Move more queries to API

api/endpoint/analyzePopularity.ts

@@ -0,0 +1,52 @@
+import {
+  VFiniteNumber,
+  VHumanString,
+  VInteger,
+  VStruct,
+  Valid,
+} from "@wzlin/valid";
+import assertInstanceOf from "@xtjs/lib/assertInstanceOf";
+import { QueryGroupByOutput, makeQuery } from "../query";
+
+const input = new VStruct({
+  query: new VHumanString(1, 512),
+  simMinHundredths: new VInteger(80, 100),
+});
+
+export const endpointAnalyzePopularity = {
+  input,
+  handler: async ({ query, simMinHundredths }: Valid<typeof input>) => {
+    const simThreshold = simMinHundredths / 100;
+    const res = await makeQuery({
+      dataset: "post",
+      queries: [query],
+      scales: {
+        sim: { min: simThreshold, max: 1.0 },
+      },
+      post_filter_clip: {
+        sim: { min: simThreshold, max: 1.0 },
+        // Some posts have UNIX timestamp 0.
+        ts_day: { min: 1, max: Number.MAX_SAFE_INTEGER },
+      },
+      weights: {
+        votes: "sim",
+      },
+      outputs: [
+        {
+          group_by: {
+            by: "ts_day",
+            bucket: 7,
+            cols: [["final_score", "sum"]],
+          },
+        },
+      ],
+    });
+    const data = assertInstanceOf(res[0], QueryGroupByOutput);
+    return {
+      timestamps: [...data.groups(new VInteger())].map(
+        (d) => new Date(d * 7 * 24 * 60 * 60 * 1000),
+      ),
+      scores: [...data.column("final_score", new VFiniteNumber())],
+    };
+  },
+};

api/endpoint/topPosts.ts

@@ -0,0 +1,46 @@
+import {
+  VFiniteNumber,
+  VHumanString,
+  VInteger,
+  VStruct,
+  Valid,
+} from "@wzlin/valid";
+import assertInstanceOf from "@xtjs/lib/assertInstanceOf";
+import { QueryItemsOutput, makeQuery } from "../query";
+
+const input = new VStruct({
+  query: new VHumanString(1, 512),
+  limit: new VInteger(1, 20),
+  simMinHundredths: new VInteger(80, 100),
+});
+
+export const endpointTopPosts = {
+  input,
+  handler: async ({ query, limit, simMinHundredths }: Valid<typeof input>) => {
+    const simThreshold = simMinHundredths / 100;
+    const res = await makeQuery({
+      dataset: "post",
+      queries: [query],
+      post_filter_clip: {
+        sim: { min: simThreshold, max: 1 },
+      },
+      outputs: [
+        {
+          items: {
+            cols: ["id", "sim"],
+            limit,
+            order_asc: false,
+            order_by: "votes",
+          },
+        },
+      ],
+    });
+    const data = assertInstanceOf(res[0], QueryItemsOutput);
+    return [
+      ...data.items({
+        id: new VInteger(1),
+        sim: new VFiniteNumber(),
+      }),
+    ];
+  },
+};

api/endpoint/topUsers.ts

@@ -0,0 +1,57 @@
+import {
+  VFiniteNumber,
+  VHumanString,
+  VInteger,
+  VString,
+  VStruct,
+  Valid,
+} from "@wzlin/valid";
+import assertInstanceOf from "@xtjs/lib/assertInstanceOf";
+import { QueryGroupByOutput, makeQuery } from "../query";
+
+const input = new VStruct({
+  query: new VHumanString(1, 512),
+  limit: new VInteger(1, 20),
+  simMinHundredths: new VInteger(80, 100),
+});
+
+export const endpointTopUsers = {
+  input,
+  handler: async ({ query, limit, simMinHundredths }: Valid<typeof input>) => {
+    const simThreshold = simMinHundredths / 100;
+    const res = await makeQuery({
+      dataset: "comment",
+      queries: [query],
+      scales: {
+        sim: {
+          min: simThreshold,
+          max: 1,
+        },
+      },
+      weights: {
+        // We can't multiply by votes, because the HN API does not expose votes for anything except posts.
+        sim_scaled: 1,
+      },
+      outputs: [
+        {
+          group_by: {
+            by: "user",
+            cols: [["final_score", "sum"]],
+            order_by: "final_score",
+            order_asc: false,
+            limit,
+          },
+        },
+      ],
+    });
+    const data = assertInstanceOf(res[0], QueryGroupByOutput);
+    return [
+      ...data.entries(new VString(), {
+        final_score: new VFiniteNumber(),
+      }),
+    ].map((e) => ({
+      user: e[0],
+      score: e[1].final_score,
+    }));
+  },
+};

api/main.ts

@@ -7,9 +7,12 @@ import readBufferStream from "@xtjs/lib/readBufferStream";
 import uint8ArrayToBuffer from "@xtjs/lib/uint8ArrayToBuffer";
 import { createSecureServer } from "http2";
 import { lg } from "../common/res";
+import { endpointAnalyzePopularity } from "./endpoint/analyzePopularity";
 import { endpointAnalyzeSentiment } from "./endpoint/analyzeSentiment";
 import { endpointHeatmap } from "./endpoint/heatmap";
 import { endpointSearch } from "./endpoint/search";
+import { endpointTopPosts } from "./endpoint/topPosts";
+import { endpointTopUsers } from "./endpoint/topUsers";
 
 const getPemEnv = (name: string) =>
   uint8ArrayToBuffer(
@@ -22,9 +25,12 @@ type Endpoint = {
 };
 
 const ENDPOINTS: Record<string, Endpoint> = {
+  analyzePopularity: endpointAnalyzePopularity,
   analyzeSentiment: endpointAnalyzeSentiment,
   heatmap: endpointHeatmap,
   search: endpointSearch,
+  topPosts: endpointTopPosts,
+  topUsers: endpointTopUsers,
 };
 
 createSecureServer(