add type annotations to most of the codeHEADmaster

1 files changed, 27 insertions, 6 deletions

diff --git a/update_sharing.py b/update_sharing.py
index 78e6171..3a86268 100755
--- a/update_sharing.py
+++ b/update_sharing.py
@@ -2,10 +2,17 @@
 
 import argparse
 import sqlite3
+import typing
 
 from dedup.utils import fetchiter
 
-def add_values(cursor, insert_key, files, size):
+
+def add_values(
+    cursor: sqlite3.Cursor,
+    insert_key: typing.Tuple[int, int, int, int],
+    files: int,
+    size: int,
+) -> None:
     cursor.execute("UPDATE sharing SET files = files + ?, size = size + ? WHERE pid1 = ? AND pid2 = ? AND fid1 = ? AND fid2 = ?;",
                    (files, size) + insert_key)
     if cursor.rowcount > 0:
@@ -13,14 +20,25 @@ def add_values(cursor, insert_key, files, size):
     cursor.execute("INSERT INTO sharing (pid1, pid2, fid1, fid2, files, size) VALUES (?, ?, ?, ?, ?, ?);",
                    insert_key + (files, size))
 
-def compute_pkgdict(rows):
-    pkgdict = dict()
+
+def compute_pkgdict(
+    rows: typing.Iterable[typing.Tuple[int, typing.Any, str, int, int]]
+) -> typing.Dict[int, typing.Dict[int, typing.List[typing.Tuple[int, str]]]]:
+    pkgdict: typing.Dict[
+        int, typing.Dict[int, typing.List[typing.Tuple[int, str]]]
+    ] = {}
     for pid, _, filename, size, fid in rows:
         funcdict = pkgdict.setdefault(pid, {})
         funcdict.setdefault(fid, []).append((size, filename))
     return pkgdict
 
-def process_pkgdict(cursor, pkgdict):
+
+def process_pkgdict(
+    cursor: sqlite3.Cursor,
+    pkgdict: typing.Dict[
+        int, typing.Dict[int, typing.List[typing.Tuple[int, str]]]
+    ],
+) -> None:
     for pid1, funcdict1 in pkgdict.items():
         for fid1, files in funcdict1.items():
             numfiles = len(files)
@@ -38,7 +56,8 @@ def process_pkgdict(cursor, pkgdict):
                     insert_key = (pid1, pid2, fid1, fid2)
                     add_values(cursor, insert_key, pkgnumfiles, pkgsize)
 
-def main(db):
+
+def main(db: sqlite3.Connection) -> None:
     cur = db.cursor()
     cur.execute("PRAGMA foreign_keys = ON;")
     cur.execute("DELETE FROM sharing;")
@@ -49,7 +68,9 @@ def main(db):
     for hashvalue, in fetchiter(readcur):
         cur.execute("SELECT function.eqclass, content.pid, content.id, content.filename, content.size, hash.fid FROM hash JOIN content ON hash.cid = content.id JOIN function ON hash.fid = function.id AND function.eqclass IS NOT NULL WHERE hash = ?;",
                     (hashvalue,))
-        rowdict = dict()
+        rowdict: typing.Dict[
+            int, typing.List[typing.Tuple[int, int, str, int, int]]
+        ] = {}
         for row in cur.fetchall():
             rowdict.setdefault(row[0], []).append(row[1:])
         for eqclass, rows in rowdict.items():