llmtap-proxy 0.2.0__tar.gz → 0.2.1__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {llmtap_proxy-0.2.0/llmtap_proxy.egg-info → llmtap_proxy-0.2.1}/PKG-INFO +1 -1
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/__init__.py +2 -2
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/cli.py +17 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1/llmtap_proxy.egg-info}/PKG-INFO +1 -1
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/pyproject.toml +1 -1
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_cli_extra.py +14 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_term.py +201 -201
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/LICENSE +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/README.md +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/__main__.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/budget.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/cache.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/cassette.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/config.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/dashboard.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/demo.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/doctor.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/guard.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/mock.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/models.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/pricing.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/providers.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/proxy.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/pytest_plugin.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/report.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/store.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/term.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap/ui/index.html +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap_proxy.egg-info/SOURCES.txt +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap_proxy.egg-info/dependency_links.txt +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap_proxy.egg-info/entry_points.txt +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap_proxy.egg-info/requires.txt +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/llmtap_proxy.egg-info/top_level.txt +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/setup.cfg +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_budget.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_cache.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_cassette.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_cli.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_dashboard.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_e2e.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_gemini.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_guard.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_hardening.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_mock.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_pricing.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_providers.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_proxy.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_pytest_plugin.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_report.py +0 -0
- {llmtap_proxy-0.2.0 → llmtap_proxy-0.2.1}/tests/test_store.py +0 -0
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: llmtap-proxy
|
|
3
|
-
Version: 0.2.
|
|
3
|
+
Version: 0.2.1
|
|
4
4
|
Summary: The flight recorder & firewall for LLM apps. See, cache, replay, budget and secure every LLM call — zero code changes, zero dependencies.
|
|
5
5
|
License: MIT
|
|
6
6
|
Project-URL: Homepage, https://github.com/AshishAgrawal101/llmtap
|
|
@@ -1,2 +1,2 @@
|
|
|
1
|
-
"""llmtap — the flight recorder & firewall for LLM apps."""
|
|
2
|
-
__version__ = "0.2.
|
|
1
|
+
"""llmtap — the flight recorder & firewall for LLM apps."""
|
|
2
|
+
__version__ = "0.2.1"
|
|
@@ -1034,8 +1034,25 @@ def parse_args(argv: list[str] | None = None) -> argparse.Namespace:
|
|
|
1034
1034
|
return args
|
|
1035
1035
|
|
|
1036
1036
|
|
|
1037
|
+
def _ensure_unicode_output() -> None:
|
|
1038
|
+
"""Switch stdout/stderr to UTF-8 when their encoding can't represent llmtap's output.
|
|
1039
|
+
|
|
1040
|
+
On Windows, output redirected to a file or pipe (CI logs, ``> out.txt``) uses the ANSI code page
|
|
1041
|
+
(e.g. cp1252), which has no box-drawing or check-mark characters, so printing a table crashed.
|
|
1042
|
+
"""
|
|
1043
|
+
for stream in (sys.stdout, sys.stderr):
|
|
1044
|
+
try:
|
|
1045
|
+
"┌─✓".encode(getattr(stream, "encoding", None) or "ascii")
|
|
1046
|
+
except (LookupError, UnicodeEncodeError):
|
|
1047
|
+
try:
|
|
1048
|
+
stream.reconfigure(encoding="utf-8", errors="replace") # type: ignore[union-attr]
|
|
1049
|
+
except Exception:
|
|
1050
|
+
pass
|
|
1051
|
+
|
|
1052
|
+
|
|
1037
1053
|
def main(argv: list[str] | None = None) -> int:
|
|
1038
1054
|
"""Entry point for the ``llmtap`` console script."""
|
|
1055
|
+
_ensure_unicode_output()
|
|
1039
1056
|
try:
|
|
1040
1057
|
args = parse_args(argv)
|
|
1041
1058
|
except SystemExit as e: # argparse --help / errors
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: llmtap-proxy
|
|
3
|
-
Version: 0.2.
|
|
3
|
+
Version: 0.2.1
|
|
4
4
|
Summary: The flight recorder & firewall for LLM apps. See, cache, replay, budget and secure every LLM call — zero code changes, zero dependencies.
|
|
5
5
|
License: MIT
|
|
6
6
|
Project-URL: Homepage, https://github.com/AshishAgrawal101/llmtap
|
|
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
|
|
|
4
4
|
|
|
5
5
|
[project]
|
|
6
6
|
name = "llmtap-proxy"
|
|
7
|
-
version = "0.2.
|
|
7
|
+
version = "0.2.1"
|
|
8
8
|
description = "The flight recorder & firewall for LLM apps. See, cache, replay, budget and secure every LLM call — zero code changes, zero dependencies."
|
|
9
9
|
readme = "README.md"
|
|
10
10
|
requires-python = ">=3.11"
|
|
@@ -232,3 +232,17 @@ def test_new_commands_listed_in_help(capsys: pytest.CaptureFixture[str]) -> None
|
|
|
232
232
|
assert cli.main(["--help"]) == 0
|
|
233
233
|
out = capsys.readouterr().out
|
|
234
234
|
assert "report" in out and "doctor" in out
|
|
235
|
+
|
|
236
|
+
|
|
237
|
+
@pytest.mark.parametrize("argv", [["prices", "claude"], ["doctor"], ["scan", "-"]])
|
|
238
|
+
def test_cli_survives_non_unicode_output_encoding(argv: list[str], tmp_path: Path) -> None:
|
|
239
|
+
"""Windows redirects output in the ANSI code page (cp1252); tables and check marks must not crash."""
|
|
240
|
+
import subprocess
|
|
241
|
+
import sys
|
|
242
|
+
|
|
243
|
+
env = dict(os.environ, PYTHONIOENCODING="cp1252", LLMTAP_HOME=str(tmp_path / "home"))
|
|
244
|
+
p = subprocess.run([sys.executable, "-m", "llmtap", *argv], input=b"nothing secret here\n",
|
|
245
|
+
capture_output=True, env=env, timeout=60)
|
|
246
|
+
assert b"codec can't encode" not in p.stderr, p.stderr
|
|
247
|
+
assert p.returncode in (0, 1), p.stderr # doctor may report setup problems (1); 2 is the crash path
|
|
248
|
+
assert p.stdout
|
|
@@ -1,201 +1,201 @@
|
|
|
1
|
-
"""Tests for llmtap.term: ANSI handling, width math, formatters, banner/summary/table, log lines."""
|
|
2
|
-
from __future__ import annotations
|
|
3
|
-
|
|
4
|
-
import io
|
|
5
|
-
|
|
6
|
-
import pytest
|
|
7
|
-
|
|
8
|
-
from llmtap import term
|
|
9
|
-
from llmtap.config import Config
|
|
10
|
-
from llmtap.models import Exchange, Usage
|
|
11
|
-
|
|
12
|
-
|
|
13
|
-
@pytest.fixture(autouse=True)
|
|
14
|
-
def _no_force(monkeypatch: pytest.MonkeyPatch) -> None:
|
|
15
|
-
monkeypatch.delenv("FORCE_COLOR", raising=False)
|
|
16
|
-
monkeypatch.delenv("NO_COLOR", raising=False)
|
|
17
|
-
monkeypatch.setenv("COLUMNS", "120")
|
|
18
|
-
|
|
19
|
-
|
|
20
|
-
class _TTY(io.StringIO):
|
|
21
|
-
def isatty(self) -> bool:
|
|
22
|
-
return True
|
|
23
|
-
|
|
24
|
-
|
|
25
|
-
def test_strip_ansi_and_vlen() -> None:
|
|
26
|
-
s = "\x1b[1;31mhello\x1b[0m world"
|
|
27
|
-
assert term.strip_ansi(s) == "hello world"
|
|
28
|
-
assert term.vlen(s) == 11
|
|
29
|
-
assert term.vlen("⚡ttft") == 6 # wide glyph counts as 2
|
|
30
|
-
assert term.vlen("") == 0
|
|
31
|
-
|
|
32
|
-
|
|
33
|
-
def test_pad_and_truncate() -> None:
|
|
34
|
-
red = term.paint("ab", "red", color=True)
|
|
35
|
-
assert term.vlen(term.pad(red, 5)) == 5
|
|
36
|
-
assert term.pad("ab", 5, ">") == " ab"
|
|
37
|
-
assert term.pad("ab", 6, "^") == " ab "
|
|
38
|
-
assert term.pad("abcdef", 3) == "abcdef"
|
|
39
|
-
assert term.truncate("abcdefghij", 5) == "abcd…"
|
|
40
|
-
assert term.truncate("abc", 5) == "abc"
|
|
41
|
-
assert term.vlen(term.truncate(term.paint("x" * 50, "red", color=True), 10)) == 10
|
|
42
|
-
|
|
43
|
-
|
|
44
|
-
def test_paint_respects_color_flag() -> None:
|
|
45
|
-
assert term.paint("x", "red", color=False) == "x"
|
|
46
|
-
assert term.paint("x", "red", color=True) == "\x1b[31mx\x1b[0m"
|
|
47
|
-
assert term.paint("x", color=True) == "x"
|
|
48
|
-
|
|
49
|
-
|
|
50
|
-
def test_color_enabled_rules(monkeypatch: pytest.MonkeyPatch) -> None:
|
|
51
|
-
assert term.color_enabled(io.StringIO()) is False
|
|
52
|
-
assert term.color_enabled(_TTY()) is True
|
|
53
|
-
monkeypatch.setenv("NO_COLOR", "1")
|
|
54
|
-
assert term.color_enabled(_TTY()) is False
|
|
55
|
-
monkeypatch.delenv("NO_COLOR")
|
|
56
|
-
monkeypatch.setenv("FORCE_COLOR", "1")
|
|
57
|
-
assert term.color_enabled(io.StringIO()) is True
|
|
58
|
-
monkeypatch.setenv("FORCE_COLOR", "0")
|
|
59
|
-
assert term.color_enabled(io.StringIO()) is False
|
|
60
|
-
|
|
61
|
-
|
|
62
|
-
@pytest.mark.parametrize("n,expected", [
|
|
63
|
-
(0, "0"), (999, "999"), (1000, "1k"), (1234, "1.2k"), (12_345, "12.3k"), (340, "340"),
|
|
64
|
-
(2_500_000, "2.5M"), (150_000, "150k"), (None, "0"), (3_000_000_000, "3B"),
|
|
65
|
-
])
|
|
66
|
-
def test_fmt_tokens(n: int | None, expected: str) -> None:
|
|
67
|
-
assert term.fmt_tokens(n) == expected
|
|
68
|
-
|
|
69
|
-
|
|
70
|
-
@pytest.mark.parametrize("ms,expected", [
|
|
71
|
-
(210, "210ms"), (1240, "1.24s"), (0, "0ms"), (None, "-"), (12_500, "12.5s"), (75_000, "1m15s"), (0.5, "0.5ms"),
|
|
72
|
-
])
|
|
73
|
-
def test_fmt_ms(ms: float | None, expected: str) -> None:
|
|
74
|
-
assert term.fmt_ms(ms) == expected
|
|
75
|
-
|
|
76
|
-
|
|
77
|
-
def test_fmt_usd() -> None:
|
|
78
|
-
assert term.fmt_usd(0) == "$0.00"
|
|
79
|
-
assert term.fmt_usd(0.0042) == "$0.0042"
|
|
80
|
-
assert term.fmt_usd(1.234) == "$1.23"
|
|
81
|
-
|
|
82
|
-
|
|
83
|
-
def test_table_alignment_ignores_ansi() -> None:
|
|
84
|
-
out = term.table([[term.paint("red", "red", color=True), "1"], ["longer-cell", "22"]], ["NAME", "N"], color=False)
|
|
85
|
-
lines = out.splitlines()
|
|
86
|
-
widths = {term.vlen(l) for l in lines}
|
|
87
|
-
assert len(widths) == 1, out
|
|
88
|
-
assert lines[0].startswith("┌") and lines[-1].startswith("└")
|
|
89
|
-
assert "│ 1 │" in term.strip_ansi(out) # numeric column right-aligned
|
|
90
|
-
|
|
91
|
-
|
|
92
|
-
def test_table_empty_and_ragged() -> None:
|
|
93
|
-
out = term.table([], ["A", "B"], color=False)
|
|
94
|
-
assert "A" in out and "B" in out
|
|
95
|
-
out = term.table([["x"], ["y", "z", "w"]], ["A"], color=False)
|
|
96
|
-
assert len({term.vlen(l) for l in out.splitlines()}) == 1
|
|
97
|
-
|
|
98
|
-
|
|
99
|
-
def test_table_shrinks_to_width() -> None:
|
|
100
|
-
out = term.table([["x" * 300, "y"]], ["A", "B"], color=False, max_width=60)
|
|
101
|
-
assert all(term.vlen(l) <= 60 for l in out.splitlines())
|
|
102
|
-
|
|
103
|
-
|
|
104
|
-
def test_box_is_rectangular() -> None:
|
|
105
|
-
out = term.box(["hello", term.paint("colored ⚡", "red", color=True), ""], title="T", color=True)
|
|
106
|
-
assert len({term.vlen(l) for l in out.splitlines()}) == 1
|
|
107
|
-
|
|
108
|
-
|
|
109
|
-
def test_banner_contents() -> None:
|
|
110
|
-
cfg = Config(cache=True, cassette="tapes/ci.jsonl", mode="replay", guard="redact",
|
|
111
|
-
budget_daily_usd=5, loop_threshold=4)
|
|
112
|
-
out = term.strip_ansi(term.banner(cfg, "http://127.0.0.1:4141", color=False))
|
|
113
|
-
assert "llmtap
|
|
114
|
-
assert "http://127.0.0.1:4141/_llmtap/" in out
|
|
115
|
-
assert "export ANTHROPIC_BASE_URL=http://127.0.0.1:4141" in out
|
|
116
|
-
assert "export OPENAI_BASE_URL=http://127.0.0.1:4141/v1" in out
|
|
117
|
-
assert "redact" in out and "ci.jsonl" in out and "replay" in out
|
|
118
|
-
assert "$5.00/day" in out and "4×" in out
|
|
119
|
-
lines = out.splitlines()
|
|
120
|
-
assert len({term.vlen(l) for l in lines}) == 1
|
|
121
|
-
|
|
122
|
-
|
|
123
|
-
def test_banner_defaults_off() -> None:
|
|
124
|
-
out = term.banner(Config(), "http://127.0.0.1:4141/", color=False)
|
|
125
|
-
assert "cache off" in out and "unlimited" in out and "loop-breaker off" in out
|
|
126
|
-
|
|
127
|
-
|
|
128
|
-
def test_log_exchange_minimal_does_not_crash() -> None:
|
|
129
|
-
buf = io.StringIO()
|
|
130
|
-
term.log_exchange(Exchange(), stream=buf)
|
|
131
|
-
line = buf.getvalue()
|
|
132
|
-
assert line.endswith("\n") and "\x1b[" not in line
|
|
133
|
-
assert "unknown" in line
|
|
134
|
-
|
|
135
|
-
|
|
136
|
-
def test_log_exchange_full_line() -> None:
|
|
137
|
-
ex = Exchange(provider="anthropic", model="claude-sonnet-4-5", status=200,
|
|
138
|
-
usage=Usage(input_tokens=1234, output_tokens=340), cost_usd=0.0123,
|
|
139
|
-
latency_ms=1240, streamed=True, ttft_ms=210,
|
|
140
|
-
guard_findings=[{"kind": "aws_access_key"}, {"kind": "openai_key"}],
|
|
141
|
-
request_body='{"x": "[REDACTED:aws_access_key]"}')
|
|
142
|
-
line = term.format_exchange(ex, color=False, width=120)
|
|
143
|
-
assert "anthropic/claude-sonnet-4-5" in line
|
|
144
|
-
assert "1.2k→340 tok" in line
|
|
145
|
-
assert "$0.0123" in line and "1.24s" in line
|
|
146
|
-
assert "⚡ttft 210ms" in line
|
|
147
|
-
assert "⚠ 2 secrets redacted" in line
|
|
148
|
-
assert "LIVE" in line and "200" in line
|
|
149
|
-
|
|
150
|
-
|
|
151
|
-
def test_log_exchange_blocked_and_cache() -> None:
|
|
152
|
-
ex = Exchange(source="blocked", status=400, guard_findings=[{"kind": "x"}])
|
|
153
|
-
assert "⛔ blocked" in term.format_exchange(ex, color=False)
|
|
154
|
-
ex = Exchange(source="cache", status=200, saved_usd=0.5, model="gpt-4o")
|
|
155
|
-
line = term.format_exchange(ex, color=False)
|
|
156
|
-
assert "CACHE" in line and "saved $0.5" in line
|
|
157
|
-
ex = Exchange(source="loop", status=429, error="loop detected: identical request sent 4 times in 60s")
|
|
158
|
-
assert "loop detected" in term.format_exchange(ex, color=False)
|
|
159
|
-
|
|
160
|
-
|
|
161
|
-
def test_log_exchange_truncates_long_model() -> None:
|
|
162
|
-
ex = Exchange(provider="openrouter", model="m" * 200, status=200)
|
|
163
|
-
line = term.format_exchange(ex, color=False, width=100)
|
|
164
|
-
assert "…" in line and term.vlen(line) < 160
|
|
165
|
-
|
|
166
|
-
|
|
167
|
-
def test_log_exchange_with_colors_aligned() -> None:
|
|
168
|
-
a = Exchange(provider="openai", model="gpt-4o", status=200, latency_ms=12)
|
|
169
|
-
b = Exchange(provider="anthropic", model="claude-haiku-4-5", status=500, latency_ms=3000)
|
|
170
|
-
la = term.format_exchange(a, color=True, width=120)
|
|
171
|
-
lb = term.format_exchange(b, color=True, width=120)
|
|
172
|
-
assert "\x1b[" in la
|
|
173
|
-
assert term.vlen(la) == term.vlen(lb)
|
|
174
|
-
|
|
175
|
-
|
|
176
|
-
def test_status_color() -> None:
|
|
177
|
-
assert term.status_color(200) == "green"
|
|
178
|
-
assert term.status_color(429) == "yellow"
|
|
179
|
-
assert term.status_color(502) == "red"
|
|
180
|
-
assert term.status_color(0) == "red"
|
|
181
|
-
|
|
182
|
-
|
|
183
|
-
def test_summary_box() -> None:
|
|
184
|
-
stats = {"requests": 12, "input_tokens": 12_000, "output_tokens": 3400, "cost_usd": 0.42, "saved_usd": 0.1,
|
|
185
|
-
"avg_latency_ms": 850.0, "errors": 1, "cache_hits": 3, "blocked": 1, "guard_findings": 2,
|
|
186
|
-
"groups": [{"key": "claude-sonnet-4-5", "requests": 8, "cost_usd": 0.4, "input_tokens": 10_000,
|
|
187
|
-
"output_tokens": 3000},
|
|
188
|
-
{"key": "gpt-4o-mini", "requests": 4, "cost_usd": 0.02, "input_tokens": 2000,
|
|
189
|
-
"output_tokens": 400}]}
|
|
190
|
-
out = term.summary(stats, 0.42, color=False)
|
|
191
|
-
for needle in ("12", "12k in", "3.4k out", "$0.42", "$0.10", "850ms", "Errors", "Blocked",
|
|
192
|
-
"2 secret findings", "claude-sonnet-4-5", "gpt-4o-mini", "Top models"):
|
|
193
|
-
assert needle in out, needle
|
|
194
|
-
assert len({term.vlen(l) for l in out.splitlines()}) == 1
|
|
195
|
-
|
|
196
|
-
|
|
197
|
-
def test_summary_empty() -> None:
|
|
198
|
-
out = term.summary({}, 0.0, color=False)
|
|
199
|
-
assert "No LLM requests" in out
|
|
200
|
-
out = term.summary({"requests": 0}, 0.0, color=False)
|
|
201
|
-
assert "No LLM requests" in out
|
|
1
|
+
"""Tests for llmtap.term: ANSI handling, width math, formatters, banner/summary/table, log lines."""
|
|
2
|
+
from __future__ import annotations
|
|
3
|
+
|
|
4
|
+
import io
|
|
5
|
+
|
|
6
|
+
import pytest
|
|
7
|
+
|
|
8
|
+
from llmtap import __version__, term
|
|
9
|
+
from llmtap.config import Config
|
|
10
|
+
from llmtap.models import Exchange, Usage
|
|
11
|
+
|
|
12
|
+
|
|
13
|
+
@pytest.fixture(autouse=True)
|
|
14
|
+
def _no_force(monkeypatch: pytest.MonkeyPatch) -> None:
|
|
15
|
+
monkeypatch.delenv("FORCE_COLOR", raising=False)
|
|
16
|
+
monkeypatch.delenv("NO_COLOR", raising=False)
|
|
17
|
+
monkeypatch.setenv("COLUMNS", "120")
|
|
18
|
+
|
|
19
|
+
|
|
20
|
+
class _TTY(io.StringIO):
|
|
21
|
+
def isatty(self) -> bool:
|
|
22
|
+
return True
|
|
23
|
+
|
|
24
|
+
|
|
25
|
+
def test_strip_ansi_and_vlen() -> None:
|
|
26
|
+
s = "\x1b[1;31mhello\x1b[0m world"
|
|
27
|
+
assert term.strip_ansi(s) == "hello world"
|
|
28
|
+
assert term.vlen(s) == 11
|
|
29
|
+
assert term.vlen("⚡ttft") == 6 # wide glyph counts as 2
|
|
30
|
+
assert term.vlen("") == 0
|
|
31
|
+
|
|
32
|
+
|
|
33
|
+
def test_pad_and_truncate() -> None:
|
|
34
|
+
red = term.paint("ab", "red", color=True)
|
|
35
|
+
assert term.vlen(term.pad(red, 5)) == 5
|
|
36
|
+
assert term.pad("ab", 5, ">") == " ab"
|
|
37
|
+
assert term.pad("ab", 6, "^") == " ab "
|
|
38
|
+
assert term.pad("abcdef", 3) == "abcdef"
|
|
39
|
+
assert term.truncate("abcdefghij", 5) == "abcd…"
|
|
40
|
+
assert term.truncate("abc", 5) == "abc"
|
|
41
|
+
assert term.vlen(term.truncate(term.paint("x" * 50, "red", color=True), 10)) == 10
|
|
42
|
+
|
|
43
|
+
|
|
44
|
+
def test_paint_respects_color_flag() -> None:
|
|
45
|
+
assert term.paint("x", "red", color=False) == "x"
|
|
46
|
+
assert term.paint("x", "red", color=True) == "\x1b[31mx\x1b[0m"
|
|
47
|
+
assert term.paint("x", color=True) == "x"
|
|
48
|
+
|
|
49
|
+
|
|
50
|
+
def test_color_enabled_rules(monkeypatch: pytest.MonkeyPatch) -> None:
|
|
51
|
+
assert term.color_enabled(io.StringIO()) is False
|
|
52
|
+
assert term.color_enabled(_TTY()) is True
|
|
53
|
+
monkeypatch.setenv("NO_COLOR", "1")
|
|
54
|
+
assert term.color_enabled(_TTY()) is False
|
|
55
|
+
monkeypatch.delenv("NO_COLOR")
|
|
56
|
+
monkeypatch.setenv("FORCE_COLOR", "1")
|
|
57
|
+
assert term.color_enabled(io.StringIO()) is True
|
|
58
|
+
monkeypatch.setenv("FORCE_COLOR", "0")
|
|
59
|
+
assert term.color_enabled(io.StringIO()) is False
|
|
60
|
+
|
|
61
|
+
|
|
62
|
+
@pytest.mark.parametrize("n,expected", [
|
|
63
|
+
(0, "0"), (999, "999"), (1000, "1k"), (1234, "1.2k"), (12_345, "12.3k"), (340, "340"),
|
|
64
|
+
(2_500_000, "2.5M"), (150_000, "150k"), (None, "0"), (3_000_000_000, "3B"),
|
|
65
|
+
])
|
|
66
|
+
def test_fmt_tokens(n: int | None, expected: str) -> None:
|
|
67
|
+
assert term.fmt_tokens(n) == expected
|
|
68
|
+
|
|
69
|
+
|
|
70
|
+
@pytest.mark.parametrize("ms,expected", [
|
|
71
|
+
(210, "210ms"), (1240, "1.24s"), (0, "0ms"), (None, "-"), (12_500, "12.5s"), (75_000, "1m15s"), (0.5, "0.5ms"),
|
|
72
|
+
])
|
|
73
|
+
def test_fmt_ms(ms: float | None, expected: str) -> None:
|
|
74
|
+
assert term.fmt_ms(ms) == expected
|
|
75
|
+
|
|
76
|
+
|
|
77
|
+
def test_fmt_usd() -> None:
|
|
78
|
+
assert term.fmt_usd(0) == "$0.00"
|
|
79
|
+
assert term.fmt_usd(0.0042) == "$0.0042"
|
|
80
|
+
assert term.fmt_usd(1.234) == "$1.23"
|
|
81
|
+
|
|
82
|
+
|
|
83
|
+
def test_table_alignment_ignores_ansi() -> None:
|
|
84
|
+
out = term.table([[term.paint("red", "red", color=True), "1"], ["longer-cell", "22"]], ["NAME", "N"], color=False)
|
|
85
|
+
lines = out.splitlines()
|
|
86
|
+
widths = {term.vlen(l) for l in lines}
|
|
87
|
+
assert len(widths) == 1, out
|
|
88
|
+
assert lines[0].startswith("┌") and lines[-1].startswith("└")
|
|
89
|
+
assert "│ 1 │" in term.strip_ansi(out) # numeric column right-aligned
|
|
90
|
+
|
|
91
|
+
|
|
92
|
+
def test_table_empty_and_ragged() -> None:
|
|
93
|
+
out = term.table([], ["A", "B"], color=False)
|
|
94
|
+
assert "A" in out and "B" in out
|
|
95
|
+
out = term.table([["x"], ["y", "z", "w"]], ["A"], color=False)
|
|
96
|
+
assert len({term.vlen(l) for l in out.splitlines()}) == 1
|
|
97
|
+
|
|
98
|
+
|
|
99
|
+
def test_table_shrinks_to_width() -> None:
|
|
100
|
+
out = term.table([["x" * 300, "y"]], ["A", "B"], color=False, max_width=60)
|
|
101
|
+
assert all(term.vlen(l) <= 60 for l in out.splitlines())
|
|
102
|
+
|
|
103
|
+
|
|
104
|
+
def test_box_is_rectangular() -> None:
|
|
105
|
+
out = term.box(["hello", term.paint("colored ⚡", "red", color=True), ""], title="T", color=True)
|
|
106
|
+
assert len({term.vlen(l) for l in out.splitlines()}) == 1
|
|
107
|
+
|
|
108
|
+
|
|
109
|
+
def test_banner_contents() -> None:
|
|
110
|
+
cfg = Config(cache=True, cassette="tapes/ci.jsonl", mode="replay", guard="redact",
|
|
111
|
+
budget_daily_usd=5, loop_threshold=4)
|
|
112
|
+
out = term.strip_ansi(term.banner(cfg, "http://127.0.0.1:4141", color=False))
|
|
113
|
+
assert f"llmtap v{__version__}" in out and "recording LLM traffic" in out
|
|
114
|
+
assert "http://127.0.0.1:4141/_llmtap/" in out
|
|
115
|
+
assert "export ANTHROPIC_BASE_URL=http://127.0.0.1:4141" in out
|
|
116
|
+
assert "export OPENAI_BASE_URL=http://127.0.0.1:4141/v1" in out
|
|
117
|
+
assert "redact" in out and "ci.jsonl" in out and "replay" in out
|
|
118
|
+
assert "$5.00/day" in out and "4×" in out
|
|
119
|
+
lines = out.splitlines()
|
|
120
|
+
assert len({term.vlen(l) for l in lines}) == 1
|
|
121
|
+
|
|
122
|
+
|
|
123
|
+
def test_banner_defaults_off() -> None:
|
|
124
|
+
out = term.banner(Config(), "http://127.0.0.1:4141/", color=False)
|
|
125
|
+
assert "cache off" in out and "unlimited" in out and "loop-breaker off" in out
|
|
126
|
+
|
|
127
|
+
|
|
128
|
+
def test_log_exchange_minimal_does_not_crash() -> None:
|
|
129
|
+
buf = io.StringIO()
|
|
130
|
+
term.log_exchange(Exchange(), stream=buf)
|
|
131
|
+
line = buf.getvalue()
|
|
132
|
+
assert line.endswith("\n") and "\x1b[" not in line
|
|
133
|
+
assert "unknown" in line
|
|
134
|
+
|
|
135
|
+
|
|
136
|
+
def test_log_exchange_full_line() -> None:
|
|
137
|
+
ex = Exchange(provider="anthropic", model="claude-sonnet-4-5", status=200,
|
|
138
|
+
usage=Usage(input_tokens=1234, output_tokens=340), cost_usd=0.0123,
|
|
139
|
+
latency_ms=1240, streamed=True, ttft_ms=210,
|
|
140
|
+
guard_findings=[{"kind": "aws_access_key"}, {"kind": "openai_key"}],
|
|
141
|
+
request_body='{"x": "[REDACTED:aws_access_key]"}')
|
|
142
|
+
line = term.format_exchange(ex, color=False, width=120)
|
|
143
|
+
assert "anthropic/claude-sonnet-4-5" in line
|
|
144
|
+
assert "1.2k→340 tok" in line
|
|
145
|
+
assert "$0.0123" in line and "1.24s" in line
|
|
146
|
+
assert "⚡ttft 210ms" in line
|
|
147
|
+
assert "⚠ 2 secrets redacted" in line
|
|
148
|
+
assert "LIVE" in line and "200" in line
|
|
149
|
+
|
|
150
|
+
|
|
151
|
+
def test_log_exchange_blocked_and_cache() -> None:
|
|
152
|
+
ex = Exchange(source="blocked", status=400, guard_findings=[{"kind": "x"}])
|
|
153
|
+
assert "⛔ blocked" in term.format_exchange(ex, color=False)
|
|
154
|
+
ex = Exchange(source="cache", status=200, saved_usd=0.5, model="gpt-4o")
|
|
155
|
+
line = term.format_exchange(ex, color=False)
|
|
156
|
+
assert "CACHE" in line and "saved $0.5" in line
|
|
157
|
+
ex = Exchange(source="loop", status=429, error="loop detected: identical request sent 4 times in 60s")
|
|
158
|
+
assert "loop detected" in term.format_exchange(ex, color=False)
|
|
159
|
+
|
|
160
|
+
|
|
161
|
+
def test_log_exchange_truncates_long_model() -> None:
|
|
162
|
+
ex = Exchange(provider="openrouter", model="m" * 200, status=200)
|
|
163
|
+
line = term.format_exchange(ex, color=False, width=100)
|
|
164
|
+
assert "…" in line and term.vlen(line) < 160
|
|
165
|
+
|
|
166
|
+
|
|
167
|
+
def test_log_exchange_with_colors_aligned() -> None:
|
|
168
|
+
a = Exchange(provider="openai", model="gpt-4o", status=200, latency_ms=12)
|
|
169
|
+
b = Exchange(provider="anthropic", model="claude-haiku-4-5", status=500, latency_ms=3000)
|
|
170
|
+
la = term.format_exchange(a, color=True, width=120)
|
|
171
|
+
lb = term.format_exchange(b, color=True, width=120)
|
|
172
|
+
assert "\x1b[" in la
|
|
173
|
+
assert term.vlen(la) == term.vlen(lb)
|
|
174
|
+
|
|
175
|
+
|
|
176
|
+
def test_status_color() -> None:
|
|
177
|
+
assert term.status_color(200) == "green"
|
|
178
|
+
assert term.status_color(429) == "yellow"
|
|
179
|
+
assert term.status_color(502) == "red"
|
|
180
|
+
assert term.status_color(0) == "red"
|
|
181
|
+
|
|
182
|
+
|
|
183
|
+
def test_summary_box() -> None:
|
|
184
|
+
stats = {"requests": 12, "input_tokens": 12_000, "output_tokens": 3400, "cost_usd": 0.42, "saved_usd": 0.1,
|
|
185
|
+
"avg_latency_ms": 850.0, "errors": 1, "cache_hits": 3, "blocked": 1, "guard_findings": 2,
|
|
186
|
+
"groups": [{"key": "claude-sonnet-4-5", "requests": 8, "cost_usd": 0.4, "input_tokens": 10_000,
|
|
187
|
+
"output_tokens": 3000},
|
|
188
|
+
{"key": "gpt-4o-mini", "requests": 4, "cost_usd": 0.02, "input_tokens": 2000,
|
|
189
|
+
"output_tokens": 400}]}
|
|
190
|
+
out = term.summary(stats, 0.42, color=False)
|
|
191
|
+
for needle in ("12", "12k in", "3.4k out", "$0.42", "$0.10", "850ms", "Errors", "Blocked",
|
|
192
|
+
"2 secret findings", "claude-sonnet-4-5", "gpt-4o-mini", "Top models"):
|
|
193
|
+
assert needle in out, needle
|
|
194
|
+
assert len({term.vlen(l) for l in out.splitlines()}) == 1
|
|
195
|
+
|
|
196
|
+
|
|
197
|
+
def test_summary_empty() -> None:
|
|
198
|
+
out = term.summary({}, 0.0, color=False)
|
|
199
|
+
assert "No LLM requests" in out
|
|
200
|
+
out = term.summary({"requests": 0}, 0.0, color=False)
|
|
201
|
+
assert "No LLM requests" in out
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|