batman

src/scrapers/twitter.rs

@@ -0,0 +1,24 @@
+use anyhow::{Context, Result, bail};
+use std::{path::PathBuf, process::Command};
+
+pub fn scrape(url: &str) -> Result<PathBuf> {
+    let tweet_id = url.split('/').next_back().unwrap();
+    println!("Scraping tweet ID: {}", tweet_id);
+
+    let out = Command::new("python")
+        .arg("scrape_user_tweet_contents.py")
+        .arg("--tweet-ids")
+        .arg(tweet_id)
+        .output()
+        .with_context(|| "Failed to execute tweet scraping command")?;
+    println!("Output command: {:?}", out);
+
+    if PathBuf::from("scraped-tweets")
+        .join(format!("tweet-{}.toml", tweet_id))
+        .exists()
+    {
+        return Ok(PathBuf::from("scraped-tweets").join(format!("tweet-{}.toml", tweet_id)));
+    }
+
+    bail!("Scraping failed for tweet: {}", url)
+}