Add first version of modules/mixed_unicode.py, designed to detect when we get a

message that mixes scripts (latin, cyrillic, greek, etc) that might be spam

1 files changed, 42 insertions, 0 deletions

diff --git a/modules/mixed_unicode.py b/modules/mixed_unicode.py
new file mode 100644
index 00000000..d2ef796e
--- /dev/null
+++ b/modules/mixed_unicode.py
@@ -0,0 +1,42 @@
+import enum
+from src import ModuleManager, utils
+
+class Script(enum.Enum):
+    Unknown = 0
+    Latin = 1
+    Cyrillic = 2
+    Greek = 3
+WORD_SEPERATORS = [",", " ", "\t", "."]
+
+class Module(ModuleManager.BaseModule):
+    def _detect_script(self, char):
+        point = ord(char)
+        if   0    <= point <= 687:
+            return Script.Latin
+        elif 880  <= point <= 1023:
+            return Script.Greek
+        elif 1024 <= point <= 1327:
+            return Script.Cyrillic
+        return Script.Unknown
+
+    @utils.hook("received.message.channel")
+    def channel_message(self, event):
+        last_script = None
+        last_was_separator = False
+        score = 0
+
+        for char in event["message"]:
+            if char in WORD_SEPERATORS:
+                last_was_separator = True
+            else:
+                script = self._detect_script(char)
+                if not script == Script.Unknown:
+                    if last_script and not script == last_script:
+                        score += 1
+                        if not last_was_separator:
+                            score += 1
+
+                    last_script = script
+
+                last_was_separator = False
+        self.log.trace("Message given a mixed-unicode score of %d", [score])