From fbf3ff75c8c1a5246e40805c6afc4968232b7634 Mon Sep 17 00:00:00 2001
From: Rohan Moniz <60864468+rm03@users.noreply.github.com>
Date: Mon, 20 Nov 2023 02:27:35 -0500
Subject: [PATCH] use uuid instead of slug

---
 dbutils/mongo.go | 15 ---------------
 dbutils/utils.py |  9 ++++-----
 server.py        | 13 +++++++------
 3 files changed, 11 insertions(+), 26 deletions(-)

diff --git a/dbutils/mongo.go b/dbutils/mongo.go
index 9f39f16..5cbfb7b 100644
--- a/dbutils/mongo.go
+++ b/dbutils/mongo.go
@@ -8,7 +8,6 @@ import (
 	"log"
 	"net/http"
 	"os"
-	"strconv"
 	"strings"
 	"time"
 
@@ -54,13 +53,6 @@ func worker(id int, jobs <-chan string, db *mongo.Database) {
 			page.Articles[i].CTime = cTime.Unix()
 			page.Articles[i].MTime = mTime.Unix()
 			page.Articles[i].PTime = pTime.Unix()
-			cMonth := ""
-			if cTime.Month() < 10 {
-				cMonth = fmt.Sprintf("0%d", int(cTime.Month()))
-			} else {
-				cMonth = strconv.Itoa(int(cTime.Month()))
-			}
-			page.Articles[i].Slug = fmt.Sprintf("/%d/%s/%s", cTime.Year(),  cMonth, page.Articles[i].Slug)
 			articlesInterface = append(articlesInterface, page.Articles[i])
 		}
 		if _, err = coll.InsertMany(context.TODO(), articlesInterface); err != nil {
@@ -113,13 +105,6 @@ func UpdateArticles(db *mongo.Database) {
 					page.Articles[i].CTime = cTime.Unix()
 					page.Articles[i].MTime = mTime.Unix()
 					page.Articles[i].PTime = pTime.Unix()
-					cMonth := ""
-					if cTime.Month() < 10 {
-						cMonth = fmt.Sprintf("0%d", int(cTime.Month()))
-					} else {
-						cMonth = strconv.Itoa(int(cTime.Month()))
-					}
-					page.Articles[i].Slug = fmt.Sprintf("/%d/%s/%s", cTime.Year(),  cMonth, page.Articles[i].Slug)
 					articlesInterface = append(articlesInterface, page.Articles[i])
 				} else {
 					p = false
diff --git a/dbutils/utils.py b/dbutils/utils.py
index d95e2d4..1c7195d 100644
--- a/dbutils/utils.py
+++ b/dbutils/utils.py
@@ -12,7 +12,7 @@
 def initialize_schema():
     redis_client = redis.Redis().from_url(os.environ.get("REDIS_URI"))
     schema = (
-        TextField("slug"),
+        TextField("uuid"),
         TextField("headline"),
         TextField("thumbnail_url"),
         NumericField("timestamp"),
@@ -47,7 +47,7 @@ def refresh_recent_articles(cron=False):
     cursor = mongo_client.Cluster.articles.find(
         {},
         projection={
-            "slug": 1,
+            "uuid": 1,
             "headline": 1,
             "dominantmedia": 1,
             "content": 1,
@@ -62,11 +62,10 @@ def refresh_recent_articles(cron=False):
             )
         ]
         byte_embedding = np.array(embedding, dtype=np.float32).tobytes()
-        key = doc["slug"]
         pipe.hset(
-            name=f"article:{key}",
+            name=f"article:{doc['uuid']}",
             mapping={
-                "slug": key,
+                "uuid": doc["uuid"],
                 "headline": doc["headline"],
                 "embedding": byte_embedding,
                 "timestamp": datetime.strptime(
diff --git a/server.py b/server.py
index fe2fd82..c3c8ff5 100644
--- a/server.py
+++ b/server.py
@@ -30,7 +30,7 @@ async def startup():
 
 @app.get("/recommend")
 @cache(expire=60 * 60)
-def recommend(slug: str = ""):
+def recommend(uuid: str = ""):
     dt = datetime.now() - timedelta(
         days=90
     )  # only show articles published in the last 90 days
@@ -39,15 +39,16 @@ def recommend(slug: str = ""):
             f"(@timestamp:[({dt.timestamp()} inf])=>[KNN {top_k + 1} @embedding $vec as score]"
         )
         .sort_by("score")
-        .return_fields("slug", "headline", "thumbnail_url")
+        .return_fields("uuid", "headline", "thumbnail_url")
         .dialect(2)
     )
-    vec = redis_client.hget(f"article:{slug}", "embedding")
+    key = f"article:{uuid}"
+    vec = redis_client.hget(key, "embedding")
     if vec:
         query_params = {"vec": vec}
         results = redis_client.ft("articles").search(query, query_params).docs
-        return results[1:] if results[0]["slug"] == slug else results[:-1]
-    doc = mongo_client.Cluster.articles.find_one({"slug": slug})
+        return results[1:] if results[0]["id"] == key else results[:-1]
+    doc = mongo_client.Cluster.articles.find_one({"uuid": uuid})
     if doc:
         embedding = [
             float(val)
@@ -58,7 +59,7 @@ def recommend(slug: str = ""):
         byte_embedding = np.array(embedding, dtype=np.float32).tobytes()
         query_params = {"vec": byte_embedding}
         results = redis_client.ft("articles").search(query, query_params).docs
-        return results[1:] if results[0]["slug"] == slug else results[:-1]
+        return results[1:] if results[0]["id"] == key else results[:-1]
     return None