Support for the new --generate-json switch for analyzer to generate json file

labs.yml

@@ -16,7 +16,9 @@ commands:
       - name: report-file
         description: (Optional) Local filesystem `path` of the analysis report file to write
       - name: source-tech
-        description: (Optional) The technology/platform of the sources to analyze
+        description: (Optional) Name of the Source System Technology you want to analyze
+      - name: generate-json
+        description: (Optional) Flag to indicate if a json file should be produced alongside the Excel file
 
   - name: transpile
     description: Transpile SQL/ETL sources to Databricks-compatible code

src/databricks/labs/lakebridge/analyzer/lakebridge_analyzer.py

@@ -2,6 +2,7 @@
 import tempfile
 from pathlib import Path
 from collections.abc import Callable
+from typing import cast
 
 from databricks.labs.blueprint.entrypoint import get_logger
 from databricks.labs.blueprint.tui import Prompts
@@ -56,24 +57,31 @@ def get_source_system(self, platform: str | None = None) -> str:
 
 class AnalyzerRunner:
     def __init__(
-        self, runnable: Callable[[Path, Path, str, bool], None], move_file: Callable[[Path, Path], None], is_debug: bool
+        self,
+        runnable: Callable[[Path, Path, str, bool, bool], None],
+        move_file: Callable[[Path, Path], None],
+        is_debug: bool,
     ):
         self._runnable = runnable
         self._move_file = move_file
         self._is_debug = is_debug
 
     @classmethod
     def create(cls, is_debug: bool = False) -> "AnalyzerRunner":
-        return cls(Analyzer.analyze, move_tmp_file, is_debug)
+        return cls(
+            cast(Callable[[Path, Path, str, bool, bool], None], Analyzer.analyze),
+            move_tmp_file,
+            is_debug,
+        )
 
-    def run(self, source_dir: Path, results_dir: Path, platform: str) -> AnalyzerResult:
+    def run(self, source_dir: Path, results_dir: Path, platform: str, generate_json: bool = False) -> AnalyzerResult:
         logger.debug(f"Starting analyzer execution in ${source_dir} for ${platform}")
 
         if not check_path(source_dir) or not check_path(results_dir):
             raise ValueError(f"Invalid path(s) provided: source_dir={source_dir}, results_dir={results_dir}")
 
         tmp_dir = self._temp_xlsx_path(results_dir)
-        self._runnable(source_dir, tmp_dir, platform, self._is_debug)
+        self._runnable(source_dir, tmp_dir, platform, self._is_debug, generate_json)
         self._move_file(tmp_dir, Path(results_dir))
         logger.info(f"Successfully Analyzed files in ${source_dir} for ${platform} and saved report to {results_dir}")
         return AnalyzerResult(Path(source_dir), Path(results_dir), platform)
@@ -90,7 +98,11 @@ def __init__(self, prompts: AnalyzerPrompts, runner: AnalyzerRunner):
         self._runner = runner
 
     def run_analyzer(
-        self, source: str | None = None, results: str | None = None, platform: str | None = None
+        self,
+        source: str | None = None,
+        results: str | None = None,
+        platform: str | None = None,
+        generate_json: bool = False,
     ) -> AnalyzerResult:
         if not source:
             source_dir = self._prompts.get_source_directory()
@@ -108,4 +120,4 @@ def run_analyzer(
 
         platform = self._prompts.get_source_system(platform)
 
-        return self._runner.run(source_dir, results_dir, platform)
+        return self._runner.run(source_dir, results_dir, platform, generate_json)

src/databricks/labs/lakebridge/cli.py

@@ -817,11 +817,12 @@ def analyze(
     source_directory: str | None = None,
     report_file: str | None = None,
     source_tech: str | None = None,
+    generate_json: bool = False,
 ):
     """Run the Analyzer"""
     ctx = ApplicationContext(w)
     try:
-        result = ctx.analyzer.run_analyzer(source_directory, report_file, source_tech)
+        result = ctx.analyzer.run_analyzer(source_directory, report_file, source_tech, generate_json)
         ctx.add_user_agent_extra("analyzer_source_tech", result.source_system)
     finally:
         exception_cls, _, _ = sys.exc_info()