Merge pull request #1 from UBC-MDS/hackathon-end-to-end

Hackathon Day One - End to end

Author: rzitomer

.gitignore

@@ -1,3 +1,4 @@
 *ipynb_checkpoints*
 *DS_Store*
 *Rhistory*
+*.json

imgs/.gitignore

@@ -0,0 +1,40 @@
+import pandas_gbq
+import networkx as nx
+import matplotlib.pyplot as plt
+import time
+
+def query_ght(queryString):
+    # https://bigquery.cloud.google.com/dataset/ghtorrent-bq:ght
+    query_result_df = pandas_gbq.read_gbq(queryString)
+
+    return query_result_df
+
+def plot_commits(commits):
+    source_target_commits = commits[["cp_parent_id", "c_id"]].dropna().astype("int64")
+    source_target_commits.columns = ["source", "target"]
+
+    g = nx.from_pandas_edgelist(source_target_commits)
+    nx.draw_kamada_kawai(g, alpha=0.5, node_color='blue', node_size = 2)
+
+if __name__ == '__main__':
+    commitQuery = """
+        select
+          c.id as c_id,
+          p.id as p_id,
+          cp.commit_id as cp_commit_id,
+          cp.parent_id as cp_parent_id
+        from `ghtorrent-bq.ght.commits` c
+        left join `ghtorrent-bq.ght.projects` p on (p.id = c.project_id)
+        left join `ghtorrent-bq.ght.commit_parents` cp on (cp.commit_id = c.id)
+        where (p.id = 5524547)
+        limit 10000
+    """
+    
+    start = time.time()
+    commits = query_ght(commitQuery)
+    getData = time.time()
+    print("Query Time:\t" + str(getData - start))
+    plot_commits(commits)
+    plotTime = time.time()
+    print("Plot Time:\t" + str(plotTime - getData))
+    plt.savefig("./imgs/branch_test")