From e436e27d2d5a8f1e060ce7d1c711ebdda9bb0740 Mon Sep 17 00:00:00 2001
From: Pat Nadolny <patnadolny@gmail.com>
Date: Tue, 29 Aug 2023 21:09:15 -0400
Subject: [PATCH] all sdk option for download metadata (#100)

* all sdk option for download metadata

* remove extension

* lint fixes
---
 README.md                 |  1 +
 hub_utils/main.py         | 39 ++++++++++-----------------------------
 hub_utils/meltano_util.py |  6 +++++-
 hub_utils/utilities.py    | 34 ++++++++++++++++++++++++++++++++++
 4 files changed, 50 insertions(+), 30 deletions(-)

diff --git a/README.md b/README.md
index 623febf..fc48b29 100644
--- a/README.md
+++ b/README.md
@@ -98,6 +98,7 @@ $ hub-utils download-metadata [OPTIONS] LOCAL_PATH
 **Options**:
 
 * `--variant-path-list TEXT`
+* `--all-sdk / --no-all-sdk`: [default: no-all-sdk]
 * `--help`: Show this message and exit.
 
 ## `hub-utils extract-sdk-metadata-to-s3`
diff --git a/hub_utils/main.py b/hub_utils/main.py
index 128bc07..05b8421 100644
--- a/hub_utils/main.py
+++ b/hub_utils/main.py
@@ -249,36 +249,9 @@ def get_variant_names(
     Generate a list of variant names for a given set of filters.
     The list will be formatted as escaped JSON to be used by Github Actions.
     """
-    formatted_output = []
     util = Utilities(True)
-    for yaml_file in find_all_yamls(f_path=f"{hub_root}/_data/meltano/"):
-        data = util._read_yaml(yaml_file)
-        if plugin_type and yaml_file.split("/")[-3] not in plugin_type.split(","):
-            continue
-
-        if metadata_type == "sdk":
-            if "meltano_sdk" not in data.get("keywords", []):
-                continue
-            suffix = "/".join(yaml_file.split("/")[-3:])
-            formatted_output.append({"plugin-name": suffix})
-
-        if metadata_type == "airbyte":
-            if "airbyte_protocol" not in data.get("keywords", []):
-                continue
-            suffix = "/".join(yaml_file.split("/")[-3:])
-            image_name = [
-                setting.get("value")
-                for setting in data.get("settings")
-                if setting.get("name") == "airbyte_spec.image"
-            ][0]
-            if not image_name:
-                continue
-            formatted_output.append(
-                {
-                    "plugin-name": suffix,
-                    "source-name": image_name.replace("airbyte/", ""),
-                }
-            )
+    util.hub_root = hub_root
+    formatted_output = util.get_variant_names(plugin_type, metadata_type)
     print(json.dumps(formatted_output).replace('"', '\\"'))
 
 
@@ -358,6 +331,7 @@ def upload_airbyte(
 def download_metadata(
     local_path: str,
     variant_path_list: str = None,
+    all_sdk: bool = False,
 ):
     """
     NOTE: USED FOR
@@ -368,6 +342,13 @@ def download_metadata(
     s3 = S3()
     if not variant_path_list:
         variant_path_list = ",".join(SDK_SUFFIX_LIST)
+    if all_sdk:
+        variant_path_list = ",".join(
+            [
+                i["plugin-name"].split(".yml")[0]
+                for i in util.get_variant_names(None, "sdk")
+            ]
+        )
     for yaml_file in variant_path_list.split(","):
         suffix = util.get_suffix(yaml_file)
         local_file_path = f"{local_path}/{suffix}.json"
diff --git a/hub_utils/meltano_util.py b/hub_utils/meltano_util.py
index e3ae468..6269d55 100644
--- a/hub_utils/meltano_util.py
+++ b/hub_utils/meltano_util.py
@@ -367,7 +367,11 @@ def _split_sentence_endings(word_list):
                 desc_list_clean.append(word)
                 continue
             elif len(word.split(".")) > 1 and not word.startswith("."):
-                if word.split(".")[0][-1].isnumeric() and word.split(".")[1] and word.split(".")[1][0].isnumeric():
+                if (
+                    word.split(".")[0][-1].isnumeric()
+                    and word.split(".")[1]
+                    and word.split(".")[1][0].isnumeric()
+                ):
                     # its numeric
                     desc_list_clean.append(word)
                     continue
diff --git a/hub_utils/utilities.py b/hub_utils/utilities.py
index 8b0e0db..4c99acd 100644
--- a/hub_utils/utilities.py
+++ b/hub_utils/utilities.py
@@ -64,6 +64,40 @@ def __init__(self, auto_accept=False):
         self.default_variants_path = f"{self.hub_root}/_data/default_variants.yml"
         self.maintainers_path = f"{self.hub_root}/_data/maintainers.yml"
 
+    def get_variant_names(self, plugin_type, metadata_type):
+        from hub_utils.yaml_lint import find_all_yamls
+
+        formatted_output = []
+        for yaml_file in find_all_yamls(f_path=f"{self.hub_root}/_data/meltano/"):
+            data = self._read_yaml(yaml_file)
+            if plugin_type and yaml_file.split("/")[-3] not in plugin_type.split(","):
+                continue
+
+            if metadata_type == "sdk":
+                if "meltano_sdk" not in data.get("keywords", []):
+                    continue
+                suffix = "/".join(yaml_file.split("/")[-3:])
+                formatted_output.append({"plugin-name": suffix})
+
+            if metadata_type == "airbyte":
+                if "airbyte_protocol" not in data.get("keywords", []):
+                    continue
+                suffix = "/".join(yaml_file.split("/")[-3:])
+                image_name = [
+                    setting.get("value")
+                    for setting in data.get("settings")
+                    if setting.get("name") == "airbyte_spec.image"
+                ][0]
+                if not image_name:
+                    continue
+                formatted_output.append(
+                    {
+                        "plugin-name": suffix,
+                        "source-name": image_name.replace("airbyte/", ""),
+                    }
+                )
+        return formatted_output
+
     def _prompt(self, question, default_val=None, type=None):
         if self.auto_accept:
             return default_val