finish timeago parser, refactor codegen

src/codegen/gen_dictionary.rs

@@ -0,0 +1,102 @@
+#![cfg(test)]
+
+use std::{
+    collections::BTreeMap,
+    fmt::Debug,
+    fs::File,
+    io::{BufReader},
+};
+
+use crate::{model::Language, timeago::TimeUnit};
+use fancy_regex::Regex;
+use once_cell::sync::Lazy;
+use serde::Deserialize;
+
+const DICT_PATH: &str = "testfiles/date/dictionary.json";
+const TARGET_FILE: &str = "src/dictionary.rs";
+
+type Dictionary = BTreeMap<Language, DictEntry>;
+
+#[derive(Debug, Deserialize)]
+struct DictEntry {
+    #[serde(default)]
+    equivalent: Vec<Language>,
+    #[serde(default)]
+    by_char: bool,
+    timeago_tokens: BTreeMap<String, String>,
+}
+
+fn parse_tu(tu: &str) -> (u8, Option<TimeUnit>) {
+    static TU_PATTERN: Lazy<Regex> = Lazy::new(|| Regex::new(r"^(\d*)(\w?)$").unwrap());
+    match TU_PATTERN.captures(tu).unwrap() {
+        Some(cap) => (
+            cap.get(1).unwrap().as_str().parse().unwrap_or(1),
+            match cap.get(2).unwrap().as_str() {
+                "s" => Some(TimeUnit::Second),
+                "m" => Some(TimeUnit::Minute),
+                "h" => Some(TimeUnit::Hour),
+                "D" => Some(TimeUnit::Day),
+                "W" => Some(TimeUnit::Week),
+                "M" => Some(TimeUnit::Month),
+                "Y" => Some(TimeUnit::Year),
+                "" => None,
+                _ => panic!("invalid time unit: {}", tu),
+            },
+        ),
+        None => panic!("invalid time unit: {}", tu),
+    }
+}
+
+fn read_dict() -> Dictionary {
+    let json_file = File::open(DICT_PATH).unwrap();
+    serde_json::from_reader(BufReader::new(json_file)).unwrap()
+}
+
+// #[test]
+fn generate_dictionary() {
+    let dict = read_dict();
+
+    let code_head = r#"// This file is automatically generated. DO NOT EDIT.
+use crate::{
+    model::Language,
+    timeago::{TaToken, TimeUnit},
+};
+"#;
+
+    let mut code_timeago_tokens = r#"#[rustfmt::skip]
+pub(crate) fn get_timeago_tokens(lang: Language) -> phf::Map<&'static str, TaToken> {
+    match lang {
+        "#
+    .to_owned();
+
+    dict.iter().for_each(|(lang, entry)| {
+        // Create a map for the language
+        let mut map = phf_codegen::Map::<&str>::new();
+
+        entry.timeago_tokens.iter().for_each(|(txt, tu_str)| {
+            let (n, unit) = parse_tu(&tu_str);
+            match unit {
+                Some(unit) => map.entry(
+                    &txt,
+                    &format!("TaToken {{ n: {}, unit: Some(TimeUnit::{:?}) }}", n, unit),
+                ),
+                None => map.entry(&txt, &format!("TaToken {{ n: {}, unit: None }}", n)),
+            };
+        });
+
+        let mut selector = format!("Language::{:?}", lang);
+        entry.equivalent.iter().for_each(|eq| {
+            selector += &format!(" | Language::{:?}", eq);
+        });
+
+        let code_map = &map.build().to_string().replace('\n', "\n        ");
+
+        code_timeago_tokens += &format!("{} => {},\n        ", selector, code_map);
+    });
+
+    code_timeago_tokens = code_timeago_tokens.trim_end().to_owned() + "\n    }\n}\n";
+
+    let code = format!("{}\n{}", code_head, code_timeago_tokens);
+
+    std::fs::write(TARGET_FILE, code).unwrap();
+}