From d84b4c3466adcc085b7df36015df4b46488ba591 Mon Sep 17 00:00:00 2001
From: Daniel Schadt <kingdread@gmx.de>
Date: Sat, 6 Mar 2021 12:12:03 +0100
Subject: add proper previous-log parsing for matrix

Doing all the "new log" insertion based on simple string operations is a
bit of madness, so the proper course of action is to parse them into a
proper intermediate representation from which we can then generate a
plain and HTML body.

In addition, this has some other minor code cleanup for the matrix
module.
---
 Cargo.lock           |  14 ++-
 Cargo.toml           |   4 +
 src/logbag.rs        | 288 +++++++++++++++++++++++++++++++++++++++++++++++++++
 src/main.rs          |   1 +
 src/matrix.rs        | 196 +++++++----------------------------
 test/logs/arkk.zevtc | Bin 465767 -> 0 bytes
 6 files changed, 342 insertions(+), 161 deletions(-)
 create mode 100644 src/logbag.rs
 delete mode 100644 test/logs/arkk.zevtc

diff --git a/Cargo.lock b/Cargo.lock
index 3f5457b..47e2400 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1,5 +1,7 @@
 # This file is automatically @generated by Cargo.
 # It is not intended for manual editing.
+version = 3
+
 [[package]]
 name = "adler32"
 version = "1.2.0"
@@ -540,6 +542,7 @@ dependencies = [
  "chrono",
  "clap",
  "evtclib",
+ "itertools",
  "log",
  "matrix-sdk",
  "notify",
@@ -1077,6 +1080,15 @@ version = "2.3.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "47be2f14c678be2fdcab04ab1171db51b2762ce6f0a8ee87c8dd4a04ed216135"
 
+[[package]]
+name = "itertools"
+version = "0.10.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "37d572918e350e82412fe766d24b15e6682fb2ed2bbe018280caa810397cb319"
+dependencies = [
+ "either",
+]
+
 [[package]]
 name = "itoa"
 version = "0.4.6"
@@ -1463,8 +1475,6 @@ dependencies = [
 [[package]]
 name = "olm-sys"
 version = "1.1.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "108c8902295791150d37f0d7b200de1fdd759f34bf479dfdc63c19de8c84c584"
 dependencies = [
  "cmake",
 ]
diff --git a/Cargo.toml b/Cargo.toml
index 7911ff8..b40fb4a 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -22,6 +22,7 @@ zip = "0.5.5"
 toml = "0.5.6"
 tokio = { version = "0.2", features = ["rt-core"] }
 matrix-sdk = "0.2.0"
+itertools = "0.10.0"
 
 [dependencies.serenity]
 default-features = false
@@ -36,3 +37,6 @@ features = [
     "rustls_backend",
 ]
 version = "0.9"
+
+[patch.crates-io]
+olm-sys = { path = "/home/daniel/src/olm-sys" }
diff --git a/src/logbag.rs b/src/logbag.rs
new file mode 100644
index 0000000..f461007
--- /dev/null
+++ b/src/logbag.rs
@@ -0,0 +1,288 @@
+use std::iter;
+use std::str::FromStr;
+
+use evtclib::{Log, Outcome};
+use itertools::Itertools;
+
+/// A [`LogBag`] is a struct that holds multiple logs in their categories.
+///
+/// This is similar to hash map mapping a category to a list of logs, but the [`LogBag`] saves its
+/// insertion order. The logs are saved as a line, that way we don't have to re-parse or re-upload
+/// them and we can just handle arbitrary data.
+#[derive(Default, Debug, Clone, PartialEq, Eq, Hash)]
+pub struct LogBag {
+    data: Vec<(String, Vec<String>)>,
+}
+
+impl LogBag {
+    /// Construct a new, empty [`LogBag`].
+    pub fn new() -> Self {
+        LogBag { data: Vec::new() }
+    }
+
+    /// Return an iterator over all available categories.
+    pub fn categories(&self) -> impl Iterator<Item = &str> {
+        self.data.iter().map(|x| &x.0 as &str)
+    }
+
+    /// Return an iterator over (category, items).
+    pub fn iter(&self) -> impl Iterator<Item = (&str, impl Iterator<Item = &str>)> {
+        self.data
+            .iter()
+            .map(|(cat, lines)| (cat as &str, lines.iter().map(|l| l as &str)))
+    }
+
+    /// Insert an item into the given category.
+    ///
+    /// If the category does not exist yet, it will be appended at the bottom.
+    pub fn insert(&mut self, category: &str, line: String) {
+        for (cat, lines) in self.data.iter_mut() {
+            if cat == category {
+                lines.push(line);
+                return;
+            }
+        }
+
+        // When we reach here, we don't have the category yet, so we gotta insert it.
+        self.data.push((String::from(category), vec![line]));
+    }
+
+    /// Tries to parse the given text as a plain [`LogBag`].
+    pub fn parse_plain(input: &str) -> Option<LogBag> {
+        input.parse().ok()
+    }
+
+    /// Renders the contents of this [`LogBag`] as plain text.
+    ///
+    /// The output of this can be fed back into [`LogBag::parse_plain`] to round-trip.
+    pub fn render_plain(&self) -> String {
+        self.iter()
+            .map(|(category, lines)| iter::once(category).chain(lines).join("\n"))
+            .join("\n\n")
+    }
+
+    /// Renders the contents of this [`LogBag`] as HTML.
+    ///
+    /// Useful for posting to Matrix chats.
+    pub fn render_html(&self) -> String {
+        self.iter()
+            .map(|(category, mut lines)| {
+                format!("<b>{}</b><br>\n{}", category, lines.join("<br>\n"))
+            })
+            .join("<br>\n<br>\n")
+    }
+
+    /// Renders the contents of this [`LogBag`] as Markdown.
+    ///
+    /// Useful for posting to Discord chats.
+    pub fn render_markdown(&self) -> String {
+        self.iter()
+            .map(|(category, mut lines)| format!("**{}**\n{}", category, lines.join("\n")))
+            .join("\n\n")
+    }
+}
+
+impl FromStr for LogBag {
+    type Err = ();
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        let data = s
+            .trim()
+            .split("\n\n")
+            .map(|chunk| {
+                let mut lines = chunk.split('\n');
+                let category = lines.next().unwrap();
+                (
+                    category.to_string(),
+                    lines.map(ToString::to_string).collect::<Vec<_>>(),
+                )
+            })
+            .collect();
+        Ok(LogBag { data })
+    }
+}
+
+impl From<Vec<(String, Vec<String>)>> for LogBag {
+    fn from(data: Vec<(String, Vec<String>)>) -> Self {
+        LogBag { data }
+    }
+}
+
+/// A helper function to return the right emoji for a given log.
+pub fn state_emoji(log: &Log) -> &'static str {
+    let outcome = log.analyzer().and_then(|a| a.outcome());
+    match outcome {
+        Some(Outcome::Success) => "✅",
+        Some(Outcome::Failure) => "❌",
+        None => "❓",
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+
+    #[test]
+    fn insert() {
+        let mut logbag = LogBag::new();
+        logbag.insert("cat 1", "line 1".to_string());
+        assert_eq!(logbag.categories().count(), 1);
+        logbag.insert("cat 1", "line 2".to_string());
+        assert_eq!(logbag.categories().count(), 1);
+        logbag.insert("cat 2", "line 1".to_string());
+        assert_eq!(logbag.categories().count(), 2);
+
+        assert_eq!(
+            logbag.categories().collect::<Vec<_>>(),
+            vec!["cat 1", "cat 2"]
+        );
+    }
+
+    #[test]
+    fn parse_single() {
+        let mut logbag = LogBag::new();
+        logbag.insert("cat 1", "line 1".to_string());
+        logbag.insert("cat 1", "line 2".to_string());
+
+        assert_eq!(
+            LogBag::parse_plain(
+                "\
+cat 1
+line 1
+line 2"
+            ),
+            Some(logbag)
+        );
+    }
+
+    #[test]
+    fn parse_multi() {
+        let mut logbag = LogBag::new();
+        logbag.insert("cat 1", "line 1".to_string());
+        logbag.insert("cat 1", "line 2".to_string());
+        logbag.insert("cat 2", "line 1".to_string());
+        logbag.insert("cat 2", "line 2".to_string());
+
+        assert_eq!(
+            LogBag::parse_plain(
+                "\
+cat 1
+line 1
+line 2
+
+cat 2
+line 1
+line 2"
+            ),
+            Some(logbag)
+        );
+    }
+
+    #[test]
+    fn render_plain_single() {
+        let mut logbag = LogBag::new();
+        logbag.insert("category", "line 1".to_string());
+        logbag.insert("category", "line 2".to_string());
+
+        assert_eq!(
+            logbag.render_plain(),
+            "\
+category
+line 1
+line 2"
+        );
+    }
+
+    #[test]
+    fn render_plain_multi() {
+        let mut logbag = LogBag::new();
+        logbag.insert("category 1", "line 1".to_string());
+        logbag.insert("category 1", "line 2".to_string());
+        logbag.insert("category 2", "enil 1".to_string());
+        logbag.insert("category 2", "enil 2".to_string());
+
+        assert_eq!(
+            logbag.render_plain(),
+            "\
+category 1
+line 1
+line 2
+
+category 2
+enil 1
+enil 2"
+        );
+    }
+
+    #[test]
+    fn render_html_single() {
+        let mut logbag = LogBag::new();
+        logbag.insert("category", "line 1".to_string());
+        logbag.insert("category", "line 2".to_string());
+
+        assert_eq!(
+            logbag.render_html(),
+            "\
+<b>category</b><br>
+line 1<br>
+line 2"
+        );
+    }
+
+    #[test]
+    fn render_html_multi() {
+        let mut logbag = LogBag::new();
+        logbag.insert("category 1", "line 1".to_string());
+        logbag.insert("category 1", "line 2".to_string());
+        logbag.insert("category 2", "enil 1".to_string());
+        logbag.insert("category 2", "enil 2".to_string());
+
+        assert_eq!(
+            logbag.render_html(),
+            "\
+<b>category 1</b><br>
+line 1<br>
+line 2<br>
+<br>
+<b>category 2</b><br>
+enil 1<br>
+enil 2"
+        );
+    }
+
+    #[test]
+    fn render_markdown_single() {
+        let mut logbag = LogBag::new();
+        logbag.insert("category", "line 1".to_string());
+        logbag.insert("category", "line 2".to_string());
+
+        assert_eq!(
+            logbag.render_markdown(),
+            "\
+**category**
+line 1
+line 2"
+        );
+    }
+
+    #[test]
+    fn render_markdown_multi() {
+        let mut logbag = LogBag::new();
+        logbag.insert("category 1", "line 1".to_string());
+        logbag.insert("category 1", "line 2".to_string());
+        logbag.insert("category 2", "enil 1".to_string());
+        logbag.insert("category 2", "enil 2".to_string());
+
+        assert_eq!(
+            logbag.render_markdown(),
+            "\
+**category 1**
+line 1
+line 2
+
+**category 2**
+enil 1
+enil 2"
+        );
+    }
+}
diff --git a/src/main.rs b/src/main.rs
index 43c98b1..ab40d86 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -21,6 +21,7 @@ use categories::Categorizable;
 mod config;
 use config::Config;
 mod discord;
+mod logbag;
 mod matrix;
 
 const DPS_REPORT_API: &str = "https://dps.report/uploadContent";
diff --git a/src/matrix.rs b/src/matrix.rs
index bd83a1f..ffd223d 100644
--- a/src/matrix.rs
+++ b/src/matrix.rs
@@ -1,11 +1,12 @@
 use super::categories::Categorizable;
 use super::config;
+use super::logbag::{state_emoji, LogBag};
 
 use std::convert::TryFrom;
 use std::time::{Duration, SystemTime};
 
 use anyhow::Result;
-use evtclib::{Log, Outcome};
+use evtclib::Log;
 use log::{debug, info};
 use tokio::runtime::Runtime;
 
@@ -37,6 +38,15 @@ impl From<config::Matrix> for MatrixUser {
     }
 }
 
+/// Maximum age of the message to still be edited.
+const MAX_HOURS: u64 = 5;
+
+/// Posts a link to the log to a Matrix room.
+///
+/// The user identification is given in the `user` parameter.
+///
+/// This function blocks until all API calls have been made, that is until the message has reached
+/// the homeserver.
 pub fn post_link(user: MatrixUser, room_id: &str, log: &Log, link: &str) -> Result<()> {
     let mut rt = Runtime::new()?;
     let room_id = RoomId::try_from(room_id)?;
@@ -62,8 +72,9 @@ pub fn post_link(user: MatrixUser, room_id: &str, log: &Log, link: &str) -> Resu
             }
             Some((old_id, old_text)) => {
                 debug!("Updating message {:?}", old_id);
-                let new_text = insert_log(&old_text, log, link);
-                let new_html = htmlify(&new_text);
+                let logbag = insert_log(&old_text, log, link);
+                let new_text = logbag.render_plain();
+                let new_html = logbag.render_html();
                 update_message(&client, &room_id, &old_id, &new_text, &new_html).await?;
             }
         }
@@ -81,15 +92,15 @@ async fn find_message(
     room_id: &RoomId,
 ) -> Result<Option<(EventId, String)>> {
     let request = get_message_events::Request::backward(room_id, "");
-    let five_h_ago = SystemTime::now() - Duration::from_secs(5 * 60 * 60);
+    let time_limit = SystemTime::now() - Duration::from_secs(MAX_HOURS * 60 * 60);
     for raw_message in client.room_messages(request).await?.chunk {
-        if let Ok(message) = raw_message.deserialize() {
-            if let AnyRoomEvent::Message(AnyMessageEvent::RoomMessage(msg)) = message {
-                if &msg.sender == my_id && msg.origin_server_ts >= five_h_ago {
-                    if let MessageEventContent::Text(text) = msg.content {
-                        if text.relates_to.is_none() {
-                            return Ok(Some((msg.event_id, text.body)));
-                        }
+        if let Ok(AnyRoomEvent::Message(AnyMessageEvent::RoomMessage(msg))) =
+            raw_message.deserialize()
+        {
+            if &msg.sender == my_id && msg.origin_server_ts >= time_limit {
+                if let MessageEventContent::Text(text) = msg.content {
+                    if text.relates_to.is_none() {
+                        return Ok(Some((msg.event_id, text.body)));
                     }
                 }
             }
@@ -98,11 +109,12 @@ async fn find_message(
     Ok(None)
 }
 
+/// Post a new message to the given Matrix channel.
 async fn post_new(client: &Client, room_id: &RoomId, log: &Log, link: &str) -> Result<()> {
-    let title = log.category();
     let line = format!("{} {}", state_emoji(log), link);
-    let body = format!("{}\n{}\n", title, line);
-    let html = format!("<b>{}</b><br>\n{}\n", title, line);
+    let logbag: LogBag = vec![(log.category().to_string(), vec![line])].into();
+    let body = logbag.render_plain();
+    let html = logbag.render_html();
 
     let text_message = TextMessageEventContent::html(body, html);
     client
@@ -115,6 +127,9 @@ async fn post_new(client: &Client, room_id: &RoomId, log: &Log, link: &str) -> R
     Ok(())
 }
 
+/// Updates the given message with some new text
+///
+/// This constructs and sends the right Matrix API message.
 async fn update_message(
     client: &Client,
     room_id: &RoomId,
@@ -139,149 +154,12 @@ async fn update_message(
     Ok(())
 }
 
-fn insert_log(old_text: &str, log: &Log, link: &str) -> String {
-    let chunks = old_text.split("\n\n");
-    let mut found = false;
-    let result = chunks
-        .map(|chunk| {
-            let category = chunk.split("\n").next().unwrap();
-            if category == log.category() {
-                found = true;
-                format!("{}\n{} {}", chunk.trim(), state_emoji(log), link)
-            } else {
-                chunk.to_string()
-            }
-        })
-        .collect::<Vec<_>>()
-        .join("\n\n");
-    if found {
-        result
-    } else {
-        format!(
-            "{}\n\n{}\n{} {}",
-            result.trim(),
-            log.category(),
-            state_emoji(log),
-            link
-        )
-    }
-}
-
-fn htmlify(text: &str) -> String {
-    text.split("\n\n")
-        .map(|chunk| {
-            let lines = chunk.split("\n").collect::<Vec<_>>();
-            format!("<b>{}</b><br>\n{}", lines[0], lines[1..].join("<br>\n"))
-        })
-        .collect::<Vec<_>>()
-        .join("<br>\n<br>\n")
-}
-
-fn state_emoji(log: &Log) -> &'static str {
-    let outcome = log.analyzer().and_then(|a| a.outcome());
-    match outcome {
-        Some(Outcome::Success) => "✅",
-        Some(Outcome::Failure) => "❌",
-        None => "❓",
-    }
-}
-
-#[cfg(test)]
-mod test {
-    use super::*;
-    use std::io::Cursor;
-    use evtclib::Compression;
-
-    static ARKK_LOG: &[u8] = include_bytes!("../test/logs/arkk.zevtc");
-
-    fn arkk_log() -> Log {
-        evtclib::process_stream(Cursor::new(ARKK_LOG), Compression::Zip).unwrap()
-    }
-
-    #[test]
-    fn test_insert_log_new_category() {
-        let log = arkk_log();
-        let old_text = "\
-Unknown
-x https://dps.report.example";
-        let new_text = insert_log(old_text, &log, "foobar");
-
-        assert_eq!("\
-Unknown
-x https://dps.report.example
-
-99 CM (Shattered Observatory)
-✅ foobar", new_text);
-    }
-
-    #[test]
-    fn test_insert_log_appending() {
-        let log = arkk_log();
-        let old_text = "\
-99 CM (Shattered Observatory)
-x old log here";
-        let new_text = insert_log(old_text, &log, "foobar");
-        assert_eq!("\
-99 CM (Shattered Observatory)
-x old log here
-✅ foobar", new_text);
-    }
-
-    #[test]
-    fn test_insert_log_multiple() {
-        let log = arkk_log();
-        let old_text = "\
-100 CM (Sunqua Peak)
-y some old log here
-
-99 CM (Shattered Observatory)
-x old log here
-
-98 CM (Nightmare)
-z raboof";
-        let new_text = insert_log(old_text, &log, "foobar");
-        assert_eq!("\
-100 CM (Sunqua Peak)
-y some old log here
-
-99 CM (Shattered Observatory)
-x old log here
-✅ foobar
-
-98 CM (Nightmare)
-z raboof", new_text);
-    }
-
-    #[test]
-    fn test_htmlify_single() {
-        let plain = "\
-Header
-Log 1
-Log 2";
-        assert_eq!(htmlify(plain), "\
-<b>Header</b><br>
-Log 1<br>
-Log 2");
-    }
-
-    #[test]
-    fn test_htmlify_multiple() {
-        let plain = "\
-Header 1
-Log 1
-Log 2
-
-Header 2
-Log 3
-Log 4";
-        assert_eq!(htmlify(plain), "\
-<b>Header 1</b><br>
-Log 1<br>
-Log 2<br>
-<br>
-<b>Header 2</b><br>
-Log 3<br>
-Log 4");
-    }
-
+/// Inserts the given log into the text.
+///
+/// The text is first parsed as a [`LogBag`], then the link is formatted and inserted.
+fn insert_log(old_text: &str, log: &Log, link: &str) -> LogBag {
+    let line = format!("{} {}", state_emoji(log), link);
+    let mut logbag = LogBag::parse_plain(old_text).unwrap();
+    logbag.insert(log.category(), line);
+    logbag
 }
diff --git a/test/logs/arkk.zevtc b/test/logs/arkk.zevtc
deleted file mode 100644
index 510df20..0000000
Binary files a/test/logs/arkk.zevtc and /dev/null differ
-- 
cgit v1.2.3