Support integration test repair and coverage

README.md

@@ -69,7 +69,7 @@ To measure existing tests before preprocessing, run `utgen coverage -p <original
 
 On a fresh working copy of a standalone crate, run `utgen pre-process -p <target-crate-path>` before `utgen analyze -p <target-crate-path>`. Analysis clears Cargo's check cache, runs brinfo and focxt, and verifies their outputs. It rejects existing `brinfo/` or `focxt/` directories so previous analysis results cannot be mixed into the run.
 
-Then use `utgen gen` and `utgen fix` for generation and compilation repair. Pass `--requirement --context` to `utgen gen` to include path constraints and focal context. Preprocessing and test checks modify the target tree; use a working copy of the target project.
+Then use `utgen gen` and `utgen fix` for generation and compilation repair. Pass `--requirement --context` to `utgen gen` to include path constraints and focal context. For integration tests, pass `--integration` to both `gen` and `fix`. Preprocessing and test checks modify the target tree; use a working copy of the target project.
 
 Start with the [minimal pipeline check](docs/minimal-pipeline.md), which uses fixed local model responses and verifies analysis, generation, repair, source restoration, and coverage. The [bytes analysis check](docs/bytes-analysis.md) validates analysis on the larger bundled target without a model service. For model-based generation and repair, follow the [small model trial](examples/README.md#prepare-a-small-model-trial) using an explicit function list and separate request limits, then inspect the request reports and test statistics. See [utgen usage](utgen/README.md) for command details and supported scope.
 

docs/minimal-pipeline.md

@@ -41,6 +41,7 @@ The default binary directory is `target/debug`. The coverage tool must be availa
 | Coverage | Excludes test bodies and test-only helpers, preserves production coverage, and executes tests once when exporting XML and JSON. |
 | Cleanup | Restores sources after the covered success, assertion-failure, compilation-failure, and report-export-failure cases. |
 | Candidate timeout | Ordinary and `should_panic` infinite loops count as failures; an expected panic and a later ordinary test pass in unit and integration statistics. |
+| Integration repair | Repairs alias/trait imports, helpers, and test bodies in integration targets; keeps sibling candidates independent; checks cached repair, mode mismatch, unselected targets, and cleanup after request failure. |
 | Integration directories | Original and generated tests coexist after normal runs and cached reruns; original-test compilation failure restores both directories; an existing staging directory is preserved and reported. |
 
 The integration-directory cases retain the original `tests.bak/` while generating `tests/`. They use one cached candidate and make no additional model requests. Separate timeout cases also check integration execution without an original-test backup.
@@ -63,7 +64,7 @@ Explicit `coverage(on)` is respected. Expression fragments are preserved; the ch
 
 `check_tasks.py` exercises function limits N=1, N=2, and the default N=4, with more than 2N functions in each scheduling case. It checks overlapping model requests, serial compiler activity, bounded-queue backpressure, and completion after queued work resumes. Cargo's target directory is reused between scenarios.
 
-Additional cases cover direct and input/prefix/oracle generation, integration imports, function selection, request budgets, and worker/consumer failures. Checks require source restoration, preservation of unrelated backups, retention of repair backups on failure, and skipped statistics after a candidate-stage error.
+Additional cases cover direct and input/prefix/oracle generation, integration imports and repair concurrency, function selection, request budgets, and worker/consumer failures. Checks require source restoration, preservation of unrelated backups, retention of repair backups on failure, and skipped statistics after a candidate-stage error.
 
 ## Model Request Regression
 

docs/palm-rust-unit-test-generation.md

@@ -262,7 +262,7 @@ Rust 的一个核心挑战是 MIR 的去糖。比如高级语法 `match color` 
 utgen pre-process --project-dir <PROJECT_DIR> [--work-dir <WORK_DIR>...]
 utgen analyze --project-dir <STANDALONE_CRATE_DIR>
 utgen gen --project-dir <PROJECT_DIR> [--work-dir <WORK_DIR>...] [--tasks <N>] [--integration] [--requirement] [--context] [--oracle]
-utgen fix --project-dir <PROJECT_DIR> [--work-dir <WORK_DIR>...] [--tasks <N>]
+utgen fix --project-dir <PROJECT_DIR> [--work-dir <WORK_DIR>...] [--tasks <N>] [--integration]
 ```
 
 `analyze` 现在执行 `cargo clean`、`cargo brinfo` 和 `focxt`，随后核对分析索引及上下文文件。当前支持用 `-p` 指定单个独立 crate；目录中已有 `brinfo/` 或 `focxt/` 时会拒绝执行，应使用新的工作副本，避免混入旧分析结果。
@@ -358,7 +358,7 @@ export PALM_CONFIG="$(pwd)/utgen/res/api.json"
   - `gen_full_tests` 使用 `test_prompt.json` 一次性生成测试。
   - `gen_tests_cot` 使用 `input -> prefix -> oracle` 三步流程。
   - `check_unit` 会临时把测试插入被测源码文件，运行 `cargo build --tests`，记录每个候选是否可编译，然后恢复文件。
-  - `check_integration` 会写入 `tests/test_check.rs` 做编译检查。
+  - `check_integration` 会写入临时 `tests/palm_candidate.rs`，只编译该集成测试目标。
 
 - `utgen/src/gene/cot/*.rs`
   - `input_infer.rs`：生成输入范围。
@@ -620,7 +620,9 @@ utgen fix -p <target-crate-path>
 
 修复阶段读取 `utgen/generation/pre_fix`，针对不可编译测试调用 LLM 生成 ChangeLog 并重新验证。修复结果写入 `utgen/generation/llm_fix`，覆盖率统计写入 `utgen/fixed_result`。
 
-`fix` 当前将测试插入源码，以单元测试方式修复和统计，没有独立的 integration 修复选项。因此 `gen --integration` 后接 `fix` 并不表示全过程保留集成测试模式。运行时断言失败也不属于当前编译修复流程的目标。
+`fix` 默认将测试插入源码，以单元测试方式修复和统计。集成测试需要在生成和修复时都传入 `--integration`：修复阶段把单个候选写入临时集成测试目标，只编译该目标；可以修改导入、模块级辅助代码和测试体，并单独保存各候选的导入与辅助代码。整体执行和逐候选覆盖率统计都保持集成测试模式，不再把候选插入源码。覆盖率阶段仍会临时添加排除测试代码的属性并恢复源文件。
+
+候选缓存记录生成模式，模式不匹配时会报错；没有模式字段的旧缓存按单元测试处理，旧集成测试缓存应在新的工作副本中重新生成。两条命令还应传入相同的函数清单。运行时断言失败不属于编译修复流程的目标。
 
 ### 10.7 示例：`examples/bytes`
 
@@ -709,7 +711,7 @@ integration 模式先按 `brinfo` 的 `visible` 标志筛选函数，并且 `utg
 
 - `utgen pre-process`：重命名所选 crate 的 `tests`，将测试源码区间替换为空白。
 - `utgen gen`：可能追加 `ntest` 依赖，写入临时测试文件或插入测试做编译检查。
-- `utgen fix`：临时插入测试并恢复，结束时删除备份。
+- `utgen fix`：默认临时插入单元测试并恢复，成功时删除本轮备份；`--integration` 使用临时集成测试文件，恢复或删除本轮临时输入。
 
 建议用 git 查看差异：
 

examples/README.md

@@ -72,9 +72,10 @@ Use a fresh working copy and repeat preprocessing and analysis before choosing t
 
 ```sh
 utgen gen -p <target-crate-path> --integration --requirement --context
+utgen fix -p <target-crate-path> --integration
 ```
 
-Integration mode generates files under the target's `tests/` directory after filtering functions using the analysis visibility flag and checking compilation. `utgen fix` currently uses source-inserted unit tests for repair and post-repair statistics; it does not provide a separate integration-mode repair pipeline.
+Integration mode generates files under the target's `tests/` directory after filtering functions using the analysis visibility flag and checking compilation. Pass `--integration` to `fix` as well to repair and evaluate candidates in integration-test targets. Imports and helpers stay with each candidate, and both combined and per-candidate coverage retain integration scope. Keep the same function selection for both commands; mixing cached generation modes is rejected.
 
 ### Results
 

scripts/check_minimal.py

@@ -202,7 +202,7 @@ def run(label, command, cwd=target, expected=0, timeout=240):
     finally:
         lib.write_bytes(saved_lib)
     assert snapshot(original) == baseline
-    state = {'generation': 0, 'repair': 0, 'errors': []}
+    state = {'generation': 0, 'repair': 0, 'errors': [], 'integration': False, 'reject_repair': False}
 
     class Model(http.server.BaseHTTPRequestHandler):
         def log_message(self, *_):
@@ -226,19 +226,60 @@ def do_POST(self):
                         'double': 'assert_eq!(double("bad"), 4);',
                     }[function]
                     answer = 'fn generated_helper() -> i32 { 1 }\n#[test]\nfn generated() {\n    assert_eq!(generated_helper(), 1);\n    ' + body + '\n}\n'
+                    if state['integration']:
+                        answer = {
+                            'classify': '''use std::cmp::max as target;
+use std::fmt::Write;
+fn helper() -> i32 { 1 }
+#[test]
+fn alias_case() {
+    let mut buffer = Vec::new();
+    buffer.write_all(b"x").unwrap();
+    assert_eq!(target(2), helper());
+}
+#[test]
+fn sibling_case() {
+    assert_eq!(target(2, 1), 2);
+    assert_eq!(palm_fixture::classify(-1), 0);
+}
+''',
+                            'double': '''use palm_fixture::nested::double as twice;
+fn helper() -> i32 { "bad-helper" }
+#[test]
+fn body_case() {
+    assert_eq!(twice("bad-body") + helper(), 5);
+}
+''',
+                        }[function]
                     state['generation'] += 1
                 else:
                     file = re.search(r'You can only modify lines \d+ to \d+ in file (.+?)\. For your answer', user).group(1)
-                    match = re.search(r'^\[(\d+)\](.*assert_eq!\(double\("bad"\), 4\);.*)$', user, re.M)
-                    assert match, 'repair prompt lacks numbered failing statement'
-                    number, old = match.groups()
-                    new = old.replace('double("bad")', 'double(2)')
-                    answer = f'ChangeLog:1@{file}\nFixDescription: Use an integer argument.\nOriginalCode@{number}-{number}:\n[{number}]{old}\nFixedCode@{number}-{number}:\n[{number}]{new}\n'
+                    replacements = {
+                        'double("bad")': 'double(2)',
+                        'use std::cmp::max as target;': 'use palm_fixture::classify as target;',
+                        'use std::fmt::Write;': 'use std::io::Write;',
+                        '"bad-helper"': '1',
+                        '"bad-body"': '2',
+                    }
+                    changes = {}
+                    for number, old in re.findall(r'^\[(\d+)\](.*)$', user, re.M):
+                        new = old
+                        for before, after in replacements.items():
+                            new = new.replace(before, after)
+                        if new != old:
+                            changes[number] = (old, new)
+                    assert changes, 'repair prompt lacks numbered failing code'
+                    answer = f'ChangeLog:1@{file}\nFixDescription: Fix candidate imports and arguments.\n'
+                    for number, (old, new) in changes.items():
+                        answer += f'OriginalCode@{number}-{number}:\n[{number}]{old}\nFixedCode@{number}-{number}:\n[{number}]{new}\n'
                     state['repair'] += 1
                 response = {'id': 'fixture', 'object': 'chat.completion', 'created': 0, 'model': 'fixture',
                             'choices': [{'index': 0, 'message': {'role': 'assistant', 'content': answer}, 'finish_reason': 'stop'}],
                             'usage': {'prompt_tokens': 1, 'completion_tokens': 1, 'total_tokens': 2}}
                 code = 200
+                if state['reject_repair']:
+                    response = {'error': {'message': 'fixture request rejected'}}
+                    code = 401
             except Exception as error:
                 state['errors'].append(str(error))
                 response = {'error': {'message': str(error), 'type': 'fixture_error'}}
@@ -307,6 +348,8 @@ def do_POST(self):
         timeout_prepared = snapshot(target)
         requests_before = (state['generation'], state['repair'])
         for integration in [False, True]:
+            data['integration'] = integration
+            candidate.write_text(json.dumps(data, indent=2) + '\n')
             run(f'candidate-timeouts-{integration}',
                 ['utgen', 'gen', '-p', str(target), '--functions-file', str(selection),
                  *(['--integration'] if integration else [])], timeout=90)
@@ -376,6 +419,89 @@ def integration_snapshot():
         assert report['attempts'] == 0, report
         shutil.rmtree(target / 'tests')
         assert snapshot(target) == prepared
+
+        # Exercise the complete integration path with real Rust diagnostics.
+        # One answer has two candidates needing different meanings of `target`.
+        state['integration'] = True
+        selected_names = [selected, 'palm_fixture::nested::double']
+        integration_selection = work / 'integration-functions.txt'
+        integration_selection.write_text('\n'.join(selected_names) + '\n')
+        for directory in ['generation/pre_fix', 'generation/llm_fix']:
+            for name in selected_names:
+                (target / 'utgen' / directory / (names[name] + '.json')).unlink()
+        integration_args = ['-p', str(target), '--functions-file', str(integration_selection),
+                            '--integration', '--tasks', '2', '--max-requests', '8']
+        run('integration-generate-imports', ['utgen', 'gen', *integration_args])
+        for name, compiled in zip(selected_names, [1, 0]):
+            result = json.loads((target / f'utgen/result/{names[name]}.json').read_text())
+            assert result['tests_compiled'] == result['tests_run'] == compiled, result
+        source_before = {p: content for p, content in snapshot(target).items() if not p.startswith('tests/')}
+        assert source_before == prepared
+        # Unselected targets must neither pollute diagnostics nor be executed.
+        unrelated = target / 'tests/unrelated.rs'
+        unrelated.write_text('compile_error!("unselected test must not compile");\n')
+        temporary_file = target / 'tests/palm_candidate.rs'
+        temporary_file.write_text('compile_error!("pre-existing temporary path");\n')
+        before_fix = snapshot(target)
+        requests_before = (state['generation'], state['repair'])
+        wrong_mode = run('integration-reject-unit-fix',
+                         ['utgen', 'fix', '-p', str(target), '--functions-file', str(integration_selection)], expected=1)
+        assert 'Test mode mismatch' in wrong_mode.stderr
+        assert snapshot(target) == before_fix
+        assert (state['generation'], state['repair']) == requests_before
+        wrong_mode = run('integration-reject-unit-gen',
+                         ['utgen', 'gen', '-p', str(target), '--functions-file', str(integration_selection)], expected=1)
+        assert 'Test mode mismatch' in wrong_mode.stderr
+        assert snapshot(target) == before_fix
+        assert (state['generation'], state['repair']) == requests_before
+        run('integration-repair-imports', ['utgen', 'fix', *integration_args])
+        assert state['repair'] > requests_before[1]
+        for name, count in zip(selected_names, [2, 1]):
+            result = json.loads((target / f'utgen/fixed_result/{names[name]}.json').read_text())
+            assert result['tests'] == result['tests_compiled'] == result['tests_run'] == result['tests_passed'] == count, result
+            assert len(result['codes_lines_covered']) == count, result
+            assert result['lines_covered'] > 0, result
+        fixed_path = target / f'utgen/generation/llm_fix/{names[selected]}.json'
+        fixed = json.loads(fixed_path.read_text())
+        tests = fixed['fn_tests'][0]['answers'][0]['chain_tests']
+        assert 'use palm_fixture::classify as target;' in tests[0]['integration_contexts'][0]['uses']
+        assert 'use std::io::Write;' in tests[0]['integration_contexts'][0]['uses']
+        assert 'use std::cmp::max as target;' in tests[1]['integration_contexts'][0]['uses']
+        assert {p: content for p, content in snapshot(target).items() if not p.startswith('tests/')} == prepared
+        assert unrelated.read_bytes() == before_fix['tests/unrelated.rs']
+        assert temporary_file.read_bytes() == before_fix['tests/palm_candidate.rs']
+        assert not list((target / 'src').rglob('*.bak'))
+        stable = snapshot(target)
+        requests_before = (state['generation'], state['repair'])
+        run('integration-repair-cached', ['utgen', 'fix', *integration_args])
+        assert snapshot(target) == stable
+        assert (state['generation'], state['repair']) == requests_before
+        report = json.loads((target / 'utgen/generation/fix-requests.json').read_text())
+        assert report['attempts'] == 0 and report['invocation']['integration'], report
+        source_file = target / 'src/lib.rs'
+        original_source = source_file.read_bytes()
+        source_file.write_bytes(original_source + b'\ncompile_error!("production compilation failure");\n')
+        broken = snapshot(target)
+        failure = run('integration-repair-build-error', ['utgen', 'fix', *integration_args], expected=1)
+        assert 'production compilation failure' in failure.stderr
+        assert snapshot(target) == broken
+        assert (state['generation'], state['repair']) == requests_before
+        assert not list((target / 'src').rglob('*.bak'))
+        source_file.write_bytes(original_source)
+        # Request failure must restore the temporary input without touching src/.
+        saved_fixed = {}
+        for name in selected_names:
+            path = target / f'utgen/generation/llm_fix/{names[name]}.json'
+            saved_fixed[path] = path.read_bytes()
+            path.unlink()
+        state['reject_repair'] = True
+        run('integration-repair-request-error', ['utgen', 'fix', *integration_args], expected=1)
+        state['reject_repair'] = False
+        assert snapshot(target) == stable
+        assert not list((target / 'src').rglob('*.bak'))
+        for path, content in saved_fixed.items():
+            path.write_bytes(content)
+        assert not state['errors'], state
         (work / 'summary.json').write_text(json.dumps(state, indent=2) + '\n')
         print('Minimal analysis/generation/repair/coverage checks passed.', flush=True)
     finally: