refactor: move get_langs out to g2p

joanise · joanise · commit cd5d6dff3e56 · 2023-04-17T11:03:00.000-04:00
diff --git a/readalongs/util.py b/readalongs/util.py
@@ -3,9 +3,6 @@
 
 import click
 
-LANGS = None
-LANG_NAMES = None
-
 
 def get_langs_deferred() -> Iterable:
     """Lazilly get the list of language codes supported by g2p library
@@ -24,69 +21,9 @@ def get_langs():
             LANGS is the list of valid language codes supported
             LANG_NAMES maps each code to its full language name
     """
+    import g2p  # Defer expensive import, which loads the whole g2p database
 
-    global LANGS
-    global LANG_NAMES
-
-    if LANGS is not None and LANG_NAMES is not None:
-        # Cache the results so we only calculate this information once.
-        return LANGS, LANG_NAMES
-    else:
-        # Imports deferred to when the function is actually first called, because
-        # they load the g2p database and take a fair bit of time. And importing
-        # networkx is also quite expensive.
-        import g2p.mappings.langs as g2p_langs
-        from networkx import has_path
-
-        # langs_available in g2p lists langs inferred by the directory structure of
-        # g2p/mappings/langs, but in ReadAlongs, we need all input languages to any mappings.
-        # E.g., for Michif, we need to allow crg-dv and crg-tmd, but not crg, which is what
-        # langs_available contains. So we define our own list of languages here.
-        langs_available = []
-
-        # this will be the set of all langs in g2p + "eng", which we need temporarily
-        full_lang_names = {"eng": "English"}
-
-        for _, v in g2p_langs.LANGS.items():
-            for mapping in v["mappings"]:
-                # add mapping to names hash table
-                full_lang_names[mapping["in_lang"]] = mapping["language_name"]
-                # add input id to all available langs list
-                if mapping["in_lang"] not in langs_available:
-                    langs_available.append(mapping["in_lang"])
-
-        # get the key from all networks in g2p module that have a path to 'eng-arpabet',
-        # which is needed for the readalongs
-        # Filter out <lang>-ipa: we only want "normal" input languages.
-        # Filter out *-norm and crk-no-symbols, these are just intermediate representations.
-        LANGS = [
-            x
-            for x in langs_available
-            if not x.endswith("-ipa")
-            and not x.endswith("-equiv")
-            and not x.endswith("-no-symbols")
-            and x not in ["und-ascii", "moh-festival"]
-            and g2p_langs.LANGS_NETWORK.has_node(x)
-            and has_path(g2p_langs.LANGS_NETWORK, x, "eng-arpabet")
-        ]
-
-        # Hack to allow old English LexiconG2P
-        LANGS += ["eng"]
-        # Sort LANGS so the -h messages list them alphabetically
-        LANGS = sorted(LANGS)
-
-        # Set up LANG_NAMES hash table for studio UI to properly name the dropdown options
-        LANG_NAMES = {lang_code: full_lang_names[lang_code] for lang_code in LANGS}
-
-        return LANGS, LANG_NAMES
-
-
-# For backwards compatibility, we keep the old names getLangs and getLangsDeferred around.
-# For example, ReadAlongsDesktop
-# (https://github.com/tobyatgithub/ReadalongsDesktop) depended on the old name,
-# and even when it's updated, it'll be helpful to avoid breaking older versions.
-getLangs = get_langs
-getLangsDeferred = get_langs_deferred
+    return g2p.get_langs()
 
 
 class JoinerCallbackForClick:
