initial commit

.gitignore

@@ -0,0 +1 @@
+/target

Cargo.toml

@@ -0,0 +1,17 @@
+[package]
+name = "flarumdown"
+version = "0.1.0"
+edition = "2024"
+license = "MIT"
+readme = "README.md"
+description = "Flarum is down - read as Markdown"
+keywords = ["archive", "conversion", "backup", "static", "dump"]
+categories = ["command-line-utilities", "parsing", "text-processing", "value-formatting"]
+repository = "https://github.com/YGGverse/flarumdown"
+
+[dependencies]
+anyhow = "1.0.102"
+chrono = "0.4.44"
+clap = { version = "4.6.0", features = ["derive"] }
+html-to-markdown-rs = "2.28.2"
+rusqlite = { version = "0.39.0", features = ["chrono"]}

README.md

@@ -1,2 +1,8 @@
 # flarumdown
+
 Flarum is down - read as Markdown
+
+CLI tool for Flarum v2 that allows to export public DB into the Markdown format.
+
+> [!IMPORTANT]
+> In development

src/config.rs

@@ -0,0 +1,19 @@
+use clap::Parser;
+use std::path::PathBuf;
+
+#[derive(Parser, Debug)]
+#[command(version, about, long_about = None)]
+pub struct Config {
+    /// Path to database
+    #[arg(short, long)]
+    pub source: PathBuf,
+
+    /// Path to export markdown
+    #[arg(short, long)]
+    pub target: PathBuf,
+
+    /// Collect discussions with given tag slug only
+    /// * keep empty to export all
+    #[arg(short, long)]
+    pub filter_tag: Vec<String>,
+}

src/database.rs

@@ -0,0 +1,104 @@
+use chrono::{DateTime, Utc};
+use rusqlite::{Connection, Error};
+use std::path::PathBuf;
+
+pub struct User {
+    pub id: i64,
+    pub username: String,
+}
+
+pub struct Tag {
+    pub id: i64,
+    //pub name: String,
+    pub slug: String,
+}
+
+pub struct Discussion {
+    pub id: i64,
+    pub user_id: i64,
+    pub first_post_id: i64,
+    pub created_at: DateTime<Utc>,
+    pub title: String,
+    //pub slug: String,
+}
+
+pub struct Post {
+    pub id: i64,
+    pub user_id: i64,
+    pub created_at: DateTime<Utc>,
+    pub edited_at: Option<DateTime<Utc>>,
+    pub content: String,
+}
+
+pub struct Database(Connection);
+
+impl Database {
+    pub fn connect(path: PathBuf) -> Result<Self, Error> {
+        Ok(Self(Connection::open(path)?))
+    }
+
+    pub fn users(&mut self) -> Result<Vec<User>, Error> {
+        self.0
+            .prepare("SELECT `id`, `username` FROM `users`")?
+            .query_map([], |row| {
+                Ok(User {
+                    id: row.get(0)?,
+                    username: row.get(1)?,
+                })
+            })?
+            .collect()
+    }
+
+    pub fn tags(&mut self) -> Result<Vec<Tag>, Error> {
+        self.0
+            .prepare("SELECT `id`, `name`, `slug` FROM `tags`")?
+            .query_map([], |row| {
+                Ok(Tag {
+                    id: row.get(0)?,
+                    //name: row.get(1)?,
+                    slug: row.get(2)?,
+                })
+            })?
+            .collect()
+    }
+
+    pub fn discussions(&mut self) -> Result<Vec<Discussion>, Error> {
+        self.0.prepare(
+            "SELECT `id`, `user_id`, `first_post_id`, `created_at`, `title`, `slug`
+                FROM `discussions` WHERE `is_private` <> 1 AND `is_approved` <> 0 AND `hidden_at` IS NULL",
+        )?.query_map([], |row| {
+            Ok(Discussion {
+                id: row.get(0)?,
+                user_id: row.get(1)?,
+                first_post_id: row.get(2)?,
+                created_at: row.get(3)?,
+                title: row.get(4)?,
+                //slug: row.get(5)?,
+            })
+        })?
+        .collect()
+    }
+
+    pub fn discussion_tag_ids(&mut self, discussion_id: i64) -> Result<Vec<i64>, Error> {
+        self.0
+            .prepare("SELECT `tag_id` FROM `discussion_tag` WHERE `discussion_id` = ?")?
+            .query_map([discussion_id], |row| row.get(0))?
+            .collect()
+    }
+
+    pub fn posts(&mut self, discussion_id: i64) -> Result<Vec<Post>, Error> {
+        self.0.prepare(
+            "SELECT `id`, `user_id`, `created_at`, `edited_at`, `content`
+                FROM `posts` WHERE `discussion_id` = ? AND `is_private` <> 1 AND `is_approved` <> 0 AND `hidden_at` IS NULL
+                ORDER BY `number` ASC",
+        )?.query_map([discussion_id], |row| {
+            Ok(Post {
+                id: row.get(0)?,
+                user_id: row.get(1)?,
+                created_at: row.get(2)?,
+                edited_at: row.get(3)?,
+                content: row.get(4)?,
+            })
+        })?.collect()
+    }
+}

src/main.rs

@@ -0,0 +1,131 @@
+mod config;
+mod database;
+
+use anyhow::Result;
+use chrono::{DateTime, Utc};
+use clap::Parser;
+use config::Config;
+use database::Database;
+use html_to_markdown_rs::convert;
+use std::{
+    collections::HashMap,
+    fs::{File, create_dir_all, remove_dir_all},
+    io::Write,
+    path::PathBuf,
+};
+
+pub struct User {
+    pub username: String,
+}
+
+pub struct Post {
+    pub id: i64,
+    pub user_id: i64,
+    pub created_at: DateTime<Utc>,
+    pub edited_at: Option<DateTime<Utc>>,
+    pub content: String,
+}
+
+pub struct Discussion {
+    pub id: i64,
+    pub created_at: DateTime<Utc>,
+    pub title: String,
+    pub posts: Vec<Post>,
+}
+
+fn main() -> Result<()> {
+    let config = Config::parse();
+
+    if config.target.exists() {
+        remove_dir_all(&config.target)?
+    }
+    create_dir_all(&config.target)?;
+
+    let mut db = Database::connect(config.source)?;
+
+    let mut users = HashMap::new();
+    for user in db.users()? {
+        assert!(
+            users
+                .insert(
+                    user.id,
+                    User {
+                        username: user.username,
+                    }
+                )
+                .is_none()
+        )
+    }
+
+    let mut tags = HashMap::new();
+    for tag in db.tags()? {
+        if !config.filter_tag.is_empty() && !config.filter_tag.contains(&tag.slug) {
+            continue;
+        }
+        assert!(tags.insert(tag.id, tag.slug).is_none())
+    }
+
+    let mut discussions = Vec::new();
+    for discussion in db.discussions()? {
+        if !db
+            .discussion_tag_ids(discussion.id)?
+            .iter()
+            .any(|id| tags.contains_key(id))
+        {
+            continue;
+        }
+        assert!(users.contains_key(&discussion.user_id));
+
+        let mut posts = Vec::new();
+        for post in db.posts(discussion.id)? {
+            posts.push(Post {
+                id: post.id,
+                user_id: post.user_id,
+                created_at: post.created_at,
+                edited_at: post.edited_at,
+                content: post.content,
+            })
+        }
+        assert_eq!(discussion.first_post_id, posts.first().unwrap().id);
+        discussions.push(Discussion {
+            id: discussion.id,
+            created_at: discussion.created_at,
+            title: discussion.title,
+            posts,
+        })
+    }
+
+    for discussion in discussions {
+        let mut file = File::create_new({
+            let mut path = PathBuf::from(&config.target);
+            path.push(format!("{}.md", discussion.id));
+            path
+        })?;
+        file.write_all(
+            {
+                let mut page = Vec::new();
+                page.push(format!("# {}", discussion.title));
+                page.push({
+                    let mut content = Vec::new();
+                    for post in discussion.posts {
+                        content.push(format!(
+                            "@{} / {}{}",
+                            users.get(&post.user_id).unwrap().username,
+                            post.created_at,
+                            post.edited_at
+                                .map(|edited_at| format!(" / {}", edited_at))
+                                .unwrap_or_default()
+                        ));
+                        content.push("---".into());
+                        content.push(convert(&post.content, None)?)
+                    }
+                    content.join("\n")
+                });
+                page.join("\n")
+            }
+            .as_bytes(),
+        )?
+    }
+
+    Ok(())
+}