grzegorz 0.6.0__py3-none-any.whl → 0.6.1__py3-none-any.whl

grzegorz/fetcher.py

@@ -15,7 +15,8 @@
 
 from grzegorz.word import Word
 
-from wiktionaryparser import WiktionaryParser
+import requests
+from bs4 import BeautifulSoup
 import re
 
 ### HELPER FUNCTIONS ###
@@ -26,21 +27,19 @@ def get_ipa_for_word(word: str, language: str) -> Word:
     and return a `Word` binding it to the letters. If no transcription was
     found, then the `ipa` field of the result is empty.
     """
-    parser = WiktionaryParser()
-    parser.set_default_language(language)
+    language = language.capitalize()
+    url = f"https://en.wiktionary.org/wiki/{word}"
+    webpage = requests.get(url)
+    soup= BeautifulSoup(webpage.text, "html.parser")
+    pronunciations= soup.select(f'li:has(sup:has(a[href="/wiki/Appendix:{language}_pronunciation"]))' )
+
     ipa = ""
-    fetched = parser.fetch(word)
-    if len(fetched):
-        first_entry = fetched[0]
-        pronunciations = first_entry.get('pronunciations')
-        text = pronunciations.get('text')
-        if len(text):
-            ipa = first_ipa_pronunciation(text[0])
-    # Not all words have their IPAs on wiktionary, but they might have a
-    # "Rhymes" section (many German words do, for example). If we did fetch a
-    # rhyme, don't add it as a valid IPA
-    if len(ipa) and ipa[0] == '-':
-        ipa = ""
+    # maybe blindly choosing the first IPA transliteration is not the wisest
+    # choice in the world?
+    if len(pronunciations):
+        first_entry = pronunciations[0].find("span", {"class": "IPA"})
+        if first_entry is not None:
+            ipa = first_entry.text
 
     return Word(word, ipa)
 

grzegorz/generator.py

@@ -46,17 +46,21 @@ class MinPairGenerator:
                 lists_of_phonemes.append(line.replace(" ", "").split(","))
         self.filter_pairs = phoneme_lists_to_phoneme_pairs(lists_of_phonemes)
 
-    def generate(self, words: list[Word]) -> list[WordPair]:
+    def generate(self, words: list[Word], silent: bool = True) -> list[WordPair]:
         """
         Generate minimal pairs from the given parameters
         """
         minpairs = []
 
-        for i in tqdm(range(0,len(words))):
-            for j in range(i+1,len(words)):
+        progress_bar = tqdm(total=int(len(words) * (len(words) - 1) / 2), disable=silent)
+        for i in range(0, len(words)):
+            words_after = range(i+1, len(words))
+            for j in words_after:
                 pair = (words[i], words[j])
                 if self.check_minpair(pair):
                     minpairs.append(pair)
+            progress_bar.update(len(words_after))
+        progress_bar.close()
 
         return minpairs
 

grzegorz/subcommands.py

@@ -95,7 +95,7 @@ def wordlist_command(language: str, bounds: str, outfile: str) -> int:
     else:
         return 1
 
-def fetchipa(infile: str, outfile: str, keep_failed: bool, numproc: int = 10) -> None:
+def fetchipa(infile: str, outfile: str, keep_failed: bool, numproc: int = 20) -> None:
     """
     Given an input file containing a list of words separated, fetch the IPAs and
     create a text file with their IPA spellings matched to their text
@@ -148,7 +148,8 @@ def generate_command(infile, outfile, nooptimise, no_phonemes, no_chronemes,
     if no_stress:
         print("Generator: syllable stress contrasts will be ignored")
 
-    minpairs = g.generate(words)
+    print('Generating minimal pairs from:', len(words), 'words')
+    minpairs = g.generate(words, False)
     writefile(outfile, encode_format(encode_minpair, minpairs))
     print('Done! Generated', len(minpairs), 'minimal pairs')
 

{grzegorz-0.6.0.dist-info → grzegorz-0.6.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.4
 Name: grzegorz
-Version: 0.6.0
+Version: 0.6.1
 Summary: Minimal pair generator and phonetics tool
 Home-page: https://github.com/xylous/grzegorz
 Author: xylous
@@ -16,10 +16,11 @@ Classifier: Topic :: Text Processing :: Linguistic
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: wiktionaryparser
+Requires-Dist: beautifulsoup4
 Requires-Dist: tqdm
 Requires-Dist: requests
 Requires-Dist: genanki
+Dynamic: license-file
 
 # grzegorz
 

grzegorz-0.6.1.dist-info/RECORD

@@ -0,0 +1,16 @@
+grzegorz/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+grzegorz/__main__.py,sha256=jYpB9FB0uuJwrpxBpU-d13vIw1vhNg2kYX4yC_-UWrg,7133
+grzegorz/anki_integration.py,sha256=eMFdFNd0NsqLxX23NtlEzinhGMCecEFyoklfFkMqQOk,3933
+grzegorz/fetcher.py,sha256=oAZRDZVqH93HgLFFffJ-dl6Qc83aD43ZuNVK9boy7F0,1902
+grzegorz/generator.py,sha256=oCz9TKg9wPN3VIGGa2H8L2Ex4Uf2_gX_XFrlxiB4RSw,9320
+grzegorz/io.py,sha256=JM2pOKgECmnVxCZplgRt1gEiyYWXUn_Z6OanmGSaab0,2221
+grzegorz/subcommands.py,sha256=QQQX1LraTi9Lfo28N1s4G1j-j_z4HtiUsAYsVNyt5FI,6101
+grzegorz/test.py,sha256=znHJFiV0Q1qP0kJYtoweMTNqJH1eX9ZHWFZedOJIuGo,3866
+grzegorz/word.py,sha256=bXNTq_sjrn7CTOWBGkKdQXky_j0c-OzxhhgJWDh0BR0,7899
+grzegorz/wordlist.py,sha256=SqKkZoyY8Ol1vp8Rt0PeNWtxL8ND3qE_yWGl5yiKZ_M,4058
+grzegorz-0.6.1.dist-info/licenses/LICENSE,sha256=STF0KkBB_RpcXwp43xCvRIKKe_4V-zrq1lU1OsTgapY,35148
+grzegorz-0.6.1.dist-info/METADATA,sha256=sZkAm0W5qIsh8fcWOpiy-cV6L64p5RsTWm06VYKbf5U,3980
+grzegorz-0.6.1.dist-info/WHEEL,sha256=wXxTzcEDnjrTwFYjLPcsW_7_XihufBwmpiBeiXNBGEA,91
+grzegorz-0.6.1.dist-info/entry_points.txt,sha256=rZ-JLt-sbS1rZ5YwodMyf9o80C6sN4AfuSCb0sFNVJ8,52
+grzegorz-0.6.1.dist-info/top_level.txt,sha256=W2SodvLxGhkJfWfNhDO0Vh7prBehEXdE9sHWJ1mZXTA,9
+grzegorz-0.6.1.dist-info/RECORD,,

{grzegorz-0.6.0.dist-info → grzegorz-0.6.1.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: bdist_wheel (0.41.0)
+Generator: setuptools (80.1.0)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

grzegorz-0.6.0.dist-info/RECORD

@@ -1,16 +0,0 @@
-grzegorz/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-grzegorz/__main__.py,sha256=jYpB9FB0uuJwrpxBpU-d13vIw1vhNg2kYX4yC_-UWrg,7133
-grzegorz/anki_integration.py,sha256=eMFdFNd0NsqLxX23NtlEzinhGMCecEFyoklfFkMqQOk,3933
-grzegorz/fetcher.py,sha256=quIpygekLCtXDS5Yx_2eHkJUeDaJDMdNcqI2P5HPxos,1929
-grzegorz/generator.py,sha256=IQyqY_dPUHmkaCcWnuN5QCQfvRiCS9B_8oOLuiNs2dM,9096
-grzegorz/io.py,sha256=JM2pOKgECmnVxCZplgRt1gEiyYWXUn_Z6OanmGSaab0,2221
-grzegorz/subcommands.py,sha256=uO0StbCG081rXPp71DMm--anb2pyA2QHNBCwDEwh6vk,6029
-grzegorz/test.py,sha256=znHJFiV0Q1qP0kJYtoweMTNqJH1eX9ZHWFZedOJIuGo,3866
-grzegorz/word.py,sha256=bXNTq_sjrn7CTOWBGkKdQXky_j0c-OzxhhgJWDh0BR0,7899
-grzegorz/wordlist.py,sha256=SqKkZoyY8Ol1vp8Rt0PeNWtxL8ND3qE_yWGl5yiKZ_M,4058
-grzegorz-0.6.0.dist-info/LICENSE,sha256=STF0KkBB_RpcXwp43xCvRIKKe_4V-zrq1lU1OsTgapY,35148
-grzegorz-0.6.0.dist-info/METADATA,sha256=IbAHh2s0xdwqEhLehK8SAImnm14mHO3k8nTnNfyOGtg,3960
-grzegorz-0.6.0.dist-info/WHEEL,sha256=AtBG6SXL3KF_v0NxLf0ehyVOh0cold-JbJYXNGorC6Q,92
-grzegorz-0.6.0.dist-info/entry_points.txt,sha256=rZ-JLt-sbS1rZ5YwodMyf9o80C6sN4AfuSCb0sFNVJ8,52
-grzegorz-0.6.0.dist-info/top_level.txt,sha256=W2SodvLxGhkJfWfNhDO0Vh7prBehEXdE9sHWJ1mZXTA,9
-grzegorz-0.6.0.dist-info/RECORD,,