOKEN = 'token.secret'
 # in seconds
 POLL_INTERVAL = 20
 
+# in case of DoS, we won't handle too long a status
+MAX_STATUS_LENGTH = 1024
+
 # account username and PW
 UNAME = ""
 PW = ""
@@ -21,3 +24,4 @@ except:
     print("account secret not found, please manually input:")
     UNAME = input("username or email")
     PW = input("password (not concealed)")
+
diff --git a/main.py b/main.py
index e6bf403..73584f5 100644
--- a/main.py
+++ b/main.py
@@ -9,44 +9,97 @@ from patterns import *
 
 def login_refresh_token():
     session = Mastodon(client_id=config.CLIENTID)
-    session.log_in(username=config.UNAME, password=config.PW, to_file=config.TOKEN)
+    session.log_in(username=config.UNAME,
+                   password=config.PW, to_file=config.TOKEN)
     return session
 
+
 def restore_session():
     return Mastodon(client_id=config.CLIENTID, access_token=config.TOKEN)
 
 
 class VnilCat:
-    def __init__(self,config):
+    def __init__(self, config):
         self.session = restore_session()
+        self.tl_lastseen_sid = None
+        self.config = config
         try:
             self.session.account_verify_credentials()
         except:
             try:
                 self.session = login_refresh_token()
+
             except:
                 exit()
+        # try to fetch the lastest status from the timeline
+        # so that we can skip the ones before we start.
+        # This is not a persistent state machine,
+        # we don't know if a older status has been processed
+        # or not!
+        try:
+            tl = self.session.timeline_home(limit=1)
+            if (len(tl)) != 0:
+                print("set init status id to begin with...", tl[0]["id"])
+                self.tl_lastseen_sid = tl[0]["id"]
+        except Exception as e:
+            print("can't init timeline, continue anyways", e)
 
+    def not_mine(self, status):
+        return status['account']['acct'] != self.config.UNAME
+
+    def is_mine(self, status):
+        return status['account']['acct'] != self.config.UNAME
 
     def reply_meow(self, ori_status):
         print("replying meow to ", ori_status["id"])
-        self.session.status_reply(to_status=ori_status, status=random.choice(cat_sounds))
+        self.session.status_reply(
+            to_status=ori_status, status=random.choice(cat_sounds))
+
+    def handle_command(self, status, content):
+        print("we have a command", content)
+        cmd = content[1:].strip().split()
+
+    def handle_home_status(self, status):
+        self.tl_lastseen_sid = status["id"]
+        if not self.is_mine(status):
+            return
+        acc = status["account"]
+        content = cleanhtml(status["content"])
+        if re_contains_bird.search(content) is not None and len(content) < 10:
+            print("i see a bird", status["id"])
+            try:
+                s = self.session.status_reply(
+                    to_status=status, status=random.choice(bird_sounds))
+            except:
+                print("fail to post")
 
+    def scantimeline(self):
+        # print("scanning timeline, lastseen=",self.tl_lastseen_sid)
+        tl = self.session.timeline_home(since_id=self.tl_lastseen_sid)
+        # important! make sure to iterate from older status to newer
+        # otherwise the 'last_seen' won't be updated correctly
+        for status in reversed(tl):
+            # skip if it's from myself
+            self.handle_home_status(status=status)
 
     # For now we don't handle other interaction types
+
     def handle_mention(self, notification):
         nid = notification["id"]
-        # in the case of mention,
-        # notification['status']['account'] and
-        # notification['account'] are the same thing
         acc = notification["account"]
         status = notification["status"]
         sid = status["id"]
-        # created_at = notification["created_at"]
-        # pleroma = notification["pleroma"]
-        print("we have a mention > ",nid)
-        content = status["content"]
-        if re_contains_meow.search(content) is not None:
+        print("we have a mention > ", nid)
+        content = cleanhtml(status["content"]).replace("@cat", "").lstrip()
+        print(content)
+        if len(content) == 0 or len(content) > config.MAX_STATUS_LENGTH:
+            print("invalid status content")
+
+        # prioritized to commands
+        elif content[0] == "!":
+            self.handle_command(status, content)
+
+        elif re_contains_meow.search(content) is not None:
             self.reply_meow(status)
 
         print("dismissing notification ", nid)
@@ -63,9 +116,6 @@ class VnilCat:
             print("already following", uid)
         self.session.notifications_dismiss(nid)
 
-    def scan_local(self,session):
-        print("TODO")
-
     def handle_notification(self):
         ns = self.session.notifications()
         for n in ns:
@@ -84,19 +134,15 @@ def init_bot():
 
 
 def run():
-    # print("HELLO")
     cat = VnilCat(config)
-    # print(cat.session.notifications())
-
-    # print(session.notifications())
-    # session.toot("WTF")
-
     while True:
-        time.sleep(config.POLL_INTERVAL)
         try:
             cat.handle_notification()
-        except:
+            cat.scantimeline()
+        except Exception as e:
             print("something wrong...")
+            print(e)
+        time.sleep(config.POLL_INTERVAL)
 
 
 if __name__ == "__main__":
@@ -106,6 +152,3 @@ if __name__ == "__main__":
         init_bot()
 
     run()
-
-
-
diff --git a/patterns.py b/patterns.py
index 9118227..a91064f 100644
--- a/patterns.py
+++ b/patterns.py
@@ -2,6 +2,14 @@ import re
 
 #### RE PATTERNS ####
 re_contains_meow = re.compile(r'(me+o+w|喵)',re.IGNORECASE)
-
+re_contains_bird = re.compile(r'(鸟|bird)',re.IGNORECASE)
 #### MISC STRING PRESETS ####
 cat_sounds = ["Meow!", "Mrrrow!", "Purr...", "Meee-OW!", "Mreoww!", "Nya~", "Mew?", "Rowr?", "Prrrr...", "Maow-maow!"]
+bird_sounds = ["鸟!", "chirp!"]
+
+#### PARSE STATUS HTMLs....
+RE_HTML_CLEANR = re.compile('<.*?>')
+
+def cleanhtml(raw_html):
+  cleantext = re.sub(RE_HTML_CLEANR, '', raw_html)
+  return cleantext