Dymas/abogen
1
0
Fork 0
mirror of https://github.com/denizsafak/abogen.git synced 2026-07-18 05:40:26 +02:00
Code Issues Packages Projects Releases Wiki Activity

Reformat using black

Browse Source
This commit is contained in:
Deniz Şafak
2026-01-09 01:36:14 +03:00
parent 3c800df450
commit 5ae153f841
48 changed files with 1530 additions and 894 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/test_date_normalization_comprehensive.py
+24 -17
View File
@@ -1,15 +1,21 @@
import pytest
from abogen.kokoro_text_normalization import _normalize_grouped_numbers, ApostropheConfig
from abogen.kokoro_text_normalization import (
_normalize_grouped_numbers,
ApostropheConfig,
)
@pytest.fixture
def cfg():
return ApostropheConfig(convert_numbers=True, year_pronunciation_mode="american")
def normalize(text, config):
return _normalize_grouped_numbers(text, config)
class TestDateNormalization:
def test_standard_years(self, cfg):
# 1990 -> nineteen hundred ninety
assert "nineteen hundred ninety" in normalize("In 1990, the web was born.", cfg)
@@ -18,8 +24,10 @@ class TestDateNormalization:
# 2023 -> twenty twenty-three
assert "twenty twenty-three" in normalize("It is currently 2023.", cfg)
# 1905 -> nineteen hundred oh five
assert "nineteen hundred oh five" in normalize("In 1905, Einstein published.", cfg)
assert "nineteen hundred oh five" in normalize(
"In 1905, Einstein published.", cfg
)
def test_future_years(self, cfg):
# 3400 -> thirty-four hundred
assert "thirty-four hundred" in normalize("In the year 3400, we fly.", cfg)
@@ -29,13 +37,13 @@ class TestDateNormalization:
def test_years_with_markers(self, cfg):
# 1021 BC -> ten twenty-one
assert "ten twenty-one" in normalize("It happened in 1021 BC.", cfg)
# 4000 BCE -> forty hundred (or four thousand?)
# _format_year_like logic:
# 4000 BCE -> forty hundred (or four thousand?)
# _format_year_like logic:
# if value % 1000 == 0: return "X thousand"
# 4000 -> four thousand.
# 4000 -> four thousand.
# Let's check 4001 -> forty oh one
assert "forty oh one" in normalize("Ancient times 4001 BCE.", cfg)
def test_addresses_explicit(self, cfg):
# "address" keyword present -> should NOT be year
# 1925 -> one thousand nine hundred twenty-five (default num2words)
@@ -46,11 +54,11 @@ class TestDateNormalization:
assert "one thousand" in res or "nineteen hundred" in res
res = normalize("Please send it to the address: 3400 North Blvd.", cfg)
assert "thirty-four hundred" not in res # Should not be year style
assert "three thousand" in res or "thirty-four hundred" in res
# Wait, "thirty-four hundred" IS how you say 3400 in num2words sometimes?
assert "thirty-four hundred" not in res # Should not be year style
assert "three thousand" in res or "thirty-four hundred" in res
# Wait, "thirty-four hundred" IS how you say 3400 in num2words sometimes?
# num2words(3400) -> "three thousand, four hundred" usually.
# Let's verify what "thirty-four hundred" implies.
# Let's verify what "thirty-four hundred" implies.
# If it's a year: "thirty-four hundred".
# If it's a number: "three thousand four hundred".
assert "three thousand" in res
@@ -62,9 +70,9 @@ class TestDateNormalization:
def test_ambiguous_numbers(self, cfg):
# Just a number, no "address", no markers. Should default to year if 4 digits 1000-9999
assert "nineteen hundred fifty" in normalize("I have 1950 apples.", cfg)
assert "nineteen hundred fifty" in normalize("I have 1950 apples.", cfg)
# This is a known limitation/feature: it aggressively identifies years.
def test_specific_user_examples(self, cfg):
# 1021
assert "ten twenty-one" in normalize("1021", cfg)
@@ -77,10 +85,10 @@ class TestDateNormalization:
# Simulating a title or sentence from the book
# "The Rise of the Robots: Technology and the Threat of a Jobless Future"
# Maybe it mentions a year like 2015 (pub date) or a future date.
# "In 2015, Martin Ford wrote..."
assert "twenty fifteen" in normalize("In 2015, Martin Ford wrote...", cfg)
# "By 2100, robots will..."
assert "twenty-one hundred" in normalize("By 2100, robots will...", cfg)
@@ -113,4 +121,3 @@ class TestDateNormalization:
res = normalize("The addresses are 1925 and 1926.", cfg)
# Expectation: should probably be numbers, not years.
assert "nineteen twenty-five" not in res