Revert "Codex/add GitHub workflow for pytest on pr"

.gitignore

@@ -8,4 +8,3 @@ build/
 .env.demo_qa
 _demo_data/*/.runs/*
 .coverage
-.DS_Store

.pre-commit-config.yaml

@@ -5,4 +5,4 @@ repos:
         name: pytest
         language: system
         pass_filenames: false
-        entry: bash -lc 'source .venv/bin/activate PYTHONPATH=".:src:${PYTHONPATH}"; python -m pytest -q -m "not slow and not known_bad"'
+        entry: bash -lc 'source .venv/bin/activate PYTHONPATH=".:src:${PYTHONPATH}"; python -m pytest -q -m "not slow"'

Makefile

@@ -66,11 +66,6 @@ COMPARE_TAG_JUNIT ?= $(DATA)/.runs/diff.tags.junit.xml
 
 MAX_FAILS ?= 5
 
-PURGE_RUNS ?= 0
-PRUNE_HISTORY ?= 0
-PRUNE_CASE_HISTORY ?= 0
-DRY ?= 0
-
 # ==============================================================================
 # 6) Настройки LLM-конфига (редактирование/просмотр)
 # ==============================================================================
@@ -99,7 +94,7 @@ LIMIT_FLAG := $(if $(strip $(LIMIT)),--limit $(LIMIT),)
         batch batch-tag batch-failed batch-failed-from \
         batch-missed batch-missed-from batch-failed-tag batch-missed-tag \
         batch-fail-fast batch-max-fails \
-        stats history-case report-tag report-tag-changes tags tag-rm case-run case-open compare compare-tag
+        stats history-case report-tag report-tag-changes tags case-run case-open compare compare-tag
 
 # ==============================================================================
 # help (на русском)
@@ -147,14 +142,6 @@ help:
 	@echo "  make case-run  CASE=case_42 - прогнать один кейс"
 	@echo "  make case-open CASE=case_42 - открыть артефакты кейса"
 	@echo ""
-	@echo "Уборка:"
-	@echo "  make tag-rm TAG=... [DRY=1] [PURGE_RUNS=1] [PRUNE_HISTORY=1] [PRUNE_CASE_HISTORY=1]"
-	@echo "    - удаляет effective snapshot тега и tag-latest* указатели"
-	@echo "    DRY=1                - dry-run: только показать, что будет удалено"
-	@echo "    PURGE_RUNS=1          - дополнительно удалить все runs, где run_meta.tag == TAG"
-	@echo "    PRUNE_HISTORY=1       - вычистить записи с этим тегом из $${DATA}/.runs/history.jsonl"
-	@echo "    PRUNE_CASE_HISTORY=1  - вычистить записи с этим тегом из $${DATA}/.runs/runs/cases/*.jsonl"
-	@echo ""
 	@echo "Сравнение результатов:"
 	@echo "  make compare BASE=... NEW=... [DIFF_OUT=...] [JUNIT=...]"
 	@echo "  make compare-tag BASE_TAG=baseline NEW_TAG=... [COMPARE_TAG_OUT=...] [COMPARE_TAG_JUNIT=...]"
@@ -353,14 +340,3 @@ compare-tag: check
 	  --new-tag "$(NEW_TAG)" \
 	  --out  "$(OUT)" \
 	  --junit "$(JUNIT)"
-
-# команды очистки
-
-tag-rm:
-	@test -n "$(strip $(TAG))" || (echo "TAG обязателен: make tag-rm TAG=..." && exit 1)
-	@TAG="$(TAG)" DATA="$(DATA)" PURGE_RUNS="$(PURGE_RUNS)" PRUNE_HISTORY="$(PRUNE_HISTORY)" PRUNE_CASE_HISTORY="$(PRUNE_CASE_HISTORY)" DRY="$(DRY)" $(PYTHON) -m scripts.tag_rm
-
-
-
-
-

examples/demo_qa/demo_qa.toml

@@ -1,5 +1,5 @@
 [llm]
-base_url = "http://localhost:8002/v1"
+base_url = "http://localhost:8000/v1"
 plan_model = "default"
 synth_model = "default"
 plan_temperature = 0.0

examples/demo_qa/runner.py

@@ -210,14 +210,6 @@ def _stringify(value: object | None) -> str | None:
     return str(value)
 
 
-def _normalize_text(value: str) -> str:
-    return value.strip().casefold()
-
-
-def _normalize_strings(values: Iterable[object]) -> list[str]:
-    return [_normalize_text(str(value)) for value in values]
-
-
 def _match_expected(case: Case, answer: str | None) -> ExpectedCheck | None:
     if not case.has_asserts:
         return None
@@ -226,15 +218,7 @@ def _match_expected(case: Case, answer: str | None) -> ExpectedCheck | None:
         return ExpectedCheck(mode="none", expected=expected_value, passed=False, detail="no answer")
     if case.expected is not None:
         expected_str = _stringify(case.expected) or ""
-        if isinstance(case.expected, (list, tuple, set)):
-            expected_items = _normalize_strings(case.expected)
-            answer_items = _normalize_strings(answer) if isinstance(answer, (list, tuple, set)) else []
-            if isinstance(case.expected, set) or isinstance(answer, set):
-                passed = set(expected_items) == set(answer_items)
-            else:
-                passed = expected_items == answer_items
-        else:
-            passed = _normalize_text(answer) == _normalize_text(expected_str)
+        passed = answer.strip() == expected_str.strip()
         detail = None if passed else f"expected={expected_str!r}, got={answer!r}"
         return ExpectedCheck(mode="exact", expected=expected_str, passed=passed, detail=detail)
     if case.expected_regex is not None:
@@ -245,7 +229,7 @@ def _match_expected(case: Case, answer: str | None) -> ExpectedCheck | None:
         return ExpectedCheck(mode="regex", expected=expected_regex, passed=passed, detail=detail)
     if case.expected_contains is not None:
         expected_contains = _stringify(case.expected_contains) or ""
-        passed = _normalize_text(expected_contains) in _normalize_text(answer)
+        passed = expected_contains in answer
         detail = None if passed else f"expected to contain {expected_contains!r}"
         return ExpectedCheck(mode="contains", expected=expected_contains, passed=passed, detail=detail)
     return None

examples/demo_qa/tests/test_demo_qa_runner.py

@@ -32,7 +32,7 @@ def test_match_expected_coerces_non_string_expected_values() -> None:
 def test_match_expected_contains_pass_and_fail() -> None:
     case = Case(id="c2", question="Q", expected_contains="bar")
 
-    match = _match_expected(case, "value BAR baz")
+    match = _match_expected(case, "value bar baz")
     assert match is not None
     assert match.passed is True
 
@@ -47,26 +47,6 @@ def test_match_expected_contains_pass_and_fail() -> None:
     assert missing_answer.detail == "no answer"
 
 
-def test_match_expected_equals_is_case_insensitive() -> None:
-    case = Case(id="c3", question="Q", expected="Alpha")
-
-    match = _match_expected(case, "alpha")
-    assert match is not None
-    assert match.passed is True
-
-
-def test_match_expected_list_comparison_normalizes_elements() -> None:
-    case = Case(id="c4", question="Q", expected=["Foo", "Bar"])
-
-    match = _match_expected(case, cast(str, ["foo", "bar"]))
-    assert match is not None
-    assert match.passed is True
-
-    mismatch = _match_expected(case, cast(str, ["foo", "baz"]))
-    assert mismatch is not None
-    assert mismatch.passed is False
-
-
 def test_diff_runs_tracks_regressions_and_improvements() -> None:
     baseline = [
         RunResult(

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "fetchgraph"
-version = "0.2.1"
+version = "0.2.0"
 description = "Graph-like planning → context fetching → synthesis agent (library-style)."
 readme = "README.md"
 requires-python = ">=3.11"

pytest.ini

@@ -45,7 +45,6 @@ markers =
     slow: slow tests (exclude via -m "not slow")
     integration: integration tests (providers / IO / external deps)
     e2e: end-to-end scenarios
-    known_bad: real-world TDD cases that are allowed to fail (excluded from CI by default)
 
 # Удобные дефолты для логов в CI и локально
 log_cli = true