Koucha/koucha/src/fetch.rs

use crate::{
  Result,
  db::{Channel, ChannelId},
  AdapterClient,
};
use reqwest::Url;
use chrono::{DateTime, Utc};
use std::hash::{Hash, Hasher};

pub struct FetchedRSSItem {
  guid: String,
  title: String,
  description: String,
  content: String,
}
impl FetchedRSSItem {
  pub fn guid(&self) -> &str { &self.guid }
  pub fn title(&self) -> &str { &self.title }
  pub fn description(&self) -> &str { &self.description }
  pub fn content(&self) -> &str { &self.content }

  fn parse(item: rss::Item) -> Self {
    FetchedRSSItem {
      guid: Self::get_or_create_guid(&item),
      title: item.title().unwrap_or("").to_string(), 
      description: item.description().unwrap_or("").to_string(),
      content: item.content().unwrap_or("").to_string(),
    }
  }

  fn get_or_create_guid(item: &rss::Item) -> String {
    if let Some(guid) = item.guid() {
      return guid.value().to_string();
    }

    let mut hasher = std::collections::hash_map::DefaultHasher::new();
    item.link().unwrap_or("").hash(&mut hasher);
    item.title().unwrap_or("").hash(&mut hasher);
    item.description().unwrap_or("").hash(&mut hasher);

    format!("gen-{:x}", hasher.finish())
  }
}
pub struct FetchedRSSChannel {
  title: String,
  link: Url,
  description: String,

  items: Vec<FetchedRSSItem>,

  fetched_at: DateTime<Utc>,
}
impl FetchedRSSChannel {
  pub fn title(&self) -> &str { &self.title }
  pub fn link(&self) -> &Url { &self.link }
  pub fn description(&self) -> &str { &self.description }
  pub fn items(&self) -> &[FetchedRSSItem] { &self.items }
  pub fn fetched_at(&self) -> &DateTime<Utc> { &self.fetched_at }

  pub async fn fetch_channel(
    client: &AdapterClient, channel: Channel
  ) -> Result<Option<Self>> {
    if channel.should_skip_fetch() {
      return Ok(None);
    }

    let bytestream = client.0.get(channel.link().clone())
      .send().await?
      .bytes().await?;

    let rss_channel = rss::Channel::read_from(&bytestream[..])?;

    let now = Utc::now();

    Ok(Some(FetchedRSSChannel::parse(rss_channel, now)?))
  }

  fn parse(rss: rss::Channel, fetched_at: DateTime<Utc>) -> Result<Self> {
    Ok(FetchedRSSChannel {
      title: rss.title,
      link: Url::parse(&rss.link)?,
      description: rss.description,

      items: rss.items.into_iter().map(FetchedRSSItem::parse).collect(),

      fetched_at: fetched_at,
    })
  }
}

#[cfg(test)]
mod tests {
  use super::*;
  use crate::test_utils::{
    ITEM_TITLE, ITEM_GUID, ITEM_GUID2, ITEM_DESC, ITEM_CONT,
    CHANNEL_TITLE, CHANNEL_DESC, FEED1,
    get_datetime
  };

  fn create_guid(value: String) -> rss::Guid {
    rss::Guid { value, permalink: false }
  }

  fn create_item(guid: rss::Guid) -> rss::Item {
    rss::ItemBuilder::default()
      .title(ITEM_TITLE.to_string())
      .guid(guid)
      .description(ITEM_DESC.to_string())
      .content(ITEM_CONT.to_string())
      .build()
  }

  fn create_channel(items: Vec<rss::Item>) -> rss::Channel {
    rss::ChannelBuilder::default()
      .title(CHANNEL_TITLE.to_string())
      .description(CHANNEL_DESC.to_string())
      .link(FEED1.to_string())
      .items(items)
      .build()
  }

  #[test]
  fn parse_item() {
    let rss_guid = create_guid(ITEM_GUID.to_string());
    let rss_item = create_item(rss_guid);
    let item = FetchedRSSItem::parse(rss_item);

    assert_eq!(item.guid, ITEM_GUID);
    assert_eq!(item.title, ITEM_TITLE);
    assert_eq!(item.description, ITEM_DESC);
    assert_eq!(item.content, ITEM_CONT);
  }

  #[test]
  fn parse_feed() {
    let rss_guid = create_guid(ITEM_GUID.to_string());
    let rss_guid2 = create_guid(ITEM_GUID2.to_string());
    let rss_item = create_item(rss_guid);
    let rss_item2 = create_item(rss_guid2);

    let rss_channel = create_channel([rss_item, rss_item2].to_vec());

    let date: DateTime<Utc> = get_datetime();

    let channel = FetchedRSSChannel::parse(rss_channel, date).unwrap();

    assert_eq!(channel.title, CHANNEL_TITLE);
    assert_eq!(channel.link.as_str(), FEED1);
    assert_eq!(channel.description, CHANNEL_DESC);
    assert_eq!(channel.fetched_at, date);
    assert_eq!(channel.items.len(), 2);
    assert!(channel.items.iter().any(|i| i.guid() == ITEM_GUID));
    assert!(channel.items.iter().any(|i| i.guid() == ITEM_GUID2));
  }
}