From b31fbe34977c39dd6c4b6cf340b10819269d58a5 Mon Sep 17 00:00:00 2001
From: Anna Scholtz <anna@scholtzan.net>
Date: Fri, 17 Jan 2020 15:46:47 -0800
Subject: [PATCH] Metadata publish improvements and update clients_daily_v6
 metadata

---
 script/generate_sql                           |  9 -----
 script/publish_metadata                       | 36 +++++++++++++++----
 .../clients_daily_v6/metadata.yaml            | 11 ------
 .../clients_daily_v6/metadata.yaml            |  8 ++---
 4 files changed, 32 insertions(+), 32 deletions(-)
 delete mode 100644 sql/telemetry_derived/clients_daily_v6/metadata.yaml

diff --git a/script/generate_sql b/script/generate_sql
index e5d1262926..89f7bd658c 100755
--- a/script/generate_sql
+++ b/script/generate_sql
@@ -24,9 +24,6 @@ from bigquery_etl.parse_udf import (
 )  # noqa E402
 
 
-METADATA_FILE = "metadata.yaml"
-
-
 parser = ArgumentParser(description=__doc__)
 parser.add_argument(
     "--destination",
@@ -69,12 +66,6 @@ def main():
                 with open(os.path.join(basename, filename), "a+") as output_file:
                     output_file.write(full_text)
 
-            if METADATA_FILE in files:
-                shutil.copyfile(
-                    os.path.join(root, METADATA_FILE),
-                    os.path.join(basename, METADATA_FILE),
-                )
-
         if os.path.exists(args.destination):
             shutil.rmtree(args.destination, ignore_errors=True)
         shutil.copytree(d, args.destination)
diff --git a/script/publish_metadata b/script/publish_metadata
index c7a528d16b..a415deef7c 100755
--- a/script/publish_metadata
+++ b/script/publish_metadata
@@ -1,11 +1,12 @@
 #!/usr/bin/env python3
 
-"""Update metadata of BigQuery tables."""
+"""Update metadata of BigQuery tables and views."""
 
 from argparse import ArgumentParser
 import logging
 import os
 import yaml
+import re
 
 from google.cloud import bigquery
 
@@ -13,6 +14,15 @@ from google.cloud import bigquery
 METADATA_FILE = "metadata.yaml"
 
 
+def is_valid_label(label):
+    """
+    Check if a label has the right format:
+    Only hyphens (-), underscores (_), lowercase characters, and numbers are allowed.
+    International characters are allowed.
+    """
+    return re.match(r"[\w\d_-]+", label)
+
+
 def publish_metadata(client, dataset, table, metadata_file):
     with open(metadata_file, "r") as yaml_stream:
         try:
@@ -27,10 +37,21 @@ def publish_metadata(client, dataset, table, metadata_file):
                 table.description = metadata["description"]
 
             if "labels" in metadata:
-                # only string type and lowercased labels are allowed
-                table.labels = {
-                    k: str(v).lower() for k, v in metadata["labels"].items()
-                }
+                table.labels = {}
+
+                for key, label in metadata["labels"].items():
+                    if is_valid_label(str(key)) and is_valid_label(str(label)):
+                        table.labels[str(key)] = str(label)
+                    else:
+                        print(
+                            """
+                            Invalid label format: {}: {}. Only hyphens (-),
+                            underscores (_), lowercase characters, and numbers
+                            are allowed. International characters are allowed.
+                            """.format(
+                                key, label
+                            )
+                        )
 
             client.update_table(table, ["friendly_name", "description", "labels"])
         except yaml.YAMLError as e:
@@ -59,8 +80,9 @@ def main():
         if os.path.isdir(target):
             for root, dirs, files in os.walk(target):
                 if METADATA_FILE in files:
-                    dataset = root.split("/")[-2]
-                    table = root.split("/")[-1]
+                    path = os.path.normpath(root)
+                    dataset = path.split(os.sep)[-2]
+                    table = path.split(os.sep)[-1]
                     metadata_file = os.path.join(root, METADATA_FILE)
 
                     publish_metadata(client, dataset, table, metadata_file)
diff --git a/sql/telemetry_derived/clients_daily_v6/metadata.yaml b/sql/telemetry_derived/clients_daily_v6/metadata.yaml
deleted file mode 100644
index f406fc97b3..0000000000
--- a/sql/telemetry_derived/clients_daily_v6/metadata.yaml
+++ /dev/null
@@ -1,11 +0,0 @@
-friendly_name: Clients Daily
-description: >
-  A daily aggregate of each Firefox client,
-  partitioned by day
-owners:
-  - relud@mozilla.com
-labels:
-  application: firefox
-  refresh: daily
-  version: 6
-  incremental: true
diff --git a/templates/telemetry_derived/clients_daily_v6/metadata.yaml b/templates/telemetry_derived/clients_daily_v6/metadata.yaml
index f406fc97b3..c5835b42a1 100644
--- a/templates/telemetry_derived/clients_daily_v6/metadata.yaml
+++ b/templates/telemetry_derived/clients_daily_v6/metadata.yaml
@@ -1,11 +1,9 @@
 friendly_name: Clients Daily
 description: >
-  A daily aggregate of each Firefox client,
+  A daily aggregate of main pings from each Firefox desktop client,
   partitioned by day
 owners:
-  - relud@mozilla.com
+  - dthorn@mozilla.com
 labels:
   application: firefox
-  refresh: daily
-  version: 6
-  incremental: true
+  schedule: daily