Initial commit

.gitignore

@@ -0,0 +1,2 @@
+*.iml
+.idea/

README.md

@@ -0,0 +1 @@
+Misc utility methods in Python

imhash.py

@@ -0,0 +1,8 @@
+import base64
+
+
+def b64hash(imhash, bcount):
+    """ImageHash hash to base64 string"""
+    return base64.b64encode(
+        sum(1 << i for i, b in enumerate(imhash.hash.flatten()) if b).to_bytes(bcount, "big")
+    ).decode("ascii")

misc.py

@@ -0,0 +1,24 @@
+import time
+
+last_time_called = dict()
+
+
+def rate_limit(per_second):
+    min_interval = 1.0 / float(per_second)
+
+    def decorate(func):
+        last_time_called[func] = 0
+
+        def wrapper(*args, **kwargs):
+            elapsed = time.perf_counter() - last_time_called[func]
+            wait_time = min_interval - elapsed
+            if wait_time > 0:
+                time.sleep(wait_time)
+
+            last_time_called[func] = time.perf_counter()
+            return func(*args, **kwargs)
+
+        return wrapper
+
+    return decorate
+

monitoring.py

@@ -0,0 +1,30 @@
+import logging
+import traceback
+
+from influxdb import InfluxDBClient
+
+
+class Monitoring:
+    def __init__(self, db, host="localhost", logger=logging.getLogger("default")):
+        self._db = db
+        self._client = InfluxDBClient(host, 8086, "", "", db)
+        self._logger = logger
+
+        self._init()
+
+    def db_exists(self, name):
+        for db in self._client.get_list_database():
+            if db["name"] == name:
+                return True
+        return False
+
+    def _init(self):
+        if not self.db_exists(self._db):
+            self._client.create_database(self._db)
+
+    def log(self, event):
+        try:
+            self._client.write_points(event)
+        except Exception as e:
+            self._logger.debug(traceback.format_exc())
+            self._logger.error(str(e))

regex.py

@@ -0,0 +1,4 @@
+import re
+
+LINK_RE = re.compile(r"(https?://[\w\-_.]+\.[a-z]{2,4}([^\s<'\"]*|$))")
+HTML_HREF_RE = re.compile(r"href=\"([^\"]+)\"")

requirements.txt

@@ -0,0 +1,2 @@
+ImageHash
+influxdb