Get dictionary path from environment variables.

akaza-im · Jan 16, 2023 · 1bca98f · 1bca98f
1 parent d0792b1
commit 1bca98f
Show file tree

Hide file tree

Showing 9 changed files with 62 additions and 53 deletions.
diff --git a/akaza-data/src/main.rs b/akaza-data/src/main.rs
@@ -79,8 +79,6 @@ struct EvaluateArgs {
     load_user_config: bool,
     /// コーパスが格納されているディレクトリ
     corpus_dir: String,
-    /// 評価に利用するシステムデータのディレクトリ
-    system_data_dir: String,
 }
 
 /// 動作確認する
@@ -206,9 +204,7 @@ fn main() -> anyhow::Result<()> {
         Commands::MakeSystemDict(opt) => {
             make_system_dict(&opt.txt_file, Some(opt.vocab_file.as_str()), opt.corpus)
         }
-        Commands::Evaluate(opt) => {
-            evaluate(&opt.corpus_dir, &opt.system_data_dir, opt.load_user_config)
-        }
+        Commands::Evaluate(opt) => evaluate(&opt.corpus_dir, opt.load_user_config),
         Commands::Check(opt) => check(&opt.yomi, opt.expected, opt.user_data),
         Commands::LearnCorpus(opts) => learn_corpus(
             opts.delta,

diff --git a/akaza-data/src/subcmd/check.rs b/akaza-data/src/subcmd/check.rs
@@ -12,14 +12,12 @@ use libakaza::engine::bigram_word_viterbi_engine::BigramWordViterbiEngineBuilder
 use libakaza::user_side_data::user_data::UserData;
 
 pub fn check(yomi: &str, expected: Option<String>, user_data: bool) -> anyhow::Result<()> {
-    let datadir = env!("CARGO_MANIFEST_DIR").to_string() + "/data/";
-
     let dict = merge_dict(vec![
         read_skkdict(Path::new("skk-dev-dict/SKK-JISYO.L"), EUC_JP)?,
         read_skkdict(Path::new("data/SKK-JISYO.akaza"), UTF_8)?,
     ]);
 
-    let mut builder = BigramWordViterbiEngineBuilder::new(&datadir, Some(dict), None);
+    let mut builder = BigramWordViterbiEngineBuilder::new(Some(dict), None);
     if user_data {
         info!("Enabled user data");
         let user_data = UserData::load_from_default_path()?;

diff --git a/akaza-data/src/subcmd/evaluate.rs b/akaza-data/src/subcmd/evaluate.rs
@@ -48,11 +48,7 @@ impl SaigenRitsu {
 /// にのっている評価方法を採用。
 ///
 /// なぜこうしているかというと、mozc の論文にのっている BLEU を使用する方式より実装が楽だからです!
-pub fn evaluate(
-    corpus_dir: &String,
-    system_data_dir: &str,
-    load_user_config: bool,
-) -> anyhow::Result<()> {
+pub fn evaluate(corpus_dir: &String, load_user_config: bool) -> anyhow::Result<()> {
     /*
     # corpus.0.txt デバッグ用のファイル
     # corpus.1.txt メイン(候補割り当ても含む)
@@ -75,7 +71,7 @@ pub fn evaluate(
         read_skkdict(Path::new("data/SKK-JISYO.akaza"), UTF_8)?,
     ]);
 
-    let akaza = BigramWordViterbiEngineBuilder::new(system_data_dir, Some(dicts), None)
+    let akaza = BigramWordViterbiEngineBuilder::new(Some(dicts), None)
         .load_user_config(load_user_config)
         .build()?;
 

diff --git a/ibus-akaza/Makefile b/ibus-akaza/Makefile
@@ -4,19 +4,14 @@ DATADIR ?= $(PREFIX)/share
 DESTDIR ?=
 AKAZA_DATA_DIR ?= $(DATADIR)/akaza/
 
-all: akaza.xml config.h src/config.rs
+all: akaza.xml config.h
 
 #   ibus_akaza/config.py: ibus_akaza/config.py.in
 #   	sed -e "s:@SYSCONFDIR@:$(SYSCONFDIR):g" \
 #   		-e "s:@MODELDIR@:$(DESTDIR)/$(DATADIR)/akaza-data/:g" \
 #   		-e "s:@DICTIONARYDIR@:$(DESTDIR)/$(DATADIR)/ibus-akaza/dictionary:g" \
 #   		$< > $@
 
-src/config.rs: src/config.rs.in
-	sed \
-	    -e "s:@AKAZA_DATA_DIR@:$(AKAZA_DATA_DIR):g" \
-	    -e "s:@DATADIR@:$(DATADIR)/:g" $< > $@
-
 
 config.h: config.h.in
 	sed \

diff --git a/ibus-akaza/ibus-akaza-debug.sh b/ibus-akaza/ibus-akaza-debug.sh
@@ -8,6 +8,8 @@ umask 077
 exec 1>> ~/.ibus-akaza.log
 exec 2>&1
 
+export AKAZA_DATA_DIR="$BASEDIR/../akaza-data/data/"
+
 export RUST_BACKTRACE=4
 
 exec $BASEDIR/../target/release/ibus-akaza --ibus -vv
diff --git a/ibus-akaza/src/config.rs.in b/ibus-akaza/src/config.rs.in
diff --git a/ibus-akaza/src/main.rs b/ibus-akaza/src/main.rs
@@ -11,7 +11,6 @@ use anyhow::Result;
 use clap::Parser;
 use log::{error, info, warn};
 
-use crate::config::AKAZA_DATA_DIR;
 use ibus_sys::core::ibus_main;
 use ibus_sys::engine::IBusEngine;
 use ibus_sys::glib::{gchar, guint};
@@ -22,7 +21,6 @@ use crate::context::AkazaContext;
 use crate::wrapper_bindings::{ibus_akaza_init, ibus_akaza_set_callback};
 
 mod commands;
-mod config;
 mod context;
 mod input_mode;
 mod keymap;
@@ -103,7 +101,7 @@ fn main() -> Result<()> {
     unsafe {
         let sys_time = SystemTime::now();
         let user_data = load_user_data();
-        let akaza = BigramWordViterbiEngineBuilder::new(AKAZA_DATA_DIR, None, None)
+        let akaza = BigramWordViterbiEngineBuilder::new(None, None)
             .user_data(user_data.clone())
             .load_user_config(true)
             .build()?;

diff --git a/libakaza/src/engine/bigram_word_viterbi_engine.rs b/libakaza/src/engine/bigram_word_viterbi_engine.rs
@@ -1,16 +1,20 @@
 use std::collections::vec_deque::VecDeque;
 use std::collections::HashMap;
+use std::env;
 use std::ops::Range;
+use std::path::{Path, PathBuf};
 use std::rc::Rc;
 use std::sync::{Arc, Mutex};
 use std::time::SystemTime;
 
-use anyhow::Result;
+use anyhow::{bail, Result};
+use encoding_rs::UTF_8;
 use log::{info, warn};
 
 use crate::config::Config;
 use crate::dict::loader::load_dicts;
 use crate::dict::merge_dict::merge_dict;
+use crate::dict::skk::read::read_skkdict;
 use crate::engine::base::HenkanEngine;
 use crate::graph::graph_builder::GraphBuilder;
 use crate::graph::graph_resolver::{Candidate, GraphResolver};
@@ -23,27 +27,6 @@ use crate::lm::system_unigram_lm::MarisaSystemUnigramLM;
 use crate::romkan::RomKanConverter;
 use crate::user_side_data::user_data::UserData;
 
-pub struct SystemDataLoader {
-    pub system_unigram_lm: MarisaSystemUnigramLM,
-    pub system_bigram_lm: MarisaSystemBigramLM,
-}
-
-impl SystemDataLoader {
-    pub fn load(system_data_dir: &str) -> Result<SystemDataLoader> {
-        let system_unigram_lm = MarisaSystemUnigramLM::load(
-            (system_data_dir.to_string() + "/stats-vibrato-unigram.trie").as_str(),
-        )?;
-        let system_bigram_lm = MarisaSystemBigramLM::load(
-            (system_data_dir.to_string() + "/stats-vibrato-bigram.trie").as_str(),
-        )?;
-
-        Ok(SystemDataLoader {
-            system_unigram_lm,
-            system_bigram_lm,
-        })
-    }
-}
-
 /// バイグラムのビタビベースかな漢字変換エンジンです。
 /// 単語バイグラムを採用しています。
 pub struct BigramWordViterbiEngine<U: SystemUnigramLM, B: SystemBigramLM> {
@@ -139,7 +122,6 @@ impl<U: SystemUnigramLM, B: SystemBigramLM> BigramWordViterbiEngine<U, B> {
 }
 
 pub struct BigramWordViterbiEngineBuilder {
-    system_data_dir: String,
     user_data: Option<Arc<Mutex<UserData>>>,
     load_user_config: bool,
     dicts: Option<HashMap<String, Vec<String>>>,
@@ -148,12 +130,10 @@ pub struct BigramWordViterbiEngineBuilder {
 
 impl BigramWordViterbiEngineBuilder {
     pub fn new(
-        system_data_dir: &str,
         dicts: Option<HashMap<String, Vec<String>>>,
         single_term: Option<HashMap<String, Vec<String>>>,
     ) -> BigramWordViterbiEngineBuilder {
         BigramWordViterbiEngineBuilder {
-            system_data_dir: system_data_dir.to_string(),
             user_data: None,
             load_user_config: false,
             dicts,
@@ -176,7 +156,19 @@ impl BigramWordViterbiEngineBuilder {
     pub fn build(
         &self,
     ) -> Result<BigramWordViterbiEngine<MarisaSystemUnigramLM, MarisaSystemBigramLM>> {
-        let system_data_loader = SystemDataLoader::load(self.system_data_dir.as_str())?;
+        let system_unigram_lm = MarisaSystemUnigramLM::load(
+            Self::try_load("stats-vibrato-unigram.trie")?
+                .to_string_lossy()
+                .to_string()
+                .as_str(),
+        )?;
+        let system_bigram_lm = MarisaSystemBigramLM::load(
+            Self::try_load("stats-vibrato-bigram.trie")?
+                .to_string_lossy()
+                .to_string()
+                .as_str(),
+        )?;
+        let system_dict = read_skkdict(Self::try_load("SKK-JISYO.akaza")?.as_path(), UTF_8)?;
 
         let user_data = if let Some(d) = &self.user_data {
             d.clone()
@@ -193,6 +185,7 @@ impl BigramWordViterbiEngineBuilder {
                 Config::default()
             };
             let dicts = load_dicts(&config.dicts)?;
+            let dicts = merge_dict(vec![system_dict, dicts]);
             let single_term = if let Some(st) = &config.single_term {
                 load_dicts(st)?
             } else {
@@ -235,8 +228,8 @@ impl BigramWordViterbiEngineBuilder {
             dict,
             single_term,
             user_data.clone(),
-            Rc::new(system_data_loader.system_unigram_lm),
-            Rc::new(system_data_loader.system_bigram_lm),
+            Rc::new(system_unigram_lm),
+            Rc::new(system_bigram_lm),
         );
 
         let graph_resolver = GraphResolver::default();
@@ -273,4 +266,35 @@ impl BigramWordViterbiEngineBuilder {
         );
         Ok(config)
     }
+
+    pub fn try_load(file_name: &str) -> Result<PathBuf> {
+        if cfg!(test) {
+            let path = Path::new(env!("CARGO_MANIFEST_DIR"));
+            let path = path.join("../akaza-data/data/").join(file_name);
+            if path.exists() {
+                Ok(path)
+            } else {
+                bail!("There's no {} for testing.", path.to_string_lossy(),)
+            }
+        } else if let Ok(dir) = env::var("AKAZA_DATA_DIR") {
+            let dir = Path::new(dir.as_str());
+            let file = dir.join(file_name);
+            if file.exists() {
+                Ok(file)
+            } else {
+                bail!(
+                    "There's no {} in AKAZA_DATA_DIR({:?})",
+                    file.to_string_lossy(),
+                    dir,
+                )
+            }
+        } else {
+            let path = xdg::BaseDirectories::with_prefix("akaza")?.find_data_file(file_name);
+            if let Some(path) = path {
+                Ok(path)
+            } else {
+                bail!("There's no {} in XDG_DATA_DIRS", file_name)
+            }
+        }
+    }
 }
diff --git a/libakaza/tests/wnn.rs b/libakaza/tests/wnn.rs
@@ -2,6 +2,7 @@
 #[cfg(feature = "it")]
 mod tests {
     use std::collections::vec_deque::VecDeque;
+    use std::env;
     use std::path::Path;
 
     use anyhow::Result;
@@ -21,8 +22,8 @@ mod tests {
     {
         let datadir = env!("CARGO_MANIFEST_DIR").to_string() + "/../akaza-data/data/";
         assert!(Path::new(datadir.as_str()).exists());
+        env::set_var("AKAZA_DATA_DIR", datadir);
         BigramWordViterbiEngineBuilder::new(
-            datadir.as_str(),
             Some(read_skkdict(
                 Path::new(
                     (env!("CARGO_MANIFEST_DIR").to_string()