its-consulting/KI-Chat-Bot-Eugen
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 1

Implement performance improvements for memory caching, HTTP client reuse, and regex optimization

Browse source 
Co-authored-by: Kenearos <86194771+Kenearos@users.noreply.github.com>
This commit is contained in:
copilot-swe-agent[bot] 2026-01-27 17:33:17 +00:00
parent 860e4d5027
commit b72cd9db1c
5 changed files with 451 additions and 246 deletions
Download patch file Download diff file Expand all files Collapse all files

54
utils.py
View file

@ -20,8 +20,22 @@ class MentionDetector:
# Create patterns for various mention formats
# Include bot name and all nicknames
all_names = [bot_name] + self.nicknames
self.patterns = []
# Build a single combined pattern for efficient matching (uses alternation)
name_alternatives = "|".join(re.escape(name) for name in all_names)
# Combined pattern that matches any mention format
# This is more efficient than checking multiple patterns separately
combined_mention_pattern = (
rf"(?:@(?:{name_alternatives})\b)|" # @name
rf"(?:\b(?:{name_alternatives})[:!?.,])|" # name: name! etc.
rf"(?:^(?:{name_alternatives})\b)|" # name at start
rf"(?:\b(?:{name_alternatives})\b)" # name anywhere
)
self._mention_pattern = re.compile(combined_mention_pattern, re.IGNORECASE)
# Keep individual patterns list for backward compatibility (tests may use it)
self.patterns = []
for name in all_names:
self.patterns.extend([
rf"@{name}\b", # @name (with word boundary)
@ -29,12 +43,21 @@ class MentionDetector:
rf"^{name}\b", # name at start of message
rf"\b{name}\b", # name anywhere as whole word
])
# Case-insensitive compilation
self.compiled_patterns = [
re.compile(pattern, re.IGNORECASE) for pattern in self.patterns
]
# Pre-compile extraction patterns for each name (more efficient than recompiling each time)
self._extraction_patterns = {}
for name in all_names:
escaped_name = re.escape(name)
self._extraction_patterns[name.lower()] = {
'at_start': re.compile(rf"^@{escaped_name}\b[,:]?\s*", re.IGNORECASE),
'name_start': re.compile(rf"^{escaped_name}\b[,:]?\s*", re.IGNORECASE),
'name_end': re.compile(rf"\s*\b{escaped_name}[,!?.]?\s*$", re.IGNORECASE),
'name_middle': re.compile(rf"\s*\b{escaped_name}[,:!?]\s*", re.IGNORECASE),
}
# Patterns for ambiguous greetings (might be directed at bot)
self.greeting_patterns = [
r"^(hi|hey|hallo|hello|servus|moin)(\s|$|\W)",
@ -78,10 +101,8 @@ class MentionDetector:
if not message:
return False
for pattern in self.compiled_patterns:
if pattern.search(message):
return True
return False
# Use the optimized single combined pattern for faster matching
return bool(self._mention_pattern.search(message))
def is_ambiguous_greeting(self, message):
"""
@ -122,18 +143,21 @@ class MentionDetector:
content = message
all_names = [self.bot_name] + self.nicknames
# Use pre-compiled patterns for better performance
for name in all_names:
# Remove @mention at start
content = re.sub(rf"^@{name}\b[,:]?\s*", "", content, flags=re.IGNORECASE)
patterns = self._extraction_patterns.get(name.lower())
if patterns:
# Remove @mention at start
content = patterns['at_start'].sub("", content)
# Remove name at start with optional punctuation
content = re.sub(rf"^{name}\b[,:]?\s*", "", content, flags=re.IGNORECASE)
# Remove name at start with optional punctuation
content = patterns['name_start'].sub("", content)
# Remove name at end with optional punctuation
content = re.sub(rf"\s*\b{name}[,!?.]?\s*$", "", content, flags=re.IGNORECASE)
# Remove name at end with optional punctuation
content = patterns['name_end'].sub("", content)
# Remove name in middle with punctuation
content = re.sub(rf"\s*\b{name}[,:!?]\s*", " ", content, flags=re.IGNORECASE)
# Remove name in middle with punctuation
content = patterns['name_middle'].sub(" ", content)
return content.strip()