Update changelog

Signed-off-by: Jacob Kiers <jacob@jacobkiers.net>

.drone.jsonnet

@@ -19,7 +19,7 @@ local add_build_steps() = [
     volumes: [
       {
         name: 'dockersock',
-        path: '/var/run/docker.sock',
+        path: '/var/run',
       },
     ],
     commands: [
@@ -29,7 +29,7 @@ local add_build_steps() = [
       'rm -rf target/' + arch.target + '/release/*',
     ],
     environment: {
-      CROSS_DOCKER_IN_DOCKER: true,
+      CROSS_REMOTE: true,
     },
     depends_on: ['Wait for Docker'],
   }
@@ -57,11 +57,11 @@ local add_build_steps() = [
         'docker pull hello-world:latest',
       ],
       environment: {
-        CROSS_DOCKER_IN_DOCKER: true,
+        CROSS_REMOTE: true,
       },
       volumes: [{
         name: 'dockersock',
-        path: '/var/run/docker.sock',
+        path: '/var/run',
       }],
     }] +
     add_build_steps() +
@@ -92,12 +92,26 @@ local add_build_steps() = [
       },
     ],
 
+  services: [{
+    name: 'docker',
+    image: 'docker:dind',
+    privileged: true,
+    volumes: [
+      {
+        name: 'dockersock',
+        path: '/var/run',
+      },
+      {
+        name: 'docker-storage',
+        path: '/var/lib/docker',
+      },
+    ],
+  }],
+
   volumes: [
     {
       name: 'dockersock',
-      host: {
-        path: '/var/run/docker.sock',
-      },
+      temp: {},
     },
   ],
 

CHANGELOG.md

@@ -7,6 +7,11 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Fixed
+
+* Truncate feed file before writing, to prevent corruption from leftover data.
+* Ensure the feed file name is part of the self URL. This was still hardcoded to `feed.atom`.
+
 ## [0.2.2] - 2022-12-16
 
 ### Changed
@@ -30,4 +35,4 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Changed
 
-* By default the name of the feed is now feed.xml instead of feed.atom.
+* By default, the name of the feed is now feed.xml instead of feed.atom.

Cargo.lock

@@ -29,8 +29,7 @@ checksum = "23b62fc65de8e4e7f52534fb52b0f3ed04746ae267519eef2a83941e8085068b"
 [[package]]
 name = "atom_syndication"
 version = "0.11.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "21fb6a0b39c6517edafe46f8137e53c51742425a4dae1c73ee12264a37ad7541"
+source = "git+https://github.com/jacobkiers/atom-syndication?rev=add7083b56b9d737f0fa1d3383aa82789b6c38ad#add7083b56b9d737f0fa1d3383aa82789b6c38ad"
 dependencies = [
  "chrono",
  "derive_builder",

Cargo.toml

@@ -19,3 +19,6 @@ imap = { version = "^2.4.1", default-features = false }
 mail-parser = "^0.8.0"
 rustls-connector = { version = "^0.16.1", default-features = false, features = [ "webpki-roots-certs", "quic" ] }
 sha2 = "^0.10.2"
+
+[patch.crates-io]
+atom_syndication = { git = "https://github.com/jacobkiers/atom-syndication", rev = "add7083b56b9d737f0fa1d3383aa82789b6c38ad" }

README.md

@@ -1,6 +1,41 @@
-# Newsletter 2 Web
+# Newsletter to Web
 
-Converts a newsletter to static HTML files.
+Converts a newsletter to and Atom feed and static HTML files.
+
+## Usage
+
+Get the latest release [from the releases page](https://code.kiers.eu/newsletter-to-web/newsletter-to-web/releases/latest).
+
+### Getting help
+
+For help, use 
+
+* `newsletter-to-web help`
+* `newsletter-to-web help <subcommand>`.
+
+### Basic usage
+
+First, download all messages from the IMAP mail server 
+and store them in the `data/` directory:
+
+```sh
+newsletter-to-web fetch-from-imap -s <imap.example.com> -u <email@example.com> -p <password>
+```
+
+Then, convert them to an Atom feed, using
+`newsletters.example.com` as the base domain:
+
+```sh
+newsletter-to-web --include-html build-feed newsletters.example.org
+```
+
+This will put the output in the `output/` directory. The Atom
+feed will be in `output/feed.xml`, together with a very simple
+`index.html` file pointing to the feed. It will also add an HTML
+file for every email with the HTML content.
+
+The feed will already contain the full HTML, so it can easily be
+read from a feed reader.
 
 ## Features
 

resources/feed.xsl

@@ -26,6 +26,7 @@
         </section>
         <section>
           <h2>Recent Items</h2>
+          <p>Last updated on <xsl:apply-templates select="atom:feed/atom:updated" /></p>
           <xsl:apply-templates select="atom:feed/atom:entry" />
         </section>
       </body>

src/feed.rs

@@ -2,6 +2,7 @@ use crate::Message;
 
 use atom_syndication::{
     ContentBuilder, Entry, EntryBuilder, Feed, FeedBuilder, Generator, LinkBuilder, Person,
+    WriteConfig,
 };
 use chrono::{DateTime, TimeZone, Utc};
 
@@ -34,19 +35,13 @@ pub(crate) fn add_entry_to_feed(
                     _ => "".to_string(),
                 },
             },
-            email: match &from.address {
-                Some(e) => Some(e.to_string()),
-                _ => None,
-            },
+            email: from.address.as_ref().map(|e| e.to_string()),
             uri: None,
         },
-        title: parsed
-            .subject()
-            .expect("Expected a subject")
-            .to_string(),
+        title: parsed.subject().expect("Expected a subject").to_string(),
         content: Some(processed_html.clone()),
         id: url.clone(),
-        published: Utc.timestamp_opt(date.to_timestamp(), 0).unwrap(), //(format!("{}{}", &date.to_iso8601(), "+00:00").as_str()).`unwrap(),
+        published: Utc.timestamp_opt(date.to_timestamp(), 0).unwrap(),
         url: match include_html {
             true => url,
             false => "".to_string(),
@@ -57,10 +52,11 @@ pub(crate) fn add_entry_to_feed(
     feed.entries.push(entry);
 }
 
-pub(crate) fn build_atom_feed(hostname: &String) -> Feed {
+pub(crate) fn build_atom_feed(hostname: &String, feed_file: &str) -> Feed {
+    let feed_url = format!("https://{}/{}", hostname, feed_file);
     FeedBuilder::default()
         .title("JJKiers Newsletters")
-        .id(format!("https://{}/feed.atom", hostname))
+        .id(&feed_url)
         .link(
             LinkBuilder::default()
                 .href(format!("https://{}/", hostname))
@@ -69,7 +65,7 @@ pub(crate) fn build_atom_feed(hostname: &String) -> Feed {
         )
         .link(
             LinkBuilder::default()
-                .href(format!("https://{}/feed.atom", hostname))
+                .href(&feed_url)
                 .rel("self".to_string())
                 .build(),
         )
@@ -81,6 +77,19 @@ pub(crate) fn build_atom_feed(hostname: &String) -> Feed {
         .build()
 }
 
+pub(crate) fn write_feed<W: std::io::Write>(
+    feed: Feed,
+    mut out: W,
+) -> Result<W, atom_syndication::Error> {
+    let _ = writeln!(out, r#"<?xml version="1.0"?>"#);
+    let _ = writeln!(out, r#"<?xml-stylesheet href="feed.xsl" type="text/xsl"?>"#);
+    let config = WriteConfig {
+        write_document_declaration: false,
+        ..Default::default()
+    };
+    feed.write(out, config)
+}
+
 //#[derive(Serialize, Deserialize, Debug)]
 pub(crate) struct Newsletter {
     id: String,
@@ -104,8 +113,8 @@ impl From<Newsletter> for Entry {
 
         eb.title(post.title)
             .id(post.id)
-            .published(Some(post.published.clone().into()))
-            .author(post.author.into())
+            .published(Some(post.published.into()))
+            .author(post.author)
             .content(content);
 
         if post.url.len() > 1 {

src/main.rs

@@ -20,8 +20,8 @@ use std::{
 
 pub(crate) use message::Message;
 
-const INDEX_HTML: & 'static str = include_str!("../resources/index.html");
-const FEED_STYLESHEET: & 'static str = include_str!("../resources/feed.xsl");
+const INDEX_HTML: &str = include_str!("../resources/index.html");
+const FEED_STYLESHEET: &str = include_str!("../resources/feed.xsl");
 
 fn main() -> Result<(), Box<dyn Error>> {
     let cli = cli::Cli::parse();
@@ -41,7 +41,11 @@ fn main() -> Result<(), Box<dyn Error>> {
             username.to_owned(),
             password.to_owned(),
         ),
-        cli::Command::BuildFeed { filename , hostname, include_html } => build_feed(&filename, hostname.to_owned(), *include_html),
+        cli::Command::BuildFeed {
+            filename,
+            hostname,
+            include_html,
+        } => build_feed(filename, hostname, *include_html),
         _ => unimplemented!("This method is not yet implemented."),
     };
 
@@ -56,7 +60,11 @@ fn create_directory<P: AsRef<Path>>(dir: P) -> Result<(), std::io::Error> {
     Ok(())
 }
 
-fn build_feed(filename: &PathBuf, hostname: String, include_html: bool) -> Result<(), Box<dyn Error>> {
+fn build_feed(
+    filename: &PathBuf,
+    hostname: &String,
+    include_html: bool,
+) -> Result<(), Box<dyn Error>> {
     let dir = filename.parent().ok_or(format!(
         "Could not get parent directory of {}",
         filename.display()
@@ -70,9 +78,15 @@ fn build_feed(filename: &PathBuf, hostname: String, include_html: bool) -> Resul
 
     create_directory(dir)?;
 
-    let mut feed = feed::build_atom_feed(&hostname);
+    let feed_file = filename
+        .file_name()
+        .expect("Feed path should have a file name")
+        .to_str()
+        .expect("Feed path should be printable.");
 
-    let mut reader = DataDirectoryMessageReader::new((&Path::new("data")).to_path_buf());
+    let mut feed = feed::build_atom_feed(&hostname, feed_file);
+
+    let mut reader = DataDirectoryMessageReader::new(Path::new("data").to_path_buf());
 
     for msg in reader.read_rfc822_messages() {
         let parsed = msg.get_parsed().expect("A parsed messsage.");
@@ -82,15 +96,12 @@ fn build_feed(filename: &PathBuf, hostname: String, include_html: bool) -> Resul
             msg.get_uid()
         ))?;
 
-        let subject = match parsed.subject() {
-            Some(subject) => subject,
-            None => "No subject",
-        };
+        let subject = parsed.subject().unwrap_or("No subject");
 
         println!(
             "Processing message {} from {} with subject {}",
             msg.get_uid(),
-            date.to_string(),
+            date,
             subject
         );
 
@@ -98,26 +109,23 @@ fn build_feed(filename: &PathBuf, hostname: String, include_html: bool) -> Resul
         let processed_html = process_html(&html_body).expect("Could not process the HTML");
 
         if include_html {
-            let path : PathBuf = [dir, Path::new(&get_path(&parsed, &msg))].iter().collect();
+            let path: PathBuf = [dir, Path::new(&get_path(&parsed, &msg))].iter().collect();
             write_file(&path, processed_html.as_bytes())?;
         }
 
         feed::add_entry_to_feed(&mut feed, &msg, &processed_html, &hostname, include_html);
     }
 
-    if feed.entries.len() > 0 {
+    if !feed.entries.is_empty() {
         feed.set_updated(Utc::now());
         println!("Writing feed to {}", filename.display());
 
-        // TODO: Ugly hack because atom_syndication crate does not support style sheets.
-        let feed_str = feed.to_string().as_str().replace(">\n<feed", ">\n<?xml-stylesheet href=\"feed.xsl\" type=\"text/xsl\"?>\n<feed");
-        let _ = write_file(filename, feed_str)?;
-        let _ = write_file(dir.join("feed.xsl"), FEED_STYLESHEET)?;
-
-        // Another ugly hack, but I don't know how to do this better...
-        let file_name = format!("{:?}", filename.file_name().unwrap()).replace('"', "");
-        write_file(dir.join("index.html"), INDEX_HTML.replace("{FEED}", file_name.as_str()))?;
-
+        feed::write_feed(feed, open_file(filename).unwrap())?;
+        write_file(dir.join("feed.xsl"), FEED_STYLESHEET)?;
+        write_file(
+            dir.join("index.html"),
+            INDEX_HTML.replace("{FEED}", feed_file),
+        )?;
     }
 
     println!("Finished building the feed.");
@@ -136,12 +144,7 @@ fn fetch_from_imap(
 
     print!("Getting mail from {} for mailbox {}", server, username);
 
-    let mut reader = ImapReader::new(
-        String::from(server),
-        port,
-        String::from(username),
-        String::from(password),
-    );
+    let mut reader = ImapReader::new(server, port, username, password);
 
     for msg in reader.read_rfc822_messages() {
         let parsed = msg.get_parsed().ok_or(format!(
@@ -154,15 +157,12 @@ fn fetch_from_imap(
             msg.get_uid()
         ))?;
 
-        let subject = match parsed.subject() {
-            Some(subject) => subject,
-            None => "No subject",
-        };
+        let subject = parsed.subject().unwrap_or("No subject");
 
         println!(
             "Processing message {} from {} with subject {}",
             msg.get_uid(),
-            date.to_string(),
+            date,
             subject
         );
 
@@ -182,8 +182,6 @@ fn fetch_from_imap(
     Ok(())
 }
 
-
-
 fn get_path(parsed: &ParsedMessage, msg: &Message) -> String {
     let date = parsed.date().expect("Could not extract date");
     let date_str = format!(
@@ -192,26 +190,32 @@ fn get_path(parsed: &ParsedMessage, msg: &Message) -> String {
     );
 
     let hash = base16ct::lower::encode_string(&Sha256::digest(
-        &parsed.body_html(0).expect("Expected a body").as_bytes(),
+        parsed.body_html(0).expect("Expected a body").as_bytes(),
     ));
 
-    let uid: i32 = msg.get_uid()
+    let uid: i32 = msg
+        .get_uid()
         .parse()
-        .expect(&format!("Could not convert message id {} to an i32.", msg.get_uid()));
+        .unwrap_or_else(|_| panic!("Could not convert message id {} to an i32.", msg.get_uid()));
 
-    format!("{:05}_{}_{}.html", uid, date_str, &hash).to_owned()
+    format!("{:05}_{}_{}.html", uid, date_str, &hash)
 }
 
 fn process_html(input: &str) -> Result<String, ()> {
     Ok(input.replace("src", "data-source"))
 }
 
-fn write_file<P: Into<PathBuf>, D: AsRef<[u8]>>(html_path: P, data: D) -> Result<(), std::io::Error> {
-    let path : PathBuf = html_path.into();
+fn open_file<P: Into<PathBuf>>(path: P) -> std::io::Result<std::fs::File> {
     OpenOptions::new()
         .write(true)
+        .truncate(true)
         .create(true)
-        .open(&path)
-        .expect(format!("Could not open file '{}' for writing", &path.display()).as_str())
+        .open(path.into())
+}
+
+fn write_file<P: Into<PathBuf>, D: AsRef<[u8]>>(path: P, data: D) -> Result<(), std::io::Error> {
+    let path: PathBuf = path.into();
+    open_file(path.clone())
+        .unwrap_or_else(|_| panic!("Could not open file '{}' for writing", &path.display()))
         .write_all(data.as_ref())
-}
+}

src/message.rs

@@ -21,4 +21,4 @@ impl Message {
     pub fn get_data(&self) -> &Vec<u8> {
         &self.data
     }
-}
+}

src/message_reader.rs

@@ -44,7 +44,7 @@ impl EmailReader for DataDirectoryMessageReader {
                 Some(ext) => ext == "eml",
                 None => false,
             })
-            .map(|i| {
+            .filter_map(|i| {
                 let uid = i
                     .path()
                     .file_stem()
@@ -52,9 +52,9 @@ impl EmailReader for DataDirectoryMessageReader {
                     .to_owned()
                     .into_string()
                     .expect("Could not convert filename to string.")
-                    .split("_")
+                    .split('_')
                     .collect::<Vec<&str>>()[0]
-                    .trim_start_matches("0")
+                    .trim_start_matches('0')
                     .to_string();
 
                 if let Ok(data) = std::fs::read(i.path()) {
@@ -63,8 +63,6 @@ impl EmailReader for DataDirectoryMessageReader {
                     None
                 }
             })
-            .filter(|i| i.is_some())
-            .map(|i| i.unwrap())
             .map(|i| Message::new(i.0, i.1));
 
         let iter = items.collect::<Vec<Message>>().into_iter();