organic/src/event_count/database.rs

use std::collections::HashMap;
use std::sync::Mutex;

use super::EventType;
use crate::parser::OrgSource;

#[derive(Debug, Eq, Hash, PartialEq)]
struct EventKey {
    event_type: EventType,
    byte_offset: usize,
}

pub(crate) type EventCount = usize;

static GLOBAL_DATA: Mutex<Option<HashMap<EventKey, EventCount>>> = Mutex::new(None);

pub(crate) fn record_event(event_type: EventType, input: OrgSource<'_>) {
    let mut db = GLOBAL_DATA.lock().unwrap();
    let db = db.get_or_insert_with(HashMap::new);
    let key = EventKey {
        event_type,
        byte_offset: input.get_byte_offset(),
    };
    *db.entry(key).or_insert(0) += 1;
}

pub(crate) fn report(original_document: &str) {
    let mut db = GLOBAL_DATA.lock().unwrap();
    let db = db.get_or_insert_with(HashMap::new);
    let mut results: Vec<_> = db.iter().collect();
    results.sort_by_key(|(_k, v)| *v);
    // This would put the most common at the top, but that is a pain when there is already a lot of output from the parser.
    // results.sort_by(|(_ak, av), (_bk, bv)| bv.cmp(av));
    for (key, count) in results {
        println!(
            "{:?} {} character offset: {}    byte offset: {}",
            key.event_type,
            count,
            original_document[..key.byte_offset].chars().count() + 1,
            key.byte_offset
        )
    }
}