Initial commit of Websearch

main.ts

@@ -0,0 +1,111 @@
+import { parseArgs } from "@std/cli/parse-args";
+import type { Args } from "@std/cli/parse-args";
+import { ensureDir } from "@std/fs";
+import { parse, stringify } from "@std/yaml";
+import { answerQuery, cleanTextFromHtml, getNewsUrls } from "./websearch.ts";
+
+export interface Config {
+  model?: string;
+  search_url?: string;
+}
+
+const parsed: Args = parseArgs(Deno.args, {
+  boolean: ["help"],
+  string: ["model", "search_url"],
+});
+
+if (parsed.help) {
+  console.log(`
+    Usage:
+      websearch [--model MODEL] [--search_url SEARCH_URL] QUERY
+
+    Options:
+      --model Ollama model to use
+      --search_url URL for SearXNG endpoint
+
+    Arguments:
+      QUERY The topic to search for news to summarize
+
+    Configuration:
+      The websearch configuration file is stored in the XDG Config directory
+      /home/$USER/.config/websearch/config.yml. Both the model and search_url
+      can be customized as an alternative to providing the options for each.
+      If both the Ollama model and SearXNG endpoint are successful, the
+      configuration is automatically saved/updated.
+  `);
+  Deno.exit();
+}
+
+const configDir = `${Deno.env.get("HOME")}/.config/websearch`;
+const configPath = `${Deno.env.get("HOME")}/.config/websearch/config.yml`;
+
+async function loadConfig(): Promise<Config> {
+  try {
+    const yamlString = await Deno.readTextFile(configPath);
+    return parse(yamlString) as Config;
+  } catch {
+    return {};
+  }
+}
+
+async function saveConfig(config: Config) {
+  await ensureDir(configDir);
+  await Deno.writeTextFile(
+    configPath,
+    stringify(config),
+  );
+}
+
+async function main(args: Args) {
+  const query = args._.join(" ");
+  if (!query) {
+    throw new Error("Please provide a search query");
+  }
+  console.log(`Query: ${query}`);
+
+  try {
+    const config = await loadConfig();
+    if (!config.model) {
+      if (args.model) {
+        config.model = args.model;
+      } else {
+        throw new Error("Provide --model or add Ollama model to configuration");
+      }
+    } else if (args.model && args.model !== config.model) {
+      config.model = args.model;
+    }
+    if (!config.search_url) {
+      if (args.search_url) {
+        config.search_url = args.search_url;
+      } else {
+        throw new Error(
+          "Provide --search_url or add search_url to configuration",
+        );
+      }
+    } else if (args.search_url && args.search_url !== config.search_url) {
+      config.search_url = args.search_url;
+    }
+
+    const urls = await getNewsUrls(config, query);
+    if (!urls || urls.length === 0) {
+      console.log("No results");
+      Deno.exit(1);
+    }
+
+    const cleanedTexts = await Promise.all(
+      urls.map((url) => cleanTextFromHtml(url)),
+    );
+    await answerQuery(config, query, cleanedTexts.join("\n\n"));
+    await saveConfig(config);
+  } catch (error: any) {
+    console.error(`Error processing query "${query}":`, error.message);
+    Deno.exit(1);
+  }
+}
+
+if (import.meta.main) {
+  main(parsed).catch((error) => {
+    console.error("Unhandled exception:", error);
+    Deno.exit(1);
+  });
+}