"""Extract temporal, situational, and conditional contexts from text"""
import re
from typing import List
from loguru import logger
class ContextExtractor:
"""
Extract temporal, situational, and conditional contexts from text.
For MVP: Simple regex-based extraction.
Future: LLM-based context understanding.
"""
CONTEXT_PATTERNS = [
# Temporal: "when X", "during X", "at X"
(r"when (\w+(?:\s+\w+){0,2})", "temporal"),
(r"during (\w+(?:\s+\w+){0,2})", "temporal"),
(r"at (night|work|home|school)", "temporal"),
(r"in the (morning|evening|afternoon)", "temporal"),
(r"in (morning|evening|afternoon)", "temporal"), # Without "the"
# Conditional: "if X", "while X"
(r"if (\w+(?:\s+\w+){0,2})", "conditional"),
(r"while (\w+(?:\s+\w+){0,2})", "conditional"),
# Situational: "at work", "at home", "at parties"
(r"at (work|home|school|office|parties|the office)", "situational"),
(r"for (work|fun|leisure|business)", "situational"),
(r"when (coding|learning|relaxing|working|exercising)", "situational"),
# Purpose/Use case: "for X" (broader pattern for any purpose)
(r"for ([a-zA-Z]+(?:\s+[a-zA-Z]+){0,2})", "purpose"),
]
def extract(self, text: str) -> List[str]:
"""
Extract context tags from text.
Args:
text: Message text to extract contexts from
Returns:
List of context strings
"""
contexts = []
text_lower = text.lower()
for pattern, context_type in self.CONTEXT_PATTERNS:
matches = re.finditer(pattern, text_lower)
for match in matches:
context = match.group(1).strip()
contexts.append(context)
logger.debug(f"Extracted {context_type} context: {context}")
# Deduplicate
unique_contexts = list(set(contexts))
if unique_contexts:
logger.info(f"Extracted {len(unique_contexts)} contexts: {unique_contexts}")
return unique_contexts
