etlplus 0.4.6__py3-none-any.whl → 0.7.0__py3-none-any.whl

etlplus/cli/handlers.py

@@ -18,6 +18,8 @@ from typing import cast
 
 from ..config import PipelineConfig
 from ..config import load_pipeline_config
+from ..database import load_table_spec
+from ..database import render_tables
 from ..enums import FileFormat
 from ..extract import extract
 from ..file import File
@@ -34,19 +36,51 @@ from ..validate import validate
 
 __all__ = [
     # Functions
-    'cmd_extract',
-    'cmd_list',
-    'cmd_load',
-    'cmd_pipeline',
-    'cmd_run',
-    'cmd_transform',
-    'cmd_validate',
+    'extract_handler',
+    'check_handler',
+    'load_handler',
+    'pipeline_handler',
+    'render_handler',
+    'run_handler',
+    'transform_handler',
+    'validate_handler',
 ]
 
 
 # SECTION: INTERNAL FUNCTIONS =============================================== #
 
 
+def _collect_table_specs(
+    config_path: str | None,
+    spec_path: str | None,
+) -> list[dict[str, Any]]:
+    """
+    Load table schemas from a pipeline config and/or standalone spec.
+
+    Parameters
+    ----------
+    config_path : str | None
+        Path to a pipeline YAML config file.
+    spec_path : str | None
+        Path to a standalone table spec file.
+
+    Returns
+    -------
+    list[dict[str, Any]]
+        Collected table specification mappings.
+    """
+    specs: list[dict[str, Any]] = []
+
+    if spec_path:
+        specs.append(load_table_spec(Path(spec_path)))
+
+    if config_path:
+        cfg = load_pipeline_config(config_path, substitute=True)
+        specs.extend(getattr(cfg, 'table_schemas', []))
+
+    return specs
+
+
 def _emit_json(
     data: Any,
     *,
@@ -75,6 +109,23 @@ def _emit_json(
     print(dumped)
 
 
+def _explicit_cli_format(
+    args: argparse.Namespace,
+) -> str | None:
+    """Return the explicit CLI format hint when provided."""
+
+    if not getattr(args, '_format_explicit', False):
+        return None
+    for attr in ('format', 'target_format', 'source_format'):
+        value = getattr(args, attr, None)
+        if value is None:
+            continue
+        normalized = value.strip().lower()
+        if normalized:
+            return normalized
+    return None
+
+
 def _infer_payload_format(
     text: str,
 ) -> str:
@@ -97,12 +148,12 @@ def _infer_payload_format(
     return 'csv'
 
 
-def _list_sections(
+def _check_sections(
     cfg: PipelineConfig,
     args: argparse.Namespace,
 ) -> dict[str, Any]:
     """
-    Build sectioned metadata output for the list command.
+    Build sectioned metadata output for the check command.
 
     Parameters
     ----------
@@ -114,9 +165,11 @@ def _list_sections(
     Returns
     -------
     dict[str, Any]
-        Metadata output for the list command.
+        Metadata output for the check command.
     """
     sections: dict[str, Any] = {}
+    if getattr(args, 'jobs', False):
+        sections['jobs'] = _pipeline_summary(cfg)['jobs']
     if getattr(args, 'pipelines', False):
         sections['pipelines'] = [cfg.name]
     if getattr(args, 'sources', False):
@@ -132,23 +185,6 @@ def _list_sections(
     return sections
 
 
-def _explicit_cli_format(
-    args: argparse.Namespace,
-) -> str | None:
-    """Return the explicit CLI format hint when provided."""
-
-    if not getattr(args, '_format_explicit', False):
-        return None
-    for attr in ('format', 'target_format', 'source_format'):
-        value = getattr(args, attr, None)
-        if value is None:
-            continue
-        normalized = value.strip().lower()
-        if normalized:
-            return normalized
-    return None
-
-
 def _materialize_file_payload(
     source: object,
     *,
@@ -222,7 +258,6 @@ def _parse_text_payload(
     JSONData | str
         The parsed payload as JSON data or raw text.
     """
-
     effective = (fmt or '').strip().lower() or _infer_payload_format(text)
     if effective == 'json':
         return cast(JSONData, json_type(text))
@@ -263,7 +298,8 @@ def _pipeline_summary(
 def _presentation_flags(
     args: argparse.Namespace,
 ) -> tuple[bool, bool]:
-    """Return presentation toggles from the parsed namespace.
+    """
+    Return presentation toggles from the parsed namespace.
 
     Parameters
     ----------
@@ -340,7 +376,6 @@ def _resolve_cli_payload(
         Parsed payload or the original source value when hydration is
         disabled.
     """
-
     if isinstance(source, (os.PathLike, str)) and str(source) == '-':
         text = _read_stdin_text()
         return _parse_text_payload(text, format_hint)
@@ -388,11 +423,11 @@ def _write_json_output(
 # SECTION: FUNCTIONS ======================================================== #
 
 
-def cmd_extract(
+def check_handler(
     args: argparse.Namespace,
 ) -> int:
     """
-    Extract data from a source.
+    Print requested pipeline sections from a YAML configuration.
 
     Parameters
     ----------
@@ -404,40 +439,20 @@ def cmd_extract(
     int
         Zero on success.
     """
-    pretty, _ = _presentation_flags(args)
-    explicit_format = _explicit_cli_format(args)
-
-    if args.source == '-':
-        text = _read_stdin_text()
-        payload = _parse_text_payload(text, getattr(args, 'format', None))
-        _emit_json(payload, pretty=pretty)
-
+    cfg = load_pipeline_config(args.config, substitute=True)
+    if getattr(args, 'summary', False):
+        print_json(_pipeline_summary(cfg))
         return 0
 
-    result = extract(
-        args.source_type,
-        args.source,
-        file_format=explicit_format,
-    )
-    output_path = getattr(args, 'target', None)
-    if output_path is None:
-        output_path = getattr(args, 'output', None)
-
-    if not _write_json_output(
-        result,
-        output_path,
-        success_message='Data extracted and saved to',
-    ):
-        _emit_json(result, pretty=pretty)
-
+    print_json(_check_sections(cfg, args))
     return 0
 
 
-def cmd_validate(
+def extract_handler(
     args: argparse.Namespace,
 ) -> int:
     """
-    Validate data from a source.
+    Extract data from a source.
 
     Parameters
     ----------
@@ -449,81 +464,36 @@ def cmd_validate(
     int
         Zero on success.
     """
-    pretty, _quiet = _presentation_flags(args)
-    format_explicit: bool = getattr(args, '_format_explicit', False)
-    format_hint: str | None = getattr(args, 'source_format', None)
-    payload = cast(
-        JSONData | str,
-        _resolve_cli_payload(
-            args.source,
-            format_hint=format_hint,
-            format_explicit=format_explicit,
-        ),
-    )
-    result = validate(payload, args.rules)
-
-    target_path = getattr(args, 'target', None)
-    if target_path:
-        validated_data = result.get('data')
-        if validated_data is not None:
-            _write_json_output(
-                validated_data,
-                target_path,
-                success_message='Validation result saved to',
-            )
-        else:
-            print(
-                f'Validation failed, no data to save for {target_path}',
-                file=sys.stderr,
-            )
-    else:
-        _emit_json(result, pretty=pretty)
-
-    return 0
-
-
-def cmd_transform(
-    args: argparse.Namespace,
-) -> int:
-    """
-    Transform data from a source.
+    pretty, _ = _presentation_flags(args)
+    explicit_format = _explicit_cli_format(args)
 
-    Parameters
-    ----------
-    args : argparse.Namespace
-        Parsed command-line arguments.
+    if args.source == '-':
+        text = _read_stdin_text()
+        payload = _parse_text_payload(text, getattr(args, 'format', None))
+        _emit_json(payload, pretty=pretty)
 
-    Returns
-    -------
-    int
-        Zero on success.
-    """
-    pretty, _quiet = _presentation_flags(args)
-    format_hint: str | None = getattr(args, 'source_format', None)
-    format_explicit: bool = format_hint is not None
+        return 0
 
-    payload = cast(
-        JSONData | str,
-        _resolve_cli_payload(
-            args.source,
-            format_hint=format_hint,
-            format_explicit=format_explicit,
-        ),
+    result = extract(
+        args.source_type,
+        args.source,
+        file_format=explicit_format,
     )
-
-    data = transform(payload, args.operations)
+    output_path = getattr(args, 'target', None)
+    if output_path is None:
+        output_path = getattr(args, 'output', None)
 
     if not _write_json_output(
-        data,
-        getattr(args, 'target', None),
-        success_message='Data transformed and saved to',
+        result,
+        output_path,
+        success_message='Data extracted and saved to',
     ):
-        _emit_json(data, pretty=pretty)
+        _emit_json(result, pretty=pretty)
 
     return 0
 
 
-def cmd_load(
+def load_handler(
     args: argparse.Namespace,
 ) -> int:
     """
@@ -582,7 +552,7 @@ def cmd_load(
     return 0
 
 
-def cmd_pipeline(
+def pipeline_handler(
     args: argparse.Namespace,
 ) -> int:
     """
@@ -598,6 +568,12 @@ def cmd_pipeline(
     int
         Zero on success.
     """
+    print(
+        'DEPRECATED: use "etlplus check --summary|--jobs" or '
+        '"etlplus run --job/--pipeline" instead of "etlplus pipeline".',
+        file=sys.stderr,
+    )
+
     cfg = load_pipeline_config(args.config, substitute=True)
 
     list_flag = getattr(args, 'list', False) or getattr(args, 'jobs', False)
@@ -620,9 +596,71 @@ def cmd_pipeline(
     return 0
 
 
-def cmd_list(args: argparse.Namespace) -> int:
+def render_handler(
+    args: argparse.Namespace,
+) -> int:
+    """Render SQL DDL statements from table schema specs."""
+    _, quiet = _presentation_flags(args)
+
+    template_value = getattr(args, 'template', 'ddl') or 'ddl'
+    template_path = getattr(args, 'template_path', None)
+    table_filter = getattr(args, 'table', None)
+    spec_path = getattr(args, 'spec', None)
+    config_path = getattr(args, 'config', None)
+
+    # If the provided template points to a file, treat it as a path override.
+    file_override = template_path
+    template_key = template_value
+    if template_path is None:
+        candidate_path = Path(template_value)
+        if candidate_path.exists():
+            file_override = str(candidate_path)
+            template_key = None
+
+    specs = _collect_table_specs(config_path, spec_path)
+    if table_filter:
+        specs = [
+            spec
+            for spec in specs
+            if str(spec.get('table')) == table_filter
+            or str(spec.get('name', '')) == table_filter
+        ]
+
+    if not specs:
+        target_desc = table_filter or 'table_schemas'
+        print(
+            'No table schemas found for '
+            f'{target_desc}. Provide --spec or a pipeline --config with '
+            'table_schemas.',
+            file=sys.stderr,
+        )
+        return 1
+
+    rendered_chunks = render_tables(
+        specs,
+        template=template_key,
+        template_path=file_override,
+    )
+    sql_text = (
+        '\n'.join(chunk.rstrip() for chunk in rendered_chunks).rstrip() + '\n'
+    )
+
+    output_path = getattr(args, 'output', None)
+    if output_path and output_path != '-':
+        Path(output_path).write_text(sql_text, encoding='utf-8')
+        if not quiet:
+            print(f'Rendered {len(specs)} schema(s) to {output_path}')
+        return 0
+
+    print(sql_text)
+    return 0
+
+
+def run_handler(
+    args: argparse.Namespace,
+) -> int:
     """
-    Print requested pipeline sections from a YAML configuration.
+    Execute an ETL job end-to-end from a pipeline YAML configuration.
 
     Parameters
     ----------
@@ -635,13 +673,22 @@ def cmd_list(args: argparse.Namespace) -> int:
         Zero on success.
     """
     cfg = load_pipeline_config(args.config, substitute=True)
-    print_json(_list_sections(cfg, args))
+
+    job_name = getattr(args, 'job', None) or getattr(args, 'pipeline', None)
+    if job_name:
+        result = run(job=job_name, config_path=args.config)
+        print_json({'status': 'ok', 'result': result})
+        return 0
+
+    print_json(_pipeline_summary(cfg))
     return 0
 
 
-def cmd_run(args: argparse.Namespace) -> int:
+def transform_handler(
+    args: argparse.Namespace,
+) -> int:
     """
-    Execute an ETL job end-to-end from a pipeline YAML configuration.
+    Transform data from a source.
 
     Parameters
     ----------
@@ -653,13 +700,75 @@ def cmd_run(args: argparse.Namespace) -> int:
     int
         Zero on success.
     """
-    cfg = load_pipeline_config(args.config, substitute=True)
+    pretty, _ = _presentation_flags(args)
+    format_hint: str | None = getattr(args, 'source_format', None)
+    format_explicit: bool = format_hint is not None
 
-    job_name = getattr(args, 'job', None) or getattr(args, 'pipeline', None)
-    if job_name:
-        result = run(job=job_name, config_path=args.config)
-        print_json({'status': 'ok', 'result': result})
-        return 0
+    payload = cast(
+        JSONData | str,
+        _resolve_cli_payload(
+            args.source,
+            format_hint=format_hint,
+            format_explicit=format_explicit,
+        ),
+    )
+
+    data = transform(payload, args.operations)
+
+    if not _write_json_output(
+        data,
+        getattr(args, 'target', None),
+        success_message='Data transformed and saved to',
+    ):
+        _emit_json(data, pretty=pretty)
+
+    return 0
+
+
+def validate_handler(
+    args: argparse.Namespace,
+) -> int:
+    """
+    Validate data from a source.
+
+    Parameters
+    ----------
+    args : argparse.Namespace
+        Parsed command-line arguments.
+
+    Returns
+    -------
+    int
+        Zero on success.
+    """
+    pretty, _ = _presentation_flags(args)
+    format_explicit: bool = getattr(args, '_format_explicit', False)
+    format_hint: str | None = getattr(args, 'source_format', None)
+    payload = cast(
+        JSONData | str,
+        _resolve_cli_payload(
+            args.source,
+            format_hint=format_hint,
+            format_explicit=format_explicit,
+        ),
+    )
+    result = validate(payload, args.rules)
+
+    target_path = getattr(args, 'target', None)
+    if target_path:
+        validated_data = result.get('data')
+        if validated_data is not None:
+            _write_json_output(
+                validated_data,
+                target_path,
+                success_message='Validation result saved to',
+            )
+        else:
+            print(
+                f'Validation failed, no data to save for {target_path}',
+                file=sys.stderr,
+            )
+    else:
+        _emit_json(result, pretty=pretty)
 
-    print_json(_pipeline_summary(cfg))
     return 0