make lint

docs/create_tasks_table.py

@@ -8,7 +8,7 @@
 
 import mteb
 from mteb.abstasks.TaskMetadata import PROGRAMMING_LANGS, TASK_TYPE
-from mteb.languages import ISO_TO_LANGUAGE, ISO_TO_FAM_LEVEL0
+from mteb.languages import ISO_TO_FAM_LEVEL0, ISO_TO_LANGUAGE
 
 
 def author_from_bibtex(bibtex: str | None) -> str:
@@ -87,13 +87,15 @@ def create_task_lang_table(tasks: list[mteb.AbsTask], sort_by_sum=False) -> str:
         pl_table_dict.append(d)
 
     df = pl.DataFrame(pl_table_dict).sort(by="0-lang-code")
-    df = df.with_columns(pl.col('0-lang-code')
+    df = df.with_columns(
+        pl.col("0-lang-code")
         .replace_strict(ISO_TO_LANGUAGE, default="unknown")
-        .alias('1-lang-name')
+        .alias("1-lang-name")
     )
-    df = df.with_columns(pl.col('0-lang-code')
+    df = df.with_columns(
+        pl.col("0-lang-code")
         .replace_strict(ISO_TO_FAM_LEVEL0, default="Unclassified")
-        .alias('2-lang-fam')
+        .alias("2-lang-fam")
     )
 
     df = df.with_columns(sum=pl.sum_horizontal(get_args(TASK_TYPE)))

mteb/languages.py

@@ -28,7 +28,8 @@
 with path_to_lang_fam.open("r") as f:
     ISO_TO_FAM = json.load(f)
 
-ISO_TO_FAM_LEVEL0 = {k:v['level0'] for k, v in ISO_TO_FAM.items()}
+ISO_TO_FAM_LEVEL0 = {k: v["level0"] for k, v in ISO_TO_FAM.items()}
+
 
 @dataclass
 class LanguageScripts:

scripts/create_language_family_mapping.py

@@ -2,15 +2,15 @@
 
 import json
 from pathlib import Path
-from tqdm import tqdm
 
-from pyglottolog.api import Glottolog
-from pyglottolog.api import lls
+from pyglottolog.api import Glottolog, lls
+from tqdm import tqdm
 
 glottolog = Glottolog(
     "/home/ubuntu/isaac/work/glottolog"
 )  # Download the Glottolog repository
 
+
 def get_languages_with_iso_by_languoid(languoid, level=0, prev_fam=None):
     # Recursively gather all descendant languages with ISO codes
     if prev_fam is None:
@@ -23,7 +23,7 @@ def get_languages_with_iso_by_languoid(languoid, level=0, prev_fam=None):
         # Create a copy of `prev_fam` to avoid overwriting
         current_fam = prev_fam.copy()
         current_fam[f"level{level}"] = languoid.name
-        
+
         if descendant.level.name == "language":  # Direct languages
             if descendant.iso:
                 iso_key = descendant.iso
@@ -34,7 +34,6 @@ def get_languages_with_iso_by_languoid(languoid, level=0, prev_fam=None):
             get_languages_with_iso_by_languoid(descendant, level + 1, current_fam)
 
 
-
 all_languoids = list(glottolog.languoids())
 with Path("language_family.json").open("r") as f:
     ISO2FAMILY = json.load(f)
@@ -45,4 +44,4 @@ def get_languages_with_iso_by_languoid(languoid, level=0, prev_fam=None):
 ISO2FAMILY = dict(sorted(ISO2FAMILY.items()))
 
 with Path("language_family.json").open("w") as f:
-    json.dump(ISO2FAMILY, f, indent=3)
+    json.dump(ISO2FAMILY, f, indent=3)