Getting Started

Installation

pip install --upgrade liwca

Counting words

count takes texts and a dictionary DataFrame, and returns a documents x categories table:

texts = ["The threat of danger loomed over the city", "A calm morning"]
results = liwca.count(texts, dx)

Values are proportions: per-document, the sum of matched contributions divided by total word count. For binary dictionaries this is the fraction of doc tokens in each category (in [0, 1]). For weighted dictionaries it is the per-token mean weight (e.g., mean sentiment per word for a VADER-style lexicon). See count for options including precision rounding and custom tokenizers.

Reading and writing local files

There are six top-level reader/writer functions, each named for the file format and value-type. Pick the function whose name matches your data; calling the wrong one fails loudly:

# Binary dictionaries
dx = liwca.read_dic("my_dictionary.dic")
dx = liwca.read_dicx("my_dictionary.dicx")
liwca.write_dic(dx, "out.dic")
liwca.write_dicx(dx, "out.dicx")

# Weighted (numeric) dictionaries -- e.g. VADER, valence lexicons
dx = liwca.read_dicx_weighted("vader.dicx")
liwca.write_dicx_weighted(dx, "out.dicx")

# Combine and prune
merged = liwca.merge_dx(dx_a, dx_b)        # mixing binary + weighted promotes to float
trimmed = liwca.drop_category(dx, "Football")

The classic .dic format has no weighted variant by spec, so there is no read_dic_weighted / write_dic_weighted.