VERY BREAKING: Rename data and operations schemas

collectoss/api/metrics/commit.py

@@ -231,7 +231,7 @@ def annual_commit_count_ranked_by_repo_in_repo_group(repo_group_id, repo_id=None
         if timeframe == 'all':
             cdRgTpRankedCommitsSQL = s.sql.text("""
                 SELECT repo.repo_id, repo_name as name, SUM(added - removed - whitespace) as net, patches
-                FROM augur_data.dm_repo_annual, repo, repo_groups
+                FROM collection_data.dm_repo_annual, repo, repo_groups
                 WHERE repo.repo_group_id = :repo_group_id
                 AND repo.repo_group_id = repo_groups.repo_group_id
                 AND dm_repo_annual.repo_id = repo.repo_id

collectoss/api/metrics/deps.py

@@ -33,13 +33,13 @@ def deps(repo_group_id, repo_id=None, period='day', begin_date=None, end_date=No
 
         depsSQL = s.sql.text("""
             SELECT
-            augur_data.repo_dependencies.*,
-            augur_data.repo_groups.repo_group_id 
+            collection_data.repo_dependencies.*,
+            collection_data.repo_groups.repo_group_id 
             FROM
-            augur_data.repo_dependencies,
-            augur_data.repo_groups,
-            augur_data.repo,
-            ( SELECT MAX ( date_trunc( 'day', augur_data.repo_dependencies.data_collection_date ) ) AS data_collection_date FROM repo_dependencies WHERE repo_id = repo_id ) C 
+            collection_data.repo_dependencies,
+            collection_data.repo_groups,
+            collection_data.repo,
+            ( SELECT MAX ( date_trunc( 'day', collection_data.repo_dependencies.data_collection_date ) ) AS data_collection_date FROM repo_dependencies WHERE repo_id = repo_id ) C 
             WHERE
             repo_dependencies.repo_id = repo.repo_id 
             AND repo.repo_group_id = repo_groups.repo_group_id 
@@ -54,13 +54,13 @@ def deps(repo_group_id, repo_id=None, period='day', begin_date=None, end_date=No
 
         depsSQL = s.sql.text("""
             SELECT
-            augur_data.repo_dependencies.*,
-            augur_data.repo_groups.repo_group_id 
+            collection_data.repo_dependencies.*,
+            collection_data.repo_groups.repo_group_id 
             FROM
-            augur_data.repo_dependencies,
-            augur_data.repo_groups,
-            augur_data.repo,
-            ( SELECT MAX ( date_trunc( 'day', augur_data.repo_dependencies.data_collection_date ) ) AS data_collection_date 
+            collection_data.repo_dependencies,
+            collection_data.repo_groups,
+            collection_data.repo,
+            ( SELECT MAX ( date_trunc( 'day', collection_data.repo_dependencies.data_collection_date ) ) AS data_collection_date 
             FROM repo_dependencies, repo, repo_groups 
             WHERE repo.repo_group_id = repo_groups.repo_group_id and 
             repo_dependencies.repo_id = repo.repo_id and
@@ -134,8 +134,8 @@ def libyear(repo_group_id, repo_id=None, period='day', begin_date=None, end_date
                     f.libyear, 
                     f.data_collection_date
                 FROM
-                    ( SELECT repo_id, NAME, MAX ( data_collection_date ) AS data_collection_date FROM augur_data.repo_deps_libyear WHERE repo_id = :repo_id GROUP BY repo_id, NAME ORDER BY NAME ) e,
-                    augur_data.repo_deps_libyear f 
+                    ( SELECT repo_id, NAME, MAX ( data_collection_date ) AS data_collection_date FROM collection_data.repo_deps_libyear WHERE repo_id = :repo_id GROUP BY repo_id, NAME ORDER BY NAME ) e,
+                    collection_data.repo_deps_libyear f 
                 WHERE
                     e.data_collection_date = f.data_collection_date and 
                     e.repo_id = f.repo_id 
@@ -203,8 +203,8 @@ def libyear(repo_group_id, repo_id=None, period='day', begin_date=None, end_date
                     f.libyear, 
                     f.data_collection_date
                 FROM
-                    ( SELECT repo_id, NAME, MAX ( data_collection_date ) AS data_collection_date FROM augur_data.repo_deps_libyear GROUP BY repo_id, NAME ORDER BY NAME ) e,
-                    augur_data.repo_deps_libyear f 
+                    ( SELECT repo_id, NAME, MAX ( data_collection_date ) AS data_collection_date FROM collection_data.repo_deps_libyear GROUP BY repo_id, NAME ORDER BY NAME ) e,
+                    collection_data.repo_deps_libyear f 
                 WHERE
                     e.data_collection_date = f.data_collection_date and 
                     e.repo_id = f.repo_id 

collectoss/api/metrics/message.py

@@ -40,12 +40,12 @@ def repo_messages(repo_group_id, repo_id=None, period='day', begin_date=None, en
                 COUNT ( * ),
                 repo_name 
             FROM
-                augur_data.repo,
-                augur_data.message 
+                collection_data.repo,
+                collection_data.message 
             WHERE
-                augur_data.repo.repo_id = augur_data.message.repo_id 
+                collection_data.repo.repo_id = collection_data.message.repo_id 
                 AND
-                augur_data.repo.repo_id = :repo_id 
+                collection_data.repo.repo_id = :repo_id 
                 AND 
                 message.msg_timestamp BETWEEN :begin_date AND :end_date
             GROUP BY
@@ -69,14 +69,14 @@ def repo_messages(repo_group_id, repo_id=None, period='day', begin_date=None, en
                 COUNT ( * ),
                 rg_name 
             FROM
-                augur_data.repo,
-                augur_data.repo_groups,
-                augur_data.message 
+                collection_data.repo,
+                collection_data.repo_groups,
+                collection_data.message 
             WHERE
-                augur_data.repo.repo_id = augur_data.message.repo_id 
-                AND augur_data.repo_groups.repo_group_id = repo.repo_group_id 
+                collection_data.repo.repo_id = collection_data.message.repo_id 
+                AND collection_data.repo_groups.repo_group_id = repo.repo_group_id 
                 AND
-                augur_data.repo_groups.repo_group_id = :repo_group_id 
+                collection_data.repo_groups.repo_group_id = :repo_group_id 
                 AND 
                 message.msg_timestamp BETWEEN :begin_date AND :end_date
             GROUP BY

collectoss/api/metrics/pull_request.py

@@ -787,8 +787,8 @@ def pull_request_average_commit_counts(repo_group_id, repo_id=None, group_by='mo
                 pr_merged_at,
                 pr_closed_at,
                 pr_created_at
-            FROM augur_data.pull_request_commits, augur_data.pull_request_meta,augur_data.repo_groups, 
-            augur_data.pull_requests JOIN repo ON pull_requests.repo_id = repo.repo_id
+            FROM collection_data.pull_request_commits, collection_data.pull_request_meta,collection_data.repo_groups, 
+            collection_data.pull_requests JOIN repo ON pull_requests.repo_id = repo.repo_id
             WHERE pull_requests.repo_id IN 
                 (SELECT repo_id FROM repo WHERE repo_group_id = :repo_group_id)
                 AND pull_requests.pull_request_id = pull_request_commits.pull_request_id
@@ -821,7 +821,7 @@ def pull_request_average_commit_counts(repo_group_id, repo_id=None, group_by='mo
             pr_merged_at,
             pr_closed_at,
             pr_created_at
-        FROM augur_data.pull_request_commits, augur_data.pull_requests, augur_data.pull_request_meta
+        FROM collection_data.pull_request_commits, collection_data.pull_requests, collection_data.pull_request_meta
         WHERE pull_requests.pull_request_id = pull_request_commits.pull_request_id
             AND pull_requests.pull_request_id = pull_request_meta.pull_request_id
             AND pull_requests.repo_id = :repo_id

collectoss/api/metrics/repo_meta.py

@@ -190,7 +190,7 @@ def sbom_download(repo_group_id, repo_id=None):
     :return: dosocs sbom
     """
     dosocs_SQL = s.sql.text("""
-        select * from augur_data.repo_sbom_scans
+        select * from collection_data.repo_sbom_scans
         where repo_id = :repo_id;
     """)
 
@@ -313,15 +313,15 @@ def cii_best_practices_badge(repo_group_id, repo_id=None):
     if not repo_id:
         cii_best_practices_badge_SQL = s.sql.text("""
             SELECT data
-            FROM augur_data.repo_badging
+            FROM collection_data.repo_badging
             WHERE repo_id IN (SELECT repo_id FROM repo WHERE repo_group_id = :repo_group_id)
             ORDER BY created_at DESC
             LIMIT 1
         """)
     else:
         cii_best_practices_badge_SQL = s.sql.text("""
             SELECT data
-            FROM augur_data.repo_badging
+            FROM collection_data.repo_badging
             WHERE repo_id = :repo_id
             ORDER BY created_at DESC
             LIMIT 1
@@ -1270,7 +1270,7 @@ def clones(repo_group_id, repo_id=None, begin_date=None, end_date=None):
                 clone_data_timestamp AS date,
                 count_clones AS total_clones,
                 unique_clones
-            FROM augur_data.repo_clones_data
+            FROM collection_data.repo_clones_data
             WHERE repo_id = :repo_id
               AND clone_data_timestamp BETWEEN :begin_date AND :end_date
             ORDER BY clone_data_timestamp
@@ -1289,9 +1289,9 @@ def clones(repo_group_id, repo_id=None, begin_date=None, end_date=None):
                 clone_data_timestamp AS date,
                 count_clones AS total_clones,
                 unique_clones
-            FROM augur_data.repo_clones_data
+            FROM collection_data.repo_clones_data
             WHERE repo_id IN (
-                SELECT repo_id FROM augur_data.repo WHERE repo_group_id = :repo_group_id
+                SELECT repo_id FROM collection_data.repo WHERE repo_group_id = :repo_group_id
             )
               AND clone_data_timestamp BETWEEN :begin_date AND :end_date
             ORDER BY repo_id, clone_data_timestamp

collectoss/api/metrics/toss.py

@@ -114,7 +114,7 @@ def toss_repo_info(repo_id):
         repo_info.default_branch,
         repo.repo_git
     FROM
-        augur_data.repo_info
+        collection_data.repo_info
         JOIN repo ON repo.repo_id = repo_info.repo_id
     WHERE
         repo_info.repo_id = :repo_id

collectoss/api/routes/collection_status.py

@@ -61,10 +61,10 @@ def issue_collection_status():  # TODO: make this name automatic - wrapper?
                     (
                     CAST (( COUNT ( * )) +1 AS DOUBLE PRECISION )  / CAST ( b.issues_count + 1 AS DOUBLE PRECISION )) AS ratio_issues
                 FROM
-                    augur_data.repo A,
-                    augur_data.issues d,
-                    augur_data.repo_info b,
-                    ( SELECT repo_id, MAX ( data_collection_date ) AS last_collected FROM augur_data.repo_info GROUP BY repo_id ORDER BY repo_id ) e,
+                    collection_data.repo A,
+                    collection_data.issues d,
+                    collection_data.repo_info b,
+                    ( SELECT repo_id, MAX ( data_collection_date ) AS last_collected FROM collection_data.repo_info GROUP BY repo_id ORDER BY repo_id ) e,
                     ( SELECT repo_id, MAX ( data_collection_date ) AS most_recently_collected_issue FROM issues GROUP BY repo_id ORDER BY repo_id ) f
                 WHERE
                     A.repo_id = b.repo_id
@@ -135,11 +135,11 @@ def pull_request_collection_status():  # TODO: make this name automatic - wrappe
                 ABS ( CAST ( ( COUNT ( * ) ) + 1 AS DOUBLE PRECISION ) / CAST ( b.pull_request_count + 1 AS DOUBLE PRECISION ) ) AS ratio_abs,
                 ( CAST ( ( COUNT ( * ) ) + 1 AS DOUBLE PRECISION ) / CAST ( b.pull_request_count + 1 AS DOUBLE PRECISION ) ) AS ratio_issues
             FROM
-                augur_data.repo A,
-                augur_data.pull_requests d,
-                augur_data.repo_info b,
-                ( SELECT repo_id, MAX ( data_collection_date ) AS last_collected FROM augur_data.repo_info GROUP BY repo_id ORDER BY repo_id ) e,
-                ( SELECT repo_id, MAX ( data_collection_date ) AS last_pr_collected FROM augur_data.pull_requests GROUP BY repo_id ORDER BY repo_id ) f
+                collection_data.repo A,
+                collection_data.pull_requests d,
+                collection_data.repo_info b,
+                ( SELECT repo_id, MAX ( data_collection_date ) AS last_collected FROM collection_data.repo_info GROUP BY repo_id ORDER BY repo_id ) e,
+                ( SELECT repo_id, MAX ( data_collection_date ) AS last_pr_collected FROM collection_data.pull_requests GROUP BY repo_id ORDER BY repo_id ) f
             WHERE
                 A.repo_id = b.repo_id
                 AND LOWER ( A.repo_git ) LIKE'%github.com%'