feat(cron::update) auto update git repo and update db index

This commit is contained in:
lumither committed 2024-11-27 00:18:08 -08:00
1 parent 99a19d8d9e
commit bacaea80a3
18 files changed
+882 -39

No files matched your search

@@ -1,24 +0,0 @@
use sqlx::{Pool, Postgres};
pub fn task(db_connection: &Pool<Postgres>) {
// get dir
// let git_work_dir_path = &env::var("GIT_WORK_DIR").unwrap_or_else(|e| {
// if let Ok(default_log_path) = env::var("WORK_DIR") {
// if default_log_path.starts_with('~') {
// let home_path = home_dir().unwrap_or_else(|| panic!("Cannot find home directory"));
// home_path
// .join(default_log_path.strip_prefix("~/").unwrap())
// .into_os_string()
// .into_string()
// .expect(
// "[fatal] failed to parse env `WORK_DIR`, try to use absolute path instead",
// )
// } else {
// default_log_path
// }
// } else {
// panic!("[fatal] failed to parse env `GIT_WORK_DIR`: {}", e);
// }
// });
println!("check update");
}
@@ -0,0 +1,49 @@
use thiserror::Error;
#[derive(Error, Debug)]
pub enum PostIdxError {
#[error("missing essential field `{}` for file `{}`", field, filename)]
MissingField { field: String, filename: String },
#[error(
"invalid field `{}` for file `{}`, <{}>: {}",
field_name,
filename,
err_field,
msg
)]
InvalidField {
field_name: String,
err_field: String,
filename: String,
msg: String,
},
#[error(
"database error on writing `{}` to table `{}`: <{}>: {}",
data_desc,
db_table,
id,
msg
)]
DBWriteFailure {
data_desc: String,
db_table: String,
id: String,
msg: String,
},
#[error(
"Database error on Reading `{}` from table `{}`: <{}>: {}",
data_desc,
db_table,
id,
msg
)]
DBReadFailure {
data_desc: String,
db_table: String,
id: String,
msg: String,
},
}
@@ -0,0 +1,333 @@
use std::error::Error;
use std::path::Path;
use crate::blog::cron::check_update::utils::get_md_file_basename;
use crate::blog::cron::check_update::{error::PostIdxError, utils::parse_date};
use markdown::MdFile;
use sqlx::{query, PgPool, Pool, Postgres, Row};
use tokio::runtime::Runtime;
use tracing::{info, warn};
async fn base(db_connection: &PgPool, post: &MdFile) -> Result<(), PostIdxError> {
let post_id = post.file_id;
let title = match post.meta.get("title") {
None => {
return Err(PostIdxError::MissingField {
field: "title".to_string(),
filename: post.filename.clone(),
})
}
Some(title) => title.as_str().unwrap_or_else(|| {
warn!("invalid title for `{}`", &post.filename);
"invalid title"
}),
};
let sub_title = post
.meta
.get("sub_title")
.and_then(|v| v.as_str())
.unwrap_or("");
let summary = post
.meta
.get("summary")
.and_then(|v| v.as_str())
.unwrap_or("");
let content = post.content.as_str();
let category = match post.meta.get("category") {
None => {
return Err(PostIdxError::MissingField {
field: "category".to_string(),
filename: post.filename.clone(),
})
}
Some(title) => title.as_str().unwrap_or_else(|| {
warn!("invalid category for `{}`", &post.filename);
"invalid category"
}),
};
let last_update = {
if let Some(last_update) = post.meta.get("last_update") {
match parse_date(last_update.as_str().unwrap()) {
Ok(date) => Some(date),
Err(e) => {
return Err(PostIdxError::InvalidField {
field_name: "last_update".to_string(),
err_field: last_update.to_string(),
filename: post.filename.clone(),
msg: e.to_string(),
})
}
}
} else {
None
}
};
let first_update = match post.meta.get("date") {
None => {
return Err(PostIdxError::MissingField {
field: "date".to_string(),
filename: post.filename.clone(),
})
}
Some(first_update) => match parse_date(first_update.as_str().unwrap()) {
Ok(date) => date,
Err(e) => {
return Err(PostIdxError::InvalidField {
field_name: "date".to_string(),
err_field: first_update.to_string(),
filename: post.filename.clone(),
msg: e.to_string(),
})
}
},
};
let filename = get_md_file_basename(&post.filename);
if let Err(e) = query(
r##"
INSERT INTO Post (post_id, title, summary, content, last_update, first_update, sub_title, category, filename)
VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9)
ON CONFLICT (post_id) DO
UPDATE SET
title = EXCLUDED.title,
summary = EXCLUDED.summary,
content = EXCLUDED.content,
last_update = EXCLUDED.last_update,
first_update = EXCLUDED.first_update,
sub_title = EXCLUDED.sub_title,
category = EXCLUDED.category,
filename = EXCLUDED.filename;
"##,
)
.bind(post_id)
.bind(title)
.bind(summary)
.bind(content)
.bind(last_update)
.bind(first_update)
.bind(sub_title)
.bind(category)
.bind(filename)
.execute(db_connection)
.await
{
return Err(
PostIdxError::DBWriteFailure {
data_desc: "post base information".to_string(),
db_table: "Post".to_string(),
id: post_id.to_string(),
msg: e.to_string(),
}
);
}
Ok(())
}
async fn tag(db_connection: &PgPool, post: &MdFile) -> Result<(), PostIdxError> {
if let Some(tags_string) = post.meta.get("tags").to_owned() {
let tags = match tags_string.as_array().map(|vec| {
vec.iter()
.map(|val| val.as_str().unwrap())
.collect::<Vec<_>>()
}) {
None => {
return Err(PostIdxError::MissingField {
field: "tags".to_string(),
filename: post.filename.clone(),
})
}
Some(val) => val,
};
let post_id = post.file_id;
// remove old record
if let Err(e) = query("DELETE FROM Tag WHERE post_id = $1")
.bind(post_id)
.execute(db_connection)
.await
{
return Err(PostIdxError::DBWriteFailure {
data_desc: "post tag information".to_string(),
db_table: "Tag".to_string(),
id: post_id.to_string(),
msg: e.to_string(),
});
}
// add new record
for tag in tags {
match query("INSERT INTO Tag VALUES ($1, $2);")
.bind(post_id)
.bind(tag)
.execute(db_connection)
.await
{
Ok(_) => {}
Err(e) => {
return Err(PostIdxError::DBWriteFailure {
data_desc: "post tag information".to_string(),
db_table: "Tag".to_string(),
id: post_id.to_string(),
msg: e.to_string(),
})
}
}
}
}
Ok(())
}
async fn hash(db_connection: &PgPool, post: &MdFile) -> Result<bool, PostIdxError> {
let base64_form = post.hash.clone();
match query("SELECT hash FROM Hash WHERE post_id = $1")
.bind(post.file_id)
.fetch_optional(db_connection)
.await
{
Ok(row) => {
let db_record: String = match row {
None => "".to_string(),
Some(row) => row.try_get("hash").unwrap_or("".to_string()),
};
if base64_form == db_record {
Ok(false)
} else {
if let Err(e) = query(
r#"
INSERT INTO Hash (post_id, hash) VALUES ($1, $2)
ON CONFLICT (post_id) DO UPDATE SET hash = EXCLUDED.hash"#,
)
.bind(post.file_id)
.bind(&base64_form)
.execute(db_connection)
.await
{
return Err(PostIdxError::DBWriteFailure {
data_desc: "post hash".to_string(),
db_table: "Hash".to_string(),
id: post.file_id.to_string(),
msg: e.to_string(),
});
}
Ok(true)
}
}
Err(e) => Err(PostIdxError::DBReadFailure {
data_desc: "post hash".to_string(),
db_table: "Hash".to_string(),
id: post.file_id.to_string(),
msg: e.to_string(),
}),
}
}
async fn meta(db_connection: &PgPool, post: &MdFile) -> Result<(), PostIdxError> {
if let Err(e) = query(
// todo: sanitization
format!(
r#"
DO
$$
DECLARE
_post_id UUID := '{}';
_meta_cols JSONB := '{}';
_exists BOOLEAN;
_col_name TEXT;
_col_value TEXT;
_sql TEXT;
_insert_columns TEXT := 'post_id';
_insert_values TEXT := quote_literal(_post_id);
_update_set TEXT := '';
_blacklist TEXT[] := ARRAY['tags'];
BEGIN
SELECT EXISTS(SELECT 1 FROM meta WHERE post_id = _post_id) INTO _exists;
FOR _col_name, _col_value IN SELECT key, value FROM jsonb_each_text(_meta_cols)
LOOP
IF _col_name = ANY(_blacklist) THEN
CONTINUE;
END IF;
IF NOT EXISTS (SELECT 1
FROM information_schema.columns
WHERE table_name = 'meta'
AND column_name = _col_name
-- may change in the future
AND table_schema = 'public') THEN
_sql := 'ALTER TABLE meta ADD COLUMN ' || quote_ident(_col_name) || ' TEXT';
EXECUTE _sql;
END IF;
_insert_columns := _insert_columns || ', ' || quote_ident(_col_name);
_insert_values := _insert_values || ', ' || quote_literal(_col_value);
_update_set := _update_set || ', ' || quote_ident(_col_name) || ' = ' || quote_literal(_col_value);
END LOOP;
_update_set := substring(_update_set FROM 3);
IF _exists THEN
_sql := 'UPDATE meta SET ' || _update_set || ' WHERE post_id = ' || quote_literal(_post_id);
EXECUTE _sql;
ELSE
_sql := 'INSERT INTO meta (' || _insert_columns || ') VALUES (' || _insert_values || ')';
EXECUTE _sql;
END IF;
END;
$$ LANGUAGE plpgsql;
"#, post.file_id.to_string().as_str(), post.meta
).as_str()
)
.execute(db_connection)
.await {
Err(PostIdxError::DBWriteFailure {
data_desc: "post meta data".to_string(),
db_table: "Meta".to_string(),
id: post.file_id.to_string(),
msg: e.to_string(),
})
} else {
Ok(())
}
}
pub async fn index_file(path: &Path, db_conn: &Pool<Postgres>) -> Result<(), Box<dyn Error>> {
let md_file = MdFile::from_file(path)?;
base(db_conn, &md_file).await?;
tag(db_conn, &md_file).await?;
meta(db_conn, &md_file).await?;
hash(db_conn, &md_file).await?;
info!("indexed markdown file: {}", path.display());
Ok(())
}
pub fn drop_index(file_basename: &str, db_conn: &Pool<Postgres>) -> Result<(), Box<dyn Error>> {
let rt = Runtime::new().unwrap();
rt.block_on(async {
if let Err(e) = query(
r#"
WITH deleted_post AS (
DELETE FROM post
WHERE id = (SELECT id FROM post WHERE filename = $1)
RETURNING id
)
DELETE FROM tag WHERE post_id IN (SELECT id FROM deleted_post);
DELETE FROM meta WHERE post_id IN (SELECT id FROM deleted_post);
DELETE FROM hash WHERE post_id IN (SELECT id FROM deleted_post);
"#,
)
.bind(file_basename)
.execute(db_conn)
.await
{
Err(Box::from(
format!("failed to drop index: {}", e).to_string(),
))
} else {
Ok(())
}
})
}
@@ -0,0 +1,144 @@
use std::env;
use std::fs::remove_dir_all;
use std::path::{Path, PathBuf};
use crate::{
blog::cron::check_update::{
index::drop_index,
utils::{expand_path, get_md_file_basename, index_files, search_md},
},
utils::git,
};
use constants::config::general::{default_value, var_name};
use macros::panic_with_log;
use git2::{Delta, Repository};
use sqlx::{Pool, Postgres};
use tracing::{error, info, warn, Level};
mod error;
mod index;
mod utils;
pub fn task(db_connection: &Pool<Postgres>) {
// init
let git_work_dir = {
let path = if let Ok(default_log_path) = env::var(var_name::WORK_DIR) {
expand_path(default_log_path)
} else {
let fallback_path = expand_path(default_value::WORK_DIR.to_string());
warn!(
"unset env var `{}`, using default path: {}/blog_git",
var_name::WORK_DIR,
fallback_path
);
fallback_path
};
Path::new(&path).join("blog_git")
};
// clone & check update
let git_url = match env::var(var_name::BLOG_GIT_URL) {
Ok(url) => url,
Err(e) => {
panic_with_log!(
Level::ERROR,
"failed to read `{}`: {}",
var_name::BLOG_GIT_URL,
e
)
}
};
let git_remote_name = env::var(var_name::BLOG_GIT_REMOTE_NAME)
.unwrap_or(default_value::BLOG_GIT_REMOTE_NAME.to_string());
let git_remote_branch = env::var(var_name::BLOG_GIT_REMOTE_BRANCH)
.unwrap_or(default_value::BLOG_GIT_REMOTE_BRANCH.to_string());
match Repository::open(&git_work_dir) {
Ok(repo) => {
// prevent git remote url mismatch (clean cache)
match repo.find_remote(&git_remote_name) {
Ok(remote) => {
if remote.url().is_some_and(|url| url != git_url) {
remove_dir_all(&git_work_dir).unwrap();
warn!("blog content git({}) url mismatch, reinitialized", &git_url);
return;
}
}
Err(e) => {
remove_dir_all(&git_work_dir).unwrap();
error!(
"unexpected error on indexing blog content git({}), reinitialized: {}",
&git_url, e
);
}
}
match git::sync(&git_remote_name, &git_remote_branch, &repo) {
Ok(updates) => {
if !updates.is_empty() {
// incremental update
info!(
"blog content git({}) update detected: {:?}",
&git_url, updates
);
let mut update_md: Vec<PathBuf> = vec![];
for update in updates {
if [Delta::Deleted, Delta::Untracked].contains(&update.status) {
if let Some(path) = &update.old_path {
if drop_index(
&get_md_file_basename(
path.file_name().unwrap().to_str().unwrap(),
),
db_connection,
)
.is_ok()
{
info!(
"blog content git({}) file `{}` is detected as deleted/untracked, index dropped",
&git_url,
&path.display()
);
} else {
error!(
"blog content git({}) file `{}` is detected as deleted/untracked, failed to drop index",
&git_url,
&path.display()
)
}
}
} else if let Some(path) = &update.new_path {
if path.extension().is_some_and(|ext| ext == "md") {
update_md.push(path.to_path_buf());
}
}
}
index_files(&update_md, db_connection);
}
}
Err(e) => {
error!(
"failed to sync blog content git({}) updates: {}",
&git_url, e
);
}
}
}
Err(_) => {
match Repository::clone(&git_url, &git_work_dir) {
Ok(_repo) => {
// full repo index
let files = search_md(git_work_dir);
index_files(&files, db_connection);
}
Err(e) => {
error!("failed to clone blog content git({}): {}", &git_url, e);
}
}
}
};
}
@@ -0,0 +1,86 @@
use dirs::home_dir;
use std::path::PathBuf;
use crate::blog::cron::check_update::index::index_file;
use chrono::{DateTime, FixedOffset, ParseResult};
use constants::config::general::{default_value, var_name};
use constants::time::DEFAULT_DATE_FORMAT;
use futures::executor::block_on;
use futures::future::join_all;
use sqlx::{Pool, Postgres};
use tokio::runtime::Runtime;
use tokio::spawn;
use tracing::{error, warn};
pub fn search_md(entry: PathBuf) -> Vec<PathBuf> {
if entry.ends_with(".git") {
return vec![];
}
if entry.is_dir() {
if let Ok(files) = entry.read_dir() {
files
.flat_map(|file| {
let file_path = file.unwrap().path();
search_md(file_path)
})
.collect()
} else {
vec![]
}
} else if entry.extension().is_some_and(|ext| ext == "md") {
vec![entry]
} else {
vec![]
}
}
pub fn index_files(files: &[PathBuf], db_conn: &Pool<Postgres>) {
let rt = Runtime::new().unwrap();
rt.block_on(async {
let files_clone = files.to_owned();
let handlers = files_clone
.into_iter()
.map(|file: PathBuf| {
let conn_clone = db_conn.clone();
spawn(async move {
if let Err(e) = index_file(&file, &conn_clone).await {
error!("failed to index markdown file `{}`: {}", file.display(), e);
}
})
})
.collect::<Vec<_>>();
block_on(join_all(handlers));
});
}
pub fn expand_path(path: String) -> String {
if path.starts_with('~') {
home_dir()
.map(|dir| {
dir.join(path.strip_prefix("~/").unwrap())
.into_os_string()
.into_string()
.unwrap()
})
.unwrap_or_else(|| {
warn!(
"failed to parse env var `{}`, using default value `{}`",
var_name::WORK_DIR,
default_value::WORK_DIR
);
default_value::WORK_DIR.to_string()
})
} else {
path
}
}
pub fn parse_date(date_str: &str) -> ParseResult<DateTime<FixedOffset>> {
DateTime::parse_from_str(format!("{}T120000", date_str).as_str(), DEFAULT_DATE_FORMAT)
}
pub fn get_md_file_basename(f_name: &str) -> String {
f_name.replace(".md", "")
}
+14
View File
@@ -0,0 +1,14 @@
use git2::opts::{set_server_connect_timeout_in_milliseconds, set_server_timeout_in_milliseconds};
use tracing::warn;
pub fn init() {
// git2 option
unsafe {
if let Err(e) = set_server_connect_timeout_in_milliseconds(10_000) {
warn!("failed to set git2 server connect timeout: {}", e);
}
if let Err(e) = set_server_timeout_in_milliseconds(60_000) {
warn!("failed to set git2 server timeout: {}", e);
}
}
}
+3
View File
@@ -1,6 +1,7 @@
mod cron;
mod database;
pub mod logger;
mod misc;
mod schema;
pub mod server;
@@ -24,6 +25,8 @@ pub fn load_env() {
}
pub async fn start() {
misc::init();
// connect database
let db_pool = match database::init().await {
Ok(pool) => {
+2 -2
View File
@@ -7,11 +7,11 @@ CREATE TABLE IF NOT EXISTS Post
post_id UUID PRIMARY KEY,
title VARCHAR(255) NOT NULL,
category VARCHAR(255) NOT NULL,
header_img TEXT,
filename TEXT UNIQUE NOT NULL,
summary TEXT,
sub_title TEXT,
content TEXT NOT NULL,
last_update TIMESTAMPTZ NOT NULL,
last_update TIMESTAMPTZ,
first_update TIMESTAMPTZ NOT NULL
);
"##;
+1
View File
@@ -1,6 +1,7 @@
mod blog;
mod init;
mod scheduler;
mod utils;
#[tokio::main]
async fn main() {
+195
View File
@@ -0,0 +1,195 @@
/// git merge
/// based on https://github.com/rust-lang/git2-rs/blob/master/examples/pull.rs
/// https://github.com/rust-lang/git2-rs/commit/f3b87baed1e33d6c2d94fe1fa6aa6503a071d837
///
use git2::{
build, AnnotatedCommit, AutotagOption, Delta, FetchOptions, Reference, Remote, Repository,
};
use std::path::PathBuf;
use tracing::{error, warn};
#[derive(Debug, Eq, PartialEq)]
pub struct FileDelta {
pub old_path: Option<PathBuf>,
pub new_path: Option<PathBuf>,
pub status: Delta,
}
fn do_fetch<'a>(
repo: &'a Repository,
refs: &[&str],
remote: &'a mut Remote,
) -> Result<AnnotatedCommit<'a>, git2::Error> {
let mut fetch_option = FetchOptions::new();
// Always fetch all tags.
// Perform a download and also update tips
fetch_option.download_tags(AutotagOption::All);
remote.fetch(refs, Some(&mut fetch_option), None)?;
let fetch_head = repo.find_reference("FETCH_HEAD")?;
repo.reference_to_annotated_commit(&fetch_head)
}
fn fast_forward(
repo: &Repository,
lb: &mut Reference,
rc: &AnnotatedCommit,
) -> Result<(), git2::Error> {
let name = match lb.name() {
Some(s) => s.to_string(),
None => String::from_utf8_lossy(lb.name_bytes()).to_string(),
};
let msg = format!("fast-forward({}) -> {}", name, rc.id());
// println!("{}", msg);
lb.set_target(rc.id(), &msg)?;
repo.set_head(&name)?;
repo.checkout_head(Some(
build::CheckoutBuilder::default()
// For some reason, the force is required to make the working directory actually get updated
// I suspect we should be adding some logic to handle dirty working directory states,
// but this is just an example so maybe not.
.force(),
))?;
Ok(())
}
fn normal_merge(
repo: &Repository,
local: &AnnotatedCommit,
remote: &AnnotatedCommit,
) -> Result<(), git2::Error> {
let local_tree = repo.find_commit(local.id())?.tree()?;
let remote_tree = repo.find_commit(remote.id())?.tree()?;
let ancestor = repo
.find_commit(repo.merge_base(local.id(), remote.id())?)?
.tree()?;
let mut idx = repo.merge_trees(&ancestor, &local_tree, &remote_tree, None)?;
if idx.has_conflicts() {
warn!("merge conflicts detected");
repo.checkout_index(Some(&mut idx), None)?;
return Ok(());
}
let result_tree = repo.find_tree(idx.write_tree_to(repo)?)?;
// now create the merge commit
let msg = format!("merge({}) -> {}", remote.id(), local.id());
let sig = repo.signature()?;
let local_commit = repo.find_commit(local.id())?;
let remote_commit = repo.find_commit(remote.id())?;
// Do our merge commit and set the current branch head to that commit.
let _merge_commit = repo.commit(
Some("HEAD"),
&sig,
&sig,
&msg,
&result_tree,
&[&local_commit, &remote_commit],
)?;
// Set working tree to match head.
repo.checkout_head(None)?;
Ok(())
}
fn do_merge<'a>(
repo: &'a Repository,
remote_branch: &str,
fetch_commit: AnnotatedCommit<'a>,
) -> Result<Vec<FileDelta>, git2::Error> {
// 1. do a merge analysis
let analysis = repo.merge_analysis(&[&fetch_commit])?;
let head_commit = repo.reference_to_annotated_commit(&repo.head()?)?;
// 2. Do the appropriate merge
if analysis.0.is_fast_forward() {
// println!("Doing a fast-forward");
// do a fast-forward
let ref_name = format!("refs/heads/{}", remote_branch);
match repo.find_reference(&ref_name) {
Ok(mut r) => {
fast_forward(repo, &mut r, &fetch_commit)?;
}
Err(_) => {
// The branch doesn't exist, so set the reference to the
// commit directly. Usually this is because you are pulling
// into an empty repository.
repo.reference(
&ref_name,
fetch_commit.id(),
true,
&format!("({}) -> {}", remote_branch, fetch_commit.id()),
)?;
repo.set_head(&ref_name)?;
repo.checkout_head(Some(
build::CheckoutBuilder::default()
.allow_conflicts(true)
.conflict_style_merge(true)
.force(),
))?;
}
};
} else if analysis.0.is_normal() {
// do a normal merge
normal_merge(repo, &head_commit, &fetch_commit)?;
}
// else {
// // already up-to-date
// println!("Nothing to do...");
// }
calculate_diff(repo, &head_commit, &fetch_commit)
}
fn calculate_diff(
repo: &Repository,
local: &AnnotatedCommit,
remote: &AnnotatedCommit,
) -> Result<Vec<FileDelta>, git2::Error> {
let local_tree = repo.find_commit(local.id())?.tree()?;
let remote_tree = repo.find_commit(remote.id())?.tree()?;
let diff = repo.diff_tree_to_tree(Some(&local_tree), Some(&remote_tree), None)?;
let mut updated_files: Vec<FileDelta> = Vec::new();
if let Err(e) = diff.foreach(
&mut |d, _| {
// let delta = d.status();
//
// let n_path = d.new_file().path();
// let o_path = d.old_file().path();
//
// if let Some(path) = n_path {
// updated_files.insert(path.to_owned());
// }
// if let Some(path) = o_path {
// updated_files.insert(path.to_owned());
// }
updated_files.push(FileDelta {
old_path: d.old_file().path().map(PathBuf::from),
new_path: d.new_file().path().map(PathBuf::from),
status: d.status(),
});
true // Continue
},
None,
None,
None,
) {
error!("error at calculating diff: {:?}", e);
}
Ok(updated_files.into_iter().collect())
}
pub fn sync(
remote_name: &str,
remote_branch: &str,
repo: &Repository,
) -> Result<Vec<FileDelta>, git2::Error> {
let mut remote = repo.find_remote(remote_name)?;
let fetch_commit = do_fetch(repo, &[remote_branch], &mut remote)?;
do_merge(repo, remote_branch, fetch_commit)
}
+1
View File
@@ -0,0 +1 @@
pub mod git;