action and postscript updated

.github/workflows/scrape-scripts.yml

@@ -22,7 +22,7 @@ jobs:
  with:
  http_url: https://forum.obsidian.md/t/dataview-plugin-snippet-showcase/13673
  downloaded_filename: dataview-plugin-snippet-showcase.html
- postprocess: ./scripts/scraper-post-process.ts
-
-
+ postprocess: ./scraper-post-process.ts
+ - name: Create Pull Request
+ uses: peter-evans/create-pull-request@v3
 

scraper-post-process.ts

@@ -0,0 +1,43 @@
+import { readJSON, writeJSON } from 'https://deno.land/x/flat/mod.ts'
+import { cheerio } from "https://deno.land/x/[email protected]/mod.ts";
+
+interface Post {
+ creator?: {
+ name: string,
+ url: string,
+ }
+ datePublished?: Date,
+ content?: string,
+}
+
+// The filename is the first invocation argument
+const filename = Deno.args[0] // Same name as downloaded_filename
+//const data = await readJSON(filename)
+const html = await Deno.readTextFile("./dataview-plugin-snippet-showcase.html");
+const $ = cheerio.load(html);
+
+const posts = $('div[class="topic-body crawler-post"]');
+var rawData: Array<Post> = []
+await posts.each(function(index, data) {
+ const post = cheerio.load(data);
+ const creatorName = post('a[itemprop=url] > span[itemprop=name]');
+ const creatorUrl = post('span[class=creator] > a[itemprop=url]');
+ const publishedDate = post("span[class=crawler-post-infos] > time[itemprop=datePublished]");
+ const publishDate = publishedDate.first().attr('datetime')?.toString();
+ const content = post('div[class=post]')?.html() || '';
+ rawData.push({
+ creator: {
+ url: creatorUrl.first().attr('href')?.toString() || '',
+ name: creatorName.text(),
+ },
+ datePublished: publishDate ? new Date(publishDate): new Date(),
+ content,
+ })
+ }
+);
+const dataviewCodes = rawData.filter(post => post.content?.includes('```dataview'));
+for (const codePosts of dataviewCodes){
+ const fileName =`${codePosts.creator?.name || ''}_${codePosts.datePublished?.getTime()}.json` 
+ await writeJSON(`./scripts/${fileName}`, codePosts);
+}
+console.log(JSON.stringify(dataviewCodes));// Pluck a specific key off