codeboarding 0.9.0__tar.gz → 0.9.3__tar.gz

{codeboarding-0.9.0/codeboarding.egg-info → codeboarding-0.9.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: codeboarding
-Version: 0.9.0
+Version: 0.9.3
 Summary: Interactive Diagrams for Code
 Author: CodeBoarding Team
 License: MIT
@@ -28,6 +28,7 @@ Requires-Dist: exceptiongroup>=1.2
 Requires-Dist: fastapi>=0.115
 Requires-Dist: filelock>=3.12
 Requires-Dist: gitpython>=3.1
+Requires-Dist: google-api-core>=2.10
 Requires-Dist: google-genai>=1.10
 Requires-Dist: gql>=3.5
 Requires-Dist: injector>=0.21

{codeboarding-0.9.0 → codeboarding-0.9.3}/agents/agent.py

@@ -32,6 +32,10 @@ from static_analyzer.reference_resolve_mixin import ReferenceResolverMixin
 logger = logging.getLogger(__name__)
 
 
+class EmptyExtractorMessageError(ValueError):
+    """Raised when extractor returns an empty message payload."""
+
+
 class CodeBoardingAgent(ReferenceResolverMixin, MonitoringMixin):
     def __init__(
         self,
@@ -281,9 +285,14 @@ class CodeBoardingAgent(ReferenceResolverMixin, MonitoringMixin):
             if "messages" in result and len(result["messages"]) != 0:
                 message = result["messages"][0].content
                 parser = PydanticOutputParser(pydantic_object=return_type)
+                if not message:
+                    raise EmptyExtractorMessageError("Extractor returned empty message content")
                 return self._try_parse(message, parser)
             parser = PydanticOutputParser(pydantic_object=return_type)
             return self._try_parse(response, parser)
+        except EmptyExtractorMessageError as e:
+            logger.warning(f"{e} (attempt {attempt + 1}/{max_retries})")
+            return self._parse_response(prompt, response, return_type, max_retries, attempt + 1)
         except AttributeError as e:
             # Workaround for trustcall bug: https://github.com/hinthornw/trustcall/issues/47
             # 'ExtractionState' object has no attribute 'tool_call_id' occurs during validation retry

{codeboarding-0.9.0 → codeboarding-0.9.3}/agents/llm_config.py

@@ -235,6 +235,19 @@ LLM_PROVIDERS = {
             "max_retries": 0,
         },
     ),
+    "openrouter": LLMConfig(
+        chat_class=ChatOpenAI,
+        api_key_env="OPENROUTER_API_KEY",
+        agent_model="google/gemini-2.5-flash",
+        parsing_model="google/gemini-2.5-flash",
+        llm_type=LLMType.GEMINI_FLASH,
+        extra_args={
+            "base_url": lambda: os.getenv("OPENROUTER_BASE_URL", "https://openrouter.ai/api/v1"),
+            "max_tokens": None,
+            "timeout": None,
+            "max_retries": 0,
+        },
+    ),
 }
 
 

{codeboarding-0.9.0 → codeboarding-0.9.3}/agents/planner_agent.py

@@ -79,7 +79,7 @@ def should_expand_component(
     return False
 
 
-def plan_analysis(
+def get_expandable_components(
     analysis: AnalysisInsights,
     parent_had_clusters: bool = True,
     min_files: int = DEFAULT_MIN_FILES,

{codeboarding-0.9.0 → codeboarding-0.9.3/codeboarding.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: codeboarding
-Version: 0.9.0
+Version: 0.9.3
 Summary: Interactive Diagrams for Code
 Author: CodeBoarding Team
 License: MIT
@@ -28,6 +28,7 @@ Requires-Dist: exceptiongroup>=1.2
 Requires-Dist: fastapi>=0.115
 Requires-Dist: filelock>=3.12
 Requires-Dist: gitpython>=3.1
+Requires-Dist: google-api-core>=2.10
 Requires-Dist: google-genai>=1.10
 Requires-Dist: gql>=3.5
 Requires-Dist: injector>=0.21

{codeboarding-0.9.0 → codeboarding-0.9.3}/codeboarding.egg-info/requires.txt

@@ -9,6 +9,7 @@ exceptiongroup>=1.2
 fastapi>=0.115
 filelock>=3.12
 gitpython>=3.1
+google-api-core>=2.10
 google-genai>=1.10
 gql>=3.5
 injector>=0.21

{codeboarding-0.9.0 → codeboarding-0.9.3}/diagram_analysis/diagram_generator.py

@@ -13,7 +13,7 @@ from agents.agent_responses import AnalysisInsights, Component
 from agents.details_agent import DetailsAgent
 from agents.llm_config import initialize_llms
 from agents.meta_agent import MetaAgent
-from agents.planner_agent import plan_analysis
+from agents.planner_agent import get_expandable_components
 from diagram_analysis.analysis_json import (
     FileCoverageReport,
     FileCoverageSummary,
@@ -35,9 +35,10 @@ from monitoring.mixin import MonitoringMixin
 from monitoring.paths import generate_run_id, get_monitoring_run_dir
 from repo_utils import get_git_commit_hash, get_repo_state_hash
 from repo_utils.ignore import RepoIgnoreManager
-from static_analyzer import get_static_analysis
+from static_analyzer import StaticAnalyzer, get_static_analysis
 from static_analyzer.analysis_result import StaticAnalysisResults
 from static_analyzer.scanner import ProjectScanner
+from utils import get_cache_dir
 
 logger = logging.getLogger(__name__)
 
@@ -53,6 +54,7 @@ class DiagramGenerator:
         project_name: str | None = None,
         run_id: str | None = None,
         monitoring_enabled: bool = False,
+        static_analyzer: StaticAnalyzer | None = None,
     ):
         self.repo_location = repo_location
         self.temp_folder = temp_folder
@@ -63,6 +65,10 @@ class DiagramGenerator:
         self.run_id = run_id
         self.monitoring_enabled = monitoring_enabled
         self.force_full_analysis = False  # Set to True to skip incremental updates
+        # Optional pre-started StaticAnalyzer injected by long-lived callers (e.g. the
+        # wrapper). When set, pre_analysis() uses it directly instead of creating a new
+        # one-shot analyzer via get_static_analysis().
+        self._static_analyzer = static_analyzer
 
         self.details_agent: DetailsAgent | None = None
         self.static_analysis: StaticAnalysisResults | None = None  # Cache static analysis for reuse
@@ -87,7 +93,7 @@ class DiagramGenerator:
             parent_had_clusters = bool(component.source_cluster_ids)
 
             # Get new components to analyze (deterministic, no LLM)
-            new_components = plan_analysis(analysis, parent_had_clusters=parent_had_clusters)
+            new_components = get_expandable_components(analysis, parent_had_clusters=parent_had_clusters)
 
             return component.component_id, analysis, new_components
         except Exception as e:
@@ -107,7 +113,7 @@ class DiagramGenerator:
             repo_path=self.repo_location,
         )
         if health_report is not None:
-            health_path = os.path.join(self.output_dir, "health", "health_report.json")
+            health_path = Path(self.output_dir) / "health" / "health_report.json"
             with open(health_path, "w") as f:
                 f.write(health_report.model_dump_json(indent=2, exclude_none=True))
             logger.info(f"Health report written to {health_path} (score: {health_report.overall_score:.3f})")
@@ -138,11 +144,18 @@ class DiagramGenerator:
             summary=FileCoverageSummary(**self.file_coverage_data["summary"]),
         )
 
-        coverage_path = os.path.join(self.output_dir, "file_coverage.json")
+        coverage_path = Path(self.output_dir) / "file_coverage.json"
         with open(coverage_path, "w") as f:
             f.write(report.model_dump_json(indent=2, exclude_none=True))
         logger.info(f"File coverage report written to {coverage_path}")
 
+    def _get_static_from_injected_analyzer(self, cache_dir: Path | None) -> StaticAnalysisResults:
+        result = self._static_analyzer.analyze(  # type: ignore[union-attr]
+            cache_dir=cache_dir,
+        )
+        result.diagnostics = self._static_analyzer.collected_diagnostics  # type: ignore[union-attr]
+        return result
+
     def pre_analysis(self):
         analysis_start_time = time.time()
 
@@ -157,14 +170,27 @@ class DiagramGenerator:
         )
         self._monitoring_agents["MetaAgent"] = self.meta_agent
 
-        # Run static analysis and meta analysis in parallel
-        with ThreadPoolExecutor(max_workers=2) as executor:
+        def get_static_with_injected_analyzer() -> StaticAnalysisResults:
+            cache_dir = None if self.force_full_analysis else get_cache_dir(self.repo_location)
+            return self._get_static_from_injected_analyzer(cache_dir)
+
+        def get_static_with_new_analyzer() -> StaticAnalysisResults:
             skip_cache = self.force_full_analysis
             if skip_cache:
                 logger.info("Force full analysis: skipping static analysis cache")
-            static_future = executor.submit(get_static_analysis, self.repo_location, skip_cache=skip_cache)
-            meta_future = executor.submit(self.meta_agent.get_meta_context, refresh=self.force_full_analysis)
+            return get_static_analysis(self.repo_location, skip_cache=skip_cache)
+
+        # Decide how to obtain static analysis results, then run it in parallel
+        # with the meta-context computation so neither blocks the other.
+        if self._static_analyzer is not None:
+            logger.info("Using injected StaticAnalyzer (clients already running)")
+            static_callable = get_static_with_injected_analyzer
+        else:
+            static_callable = get_static_with_new_analyzer
 
+        with ThreadPoolExecutor(max_workers=2) as executor:
+            static_future = executor.submit(static_callable)
+            meta_future = executor.submit(self.meta_agent.get_meta_context, refresh=self.force_full_analysis)
             static_analysis = static_future.result()
             meta_context = meta_future.result()
 
@@ -205,7 +231,7 @@ class DiagramGenerator:
         )
         self._monitoring_agents["AbstractionAgent"] = self.abstraction_agent
 
-        version_file = os.path.join(self.output_dir, "codeboarding_version.json")
+        version_file = Path(self.output_dir) / "codeboarding_version.json"
         with open(version_file, "w") as f:
             f.write(
                 Version(
@@ -317,7 +343,7 @@ class DiagramGenerator:
 
         return expanded_components, sub_analyses
 
-    def generate_analysis(self):
+    def generate_analysis(self) -> list[Path]:
         """
         Generate the graph analysis for the given repository.
         The output is stored in a single analysis.json file in output_dir.
@@ -337,7 +363,7 @@ class DiagramGenerator:
             analysis, cluster_results = self.abstraction_agent.run()
 
             # Get the initial components to analyze (deterministic, no LLM)
-            root_components = plan_analysis(analysis)
+            root_components = get_expandable_components(analysis)
             logger.info(f"Found {len(root_components)} components to analyze at level 1")
 
             # Process components using a frontier queue: submit children as soon as parent finishes.
@@ -355,18 +381,15 @@ class DiagramGenerator:
                 )
 
             # Final write of unified analysis.json
-            analysis_path = str(
-                save_analysis(
-                    analysis=analysis,
-                    output_dir=Path(self.output_dir),
-                    sub_analyses=sub_analyses,
-                    repo_name=self.repo_name,
-                    file_coverage_summary=file_coverage_summary,
-                )
+            analysis_path = save_analysis(
+                analysis=analysis,
+                output_dir=Path(self.output_dir),
+                sub_analyses=sub_analyses,
+                repo_name=self.repo_name,
+                file_coverage_summary=file_coverage_summary,
             )
 
             logger.info(f"Analysis complete. Written unified analysis to {analysis_path}")
-            print("Generated analysis file: %s", os.path.abspath(analysis_path))
 
             # Write file_coverage.json
             self._write_file_coverage()
@@ -396,7 +419,7 @@ class DiagramGenerator:
         except Exception as e:
             logger.warning(f"Failed to save manifest: {e}")
 
-    def try_incremental_update(self) -> list[str] | None:
+    def try_incremental_update(self) -> list[Path] | None:
         """
         Attempt an incremental update if possible.
 
@@ -416,7 +439,11 @@ class DiagramGenerator:
         # recompute file assignments with cluster matching. Load it first.
         static_analysis = None
         try:
-            static_analysis = get_static_analysis(self.repo_location)
+            if self._static_analyzer is not None:
+                static_analysis = self._static_analyzer.analyze(cache_dir=get_cache_dir(self.repo_location))
+                static_analysis.diagnostics = self._static_analyzer.collected_diagnostics
+            else:
+                static_analysis = get_static_analysis(self.repo_location)
             logger.info("Loaded static analysis for incremental update")
         except Exception as e:
             logger.warning(f"Could not load static analysis: {e}")
@@ -442,7 +469,7 @@ class DiagramGenerator:
 
         if impact.action == UpdateAction.NONE:
             logger.info("No changes detected, analysis is up to date")
-            return [str(self.output_dir / "analysis.json")]
+            return [self.output_dir / "analysis.json"]
 
         # For structural changes, recompute which components are actually affected
         # after static analysis has been updated with cluster matching
@@ -465,13 +492,13 @@ class DiagramGenerator:
                 )
                 self._write_file_coverage()
 
-            return [str(self.output_dir / "analysis.json")]
+            return [self.output_dir / "analysis.json"]
 
         # Incremental update failed or not possible
         logger.info("Incremental update not possible, falling back to full analysis")
         return None
 
-    def generate_analysis_smart(self) -> list[str]:
+    def generate_analysis_smart(self) -> list[Path]:
         """
         Smart analysis that tries incremental first, falls back to full.
 

{codeboarding-0.9.0 → codeboarding-0.9.3}/diagram_analysis/incremental/reexpansion.py

@@ -12,7 +12,7 @@ from agents.llm_config import initialize_llms
 from agents.agent_responses import AnalysisInsights
 from agents.details_agent import DetailsAgent
 from agents.meta_agent import MetaAgent
-from agents.planner_agent import plan_analysis
+from agents.planner_agent import get_expandable_components
 from diagram_analysis.incremental.io_utils import load_sub_analysis, save_sub_analysis
 from diagram_analysis.incremental.models import ChangeImpact
 from diagram_analysis.incremental.path_patching import patch_sub_analysis

{codeboarding-0.9.0 → codeboarding-0.9.3}/github_action.py

@@ -154,7 +154,7 @@ def generate_analysis(
     analysis_files = generator.generate_analysis_smart()
 
     # The generator now returns a single analysis.json path
-    analysis_path = Path(analysis_files[0])
+    analysis_path = analysis_files[0]
 
     # Now generate the output docs:
     match extension:

{codeboarding-0.9.0 → codeboarding-0.9.3}/health/checks/unused_code_diagnostics.py

@@ -136,13 +136,17 @@ class LSPDiagnosticsCollector:
         """Process all collected diagnostics and convert to issues."""
         self.issues = []
         skipped = 0
-        seen_issues: set[tuple[str, str, int]] = set()  # (file_path, code, line_start)
+        seen_issues: set[tuple[str, str, int]] = set()  # (file_path, category, line_start)
 
         for item in self.diagnostics:
             issue = self._convert_to_issue(item.file_path, item.diagnostic)
             if issue:
-                # Deduplicate: same file, code, and line
-                key = (issue.file_path, issue.code or "", issue.line_start)
+                # Deduplicate: same file, category, and line.
+                # Using category instead of code prevents duplicates from the same
+                # LSP server reporting the same issue under different diagnostic codes
+                # (e.g. Pyright reports unused variables both via reportUnusedVariable
+                # and via the "Unnecessary" tag with different messages).
+                key = (issue.file_path, issue.category.value, issue.line_start)
                 if key not in seen_issues:
                     seen_issues.add(key)
                     self.issues.append(issue)
@@ -187,8 +191,8 @@ class LSPDiagnosticsCollector:
 
         return DiagnosticIssue(
             file_path=file_path,
-            line_start=diagnostic.range.start.line,
-            line_end=diagnostic.range.end.line,
+            line_start=diagnostic.range.start.line + 1,  # LSP lines are 0-based, convert to 1-based
+            line_end=diagnostic.range.end.line + 1,  # LSP lines are 0-based, convert to 1-based
             message=message,
             code=code if code else None,
             category=category,

{codeboarding-0.9.0 → codeboarding-0.9.3}/main.py

@@ -66,7 +66,7 @@ def generate_analysis(
     )
     generator.force_full_analysis = force_full
     generated_files = generator.generate_analysis()
-    return [Path(path) for path in generated_files]
+    return generated_files
 
 
 def generate_markdown_docs(

{codeboarding-0.9.0 → codeboarding-0.9.3}/monitoring/context.py

@@ -3,7 +3,7 @@ import logging
 import functools
 import time
 import contextlib
-from contextvars import ContextVar
+from contextvars import ContextVar, Token
 from pathlib import Path
 from typing import Callable, Any
 
@@ -71,25 +71,40 @@ def monitor_execution(
 
     # Allow the user to manually log steps via the yielded context
     class MonitorContext:
+        def __init__(self):
+            self._step_tokens: list[Token[str]] = []
+
         def step(self, name):
             trace_logger.info(json.dumps({"event": "phase_change", "step": name, "timestamp": time.time()}))
             # Also update the ContextVar for other components
-            self._token = current_step.set(name)
+            token = current_step.set(name)
+            self._step_tokens.append(token)
 
-            def end_step(self):
-                pass
+        def end_step(self):
+            if not self._step_tokens:
+                return
+            token = self._step_tokens.pop()
+            current_step.reset(token)
+
+        def close(self):
+            while self._step_tokens:
+                self.end_step()
 
     # Initialize stats for this run
     run_stats = RunStats()
     stats_token = current_stats.set(run_stats)
 
+    monitor_context = MonitorContext()
+
     try:
         # Log start of run
         trace_logger.info(json.dumps({"event": "run_start", "run_id": run_id, "timestamp": time.time()}))
 
-        yield MonitorContext()
+        yield monitor_context
 
     finally:
+        monitor_context.close()
+
         # Log end of run
         trace_logger.info(json.dumps({"event": "run_end", "run_id": run_id, "timestamp": time.time()}))
 

{codeboarding-0.9.0 → codeboarding-0.9.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "codeboarding"
-version = "0.9.0"
+version = "0.9.3"
 description = "Interactive Diagrams for Code"
 readme = "PYPI.md"
 license = {text = "MIT"}
@@ -34,6 +34,7 @@ dependencies = [
     "fastapi>=0.115",
     "filelock>=3.12",
     "gitpython>=3.1",
+    "google-api-core>=2.10",
     "google-genai>=1.10",
     "gql>=3.5",
     "injector>=0.21",
@@ -106,7 +107,7 @@ warn_return_any = false
 warn_unused_configs = true
 disallow_untyped_defs = false
 check_untyped_defs = true
-exclude = "(?x)(^repos/|^build/|^dist/|^temp/|^static_analyzer/servers/)"
+exclude = "(?x)(^repos/|^build/|^dist/|^temp/|^static_analyzer/servers/|^tests/integration/projects/)"
 explicit_package_bases = true
 disable_error_code = ["import-untyped", "no-any-return", "assignment"]
 
@@ -116,6 +117,8 @@ dev-dependencies = [
     "pre-commit==3.8.0",
     "mypy==1.19.0",
     "pytest-cov==7.0.0",
+    "build>=1.4.0",
+    "twine>=6.2.0",
 ]
 
 [tool.pytest.ini_options]
@@ -133,4 +136,4 @@ markers = [
     "typescript_lang: marks tests for TypeScript language",
     "php_lang: marks tests for PHP language",
     "javascript_lang: marks tests for JavaScript language",
-]
+]

{codeboarding-0.9.0 → codeboarding-0.9.3}/repo_utils/change_detector.py

@@ -125,7 +125,7 @@ def detect_changes(
     Example:
         changes = detect_changes(repo_path, "abc1234")
         for rename_old, rename_new in changes.renames.items():
-            print(f"Renamed: {rename_old} -> {rename_new}")
+            logger.info(f"Renamed: {rename_old} -> {rename_new}")
     """
     changes: list[DetectedChange] = []