hist

store_history.py

@@ -3,14 +3,15 @@ import sqlite3
 import getopt
 import pygit2
 import sys
+import time
 
 if len(sys.argv) != 5:
     print("usage: store_history.py -i repo_file -d database.db")
     sys.exit(1)
 
 optlist, args = getopt.getopt(sys.argv[1:], 'i:d:')
-repopath = ""
-dbpath = ""
+repopath = "."
+dbpath = "example.db"
 
 
 for o, a in optlist:
@@ -20,8 +21,8 @@ for o, a in optlist:
         dbpath = a
 
 
-print(repopath)
-print(dbpath)
+print("repo path:", repopath)
+print("databse path:", dbpath)
 
 repo = pygit2.Repository(repopath)
 
@@ -38,12 +39,14 @@ CREATE TABLE IF NOT EXISTS commits(
 init_query2 = """
 CREATE TABLE IF NOT EXISTS deltas(
     id INTEGER PRIMARY KEY,
-    sha TEXT,
+    from_sha TEXT,
+    to_sha TEXT,
     new_file_name TEXT,
     addition_lines INTEGER,
     deletion_lines INTEGER,
     type TEXT,
-    FOREIGN KEY (sha) REFERENCES commits(sha)
+    FOREIGN KEY (from_sha) REFERENCES commits(sha),
+    FOREIGN KEY (to_sha) REFERENCES commits(sha)
 );
 """
 
@@ -62,9 +65,9 @@ insert_query = """
 """
 
 insert_delta_query = """
-    INSERT INTO deltas(sha, new_file_name, addition_lines, deletion_lines, type)
+    INSERT INTO deltas(from_sha, to_sha, new_file_name, addition_lines, deletion_lines, type)
     VALUES (
-        ?, ?, ?, ?, ? 
+        ?, ?, ?, ?, ?, ? 
     );
 """
 
@@ -82,34 +85,55 @@ data = []
 deltas = []
 
 last_commit = None
-for commit in repo.walk(repo.head.target, pygit2.GIT_SORT_REVERSE): 
+first_time = None
+last_time = 0
+for commit in repo.walk(repo.head.target, pygit2.GIT_SORT_TIME): 
+    time_t = commit.commit_time
+
+
     sha = str(commit.id)
     author_name = commit.author.name
     author_email = commit.author.email
-    time_t = commit.commit_time
-    message = commit.message
-    data.append((sha, author_name, author_email, time_t, message))
+    print("Found commit", sha, "for", time.asctime(time.gmtime(time_t)))
 
-    if last_commit is not None:
-        diff = repo.diff(last_commit, commit)
+    if first_time is None:
+        first_time = time_t
 
-        for patch in diff:
-            file_name = str(patch.delta.new_file.path)
-            status_char = patch.delta.status_char()
-            _, additions, deletions = patch.line_stats
-            deltas.append(
-                    (sha, file_name, additions, deletions, status_char)
-                    )
+    # stop after one year
+    if (abs(first_time - time_t) > 31536000):
+        break
 
-    last_commit = commit
+    message = commit.message
+    data.append((sha, author_name, author_email, time_t, message))
 
-    if (len(data) > 1000):
+    if (abs(last_time - time_t) > 24 * 60 * 60):
+        last_time = time_t
+        if last_commit is not None:
+            print("Finding deltas")
+            diff = repo.diff(last_commit, commit)
+
+            for patch in diff:
+                file_name = str(patch.delta.new_file.path)
+                status_char = patch.delta.status_char()
+                _, additions, deletions = patch.line_stats
+                deltas.append(
+                        (str(last_commit.id), sha, file_name, additions, deletions, status_char)
+                        )
+            last_commit = commit
+
+    if (last_commit is None):
+        last_commit = commit
+
+    if (len(data) + len(deltas)  >= 1000):
+        print("Storing", len(deltas) + len(data), "rows.")
         cursor.executemany(insert_query, data)
         cursor.executemany(insert_delta_query, deltas)
         con.commit()
         data = []
+        deltas = []
 
 
+print("Storing", len(deltas) + len(data), "rows.")
 cursor.executemany(insert_delta_query, deltas)
 cursor.executemany(insert_query, data)
 con.commit()