From 0083920e84483ab18b4b2f9ecf736e57d75e6944 Mon Sep 17 00:00:00 2001
From: p-goulart <pedro.goulart@languagetool.com>
Date: Fri, 2 Feb 2024 14:15:16 +0100
Subject: [PATCH] Sundry fixes

 - custom install version env var must be formed from language;

 - frequency info added to POS dict compilation.
---
 lib/languagetool_utils.py     | 1 +
 lib/utils.py                  | 6 +++---
 lib/variant.py                | 6 +++++-
 scripts/build_tagger_dicts.py | 4 +++-
 4 files changed, 12 insertions(+), 5 deletions(-)

diff --git a/lib/languagetool_utils.py b/lib/languagetool_utils.py
index 1462aac..93aa088 100644
--- a/lib/languagetool_utils.py
+++ b/lib/languagetool_utils.py
@@ -96,6 +96,7 @@ def build_pos_binary(self) -> None:
             f"org.languagetool.tools.POSDictionaryBuilder "
             f"-i {gd.DIRS.RESULT_POS_DICT_FILEPATH} "
             f"-info {self.variant.pos_info_java_input_path()} "
+            f"-freq {self.variant.freq()} "
             f"-o {self.variant.pos_dict_java_output_path()}"
         )
         ShellCommand(cmd_build).run()
diff --git a/lib/utils.py b/lib/utils.py
index 8223427..6bfca80 100644
--- a/lib/utils.py
+++ b/lib/utils.py
@@ -24,13 +24,13 @@ def compile_lt():
     ShellCommand("mvn clean install -DskipTests", cwd=gd.DIRS.LT_DIR).run()
 
 
-def install_dictionaries(custom_version: Optional[str]):
+def install_dictionaries(custom_version: Optional[tuple[str, str]]):
     """Install our dictionaries to the local ~/.m2."""
     LOGGER.info("Installing dictionaries...")
     env: dict = {}
     if custom_version is not None:
-        LOGGER.info(f"Installing custom version \"{custom_version}\"")
-        env['PT_DICT_VERSION'] = custom_version
+        LOGGER.info(f"Installing custom version \"{custom_version[1]}\"")
+        env[custom_version[0]] = custom_version[1]
     else:
         LOGGER.info(f"Installing environment-defined version \"{env['PT_DICT_VERSION']}\"")
     ShellCommand("mvn clean install", env=env, cwd=gd.DIRS.JAVA_RESULTS_DIR).run()
diff --git a/lib/variant.py b/lib/variant.py
index cdb6e87..17d6f93 100644
--- a/lib/variant.py
+++ b/lib/variant.py
@@ -57,7 +57,11 @@ def compounds(self) -> str:
         return path.join(gd.DIRS.COMPOUNDS_DIR, f"{self.underscored}.dic")
 
     def freq(self) -> str:
-        return path.join(gd.DIRS.SPELLING_DICT_DIR, f"{self.lang}_{self.country}_wordlist.xml")
+        if self.country:
+            filename = f"{self.lang}_{self.country}_wordlist.xml"
+        else:
+            filename = f"{self.lang}_wordlist.xml"
+        return path.join(gd.DIRS.SPELLING_DICT_DIR, filename)
 
     def java_output_dir(self) -> str:
         return path.join(gd.DIRS.JAVA_RESULTS_DIR, "src/main/resources/org/languagetool/resource", self.lang)
diff --git a/scripts/build_tagger_dicts.py b/scripts/build_tagger_dicts.py
index 631407c..5ad2fc9 100644
--- a/scripts/build_tagger_dicts.py
+++ b/scripts/build_tagger_dicts.py
@@ -69,7 +69,9 @@ def main():
     lt.build_pos_binary()
     lt.build_synth_binary()
     if FORCE_INSTALL:
-        install_dictionaries(custom_version=CUSTOM_INSTALL_VERSION)
+        custom_install_env_var_name = LANGUAGE.lang.upper() + "_DICT_VERSION"
+        custom_version: tuple[str, str] = (custom_install_env_var_name, CUSTOM_INSTALL_VERSION)
+        install_dictionaries(custom_version)
     if LOGGER.level == 10:  # DEBUG
         lt.dump_dictionary()
     end_time = datetime.now()