procmail2notmuch: add sql rule loader

This commit is contained in:
Bill Thiede 2025-04-20 08:40:06 -07:00
parent 1c4f27902e
commit 7de36bbc3d
5 changed files with 140 additions and 92 deletions

3
Cargo.lock generated
View File

@ -3014,6 +3014,9 @@ version = "0.15.9"
dependencies = [
"anyhow",
"clap",
"serde",
"sqlx",
"tokio 1.44.2",
]
[[package]]

View File

@ -12,4 +12,7 @@ version.workspace = true
[dependencies]
anyhow = "1.0.69"
clap = { version = "4.5.37", features = ["derive"] }
clap = { version = "4.5.37", features = ["derive", "env"] }
serde = { version = "1.0.219", features = ["derive"] }
sqlx = { version = "0.8.5", features = ["postgres", "runtime-tokio"] }
tokio = { version = "1.44.2", features = ["rt", "macros", "rt-multi-thread"] }

View File

@ -1,8 +1,12 @@
use std::{collections::HashMap, convert::Infallible, io::Write, str::FromStr};
use clap::{Parser, ValueEnum};
use clap::{Parser, Subcommand};
use serde::{Deserialize, Serialize};
use sqlx::{types::Json, PgPool};
#[derive(Copy, Clone, Debug, Default, PartialEq, Eq, Hash, Ord, PartialOrd)]
#[derive(
Copy, Clone, Debug, Default, PartialEq, Eq, Hash, Ord, PartialOrd, Serialize, Deserialize,
)]
enum MatchType {
From,
Sender,
@ -19,16 +23,16 @@ enum MatchType {
#[default]
Unknown,
}
#[derive(Debug, Default)]
#[derive(Debug, Default, Serialize, Deserialize)]
struct Match {
match_type: MatchType,
needle: String,
}
#[derive(Debug, Default)]
#[derive(Debug, Default, Serialize, Deserialize)]
struct Rule {
matches: Vec<Match>,
tags: Option<String>,
tag: Option<String>,
}
fn unescape(s: &str) -> String {
@ -155,13 +159,109 @@ impl FromStr for Match {
}
}
#[derive(Debug, Subcommand)]
enum Mode {
Debug,
Notmuchrc,
LoadSql {
#[arg(short, long, default_value = env!("DATABASE_URL"))]
dsn: String,
},
}
/// Simple program to greet a person
#[derive(Parser, Debug)]
#[command(version, about, long_about = None)]
struct Args {
#[arg(short, long, default_value = "/home/wathiede/dotfiles/procmailrc")]
input: String,
#[command(subcommand)]
mode: Mode,
}
#[tokio::main]
async fn main() -> anyhow::Result<()> {
let args = Args::parse();
let mut rules = Vec::new();
let mut cur_rule = Rule::default();
for l in std::fs::read_to_string(args.input)?.lines() {
let l = if let Some(idx) = l.find('#') {
&l[..idx]
} else {
l
}
.trim();
if l.is_empty() {
continue;
}
if l.find('=').is_some() {
// Probably a variable assignment, skip line
continue;
}
let first = l.chars().nth(0).unwrap_or(' ');
match first {
':' => {
// start of rule
}
'*' => {
// add to current rule
let m: Match = l.parse()?;
cur_rule.matches.push(m);
}
'.' => {
// delivery to folder
cur_rule.tag = Some(cleanup_match(
"",
&l.replace('.', "/")
.replace(' ', "")
.trim_matches('/')
.to_string(),
));
rules.push(cur_rule);
cur_rule = Rule::default();
}
'/' => cur_rule = Rule::default(), // Ex. /dev/null
'|' => cur_rule = Rule::default(), // external command
'$' => {
// TODO(wathiede): tag messages with no other tag as 'inbox'
cur_rule.tag = Some(cleanup_match("", "inbox"));
rules.push(cur_rule);
cur_rule = Rule::default();
} // variable, should only be $DEFAULT in my config
_ => panic!("Unhandled first character '{}'\nLine: {}", first, l),
}
}
match args.mode {
Mode::Debug => print_rules(&rules),
Mode::Notmuchrc => notmuch_from_rules(std::io::stdout(), &rules)?,
Mode::LoadSql { dsn } => load_sql(&dsn, &rules).await?,
}
Ok(())
}
fn print_rules(rules: &[Rule]) {
let mut tally = HashMap::new();
for r in rules {
for m in &r.matches {
*tally.entry(m.match_type).or_insert(0) += 1;
}
}
let mut sorted: Vec<_> = tally.iter().map(|(k, v)| (v, k)).collect();
sorted.sort();
sorted.reverse();
for (v, k) in sorted {
println!("{k:?}: {v}");
}
}
fn notmuch_from_rules<W: Write>(mut w: W, rules: &[Rule]) -> anyhow::Result<()> {
// TODO(wathiede): if reindexing this many tags is too slow, see if combining rules per tag is
// faster.
let mut lines = Vec::new();
for r in rules {
for m in &r.matches {
if let Some(t) = &r.tags {
if let Some(t) = &r.tag {
if let MatchType::Unknown = m.match_type {
eprintln!("rule has unknown match {:?}", r);
continue;
@ -206,92 +306,26 @@ fn notmuch_from_rules<W: Write>(mut w: W, rules: &[Rule]) -> anyhow::Result<()>
Ok(())
}
#[derive(Debug, Copy, Clone, PartialEq, Eq, PartialOrd, Ord, ValueEnum)]
enum Mode {
Debug,
Notmuchrc,
}
async fn load_sql(dsn: &str, rules: &[Rule]) -> anyhow::Result<()> {
let pool = PgPool::connect(dsn).await?;
sqlx::migrate!("../server/migrations").run(&pool).await?;
println!("clearing email_rule table");
sqlx::query!("DELETE FROM email_rule")
.execute(&pool)
.await?;
/// Simple program to greet a person
#[derive(Parser, Debug)]
#[command(version, about, long_about = None)]
struct Args {
#[arg(short, long, default_value = "/home/wathiede/dotfiles/procmailrc")]
input: String,
#[arg(value_enum)]
mode: Mode,
}
fn main() -> anyhow::Result<()> {
let args = Args::parse();
let mut rules = Vec::new();
let mut cur_rule = Rule::default();
for l in std::fs::read_to_string(args.input)?.lines() {
let l = if let Some(idx) = l.find('#') {
&l[..idx]
} else {
l
}
.trim();
if l.is_empty() {
continue;
}
if l.find('=').is_some() {
// Probably a variable assignment, skip line
continue;
}
let first = l.chars().nth(0).unwrap_or(' ');
match first {
':' => {
// start of rule
}
'*' => {
// add to current rule
let m: Match = l.parse()?;
cur_rule.matches.push(m);
}
'.' => {
// delivery to folder
cur_rule.tags = Some(cleanup_match(
"",
&l.replace('.', "/")
.replace(' ', "")
.trim_matches('/')
.to_string(),
));
rules.push(cur_rule);
cur_rule = Rule::default();
}
'/' => cur_rule = Rule::default(), // Ex. /dev/null
'|' => cur_rule = Rule::default(), // external command
'$' => {
// TODO(wathiede): tag messages with no other tag as 'inbox'
cur_rule.tags = Some(cleanup_match("", "inbox"));
rules.push(cur_rule);
cur_rule = Rule::default();
} // variable, should only be $DEFAULT in my config
_ => panic!("Unhandled first character '{}'\nLine: {}", first, l),
}
}
match args.mode {
Mode::Debug => print_rules(&rules),
Mode::Notmuchrc => notmuch_from_rules(std::io::stdout(), &rules)?,
for (order, rule) in rules.iter().enumerate() {
println!("inserting {order}: {rule:?}");
sqlx::query!(
r#"
INSERT INTO email_rule (sort_order, rule)
VALUES ($1, $2)
"#,
order as i32,
Json(rule) as _
)
.execute(&pool)
.await?;
}
Ok(())
}
fn print_rules(rules: &[Rule]) {
let mut tally = HashMap::new();
for r in rules {
for m in &r.matches {
*tally.entry(m.match_type).or_insert(0) += 1;
}
}
let mut sorted: Vec<_> = tally.iter().map(|(k, v)| (v, k)).collect();
sorted.sort();
sorted.reverse();
for (v, k) in sorted {
println!("{k:?}: {v}");
}
}

View File

@ -0,0 +1,3 @@
DROP TABLE IF NOT EXISTS email_rule;
-- Add down migration script here

View File

@ -0,0 +1,5 @@
CREATE TABLE IF NOT EXISTS email_rule (
id integer NOT NULL GENERATED ALWAYS AS IDENTITY,
sort_order integer NOT NULL,
rule jsonb NOT NULL
);