Fix tests

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = vocabsieve
-version = 0.11.2
+version = 0.12.0
 author = FreeLanguageTools
 author_email = pypi@5f37.8shield.net
 description = A simple, effective sentence mining tool.

tests/test_local_dictionary.py

@@ -1,124 +1,145 @@
 from vocabsieve.local_dictionary import LocalDictionary
 
-def test_local_dictionary():
-    dictdb = LocalDictionary("testdir/basic")
-    assert dictdb.countDicts() == 0
-    dictdb.importdict({"test": "a test is a test"}, "de", "test-dict")
-    assert dictdb.countDicts() == 1
-    assert dictdb.define("test", "de", "test-dict") == "a test is a test"
-    dictdb.deletedict("test-dict")
-    assert dictdb.countDicts() == 0
 
-def test_import_stardict_normal():
-    dictdb = LocalDictionary("testdir/stardict_normal")
-    assert dictdb.countDicts() == 0
-    dictdb.dictimport("testdata/stardict/quick_eng-rus-2.4.2/quick_english-russian.ifo",
-                      dicttype="stardict",
-                      lang="en",
-                      name="quick_eng-rus-2.4.2")
-    assert dictdb.countDicts() == 1
-    assert dictdb.define("abdominous", "en", "quick_eng-rus-2.4.2") == "толстый"
-    assert dictdb.define("loophole", "en", "quick_eng-rus-2.4.2") == "бойница"
-    assert dictdb.define("luggage", "en", "quick_eng-rus-2.4.2") == "багаж"
-
+def test_local_dictionary(tmp_path):
+    db = LocalDictionary(tmp_path)
+    print(tmp_path)
+    assert db.countDicts() == 0
+    db.importdict({"test": "a test is a test"}, "de", "test-dict")
+    assert db.countDicts() == 1
+    assert db.define("test", "de", "test-dict") == "a test is a test"
+    db.deletedict("test-dict")
+    assert db.countDicts() == 0
 
-def test_import_stardict_xdxf():
-    dictdb = LocalDictionary("testdir/stardict_xdxf")
-    assert dictdb.countDicts() == 0
-    dictdb.dictimport("testdata/stardict/stardict-FR-LingvoUniversal-2.4.2/FR-Universal.ifo",
-                      dicttype="stardict",
-                      lang="fr",
-                      name="fr-universal")
-    assert dictdb.countDicts() == 1
-    assert dictdb.define("accouchement", "fr", "fr-universal") == '''<i>m</i>
+
+def test_import_stardict_normal(tmp_path):
+    db = LocalDictionary(tmp_path)
+    assert db.countDicts() == 0
+    db.dictimport("testdata/stardict/quick_eng-rus-2.4.2/quick_english-russian.ifo",
+                  dicttype="stardict",
+                  lang="en",
+                  name="quick_eng-rus-2.4.2")
+    assert db.countDicts() == 1
+    assert db.define("abdominous", "en", "quick_eng-rus-2.4.2") == "толстый"
+    assert db.define("loophole", "en", "quick_eng-rus-2.4.2") == "бойница"
+    assert db.define("luggage", "en", "quick_eng-rus-2.4.2") == "багаж"
+
+
+def test_import_stardict_xdxf(tmp_path):
+    db = LocalDictionary(tmp_path)
+    assert db.countDicts() == 0
+    db.dictimport("testdata/stardict/stardict-FR-LingvoUniversal-2.4.2/FR-Universal.ifo",
+                  dicttype="stardict",
+                  lang="fr",
+                  name="fr-universal")
+    assert db.countDicts() == 1
+    assert db.define("accouchement", "fr", "fr-universal") == '''<i>m</i>
  1) р'оды
   accouchement avant terme, accouchement prématuré — преждевр'еменные р'оды
   accouchement après terme, accouchement tardif — запозд'алые р'оды
   accouchement sans douleur — обезб'оливание р'одов
   douleurs de l'accouchement — родов'ые б'оли
  2) <i>перен.</i> дл'ительное созрев'ание, тр'удное осуществл'ение'''
-    assert dictdb.define("persévérant", "fr", "fr-universal") == '''<i>adj</i>, <i>subst</i> (<i>fém</i> - persévérante)
+    assert db.define("persévérant", "fr", "fr-universal") == '''<i>adj</i>, <i>subst</i> (<i>fém</i> - persévérante)
  1) наст'ойчивый [наст'ойчивая], уп'орный [уп'орная]; твёрдый [твёрдая]
  2) посто'янный [посто'янная]'''
-    assert dictdb.define("pièce-raccord", "fr", "fr-universal") == '''pièce-raccord
+    assert db.define("pièce-raccord", "fr", "fr-universal") == '''pièce-raccord
  <i>m</i>
  <i>(pl s + s</i> ) соедин'ительная часть, соедин'ительная дет'аль'''
 
-def test_import_dsl():
-    dictdb = LocalDictionary("testdir/dsl")
-    assert dictdb.countDicts() == 0
-    dictdb.dictimport("testdata/dsl/ru_en.dsl",
-                      dicttype="dsl",
-                      lang="ru",
-                      name="dsl_test"
-                      )
-    dictdb.dictimport("testdata/dsl/ru_en.dsl.dz",
-                      dicttype="dsl",
-                      lang="ru",
-                      name="dsl_test2"
-                      )
-    assert dictdb.countDicts() == 2
-    assert dictdb.define("зубчатый", "ru", "dsl_test") == '''serrated, toothed'''
-    assert dictdb.define("лиственный", "ru", "dsl_test") == '''broadleaf; deciduous; leafy'''
-    assert dictdb.define("окорять", "ru", "dsl_test") == '''bark, peel'''
-    assert dictdb.define("зубчатый", "ru", "dsl_test2") == '''serrated, toothed'''
-    assert dictdb.define("лиственный", "ru", "dsl_test2") == '''broadleaf; deciduous; leafy'''
-    assert dictdb.define("окорять", "ru", "dsl_test2") == '''bark, peel'''
-    dictdb.dictimport("testdata/dsl/universal.dsl.dz",
-                      dicttype="dsl",
-                      lang="ru",
-                      name="dsl_test3"
-                      )
-    assert dictdb.countDicts() == 3
-    assert dictdb.define("ямчатость", "ru", "dsl_test3") == '''ж. с.-х.<br>  (патологическое свойство плодов) pit<br>'''
-    assert dictdb.define("эмиграция", "ru", "dsl_test3") == '''ж.<br>  1) (переселение из своего отечества) emigration<br>  2) (пребывание в другой стране) life in emigration<br>    жить в эмиграции — live as an emigrant / émigré (фр.) /<'emɪgreɪ/><br>  3) собир. emigrants pl; émigrés (фр.) /<'emɪgreɪz/> pl<br>'''
-    assert dictdb.define("щемящий", "ru", "dsl_test3") == '''1) (ноющий, тупой) aching /<'eɪk-/>, nagging<br>    щемящая боль — nagging ache /<eɪk/><br>  2) (мучительный, гнетущий) painful, melancholy, oppressive<br>    щемящий душу напев — plaintive / melancholy /<-k-/> tune<br>'''
 
-def test_import_cognates():
-    dictdb = LocalDictionary("testdir/cognates")
-    assert dictdb.countDicts() == 0
-    dictdb.dictimport("testdata/cognates/cognates.json.gz",
-                      dicttype="cognates",
-                      lang="<all>",
-                      name="cognates"
-                      )
-    assert dictdb.countDicts() == 1
-    assert dictdb.define("chodník", "cs", "cognates") == '''["sk", "pl"]'''
-    assert dictdb.define("beluga", "hr", "cognates") == '''["fi", "hu", "ru", "nl", "en", "de", "bg", "fr", "ro", "ca", "mhr", "kk", "pt", "eo", "uk", "cs", "es"]'''
-    assert dictdb.define("apple", "en", "cognates") == '''["nl", "ksh", "xh", "nso", "da", "kn", "hsb", "pl", "dsb", "uk", "ltg", "hr", "af", "ru", "nb", "lb", "pap", "bg", "ml", "tn", "brx", "gd", "jam", "sah", "gv", "ve", "zu", "cs", "wym", "si", "cy", "fo", "sco", "bn", "sk", "ga", "sv", "zsm", "fy", "be", "mk", "as", "mi", "cu", "lt", "abe", "de", "nn", "br", "id", "ta", "st", "kok", "te", "ms", "sl", "is"]'''
-    assert dictdb.define("tragisch", "de", "cognates") == '''["nl", "fi", "ro", "pt", "pl", "hr", "hu", "nb", "ast", "fr", "ms", "eu", "eo", "cs", "ca", "sk", "sv", "lij", "es", "en", "id", "oc", "gl", "sl"]'''
+def test_import_dsl(tmp_path):
+    db = LocalDictionary(tmp_path)
+    assert db.countDicts() == 0
+    db.dictimport("testdata/dsl/ru_en.dsl",
+                  dicttype="dsl",
+                  lang="ru",
+                  name="dsl_test"
+                  )
+    db.dictimport("testdata/dsl/ru_en.dsl.dz",
+                  dicttype="dsl",
+                  lang="ru",
+                  name="dsl_test2"
+                  )
+    assert db.countDicts() == 2
+    assert db.define("зубчатый", "ru", "dsl_test") == '''serrated, toothed'''
+    assert db.define("лиственный", "ru", "dsl_test") == '''broadleaf; deciduous; leafy'''
+    assert db.define("окорять", "ru", "dsl_test") == '''bark, peel'''
+    assert db.define("зубчатый", "ru", "dsl_test2") == '''serrated, toothed'''
+    assert db.define("лиственный", "ru", "dsl_test2") == '''broadleaf; deciduous; leafy'''
+    assert db.define("окорять", "ru", "dsl_test2") == '''bark, peel'''
+    db.dictimport("testdata/dsl/universal.dsl.dz",
+                  dicttype="dsl",
+                  lang="ru",
+                  name="dsl_test3"
+                  )
+    assert db.countDicts() == 3
+    assert db.define("ямчатость", "ru", "dsl_test3") == '''ж. с.-х.<br>  (патологическое свойство плодов) pit<br>'''
+    assert db.define("эмиграция", "ru", "dsl_test3") == '''ж.<br>  1) (переселение из своего отечества) emigration<br>  2) (пребывание в другой стране) life in emigration<br>    жить в эмиграции — live as an emigrant / émigré (фр.) /<'emɪgreɪ/><br>  3) собир. emigrants pl; émigrés (фр.) /<'emɪgreɪz/> pl<br>'''
+    assert db.define(
+        "щемящий",
+        "ru",
+        "dsl_test3") == '''1) (ноющий, тупой) aching /<'eɪk-/>, nagging<br>    щемящая боль — nagging ache /<eɪk/><br>  2) (мучительный, гнетущий) painful, melancholy, oppressive<br>    щемящий душу напев — plaintive / melancholy /<-k-/> tune<br>'''
+
 
-def test_kaikki():
-    dictdb = LocalDictionary("testdir/kaikki")
-    assert dictdb.countDicts() == 0
-    dictdb.dictimport("testdata/kaikki/kaikki.org-dictionary-Swedish.json.gz",
-                      dicttype="wiktdump",
-                      lang="sv",
-                      name="kaikki-swedish"
-                      )
-    assert dictdb.countDicts() == 1
-    assert dictdb.define("uppåkrakaka", "sv", "kaikki-swedish") == '''<i>Noun</i>
+def test_import_cognates(tmp_path):
+    db = LocalDictionary(tmp_path)
+    assert db.countDicts() == 0
+    db.dictimport("testdata/cognates/cognates.json.gz",
+                  dicttype="cognates",
+                  lang="<all>",
+                  name="cognates"
+                  )
+    assert db.define("chodník", "cs", "cognates") == '''["sk", "pl"]'''
+    assert db.define(
+        "beluga",
+        "hr",
+        "cognates") == '''["fi", "hu", "ru", "nl", "en", "de", "bg", "fr", "ro", "ca", "mhr", "kk", "pt", "eo", "uk", "cs", "es"]'''
+    assert db.define(
+        "apple",
+        "en",
+        "cognates") == '''["nl", "ksh", "xh", "nso", "da", "kn", "hsb", "pl", "dsb", "uk", "ltg", "hr", "af", "ru", "nb", "lb", "pap", "bg", "ml", "tn", "brx", "gd", "jam", "sah", "gv", "ve", "zu", "cs", "wym", "si", "cy", "fo", "sco", "bn", "sk", "ga", "sv", "zsm", "fy", "be", "mk", "as", "mi", "cu", "lt", "abe", "de", "nn", "br", "id", "ta", "st", "kok", "te", "ms", "sl", "is"]'''
+    assert db.define(
+        "tragisch",
+        "de",
+        "cognates") == '''["nl", "fi", "ro", "pt", "pl", "hr", "hu", "nb", "ast", "fr", "ms", "eu", "eo", "cs", "ca", "sk", "sv", "lij", "es", "en", "id", "oc", "gl", "sl"]'''
+    assert db.countDicts() == 1
+
+
+def test_kaikki(tmp_path):
+    db = LocalDictionary(tmp_path)
+    assert db.countDicts() == 0
+    db.dictimport("testdata/kaikki/swedish_short.json",
+                  dicttype="wiktdump",
+                  lang="sv",
+                  name="kaikki-swedish"
+                  )
+    assert db.countDicts() == 1
+    assert db.define("uppåkrakaka", "sv", "kaikki-swedish") == '''<i>Noun</i>
 uppåkrakaka c
 1. a biscuit made of mördeg (without egg), in a circular shape folded almost in the middle, garnished with chopped pistachios and nib sugar'''
-    assert dictdb.define("affektionsvärde", "sv", "kaikki-swedish") == '''<i>Noun</i>
+    assert db.define("affektionsvärde", "sv", "kaikki-swedish") == '''<i>Noun</i>
 affektionsvärde n
 1. sentimental value'''
-    assert dictdb.define("rådigt", "sv", "kaikki-swedish") == '''<i>Adv</i>
+    assert db.define("rådigt", "sv", "kaikki-swedish") == '''<i>Adj</i>
+rådigt
+1. indefinite neuter singular of rådig
+
+<i>Adv</i>
 rådigt (comparative rådigare, superlative rådigast)
 1. resourcefully, resolutely'''
 
-    dictdb.dictimport("testdata/kaikki/fr-extract.json.gz",
-                      dicttype="wiktdump",
-                      lang="fr",
-                      name="kaikki-french"
-                      )
-    assert dictdb.countDicts() == 2
+    db.dictimport("testdata/kaikki/fr_short.json",
+                  dicttype="wiktdump",
+                  lang="fr",
+                  name="kaikki-french"
+                  )
+    assert db.countDicts() == 2
     # french tests
-    assert dictdb.define("évhémérisassent", "fr", "kaikki-french") == """<i>Verb</i>
+    assert db.define("évhémérisassent", "fr", "kaikki-french") == """<i>Verb</i>
 1. Troisième personne du pluriel de l’imparfait du subjonctif de évhémériser."""
-    assert dictdb.define("fortitrer", "fr", "kaikki-french") == """<i>Verb</i>
+    assert db.define("fortitrer", "fr", "kaikki-french") == """<i>Verb</i>
 1. Un cerf fortitre, quand il évite de passer près des chiens frais et des relais."""
-    assert dictdb.define("géminer", "fr", "kaikki-french") == """<i>Verb</i>
+    assert db.define("géminer", "fr", "kaikki-french") == """<i>Verb</i>
 1. Se doubler.
 2. Grouper deux à deux, doubler."""
-

vocabsieve/dictformats.py

@@ -249,6 +249,7 @@ def parseKaikki(path, lang) -> dict[str, str]:
     (https://github.com/tatuylonen/wiktextract)
     The format is lines of json objects, each containing a word and its definition
     '''
+    print("Parsing Kaikki wiktionary dump at " + path)
     items: list[tuple[str, str]] = []
     with zopen(path) as f:
         logger.debug("Parsing Kaikki wiktionary dump at " + path)

vocabsieve/main.py

@@ -322,6 +322,15 @@ def setupMenu(self) -> None:
         self.setMenuBar(self.menu)
 
     def markWords(self):
+        if settings.value("freq_source", "<disabled>") == "<disabled>":
+            self.warn("No frequency source is set. Please set a frequency source in the configuration dialog.")
+            return
+        if not settings.value("lemfreq", False, type=bool):
+            self.warn("Marking words requires a lemmatized frequency list to work properly.")
+            return
+        if self.known_data is None:
+            self.warnKnownDataNotReady()
+            return
         words = self.freq_widget.getAllWords()
         dialog = WordMarkingDialog(self, words)
         dialog.exec()
@@ -423,6 +432,7 @@ def _refreshKnownData(self) -> None:
         with lock:
             self.known_data, self.known_metadata = self.rec.getKnownData()
             self.known_data_timestamp = time.time()
+            self.status("Known data is ready")
 
     def exportWordData(self):
         path, _ = QFileDialog.getSaveFileName(
@@ -1058,7 +1068,7 @@ def time(self) -> str:
         return QDateTime.currentDateTime().toString('[hh:mm:ss]')
 
     def status(self, msg: str) -> None:
-        self.status_bar.showMessage(self.time() + " " + msg, 4000)
+        self.status_bar.showMessage(self.time() + " " + msg)
 
     def warn(self, text: str) -> None:
         msg = QMessageBox()