Db: move to postgresql

src/app.py

@@ -5,8 +5,6 @@
 
 @app.route('/files', methods=['GET'])
 def files():
-   # FIXME: don't do this every time
-   db.connect("./data/db.sqlite3")
    tags = request.args.get("tags")
    if tags is None: 
       tags = []

src/db.py

@@ -1,6 +1,5 @@
 from dataclasses import dataclass
-import os
-import sqlite3
+import psycopg2
 
 
 @dataclass
@@ -14,90 +13,72 @@ def list_files(tags: list[str]) -> list[File]:
     """
     List files in database that have the tags.
     """
+    cur = _conn.cursor()
     if tags:
-        qs = ",".join(['?' for _ in tags])
-
         query = f"""
-            select f.name, f.link, group_concat(t.name) as tags
+            select f.name, f.link, array_agg(t.name) as tags
             from (
                 select ft.fid
                 from file_tags ft
-                where ft.tid in (select tid from tags where name in ({qs}))
+                where ft.tid in (select tid from tags where name = ANY(%s))
                 group by ft.fid
-                having count(ft.tid) = (?)
+                having count(ft.tid) = (%s)
             ) as f1
             join files f on f.fid = f1.fid
             join file_tags ft on f.fid = ft.fid
             join tags t on ft.tid = t.tid
-            group by f1.fid;
+            group by f.fid, f.name, f.link;
         """
-        # do not mutate the list
-        params: list[str | int] = tags.copy()
-        params.append(len(tags))
-        _cur.execute(query, params)
+        cur.execute(query, [tags, len(tags)])
     else:
         query = f"""
-            select f.name, f.link, group_concat(t.name) as tags
+            select f.name, f.link, array_agg(t.name) as tags
             from files f
             join file_tags ft on f.fid = ft.fid
             join tags t on ft.tid = t.tid
             group by f.fid;
         """
-        _cur.execute(query)
+        cur.execute(query)
 
-    results = _cur.fetchall()
-    return [File(name, link, tags.split(",")) for name, link, tags in results]
+    results = cur.fetchall()
+    return [File(name, link, tags) for name, link, tags in results]
 
 
 def add_files(files: list[File]):
     """
     Add files to database.
     """
+    cur = _conn.cursor()
 
     for file in files:
-        query1 = "INSERT INTO files(name, link) VALUES (?,?) RETURNING fid"
-        _cur.execute(query1, [file.name, file.link])
-        fid = _cur.fetchone()[0]
+        query1 = "INSERT INTO files(name, link) VALUES (%s, %s) RETURNING fid"
+        cur.execute(query1, [file.name, file.link])
+        fid = cur.fetchone()[0]
 
         for tag in file.tags:
-            tid = get_tagid_or_insert(tag)
-            query2 = "INSERT INTO file_tags(tid,fid) VALUES (?,?)"
-            _cur.execute(query2, [tid, fid])
-
+            tid = get_tagid_or_insert(cur, tag)
+            query2 = "INSERT INTO file_tags(tid,fid) VALUES (%s, %s)"
+            cur.execute(query2, [tid, fid])
 
     _conn.commit()
-
-    pass
 
-def get_tagid_or_insert(tag:str) -> int:
-    query = "SELECT tid FROM tags WHERE name = ?"
-    _cur.execute(query, [tag])
-    tid = _cur.fetchone()
+def get_tagid_or_insert(cur, tag: str) -> int:
+    """
+    Caller needs to commit to the database.
+    """
+    query = "SELECT tid FROM tags WHERE name = %s"
+    cur.execute(query, [tag])
+    tid = cur.fetchone()
     if tid is None:
-        query = "INSERT INTO tags(name) VALUES (?) RETURNING tid"
-        _cur.execute(query, [tag])
-        tid = _cur.fetchone()[0]
+        query = "INSERT INTO tags(name) VALUES (%s) RETURNING tid"
+        cur.execute(query, [tag])
+        tid = cur.fetchone()[0]
     else:
         tid = tid[0]
 
-    _conn.commit()  
-
     return tid  
 
 
 
-_conn: sqlite3.Connection
-_cur: sqlite3.Cursor
-
-
-def connect(db_path: str):
-    global _conn, _cur
-
-    self_path = os.path.dirname(os.path.realpath(__file__))
-    schema = os.path.join(self_path, "schema.sql")
-
-    _conn = sqlite3.connect(db_path)
-    _cur = _conn.cursor()
-    with open(schema, "r") as f:
-        _cur.executescript(f.read())
-        _conn.commit()        
+# export PGHOST, PGPORT, PGUSER, PGDATABASE in the shell
+_conn = psycopg2.connect()

src/parse-tags.py

@@ -12,7 +12,6 @@
 pptmimeList = ["application/vnd.ms-powerpoint","application/vnd.openxmlformats-officedocument.presentationml.presentation"]
 
 data = []
-db.connect("data/db.sqlite3")
 
 for file in x:
     id = file["id"]

src/schema.sql

@@ -1,16 +1,16 @@
 create table if not exists files(
-  fid         INTEGER primary key AUTOINCREMENT,
+  fid         serial primary key,
   name        text not null,
   link        text not null
 );
 
 create table if not exists tags(
-  tid         INTEGER primary key AUTOINCREMENT,
+  tid         serial primary key,
   name        text not null
 );
 
 create table if not exists file_tags(
-  tid         INTEGER,
-  fid         INTEGER,
+  tid         int,
+  fid         int,
   primary key (tid, fid)
 );