sonatoki 0.1.4__tar.gz → 0.1.5__tar.gz

{sonatoki-0.1.4 → sonatoki-0.1.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sonatoki
-Version: 0.1.4
+Version: 0.1.5
 Summary: ilo li moku e toki li pana e sona ni: ni li toki ala toki pona?
 Author-Email: "jan Kekan San (@gregdan3)" <gregory.danielson3@gmail.com>
 License: AGPL-3.0-or-later

{sonatoki-0.1.4 → sonatoki-0.1.5}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "sonatoki"
-version = "0.1.4"
+version = "0.1.5"
 description = "ilo li moku e toki li pana e sona ni: ni li toki ala toki pona?"
 authors = [
     { name = "jan Kekan San (@gregdan3)", email = "gregory.danielson3@gmail.com" },

{sonatoki-0.1.4 → sonatoki-0.1.5}/src/sonatoki/Filters.py

@@ -1,10 +1,11 @@
 # STL
+import re
 from abc import ABC, abstractmethod
 from typing import Set
 from functools import lru_cache as cache  # cache comes in 3.9
 
 # PDM
-import regex as re
+import regex
 from typing_extensions import override
 
 # LOCAL
@@ -13,14 +14,16 @@ from sonatoki.constants import (
     CONSONANTS,
     NIMI_PU_SET,
     ALPHABET_SET,
+    UNICODE_PUNCT,
     ALLOWABLES_SET,
     NIMI_LINKU_SET,
     NIMI_PU_ALE_SET,
     NIMI_LINKU_ALE_SET,
+    PRUNED_POSIX_PUNCT,
     NIMI_LINKU_SANDBOX_SET,
 )
 
-re.DEFAULT_VERSION = re.VERSION1
+regex.DEFAULT_VERSION = regex.VERSION1
 
 
 class Filter(ABC):
@@ -41,6 +44,16 @@ class RegexFilter(Filter):
         return not not re.fullmatch(cls.pattern, token)
 
 
+class Regex1Filter(Filter):
+    pattern: "regex.Pattern[str]"
+
+    @classmethod
+    @override
+    @cache(maxsize=None)
+    def filter(cls, token: str) -> bool:
+        return not not regex.fullmatch(cls.pattern, token)
+
+
 class SetFilter(Filter):
     tokens: Set[str]
 
@@ -148,7 +161,7 @@ class Numeric(Filter):
 
 
 class Punctuation(RegexFilter):
-    pattern = re.compile(r"[\p{Punctuation}\p{posix_punct}]+")
+    pattern = re.compile(rf"[{PRUNED_POSIX_PUNCT}{UNICODE_PUNCT}]+")
 
 
 __all__ = [

{sonatoki-0.1.4 → sonatoki-0.1.5}/src/sonatoki/Preprocessors.py

@@ -17,13 +17,14 @@ It is up to the user to order them appropriately.
 """
 
 # STL
+import re
 from abc import ABC, abstractmethod
 
 # PDM
-import regex as re
+import regex
 from typing_extensions import override
 
-re.DEFAULT_VERSION = re.VERSION1
+regex.DEFAULT_VERSION = regex.VERSION1
 
 
 class Preprocessor(ABC):
@@ -43,6 +44,16 @@ class RegexPreprocessor(Preprocessor):
         return re.sub(cls.pattern, cls.replace, msg)
 
 
+class Regex1Preprocessor(Preprocessor):
+    pattern: "regex.Pattern[str]"
+    replace: str = " "
+
+    @classmethod
+    @override
+    def process(cls, msg: str) -> str:
+        return regex.sub(cls.pattern, cls.replace, msg)
+
+
 """
 The following classes are Ignorables.
 

{sonatoki-0.1.4 → sonatoki-0.1.5}/src/sonatoki/Scorers.py

@@ -10,8 +10,6 @@ from typing_extensions import override
 # LOCAL
 from sonatoki.Filters import Filter
 
-LOG = logging.getLogger(__name__)
-
 Number = Union[int, float]
 Weights = Dict[str, Number]
 
@@ -37,12 +35,7 @@ class PassFail(Scorer):
     def score_token(cls, token: str, filters: List[Type[Filter]]) -> Number:
         for f in filters:
             if f.filter(token):
-                score = 1
-                LOG.debug(
-                    "%12s.%s('%s') = %.2f", cls.__name__, f.__name__, token, score
-                )
-                return score
-        LOG.debug("%12s('%s') = 0.00", cls.__name__, token)
+                return 1
         return 0
 
     @classmethod
@@ -86,12 +79,7 @@ class Scaling(Scorer):
     def score_token(cls, token: str, filters: List[Type[Filter]], scale: int):
         for i, f in enumerate(filters):
             if f.filter(token):
-                score = scale - i
-                LOG.debug(
-                    "%12s.%s('%s') = %.2f", cls.__name__, f.__name__, token, score
-                )
-                return score
-        LOG.debug("%12s('%s') = 0.00", cls.__name__, token)
+                return scale - i
         return 0
 
     @classmethod

{sonatoki-0.1.4 → sonatoki-0.1.5}/src/sonatoki/Tokenizers.py

@@ -1,11 +1,15 @@
 # STL
+import re
 from abc import ABC, abstractmethod
 from typing import List
 
 # PDM
-import regex as re
+import regex
 from typing_extensions import override
 
+# LOCAL
+from sonatoki.constants import UNICODE_PUNCT, PRUNED_POSIX_PUNCT
+
 try:
     # PDM
     import nltk
@@ -15,7 +19,7 @@ except ImportError as e:
     nltk = e
 
 
-LANGUAGE = "english"  # for NLTK
+regex.DEFAULT_VERSION = regex.VERSION1
 
 
 class Tokenizer(ABC):
@@ -42,15 +46,26 @@ class RegexTokenizer(Tokenizer):
         return [clean for word in re.split(cls.pattern, s) if (clean := word.strip())]
 
 
+class Regex1Tokenizer(Tokenizer):
+    pattern: "regex.Pattern[str]"
+
+    @classmethod
+    @override
+    def tokenize(cls, s: str) -> List[str]:
+        return [
+            clean for word in regex.split(cls.pattern, s) if (clean := word.strip())
+        ]
+
+
 class WordTokenizerTok(RegexTokenizer):
-    pattern = re.compile(r"""([\p{Punctuation}\p{posix_punct}]+|\s+)""")
-    # TODO: are <> or {} that common as *sentence* delims? [] are already a stretch
-    # TODO: do the typography characters matter?
-    # NOTE: | / and , are *not* sentence delimiters for my purpose
+    pattern = re.compile(rf"""([{PRUNED_POSIX_PUNCT}{UNICODE_PUNCT}]+|\s+)""")
 
 
 class SentTokenizerTok(RegexTokenizer):
-    pattern = re.compile(r"""(?<=[.?!:;·…“”"'()\[\]\-]|$)""")
+    pattern = re.compile(r"""(?<=[.?!:;·…“”"'()\[\]\-])|$""", flags=re.MULTILINE)
+    # TODO: are <> or {} that common as *sentence* delims? [] are already a stretch
+    # TODO: do the typography characters matter?
+    # NOTE: | / and , are *not* sentence delimiters for my purpose
 
 
 class WordTokenizerRe(RegexTokenizer):

sonatoki-0.1.5/src/sonatoki/constants.py

@@ -0,0 +1,83 @@
+# STL
+import json
+from typing import Dict, List
+from pathlib import Path
+
+LINKU = Path(__file__).resolve().parent / Path("linku.json")
+SANDBOX = Path(__file__).resolve().parent / Path("sandbox.json")
+
+VOWELS = "aeiou"
+CONSONANTS = "jklmnpstw"
+ALPHABET = VOWELS + CONSONANTS
+ALPHABET_SET = set(ALPHABET)
+
+LANGUAGE = "english"  # for NLTK
+
+# `\p{posix_punct}` character class
+POSIX_PUNCT = r"""-!"#$%&'()*+,./:;<=>?@[\]^_`{|}~"""
+PRUNED_POSIX_PUNCT = r"""$+<=>^`|~"""  # only those that are not in UNICODE_PUNCT
+
+# `\p{Punctuation}` character class
+UNICODE_PUNCT = r"""!"#%&'()*,-./:;?@\[\\\]_{}¡§«¶·»¿;·՚՛՜՝՞՟։֊־׀׃׆׳״؉؊،؍؛؝؞؟٪٫٬٭۔܀܁܂܃܄܅܆܇܈܉܊܋܌܍߷߸߹࠰࠱࠲࠳࠴࠵࠶࠷࠸࠹࠺࠻࠼࠽࠾࡞।॥॰৽੶૰౷಄෴๏๚๛༄༅༆༇༈༉༊་༌།༎༏༐༑༒༔༺༻༼༽྅࿐࿑࿒࿓࿔࿙࿚၊။၌၍၎၏჻፠፡።፣፤፥፦፧፨᐀᙮᚛᚜᛫᛬᛭᜵᜶។៕៖៘៙៚᠀᠁᠂᠃᠄᠅᠆᠇᠈᠉᠊᥄᥅᨞᨟᪠᪡᪢᪣᪤᪥᪦᪨᪩᪪᪫᪬᪭᭚᭛᭜᭝᭞᭟᭠᭽᭾᯼᯽᯾᯿᰻᰼᰽᰾᰿᱾᱿᳀᳁᳂᳃᳄᳅᳆᳇᳓‐‑‒–—―‖‗‘’‚‛“”„‟†‡•‣․‥…‧‰‱′″‴‵‶‷‸‹›※‼‽‾‿⁀⁁⁂⁃⁅⁆⁇⁈⁉⁊⁋⁌⁍⁎⁏⁐⁑⁓⁔⁕⁖⁗⁘⁙⁚⁛⁜⁝⁞⁽⁾₍₎⌈⌉⌊⌋〈〉❨❩❪❫❬❭❮❯❰❱❲❳❴❵⟅⟆⟦⟧⟨⟩⟪⟫⟬⟭⟮⟯⦃⦄⦅⦆⦇⦈⦉⦊⦋⦌⦍⦎⦏⦐⦑⦒⦓⦔⦕⦖⦗⦘⧘⧙⧚⧛⧼⧽⳹⳺⳻⳼⳾⳿⵰⸀⸁⸂⸃⸄⸅⸆⸇⸈⸉⸊⸋⸌⸍⸎⸏⸐⸑⸒⸓⸔⸕⸖⸗⸘⸙⸚⸛⸜⸝⸞⸟⸠⸡⸢⸣⸤⸥⸦⸧⸨⸩⸪⸫⸬⸭⸮⸰⸱⸲⸳⸴⸵⸶⸷⸸⸹⸺⸻⸼⸽⸾⸿⹀⹁⹂⹃⹄⹅⹆⹇⹈⹉⹊⹋⹌⹍⹎⹏⹒⹓⹔⹕⹖⹗⹘⹙⹚⹛⹜⹝、。〃〈〉《》「」『』【】〔〕〖〗〘〙〚〛〜〝〞〟〰〽゠・꓾꓿꘍꘎꘏꙳꙾꛲꛳꛴꛵꛶꛷꡴꡵꡶꡷꣎꣏꣸꣹꣺꣼꤮꤯꥟꧁꧂꧃꧄꧅꧆꧇꧈꧉꧊꧋꧌꧍꧞꧟꩜꩝꩞꩟꫞꫟꫰꫱꯫﴾﴿︐︑︒︓︔︕︖︗︘︙︰︱︲︳︴︵︶︷︸︹︺︻︼︽︾︿﹀﹁﹂﹃﹄﹅﹆﹇﹈﹉﹊﹋﹌﹍﹎﹏﹐﹑﹒﹔﹕﹖﹗﹘﹙﹚﹛﹜﹝﹞﹟﹠﹡﹣﹨﹪﹫！＂＃％＆＇（）＊，－．／：；？＠［＼］＿｛｝｟｠｡｢｣､･𐄀𐄁𐄂𐎟𐏐𐕯𐡗𐤟𐤿𐩐𐩑𐩒𐩓𐩔𐩕𐩖𐩗𐩘𐩿𐫰𐫱𐫲𐫳𐫴𐫵𐫶𐬹𐬺𐬻𐬼𐬽𐬾𐬿𐮙𐮚𐮛𐮜𐺭𐽕𐽖𐽗𐽘𐽙𐾆𐾇𐾈𐾉𑁇𑁈𑁉𑁊𑁋𑁌𑁍𑂻𑂼𑂾𑂿𑃀𑃁𑅀𑅁𑅂𑅃𑅴𑅵𑇅𑇆𑇇𑇈𑇍𑇛𑇝𑇞𑇟𑈸𑈹𑈺𑈻𑈼𑈽𑊩𑑋𑑌𑑍𑑎𑑏𑑚𑑛𑑝𑓆𑗁𑗂𑗃𑗄𑗅𑗆𑗇𑗈𑗉𑗊𑗋𑗌𑗍𑗎𑗏𑗐𑗑𑗒𑗓𑗔𑗕𑗖𑗗𑙁𑙂𑙃𑙠𑙡𑙢𑙣𑙤𑙥𑙦𑙧𑙨𑙩𑙪𑙫𑙬𑚹𑜼𑜽𑜾𑠻𑥄𑥅𑥆𑧢𑨿𑩀𑩁𑩂𑩃𑩄𑩅𑩆𑪚𑪛𑪜𑪞𑪟𑪠𑪡𑪢𑬀𑬁𑬂𑬃𑬄𑬅𑬆𑬇𑬈𑬉𑱁𑱂𑱃𑱄𑱅𑱰𑱱𑻷𑻸𑽃𑽄𑽅𑽆𑽇𑽈𑽉𑽊𑽋𑽌𑽍𑽎𑽏𑿿𒑰𒑱𒑲𒑳𒑴𒿱𒿲𖩮𖩯𖫵𖬷𖬸𖬹𖬺𖬻𖭄𖺗𖺘𖺙𖺚𖿢𛲟𝪇𝪈𝪉𝪊𝪋𞥞𞥟"""
+# NOTE: This list diverges slightly from the raw list, since []\ must be escaped
+# The [] need to be escaped to avoid prematurely closing the regex character class
+# The \ needs to be escaped to be considered as a raw \
+
+# https://www.compart.com/en/unicode/category
+# https://unicode.org/Public/UNIDATA/UnicodeData.txt
+
+
+"""Commonly occurring strings which are some kind of valid Toki Pona or external token"""
+ALLOWABLES = {
+    "cw",  # Content Warning
+    "x",  # ala
+    "y",  # anu
+    "kxk",  # ken ala ken
+    "wxw",  # wile ala wile
+}
+
+
+with open(LINKU) as f:
+    r: Dict[str, Dict[str, str]] = json.loads(f.read())
+    NIMI_PU: List[str] = [d["word"] for d in r.values() if d["book"] == "pu"]
+    NIMI_PU_ALE: List[str] = NIMI_PU + ["namako", "kin", "oko"]
+    NIMI_LINKU: List[str] = [
+        d["word"] for d in r.values() if d["usage_category"] in ["core", "common"]
+    ]
+    NIMI_LINKU_ALE: List[str] = [d["word"] for d in r.values()]
+
+with open(SANDBOX) as f:
+    r: Dict[str, Dict[str, str]] = json.loads(f.read())
+    NIMI_LINKU_SANDBOX: List[str] = [d["word"] for d in r.values()]
+
+
+NIMI_PU_SET = set(NIMI_PU)
+NIMI_PU_ALE_SET = set(NIMI_PU_ALE)
+NIMI_LINKU_SET = set(NIMI_LINKU)
+NIMI_LINKU_ALE_SET = set(NIMI_LINKU_ALE)
+NIMI_LINKU_SANDBOX_SET = set(NIMI_LINKU_SANDBOX)
+ALLOWABLES_SET = set(ALLOWABLES)
+
+__all__ = [
+    "VOWELS",
+    #
+    "CONSONANTS",
+    #
+    "ALPHABET",
+    "ALPHABET_SET",
+    #
+    "NIMI_PU",
+    "NIMI_PU_SET",
+    #
+    "NIMI_PU_ALE",
+    "NIMI_PU_ALE_SET",
+    #
+    "NIMI_LINKU",
+    "NIMI_LINKU_SET",
+    #
+    "NIMI_LINKU_ALE",
+    "NIMI_LINKU_ALE_SET",
+    #
+    "NIMI_LINKU_SANDBOX",
+    "NIMI_LINKU_SANDBOX_SET",
+]

{sonatoki-0.1.4 → sonatoki-0.1.5}/src/sonatoki/ilo.py

@@ -1,5 +1,4 @@
 # STL
-import logging
 from typing import List, Type, Tuple
 
 # LOCAL
@@ -9,8 +8,6 @@ from sonatoki.Cleaners import Cleaner
 from sonatoki.Tokenizers import Tokenizer
 from sonatoki.Preprocessors import Preprocessor
 
-LOG = logging.getLogger(__name__)
-
 
 class Ilo:
     __preprocessors: List[Type[Preprocessor]]
@@ -20,7 +17,6 @@ class Ilo:
     __scoring_filters: List[Type[Filter]]
     __scorer: Type[Scorer]
     __passing_score: Number
-    logging_threshold: Number = -1
 
     def __init__(
         self,
@@ -104,14 +100,6 @@ class Ilo:
         score = self.score_tokens(cleaned)
         result = score >= self.__passing_score
 
-        if score <= self.logging_threshold:
-            LOG.debug("msg: %.2f  %s", score, repr(message))
-            LOG.debug("preproc:   %s", repr(preprocessed))
-            LOG.debug("tokenized: %s", tokenized)
-            LOG.debug("filtered:  %s", filtered)
-            LOG.debug("cleaned:   %s", cleaned)
-        # TODO: Move to each function? Loses ability to control when logging occurs by threshold
-
         return preprocessed, tokenized, filtered, cleaned, score, result
 
     def is_toki_pona(self, message: str) -> bool:

{sonatoki-0.1.4 → sonatoki-0.1.5}/tests/test_filters.py

@@ -82,16 +82,15 @@ def test_ProperName(s: str):
     assert res, repr(s)
 
 
-# I use `regex`'s Unicode property feature, which Hypothesis doesn't understand
-# So I have to provide a different regex tha doesn't technically match
-@given(st.from_regex(r"[^\w\s]+", fullmatch=True))
+@given(st.from_regex(Punctuation.pattern.pattern, fullmatch=True))
+@example("[]")
+@example(r"\\")
+@example(r"\"")
 @example("⟨·⟩")
 @example("…")
-@example("「　」")
+@example("「」")  # `　`
 @example(string.punctuation)
-@settings(suppress_health_check=[HealthCheck.filter_too_much])  # FIXME
 def test_Punctuation(s: str):
-    _ = assume(re.fullmatch(Punctuation.pattern.pattern, s))
     res = Punctuation.filter(s)
     assert res, repr(s)
 

{sonatoki-0.1.4 → sonatoki-0.1.5}/tests/test_ilo.py

@@ -9,7 +9,6 @@ from sonatoki.Configs import LazyConfig, PrefConfig
 @pytest.fixture
 def ilo():
     ilo = Ilo(**PrefConfig)
-    # ilo.logging_threshold = 0.8
     return ilo
 
 

{sonatoki-0.1.4 → sonatoki-0.1.5}/tests/tokenize_cases/tokenize_sentences_tok.yml

@@ -19,6 +19,24 @@
   output:
     - "mi mu."
     - "mi wawa."
+- name: "empty"
+  input: ""
+  output: []
+- name: "whitespace"
+  input: "  \n  "
+  output: []
+- name: "newline basic"
+  input: "sina lon seme?\nmi wile lon poka...\n"
+  output:
+    - "sina lon seme?"
+    - "mi wile lon poka."
+    - "."
+    - "."
+- name: "newline alone"
+  input: "sina lon seme\nmi wile lon poka"
+  output:
+    - "sina lon seme"
+    - "mi wile lon poka"
 - name: "dash"
   input: "mi sona ala e ni- sina seme a"
   output:

sonatoki-0.1.4/src/sonatoki/constants.py

@@ -1,67 +0,0 @@
-# STL
-import json
-from typing import Dict, List
-from pathlib import Path
-
-LINKU = Path(__file__).resolve().parent / Path("linku.json")
-SANDBOX = Path(__file__).resolve().parent / Path("sandbox.json")
-
-VOWELS = "aeiou"
-CONSONANTS = "jklmnpstw"
-ALPHABET = VOWELS + CONSONANTS
-ALPHABET_SET = set(ALPHABET)
-
-"""Commonly occurring strings which are some kind of valid Toki Pona or external token"""
-ALLOWABLES = {
-    "cw",  # Content Warning
-    "x",  # ala
-    "y",  # anu
-    "kxk",  # ken ala ken
-    "wxw",  # wile ala wile
-}
-
-
-with open(LINKU) as f:
-    r: Dict[str, Dict[str, str]] = json.loads(f.read())
-    NIMI_PU: List[str] = [d["word"] for d in r.values() if d["book"] == "pu"]
-    NIMI_PU_ALE: List[str] = NIMI_PU + ["namako", "kin", "oko"]
-    NIMI_LINKU: List[str] = [
-        d["word"] for d in r.values() if d["usage_category"] in ["core", "common"]
-    ]
-    NIMI_LINKU_ALE: List[str] = [d["word"] for d in r.values()]
-
-with open(SANDBOX) as f:
-    r: Dict[str, Dict[str, str]] = json.loads(f.read())
-    NIMI_LINKU_SANDBOX: List[str] = [d["word"] for d in r.values()]
-
-
-NIMI_PU_SET = set(NIMI_PU)
-NIMI_PU_ALE_SET = set(NIMI_PU_ALE)
-NIMI_LINKU_SET = set(NIMI_LINKU)
-NIMI_LINKU_ALE_SET = set(NIMI_LINKU_ALE)
-NIMI_LINKU_SANDBOX_SET = set(NIMI_LINKU_SANDBOX)
-ALLOWABLES_SET = set(ALLOWABLES)
-
-__all__ = [
-    "VOWELS",
-    #
-    "CONSONANTS",
-    #
-    "ALPHABET",
-    "ALPHABET_SET",
-    #
-    "NIMI_PU",
-    "NIMI_PU_SET",
-    #
-    "NIMI_PU_ALE",
-    "NIMI_PU_ALE_SET",
-    #
-    "NIMI_LINKU",
-    "NIMI_LINKU_SET",
-    #
-    "NIMI_LINKU_ALE",
-    "NIMI_LINKU_ALE_SET",
-    #
-    "NIMI_LINKU_SANDBOX",
-    "NIMI_LINKU_SANDBOX_SET",
-]