More ergonomic handling of JSONL in dictinfo

.gitignore

@@ -1,6 +1,7 @@
 __pycache__/
 env/
 dist/
+.DS_Store
 *.un~
 vocabsieve.egg-info/
 build/
@@ -12,5 +13,8 @@ htmlcov
 docs/_site
 docs/.jekyll-cache
 .vscode
-testdata
-testdir
+testdir
+
+testdata/kaikki/*
+!kaikki.org-dictionary-English-by-pos-postp.jsonl
+!kaikki.org-dictionary-English-by-pos-postp.jsonl.gz

tests/test_dictformats.py

@@ -0,0 +1,20 @@
+from vocabsieve.dictformats import dictinfo
+
+
+def test_dictinfo_kaikki_jsonl():
+    info = dictinfo("testdata/kaikki/kaikki.org-dictionary-English-by-pos-postp.jsonl")
+
+    assert info == {
+        "basename": "kaikki.org-dictionary-English-by-pos-postp",
+        "path": "testdata/kaikki/kaikki.org-dictionary-English-by-pos-postp.jsonl",
+        "type": "wiktdump",
+    }
+
+def test_dictinfo_kaikki_jsonl_gz():
+    info = dictinfo("testdata/kaikki/kaikki.org-dictionary-English-by-pos-postp.jsonl.gz")
+
+    assert info == {
+        "basename": "kaikki.org-dictionary-English-by-pos-postp",
+        "path": "testdata/kaikki/kaikki.org-dictionary-English-by-pos-postp.jsonl.gz",
+        "type": "wiktdump",
+    }

vocabsieve/dictformats.py

@@ -84,7 +84,26 @@ def dictinfo(path) -> dict[str, str]:
         return {"type": "audiolib", "basename": basename, "path": path}
     if ext not in supported_dict_extensions:
         raise NotImplementedError("Unsupported format")
-    if ext in ('.json', '.jsonl', '.xz', '.bz2', '.gz'):
+
+    for jsonl_ext in (".jsonl", ".jsonl.xz", ".jsonl.gz", ".jsonl.bz2"):
+        if not path.endswith(jsonl_ext):
+           continue
+
+        with zopen(path) as f:
+            first_line = f.readline()
+            try:
+                if json.loads(first_line):
+                    logger.debug("Detected Kaikki wiktionary dump")
+                    return {
+                        "type": "wiktdump",
+                        "basename": os.path.basename(path.removesuffix(jsonl_ext)), 
+                        "path": path
+                    }
+            except json.JSONDecodeError:
+                pass
+        raise NotImplementedError(f"File {path} is not a supported jsonl format")
+
+    if ext in ('.json', '.xz', '.bz2', '.gz'):
         with zopen(path) as f:
             try:
                 d = json.load(f)
@@ -108,10 +127,7 @@ def dictinfo(path) -> dict[str, str]:
                 f.seek(0)
                 first_line = f.readline()
                 logger.debug("First line of bad json file: ", first_line)
-                logger.debug("Detected Kaikki wiktionary dump")
-                if json.loads(first_line):
-                    return {"type": "wiktdump", "basename": basename, "path": path}
-            raise NotImplementedError(f"File {path} is not a supported json format")
+                raise NotImplementedError(f"File {path} is not a supported json format")
     elif ext == ".ifo":
         return {"type": "stardict", "basename": basename, "path": path}
     elif ext == ".mdx":