Merge pull request #191 from aahnik/add-regex-support

Add regex support for text filtering and replacement

.gitignore

@@ -6,6 +6,7 @@ tgcf.config.yml
 foo.py
 bar.py
 foo.bar
+t.ipynb
 
 
 

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "tgcf"
-version = "0.2.4"
+version = "0.2.5"
 description = "The ultimate tool to automate custom telegram message forwarding."
 authors = ["aahnik <daw@aahnik.dev>"]
 license = "MIT"

tgcf/plugins/filter.py

@@ -4,6 +4,7 @@
 from pydantic import BaseModel  # pylint: disable=no-name-in-module
 
 from tgcf.plugins import FileType, TgcfMessage, TgcfPlugin
+from tgcf.utils import match
 
 
 class FilterList(BaseModel):
@@ -18,6 +19,7 @@ class FilesFilterList(BaseModel):
 
 class TextFilter(FilterList):
     case_sensitive: bool = False
+    regex: bool = False
 
 
 class Filters(BaseModel):
@@ -58,14 +60,18 @@ def text_safe(self, tm: TgcfMessage) -> bool:
         if not text and flist.whitelist == []:
             return True
 
+        # first check if any blacklisted pattern is present
         for forbidden in flist.blacklist:
-            if forbidden in text:
-                return False
+            if match(forbidden, text, self.filters.text.regex):
+                return False  # when a forbidden pattern is found
+
         if not flist.whitelist:
-            return True
+            return True  # if no whitelist is present
+
+        # if whitelist is present
         for allowed in flist.whitelist:
-            if allowed in text:
-                return True
+            if match(allowed, text, self.filters.text.regex):
+                return True  # only when atleast one whitelisted pattern is found
 
     def users_safe(self, tm: TgcfMessage) -> bool:
         flist = self.filters.users

tgcf/plugins/replace.py

@@ -4,10 +4,12 @@
 from pydantic import BaseModel  # pylint: disable=no-name-in-module
 
 from tgcf.plugins import TgcfMessage, TgcfPlugin
+from tgcf.utils import replace
 
 
 class Replace(BaseModel):
     text: Dict[str, str] = {}
+    regex: bool = False
 
 
 class TgcfReplace(TgcfPlugin):
@@ -22,6 +24,6 @@ def modify(self, tm: TgcfMessage) -> TgcfMessage:
         if not msg_text:
             return tm
         for original, new in self.replace.text.items():
-            msg_text = msg_text.replace(original, new)
+            msg_text = replace(original, new, msg_text, self.replace.regex)
         tm.text = msg_text
         return tm

tgcf/utils.py

@@ -56,3 +56,16 @@ def safe_name(string: str) -> str:
     Certain characters in the file name can cause potential problems in rare scenarios.
     """
     return re.sub(pattern=r"[-!@#$%^&*()\s]", repl="_", string=string)
+
+
+def match(pattern: str, string: str, regex: bool) -> bool:
+    if regex:
+        return bool(re.match(pattern, string))
+    return pattern in string
+
+
+def replace(pattern: str, new: str, string: str, regex: bool) -> str:
+    if regex:
+        return re.sub(pattern, new, string)
+    else:
+        return string.replace(pattern, new)