wuenlp_tools.utils.sentiment

 1from pathlib import Path
 2
 3import pandas as pd
 4
 5_EMOTION_COLUMNS = (
 6    "anticipation",
 7    "sadness",
 8    "fear",
 9    "anger",
10    "disgust",
11    "trust",
12    "surprise",
13    "joy",
14    "positive",
15    "negative",
16)
17
18
19class SentimentLexicon(object):
20    def __init__(self, lexicon: pd.DataFrame):
21        self.lexicon = lexicon
22        self._word_scores: dict[str, dict[str, float]] | None = None
23
24    def _ensure_word_scores(self) -> dict[str, dict[str, float]]:
25        if self._word_scores is None:
26            grouped = self.lexicon.groupby(level=0, sort=False).mean(numeric_only=True)
27            lookup: dict[str, dict[str, float]] = {}
28            for word, row in grouped.iterrows():
29                lookup[str(word)] = {
30                    "sentiment": float(row["positive"] - row["negative"]),
31                    **{column: float(row[column]) for column in _EMOTION_COLUMNS},
32                }
33            self._word_scores = lookup
34        return self._word_scores
35
36    def get_word_scores(self, word: str) -> dict[str, float] | None:
37        return self._ensure_word_scores().get(word)
38
39    def sentiment(self, word: str) -> float:
40        scores = self.get_word_scores(word)
41        if scores is None:
42            raise KeyError(word)
43        return scores["sentiment"]
44
45    def anger(self, word: str):
46        return self._column_score(word, "anger")
47
48    def anticipation(self, word: str):
49        return self._column_score(word, "anticipation")
50
51    def disgust(self, word: str):
52        return self._column_score(word, "disgust")
53
54    def fear(self, word: str):
55        return self._column_score(word, "fear")
56
57    def joy(self, word: str):
58        return self._column_score(word, "joy")
59
60    def sadness(self, word: str):
61        return self._column_score(word, "sadness")
62
63    def surprise(self, word: str):
64        return self._column_score(word, "surprise")
65
66    def trust(self, word: str):
67        return self._column_score(word, "trust")
68
69    def _column_score(self, word: str, column: str) -> float:
70        scores = self.get_word_scores(word)
71        if scores is None:
72            raise KeyError(word)
73        return scores[column]
74
75    def __contains__(self, item):
76        return item in self._ensure_word_scores()