Files
hermes-agent/tests/run_agent/test_vision_aware_preprocessing.py
Teknium ec671c4154 feat(image-input): native multimodal routing based on model vision capability (#16506)
* feat(image-input): native multimodal routing based on model vision capability

Attach user-sent images as OpenAI-style content parts on the user turn when
the active model supports native vision, so vision-capable models see real
pixels instead of a lossy text description from vision_analyze.

Routing decision (agent/image_routing.py::decide_image_input_mode):

  agent.image_input_mode = auto | native | text  (default: auto)

In auto mode:
  - If auxiliary.vision.provider/model is explicitly configured, keep the
    text pipeline (user paid for a dedicated vision backend).
  - Else if models.dev reports supports_vision=True for the active
    provider/model, attach natively.
  - Else fall back to text (current behaviour).

Call sites updated: gateway/run.py (all messaging platforms), tui_gateway
(dashboard/Ink), cli.py (interactive /attach + drag-drop).

run_agent.py changes:
  - _prepare_anthropic_messages_for_api now passes image parts through
    unchanged when the model supports vision — the Anthropic adapter
    translates them to native image blocks. Previous behaviour
    (vision_analyze → text) only runs for non-vision Anthropic models.
  - New _prepare_messages_for_non_vision_model mirrors the same contract
    for chat.completions and codex_responses paths, so non-vision models
    on any provider get text-fallback instead of failing at the provider.
  - New _model_supports_vision() helper reads models.dev caps.

vision_analyze description rewritten: positions it as a tool for images
NOT already visible in the conversation (URLs, tool output, deeper
inspection). Prevents the model from redundantly calling it on images
already attached natively.

Config default: agent.image_input_mode = auto.

Tests: 35 new (test_image_routing.py + test_vision_aware_preprocessing.py),
all existing tests that reference _prepare_anthropic_messages_for_api
still pass (198 targeted + new tests green).

* feat(image-input): size-cap + resize oversized images, charge image tokens in compressor

Two follow-ups that make the native image routing safer for long / heavy
sessions:

1) Oversize handling in build_native_content_parts:
   - 20 MB ceiling per image (matches vision_tools._MAX_BASE64_BYTES,
     the most restrictive provider — Gemini inline data).
   - Delegates to vision_tools._resize_image_for_vision (Pillow-based,
     already battle-tested) to downscale to 5 MB first-try.
   - If Pillow is missing or resize still overshoots, the image is
     dropped and reported back in skipped[]; caller falls back to text
     enrichment for that image.

2) Image-token accounting in context_compressor:
   - New _IMAGE_TOKEN_ESTIMATE = 1600 (matches Claude Code's constant;
     within the realistic range for Anthropic/GPT-4o/Gemini billing).
   - _content_length_for_budget() helper: sums text-part lengths and
     charges _IMAGE_CHAR_EQUIVALENT (1600 * 4 chars) per image/image_url/
     input_image part.  Base64 payload inside image_url is NOT counted
     as chars — dimensions don't matter, only image-presence.
   - Both tail-cut sites (_prune_old_tool_results L527 and
     _find_tail_cut_by_tokens L1126) now call the helper so multi-image
     conversations don't slip past compression budget.

Tests: 9 new in test_image_routing.py (oversize triggers resize,
resize-fails-returns-None, oversize-skipped-reported), 11 new in
test_compressor_image_tokens.py (flat charge per image, multiple images,
Responses-API / Anthropic-native / OpenAI-chat shapes, no-inflation on
raw base64, bounds-check on the constant, integration test that an
image-heavy tail actually gets trimmed).

* fix(image-input): replace blanket 20MB ceiling with empirically-verified per-provider limits

The previous commit imposed a hardcoded 20 MB base64 ceiling on all
providers, triggering auto-resize on anything larger. This was wrong in
both directions:

  * Too loose for Anthropic — actual limit is 5 MB (returns HTTP 400
    'image exceeds 5 MB maximum' above that).
  * Too strict for OpenAI / Codex / OpenRouter — accept 49 MB+ without
    complaint (empirically verified April 2026 with progressive PNG
    sizes).

New behaviour:

  * _PROVIDER_BASE64_CEILING table: only anthropic and bedrock have a
    ceiling (5 MB, since bedrock-on-Claude shares Anthropic's decoder).
  * Providers NOT in the table get no ceiling — images attach at native
    size and we trust the provider to return its own error if it
    disagrees. A provider-specific 400 message is clearer than us
    guessing wrong and silently degrading image quality.
  * build_native_content_parts() gains a keyword-only provider arg;
    gateway/CLI/TUI pass the active provider so Anthropic users get
    auto-resize protection while OpenAI users don't pay it.
  * Resize target dropped from 5 MB to 4 MB to slide safely under
    Anthropic's boundary with header overhead.

Empirical measurements (direct API, no Hermes in the loop):

    image b64     anthropic   openrouter/gpt5.5   codex-oauth/gpt5.5
    0.19 MB       ✓           ✓                   ✓
    12.37 MB      ✗ 400 5MB   ✓                   ✓
    23.85 MB      ✗ 400 5MB   ✓                   ✓
    49.46 MB      ✗ 413       ✓                   ✓

Tests: rewrote TestOversizeHandling (5 tests): no-ceiling pass-through,
Anthropic resize fires, Anthropic skip on resize-fail, build_native_parts
routes ceiling by provider, unknown provider gets no ceiling. All 52
targeted tests pass.

* refactor(image-input): attempt native, shrink-and-retry on provider reject

Replace proactive per-provider size ceilings with a reactive shrink path
on the provider's actual rejection. All providers now attempt native
full-size attachment first; if the provider returns an image-too-large
error, the agent silently shrinks and retries once.

Why the previous design was wrong: hardcoding provider ceilings
(anthropic=5MB, others=unlimited) meant OpenAI users on a 10MB image
paid no tax, but Anthropic users lost quality on anything >5MB even
though the empirical behaviour at provider-reject time is the same
(shrink + retry). Baking the table into the routing layer also
requires updating Hermes every time a provider's limit changes.

Reactive design:
  - image_routing.py: _file_to_data_url encodes native size, no ceiling.
    build_native_content_parts drops its provider kwarg.
  - error_classifier.py: new FailoverReason.image_too_large + pattern
    match ("image exceeds", "image too large", etc.) checked BEFORE
    context_overflow so Anthropic's 5MB rejection lands in the right
    bucket.
  - run_agent.py: new _try_shrink_image_parts_in_messages walks api
    messages in-place, re-encodes oversized data: URL image parts
    through vision_tools._resize_image_for_vision to fit under 4MB,
    handles both chat.completions (dict image_url) and Responses
    (string image_url) shapes, ignores http URLs (provider-fetched).
    New image_shrink_retry_attempted flag in the retry loop fires the
    shrink exactly once per turn after credential-pool recovery but
    before auth retries.

E2E verified live against Anthropic claude-sonnet-4-6:
  - 17.9MB PNG (23.9MB b64) attached at native size
  - Anthropic returns 400 "image exceeds 5 MB maximum"
  - Agent logs '📐 Image(s) exceeded provider size limit — shrank and
    retrying...'
  - Retry succeeds, correct response delivered in 6.8s total.

Tests: 12 new (8 shrink-helper shapes + 4 classifier signals),
replaces 5 proactive-ceiling tests with 3 simpler 'native attach works'
tests. 181 targeted tests pass. test_enum_members_exist in
test_error_classifier.py updated for the new enum value.
2026-04-27 06:27:59 -07:00

171 lines
6.7 KiB
Python

"""Tests for the vision-aware image preprocessing in run_agent.py.
Covers:
* ``_prepare_anthropic_messages_for_api`` — passes image parts through
unchanged when the active model reports ``supports_vision=True`` (the
adapter handles them natively), and falls back to text-description
replacement when the model lacks vision.
* ``_prepare_messages_for_non_vision_model`` — the mirror method for the
chat.completions / codex_responses paths. Same contract.
"""
from __future__ import annotations
from unittest.mock import MagicMock, patch
import pytest
from run_agent import AIAgent
def _make_agent() -> AIAgent:
"""Build a bare-bones AIAgent instance without running __init__.
Avoids the heavy provider/credential setup for these pure-method tests.
"""
agent = object.__new__(AIAgent)
agent.provider = "anthropic"
agent.model = "claude-sonnet-4"
agent._anthropic_image_fallback_cache = {}
return agent
IMG_PARTS_USER_MSG = {
"role": "user",
"content": [
{"type": "text", "text": "What's in this image?"},
{"type": "image_url", "image_url": {"url": "data:image/png;base64,AAAA"}},
],
}
PLAIN_USER_MSG = {"role": "user", "content": "hello, no images here"}
# ─── _prepare_anthropic_messages_for_api ─────────────────────────────────────
class TestPrepareAnthropicMessages:
def test_no_images_passes_through(self):
agent = _make_agent()
msgs = [PLAIN_USER_MSG]
out = agent._prepare_anthropic_messages_for_api(msgs)
assert out is msgs # unchanged reference
def test_vision_capable_passes_images_through(self):
"""The Anthropic adapter handles image_url/input_image natively."""
agent = _make_agent()
with patch.object(agent, "_model_supports_vision", return_value=True):
out = agent._prepare_anthropic_messages_for_api([IMG_PARTS_USER_MSG])
# Passes through unchanged — image_url parts still present.
assert out[0]["content"][1]["type"] == "image_url"
def test_non_vision_replaces_images_with_text(self):
agent = _make_agent()
with patch.object(agent, "_model_supports_vision", return_value=False), \
patch.object(
agent,
"_describe_image_for_anthropic_fallback",
return_value="[Image description: a cat]",
):
out = agent._prepare_anthropic_messages_for_api([IMG_PARTS_USER_MSG])
# Content collapsed to a string containing the description + user text.
content = out[0]["content"]
assert isinstance(content, str)
assert "[Image description: a cat]" in content
assert "What's in this image?" in content
# No more image parts.
assert "image_url" not in content
# ─── _prepare_messages_for_non_vision_model ──────────────────────────────────
class TestPrepareMessagesForNonVision:
def test_no_images_passes_through(self):
agent = _make_agent()
msgs = [PLAIN_USER_MSG]
out = agent._prepare_messages_for_non_vision_model(msgs)
assert out is msgs
def test_vision_capable_passes_through(self):
"""For vision-capable models on chat.completions path, provider handles pixels."""
agent = _make_agent()
agent.provider = "openrouter"
agent.model = "anthropic/claude-sonnet-4"
with patch.object(agent, "_model_supports_vision", return_value=True):
out = agent._prepare_messages_for_non_vision_model([IMG_PARTS_USER_MSG])
assert out[0]["content"][1]["type"] == "image_url"
def test_non_vision_strips_images(self):
agent = _make_agent()
agent.provider = "openrouter"
agent.model = "qwen/qwen3-235b-a22b"
with patch.object(agent, "_model_supports_vision", return_value=False), \
patch.object(
agent,
"_describe_image_for_anthropic_fallback",
return_value="[Image description: a dog]",
):
out = agent._prepare_messages_for_non_vision_model([IMG_PARTS_USER_MSG])
content = out[0]["content"]
assert isinstance(content, str)
assert "[Image description: a dog]" in content
assert "image_url" not in content
def test_multiple_messages_with_mixed_content(self):
agent = _make_agent()
agent.model = "qwen/qwen3-235b"
msgs = [
{"role": "user", "content": "first turn"},
{"role": "assistant", "content": "ack"},
IMG_PARTS_USER_MSG,
]
with patch.object(agent, "_model_supports_vision", return_value=False), \
patch.object(
agent,
"_describe_image_for_anthropic_fallback",
return_value="[Image: thing]",
):
out = agent._prepare_messages_for_non_vision_model(msgs)
# First two messages unchanged (no images), third stripped.
assert out[0]["content"] == "first turn"
assert out[1]["content"] == "ack"
assert isinstance(out[2]["content"], str)
assert "[Image: thing]" in out[2]["content"]
# ─── _model_supports_vision ──────────────────────────────────────────────────
class TestModelSupportsVision:
def test_missing_provider_or_model_returns_false(self):
agent = _make_agent()
agent.provider = ""
agent.model = "claude-sonnet-4"
assert agent._model_supports_vision() is False
agent.provider = "anthropic"
agent.model = ""
assert agent._model_supports_vision() is False
def test_uses_get_model_capabilities(self):
agent = _make_agent()
fake_caps = MagicMock()
fake_caps.supports_vision = True
with patch("agent.models_dev.get_model_capabilities", return_value=fake_caps):
assert agent._model_supports_vision() is True
fake_caps.supports_vision = False
with patch("agent.models_dev.get_model_capabilities", return_value=fake_caps):
assert agent._model_supports_vision() is False
def test_none_caps_returns_false(self):
agent = _make_agent()
with patch("agent.models_dev.get_model_capabilities", return_value=None):
assert agent._model_supports_vision() is False
def test_exception_returns_false(self):
agent = _make_agent()
with patch("agent.models_dev.get_model_capabilities", side_effect=RuntimeError("boom")):
assert agent._model_supports_vision() is False