ODS V3 Pre-Release: Public testing and refinement ahead of the official V3 launch. Turn your PC, Mac, or Linux box into a private AI server.
Python ai-agentsamdcomfyuidockerllama-cppllmlocal-ain8nnvidiaopen-webuiragself-hostedspeech-to-textstrix-halotext-to-speechworkflow-automation
import pytest
from env_values import parse_env_value, quote_env_value, strip_matching_quotes
@pytest.mark.parametrize(
("raw", "expected"),
[
("value", "value"),
(" value ", "value"),
('"value"', "value"),
("'value'", "value"),
('""', ""),
("''", ""),
("it's", "it's"),
('"value', '"value'),
('value"', 'value"'),
("'value", "'value"),
("value'", "value'"),
('"value\'', '"value\''),
("''value''", "'value'"),
('""value""', '"value"'),
],
)
def test_strip_matching_quotes_removes_exactly_one_complete_pair(raw, expected):
assert strip_matching_quotes(raw) == expected
@pytest.mark.parametrize(
("value", "expected"),
[
# Bare values every reader agrees on stay bare.
("", ""),
("value", "value"),
("sk-live-secret", "sk-live-secret"),
("http://llama-server:8080/v1", "http://llama-server:8080/v1"),
("it's", "it's"),
("a=b", "a=b"),
('say "hi"', 'say "hi"'),
("C:\\models\\foo.gguf", "C:\\models\\foo.gguf"),
# Compose interpolates $NAME, cuts unquoted values at " #", trims
# surrounding whitespace and treats a leading quote as a delimiter.
("hunter$two", "'hunter$two'"),
("abc${", "'abc${'"),
("token #1", "'token #1'"),
("#literal", "'#literal'"),
(" padded ", "' padded '"),
('"already"', "'\"already\"'"),
("'x", "\"'x\""),
# A single quote or a backslash needs the double-quoted escape set.
("it's $5", '"it\'s \\$5"'),
('"x" it\'s', '"\\"x\\" it\'s"'),
("back\\slash $x", '"back\\\\slash \\$x"'),
],
)
def test_quote_env_value_only_quotes_when_readers_would_disagree(value, expected):
assert quote_env_value(value) == expected
@pytest.mark.parametrize("value", ["a\nb", "a\rb"])
def test_quote_env_value_rejects_newlines(value):
with pytest.raises(ValueError):
quote_env_value(value)
@pytest.mark.parametrize(
("raw", "expected"),
[
# Rows taken from `docker compose config` on the same lines.
("11434 # llama-server API", "11434"),
("a # x", "a"),
("a# b", "a# b"),
(" # only comment", "# only comment"),
(" #x", "#x"),
(" padded value ", "padded value"),
(" ", ""),
('"quoted value" # trailing', "quoted value"),
("'single value' # trailing", "single value"),
('"keep # this"', "keep # this"),
("'keep # this too'", "keep # this too"),
("\"'literal'\"", "'literal'"),
("'\"json\"'", '"json"'),
('say "hi"', 'say "hi"'),
('trailing-quote"', 'trailing-quote"'),
(' "leading space quote"', "leading space quote"),
('"q" ', "q"),
# The writer's escape set decodes back to the original value.
('"it\'s \\$5"', "it's $5"),
('"say \\"hi\\" it\'s"', 'say "hi" it\'s'),
('"back\\\\slash \\$x"', "back\\slash $x"),
# Mismatched quotes stay data (strip_matching_quotes fallback).
("''value''", "'value'"),
('"value', '"value'),
],
)
def test_parse_env_value_matches_compose(raw, expected):
assert parse_env_value(raw) == expected
@pytest.mark.parametrize(
"value",
["plain", "hunter$two", "token #1", "it's $5", 'say "hi" it\'s', "back\\slash $x", " padded ", "#literal", "'x"],
)
def test_quote_then_parse_round_trips(value):
assert parse_env_value(quote_env_value(value)) == value