@hupan56/wlkj 2.7.12 → 3.1.2

package/templates/qoder/scripts/domain/kg/extract/test_extract.py

@@ -0,0 +1,115 @@
+# -*- coding: utf-8 -*-
+"""
+Test code extractor — E1: extract structured assertions from Test.java files.
+
+These are the "second layer" of the knowledge graph: structured business
+knowledge (Given-When-Then, assertion patterns, service dependencies) that
+production code extraction can never reach. 39+ Test.java files in fywl-ics
+contain real integration tests with @Autowired services and assertion logic.
+
+JOIN keys to other layers:
+  - injected service class → code-callgraph (who this test exercises)
+  - package path → ui-modules / prd-index (which business module)
+  - @Test method name → code-keyword (searchable)
+"""
+import os
+import re
+
+# @Test annotation (Junit 5)
+RE_TEST = re.compile(r'@Test\b', re.M)
+# @DisplayName("...")
+RE_DISPLAY = re.compile(r'@DisplayName\s*\(\s*["\']([^"\']+)["\']\s*\)')
+# method signature: void methodName() or public void methodName()
+RE_METHOD = re.compile(r'(?:public\s+|private\s+|protected\s+)?(?:static\s+)?(?:void|\w+(?:<[^>]*>)?)\s+(\w+)\s*\([^)]*\)\s*(?:throws\s+[\w.,\s]+)?\s*\{', re.M)
+# @Autowired / @Resource field: @Autowired private SomeService svc;
+RE_INJECT = re.compile(r'@(?:Autowired|Resource)\s+(?:private\s+|protected\s+)?(\w+(?:<[^>]*>)?)\s+(\w+)\s*;', re.M)
+# @SpringBootTest, @RunWith, @ExtendWith — marks integration tests
+RE_INTEGRATION = re.compile(r'@(?:SpringBootTest|RunWith|ExtendWith|WebMvcTest|DataJpaTest)', re.M)
+# Assertions: assertThat(x).isEqualTo(y), Assertions.assertEquals(x,y), Assert.X
+RE_ASSERT = re.compile(r'(?:assertThat|Assertions?|Assert)\.(\w+)\s*\(', re.M)
+# package declaration
+RE_PACKAGE = re.compile(r'package\s+([\w.]+)\s*;', re.M)
+
+
+def extract_test_assertions(proj_path):
+    """Scan a project for *Test.java files and extract structured test info.
+
+    Returns a list of dicts:
+      {file, class, package, method, display_name, injected_deps[],
+       assertions[], is_integration, comment}
+    """
+    results = []
+    if not proj_path or not os.path.isdir(proj_path):
+        return results
+
+    for root, dirs, files in os.walk(proj_path):
+        dirs[:] = [d for d in dirs if d not in ('node_modules', '.git', 'dist', '__pycache__')]
+        for f in files:
+            if not f.endswith('Test.java'):
+                continue
+            fpath = os.path.join(root, f)
+            try:
+                content = open(fpath, encoding='utf-8', errors='ignore').read()
+            except Exception:
+                continue
+            if '@Test' not in content:
+                continue
+
+            # Class-level info
+            pkg_m = RE_PACKAGE.search(content)
+            package = pkg_m.group(1) if pkg_m else ''
+            is_integration = bool(RE_INTEGRATION.search(content))
+
+            # Injected dependencies (services/mappers under test)
+            injected = []
+            for im in RE_INJECT.finditer(content):
+                dep_type = im.group(1).split('<')[0]  # strip generics
+                injected.append({'type': dep_type, 'name': im.group(2)})
+
+            # Class name (from filename, strip .java)
+            class_name = f[:-5]
+
+            # Extract each @Test method
+            test_positions = [m.start() for m in RE_TEST.finditer(content)]
+            display_map = {}
+            for dm in RE_DISPLAY.finditer(content):
+                # Associate @DisplayName with the nearest @Test after it
+                display_map[dm.end()] = dm.group(1)
+
+            for tp in test_positions:
+                # Find display name (nearest before this @Test, within 100 chars)
+                display = ''
+                for pos, name in display_map.items():
+                    if pos <= tp and tp - pos < 200:
+                        display = name
+
+                # Find method name after @Test
+                search_region = content[tp:tp + 300]
+                method_m = RE_METHOD.search(search_region)
+                if not method_m:
+                    continue
+                method_name = method_m.group(1)
+
+                # Extract method body for assertions
+                body_start = tp + method_m.end()
+                body = content[body_start:body_start + 600]
+                assertions = list(set(m.group(1) for m in RE_ASSERT.finditer(body)))
+
+                # Extract comment before @Test (/** ... */ or // ...)
+                before = content[max(0, tp - 200):tp]
+                comment_m = re.search(r'(?:/\*\*\s*\n\s*\*\s*([^*]+?)\s*\n)|(?://\s*(.+)$)', before, re.M)
+                comment = (comment_m.group(1) or comment_m.group(2)).strip() if comment_m else ''
+
+                results.append({
+                    'file': fpath.replace('\\', '/'),
+                    'class': class_name,
+                    'package': package,
+                    'method': method_name,
+                    'display_name': display,
+                    'comment': comment[:80],
+                    'injected_deps': [d['type'] for d in injected[:5]],
+                    'assertions': assertions[:5],
+                    'is_integration': is_integration,
+                })
+
+    return results