From 2ea6f5bba1ad125dc7972621c41f44445945b3dc Mon Sep 17 00:00:00 2001
From: Kan-Ru Chen <kanru@kanru.info>
Date: Mon, 15 Jul 2024 17:52:30 +0900
Subject: [PATCH] feat(dict): Automatically load extra dictionaries found in
 search path

---
 src/dictionary/loader.rs             |  25 ++++++++++++++++------
 src/path.rs                          |  30 +++++++++++++++++++++++++++
 tests/data/dictionary.d/01-extra.dat | Bin 0 -> 136 bytes
 tests/data/dictionary.d/02-empty.dat |   0
 tests/data/extra.src                 |   2 ++
 tests/test-config.c                  |  17 +++++++++++++++
 6 files changed, 68 insertions(+), 6 deletions(-)
 create mode 100644 tests/data/dictionary.d/01-extra.dat
 create mode 100644 tests/data/dictionary.d/02-empty.dat
 create mode 100644 tests/data/extra.src

diff --git a/src/dictionary/loader.rs b/src/dictionary/loader.rs
index ef60ac7dc..205b428f1 100644
--- a/src/dictionary/loader.rs
+++ b/src/dictionary/loader.rs
@@ -7,11 +7,11 @@ use std::{
     path::{Path, PathBuf},
 };
 
-use log::info;
+use log::{info, warn};
 
 use crate::{
     editor::{AbbrevTable, SymbolSelector},
-    path::{find_path_by_files, sys_path_from_env_var, userphrase_path},
+    path::{find_extra_dat_by_path, find_path_by_files, sys_path_from_env_var, userphrase_path},
 };
 
 #[cfg(feature = "sqlite")]
@@ -70,15 +70,28 @@ impl SystemDictionaryLoader {
         let sys_path = find_path_by_files(&search_path, &[SD_WORD_FILE_NAME, SD_TSI_FILE_NAME])
             .ok_or(LoadDictionaryError::NotFound)?;
 
-        let tsi_dict_path = sys_path.join(SD_TSI_FILE_NAME);
-        info!("Loading {SD_TSI_FILE_NAME}");
-        let tsi_dict = Trie::open(tsi_dict_path).map_err(io_err)?;
+        let mut results: Vec<Box<dyn Dictionary>> = vec![];
 
         let word_dict_path = sys_path.join(SD_WORD_FILE_NAME);
         info!("Loading {SD_WORD_FILE_NAME}");
         let word_dict = Trie::open(word_dict_path).map_err(io_err)?;
+        results.push(Box::new(word_dict));
+
+        let tsi_dict_path = sys_path.join(SD_TSI_FILE_NAME);
+        info!("Loading {SD_TSI_FILE_NAME}");
+        let tsi_dict = Trie::open(tsi_dict_path).map_err(io_err)?;
+        results.push(Box::new(tsi_dict));
+
+        let extra_files = find_extra_dat_by_path(&search_path);
+        for path in extra_files {
+            info!("Loading {}", path.display());
+            match Trie::open(&path) {
+                Ok(dict) => results.push(Box::new(dict)),
+                Err(e) => warn!("Failed to load {}: {e}", path.display()),
+            }
+        }
 
-        Ok(vec![Box::new(word_dict), Box::new(tsi_dict)])
+        Ok(results)
     }
     pub fn load_abbrev(&self) -> Result<AbbrevTable, LoadDictionaryError> {
         let search_path = if let Some(sys_path) = &self.sys_path {
diff --git a/src/path.rs b/src/path.rs
index 65222a72e..dc67c39bb 100644
--- a/src/path.rs
+++ b/src/path.rs
@@ -2,6 +2,7 @@
 
 use std::{
     env,
+    ffi::OsStr,
     path::{Path, PathBuf},
 };
 
@@ -17,6 +18,8 @@ const SEARCH_PATH_SEP: char = ';';
 #[cfg(target_family = "unix")]
 const SEARCH_PATH_SEP: char = ':';
 
+const DICT_FOLDER: &str = "dictionary.d";
+
 pub(crate) fn sys_path_from_env_var() -> String {
     let chewing_path = env::var("CHEWING_PATH");
     if let Ok(chewing_path) = chewing_path {
@@ -55,6 +58,33 @@ pub(crate) fn find_path_by_files(search_path: &str, files: &[&str]) -> Option<Pa
     None
 }
 
+pub(crate) fn find_extra_dat_by_path(search_path: &str) -> Vec<PathBuf> {
+    let mut results = vec![];
+    for path in search_path.split(SEARCH_PATH_SEP) {
+        let prefix = Path::new(path).join(DICT_FOLDER);
+        info!("Search dictionary files in {}", prefix.display());
+        if let Ok(read_dir) = prefix.read_dir() {
+            let mut files = vec![];
+            for entry in read_dir {
+                if let Ok(entry) = entry {
+                    let path = entry.path();
+                    let is_dat = path
+                        .extension()
+                        .and_then(OsStr::to_str)
+                        .map_or(false, |ext| ext == "dat");
+                    if path.is_file() && is_dat {
+                        info!("Found {}", path.display());
+                        files.push(path);
+                    }
+                }
+            }
+            files.sort();
+            results.extend(files);
+        }
+    }
+    results
+}
+
 /// Returns the path to the user's default chewing data directory.
 ///
 /// The returned value depends on the operating system and is either a
diff --git a/tests/data/dictionary.d/01-extra.dat b/tests/data/dictionary.d/01-extra.dat
new file mode 100644
index 0000000000000000000000000000000000000000..be03d5482f55dc5934ad5a9d02c646f25bd8068d
GIT binary patch
literal 136
zcmXqLY~^8b_HYenVq`F|;^BF@u=nYbX;0Tqf7-f(hdneeJ1@UH55ZtH)HBdC;1NvD
zNG;FIOV>@#$y6}Vv(PhT(O_U;U}Rtdk_-&YK$aAcW?^7t)B%fdK}jHOAY#D5!wj?s
MWFwFTv<}Du0Od>_qW}N^

literal 0
HcmV?d00001

diff --git a/tests/data/dictionary.d/02-empty.dat b/tests/data/dictionary.d/02-empty.dat
new file mode 100644
index 000000000..e69de29bb
diff --git a/tests/data/extra.src b/tests/data/extra.src
new file mode 100644
index 000000000..3ad635f46
--- /dev/null
+++ b/tests/data/extra.src
@@ -0,0 +1,2 @@
+額 0 ㄜˊ
+外 0 ㄨㄞˋ
diff --git a/tests/test-config.c b/tests/test-config.c
index c3263377f..0c3e9a348 100644
--- a/tests/test-config.c
+++ b/tests/test-config.c
@@ -745,6 +745,21 @@ void test_runtime_version()
     ok(strcmp(buf, version) == 0, "chewing_version can be created from components");
 }
 
+void test_dictionary_d()
+{
+    ChewingContext *ctx;
+
+    ctx = chewing_new2(TEST_DATA_DIR, NULL, logger, fd);
+    start_testcase(ctx, fd);
+
+    ok(ctx != NULL, "chewing_new2 returns `%#p' shall not be `%#p'", ctx, NULL);
+
+    type_keystroke_by_string(ctx, "k6j94<E>");
+    ok_commit_buffer(ctx, "額外");
+
+    chewing_delete(ctx);
+}
+
 int main(int argc, char *argv[])
 {
     char *logname;
@@ -785,6 +800,8 @@ int main(int argc, char *argv[])
 
     test_runtime_version();
 
+    test_dictionary_d();
+
     fclose(fd);
 
     return exit_status();