etlplus 0.9.0__tar.gz → 0.9.1__tar.gz

{etlplus-0.9.0/etlplus.egg-info → etlplus-0.9.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: etlplus
-Version: 0.9.0
+Version: 0.9.1
 Summary: A Swiss Army knife for simple ETL operations
 Home-page: https://github.com/Dagitali/ETLPlus
 Author: ETLPlus Team

{etlplus-0.9.0 → etlplus-0.9.1}/etlplus/cli/commands.py

@@ -443,9 +443,9 @@ def extract_cmd(
         Source (JSON payload, file/folder path, URL/URI, or - for STDIN)
         from which to extract data. Default is ``-``.
     source_format : SourceFormatOption, optional
-        Source data format. Overrides the inferred format (``csv``, ``json``,
-        ``parquet``, ``xml``) based on filename extension or STDIN content.
-        Default is ``None``.
+        Data source format. Overrides the inferred format (``csv``, ``json``,
+        etc.) based on filename extension or STDIN content. Default is
+        ``None``.
     source_type : SourceTypeOption, optional
         Data source type. Overrides the inferred type (``api``, ``database``,
         ``file``, ``folder``) based on URI/URL schema. Default is ``None``.
@@ -523,15 +523,15 @@ def load_cmd(
     ctx : typer.Context
         The Typer context.
     source_format : SourceFormatOption, optional
-        Source data format. Overrides the inferred format (``csv``, ``json``,
-        ``parquet``, ``xml``) based on STDIN content. Default is ``None``.
+        Data source format. Overrides the inferred format (``csv``, ``json``,
+        etc.) based on filename extension or STDIN content. Default is
+        ``None``.
     target : TargetArg, optional
         Target (file/folder path, URL/URI, or - for STDOUT) into which to load
         data. Default is ``-``.
     target_format : TargetFormatOption, optional
-        Format of the target data. Overrides the inferred format (``csv``,
-        ``json``, ``parquet``, ``xml``) based on filename extension. Default is
-        ``None``.
+        Target data format. Overrides the inferred format (``csv``, ``json``,
+        etc.) based on filename extension. Default is ``None``.
     target_type : TargetTypeOption, optional
         Data target type. Overrides the inferred type (``api``, ``database``,
         ``file``, ``folder``) based on URI/URL schema. Default is ``None``.
@@ -760,9 +760,9 @@ def transform_cmd(
         Source (JSON payload, file/folder path, URL/URI, or - for STDIN) from
         which to extract data. Default is ``-``.
     source_format : SourceFormatOption, optional
-        Source data format. Overrides the inferred format (``csv``, ``json``,
-        ``parquet``, ``xml``) based on filename extension or STDIN content.
-        Default is ``None``.
+        Data source format. Overrides the inferred format (``csv``, ``json``,
+        etc.) based on filename extension or STDIN content. Default is
+        ``None``.
     source_type : SourceTypeOption, optional
         Data source type. Overrides the inferred type (``api``, ``database``,
         ``file``, ``folder``) based on URI/URL schema. Default is ``None``.
@@ -770,9 +770,8 @@ def transform_cmd(
         Target (file/folder path, URL/URI, or - for STDOUT) into which to load
         data. Default is ``-``.
     target_format : TargetFormatOption, optional
-        Format of the target data. Overrides the inferred format (``csv``,
-        ``json``, ``parquet``, ``xml``) based on filename extension. Default is
-        ``None``.
+        Target data format. Overrides the inferred format (``csv``, ``json``,
+        etc.) based on filename extension. Default is ``None``.
     target_type : TargetTypeOption, optional
         Data target type. Overrides the inferred type (``api``, ``database``,
         ``file``, ``folder``) based on URI/URL schema. Default is ``None``.
@@ -876,11 +875,12 @@ def validate_cmd(
     source : SourceArg
         Data source to validate (path, JSON payload, or - for STDIN).
     source_format : SourceFormatOption, optional
-        Format of the source. Overrides filename-based inference when provided.
-        Default is ``None``.
-    source_type : SourceTypeOption, optional
-        Override the inferred source type (file, database, api). Default is
+        Data source format. Overrides the inferred format (``csv``, ``json``,
+        etc.) based on filename extension or STDIN content. Default is
         ``None``.
+    source_type : SourceTypeOption, optional
+        Data source type. Overrides the inferred type (``api``, ``database``,
+        ``file``, ``folder``) based on URI/URL schema. Default is ``None``.
     output : OutputOption, optional
         Output file for validated output (- for STDOUT). Default is ``None``.
 

{etlplus-0.9.0 → etlplus-0.9.1}/etlplus/enums.py

@@ -208,8 +208,19 @@ class FileFormat(CoercibleStrEnum):
 
     # -- Constants -- #
 
+    AVRO = 'avro'
     CSV = 'csv'
+    FEATHER = 'feather'
+    GZ = 'gz'
     JSON = 'json'
+    NDJSON = 'ndjson'
+    ORC = 'orc'
+    PARQUET = 'parquet'
+    TSV = 'tsv'
+    TXT = 'txt'
+    XLS = 'xls'
+    XLSX = 'xlsx'
+    ZIP = 'zip'
     XML = 'xml'
     YAML = 'yaml'
 
@@ -227,11 +238,50 @@ class FileFormat(CoercibleStrEnum):
         """
         return {
             # Common shorthand
+            'parq': 'parquet',
             'yml': 'yaml',
+            # File extensions
+            '.avro': 'avro',
+            '.csv': 'csv',
+            '.feather': 'feather',
+            '.gz': 'gz',
+            '.json': 'json',
+            '.jsonl': 'ndjson',
+            '.ndjson': 'ndjson',
+            '.orc': 'orc',
+            '.parquet': 'parquet',
+            '.pq': 'parquet',
+            '.tsv': 'tsv',
+            '.txt': 'txt',
+            '.xls': 'xls',
+            '.xlsx': 'xlsx',
+            '.zip': 'zip',
+            '.xml': 'xml',
+            '.yaml': 'yaml',
+            '.yml': 'yaml',
             # MIME types
-            'text/csv': 'csv',
+            'application/avro': 'avro',
+            'application/feather': 'feather',
+            'application/gzip': 'gz',
             'application/json': 'json',
+            'application/jsonlines': 'ndjson',
+            'application/ndjson': 'ndjson',
+            'application/orc': 'orc',
+            'application/vnd.apache.arrow.file': 'feather',
+            'application/vnd.apache.orc': 'orc',
+            'application/vnd.ms-excel': 'xls',
+            (
+                'application/vnd.openxmlformats-'
+                'officedocument.spreadsheetml.sheet'
+            ): 'xlsx',
+            'application/x-avro': 'avro',
+            'application/x-ndjson': 'ndjson',
+            'application/x-parquet': 'parquet',
             'application/xml': 'xml',
+            'application/zip': 'zip',
+            'text/csv': 'csv',
+            'text/plain': 'txt',
+            'text/tab-separated-values': 'tsv',
         }
 
 

{etlplus-0.9.0 → etlplus-0.9.1/etlplus.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: etlplus
-Version: 0.9.0
+Version: 0.9.1
 Summary: A Swiss Army knife for simple ETL operations
 Home-page: https://github.com/Dagitali/ETLPlus
 Author: ETLPlus Team