yep

src/scrapers/twitter.rs

@@ -1,5 +1,83 @@
 use anyhow::{Context, Result, bail};
-use std::{path::PathBuf, process::Command};
+use serde::Deserialize;
+use std::{fs, path::PathBuf, process::Command};
+
+#[derive(Debug, Deserialize)]
+pub struct ScrapedTweet {
+    pub id: String,
+    pub text: String,
+    pub author: String,
+}
+
+#[derive(Debug, Deserialize)]
+struct RawScrapedTweet {
+    pub id: String,
+    #[serde(rename = "full_text")]
+    pub text: String,
+    pub author: Option<TweetAuthor>,
+}
+
+#[derive(Debug, Deserialize)]
+struct TweetAuthor {
+    #[serde(rename = "screen_name")]
+    pub handle: String,
+}
+
+pub fn parse_scraped_tweet(path: &PathBuf) -> Result<ScrapedTweet> {
+    let contents = fs::read_to_string(path)
+        .with_context(|| format!("Failed to read scraped tweet file: {}", path.display()))?;
+
+    if let Ok(raw) = toml::from_str::<RawScrapedTweet>(&contents) {
+        let author = raw
+            .author
+            .map(|author| author.handle)
+            .unwrap_or_else(|| "unknown".to_string());
+        return Ok(ScrapedTweet {
+            id: raw.id,
+            text: raw.text,
+            author,
+        });
+    }
+
+    parse_scraped_tweet_fallback(&contents, path)
+}
+
+fn parse_scraped_tweet_fallback(contents: &str, path: &PathBuf) -> Result<ScrapedTweet> {
+    let mut id = None;
+    let mut text = None;
+    let mut author = None;
+
+    for line in contents.lines() {
+        let trimmed = line.trim();
+        if trimmed.starts_with("id = ") && id.is_none() {
+            id = parse_quoted_value(trimmed);
+        } else if trimmed.starts_with("full_text = ") && text.is_none() {
+            text = parse_quoted_value(trimmed).map(unescape_toml_string);
+        } else if trimmed.starts_with("screen_name = ") && author.is_none() {
+            author = parse_quoted_value(trimmed);
+        }
+    }
+
+    let id = id.with_context(|| format!("Missing id in scraped tweet: {}", path.display()))?;
+    let text =
+        text.with_context(|| format!("Missing full_text in scraped tweet: {}", path.display()))?;
+    let author = author.unwrap_or_else(|| "unknown".to_string());
+
+    Ok(ScrapedTweet { id, text, author })
+}
+
+fn parse_quoted_value(line: &str) -> Option<String> {
+    let start = line.find('"')?;
+    let end = line.rfind('"')?;
+    if end <= start {
+        return None;
+    }
+    Some(line[start + 1..end].to_string())
+}
+
+fn unescape_toml_string(value: String) -> String {
+    value.replace("\\n", "\n").replace("\\\"", "\"")
+}
 
 pub fn scrape(url: &str) -> Result<PathBuf> {
     let tweet_id = url.split('/').next_back().unwrap();