cjclassifier 1.0.2__py3-none-any.whl

cjclassifier/__init__.py

@@ -0,0 +1,25 @@
+# Copyright 2026 Jeremy Lilley (jeremy@jlilley.net)
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""CJClassifier — classify text as Chinese Simplified, Chinese Traditional, or Japanese.
+
+Uses a statistical model of unigram and bigram ideograph probabilities built from
+Japanese and Chinese language Wikipedia corpora.
+"""
+
+from cjclassifier.classifier import CJClassifier
+from cjclassifier.language import CJLanguage
+
+__all__ = ["CJClassifier", "CJLanguage"]
+__version__ = "1.0.2"

cjclassifier/classifier.py

@@ -0,0 +1,605 @@
+# Copyright 2026 Jeremy Lilley (jeremy@jlilley.net)
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Core classifier: detects whether CJ text is Japanese, Chinese Simplified,
+or Chinese Traditional.
+
+Classification of ideographs is based on a unigram + bigram statistical model
+built from Japanese and Chinese language Wikipedia corpora. For every character
+and character-pair in the CJ range, the model stores per-language
+log-probabilities. At classification time the library sums these
+log-probabilities across the input and picks the language with the highest
+score."""
+
+from __future__ import annotations
+
+import array
+import gzip
+import importlib.resources
+import logging
+from typing import Dict, List, Optional, Tuple
+
+from cjclassifier.language import CJLanguage
+
+logger = logging.getLogger(__name__)
+
+_CJ_RANGE_START = 0x3400
+_CJ_RANGE_END = 0x9FFF
+_CJ_RANGE_SIZE = _CJ_RANGE_END - _CJ_RANGE_START + 1
+_LANG_COUNT = 3  # zh-hans=0, zh-hant=1, ja=2
+
+CJ_LANGUAGES: List[CJLanguage] = [
+    CJLanguage.CHINESE_SIMPLIFIED,
+    CJLanguage.CHINESE_TRADITIONAL,
+    CJLanguage.JAPANESE,
+]
+
+# Singleton cache
+_cache: Dict[str, "CJClassifier"] = {}
+
+
+def _is_kana(c: int) -> bool:
+    """Check if a Unicode code point is Hiragana, Katakana, or half-width Katakana."""
+    if 0x3040 <= c <= 0x30FF:
+        return True
+    if 0x31F0 <= c <= 0x31FF:
+        return True
+    if 0xFF65 <= c <= 0xFF9F:
+        return True
+    return False
+
+
+def _in_main_cj_range(c: int) -> bool:
+    """Check if a Unicode code point is in the main CJ Unified Ideograph range."""
+    return _CJ_RANGE_START <= c <= _CJ_RANGE_END
+
+
+# ========================================================================
+# BigramMap: open-addressing hash map using flat array.array storage
+# ========================================================================
+# We could just use a Dict[int, int] here, but this cuts the model's
+# memory overhead from 109MB to 29MB, since all the int boxing and
+# hashtable overhead adds up.
+#
+# This is just a port of the Java BigramToFloatArrayMap. It avoids Python
+# object boxing by storing keys and value-offsets in array.array('i')
+# (4 bytes per slot) instead of a dict of Python int objects
+# (~28-32 bytes each).
+
+_EMPTY = 0
+_MAX_LOAD_FACTOR = 0.75
+
+
+def _mix(k: int) -> int:
+    """Hash mixing function to spread bits of the key."""
+    k = (k ^ (k >> 16)) & 0xFFFFFFFF
+    k = (k * 0x85EBCA6B) & 0xFFFFFFFF
+    k = (k ^ (k >> 13)) & 0xFFFFFFFF
+    return k
+
+
+def _table_size_for(expected: int) -> int:
+    """Return the smallest power of 2 that can hold expected entries at <=75% load."""
+    min_cap = int(expected / _MAX_LOAD_FACTOR) + 1
+    cap = 1
+    while cap < min_cap:
+        cap <<= 1
+    return max(cap, 16)
+
+
+class _BigramMap:
+    """Open-addressing hash map: bigram key (int) -> offset into prob_data.
+
+    Keys and value-offsets are stored in array.array('i') to avoid the overhead
+    of Python int objects. Probabilities are in a separate array.array('f').
+    """
+
+    __slots__ = ("_keys", "_value_indices", "prob_data", "_mask")
+
+    def __init__(self, keys, value_indices, prob_data, mask):
+        self._keys = keys
+        self._value_indices = value_indices
+        self.prob_data = prob_data
+        self._mask = mask
+
+    def get_offset(self, c1: int, c2: int) -> int:
+        """Look up the offset for a bigram. Returns 0 if not present."""
+        key = (c1 << 16) | c2
+        keys = self._keys
+        value_indices = self._value_indices
+        mask = self._mask
+        idx = _mix(key) & mask
+        while True:
+            k = keys[idx]
+            if k == key:
+                return value_indices[idx]
+            if k == _EMPTY:
+                return 0
+            idx = (idx + 1) & mask
+
+
+class _BigramMapBuilder:
+    """Mutable builder for constructing a _BigramMap."""
+
+    __slots__ = ("_keys", "_value_indices", "_prob_data", "_prob_data_size",
+                 "_size", "_mask", "_threshold")
+
+    def __init__(self, expected_size: int):
+        capacity = _table_size_for(expected_size)
+        self._keys = array.array('I', [0]) * capacity
+        self._value_indices = array.array('I', [0]) * capacity
+        self._mask = capacity - 1
+        self._threshold = int(capacity * _MAX_LOAD_FACTOR)
+        self._prob_data = array.array('f', [0.0])  # offset 0 reserved
+        self._prob_data_size = 1
+        self._size = 0
+
+    def put(self, c1: int, c2: int, probs: List[float]):
+        """Store probabilities for a bigram key. Copies from probs list."""
+        key = (c1 << 16) | c2
+        if self._size >= self._threshold:
+            self._resize()
+        idx = _mix(key) & self._mask
+        while True:
+            k = self._keys[idx]
+            if k == _EMPTY:
+                self._keys[idx] = key
+                self._value_indices[idx] = len(self._prob_data)
+                self._prob_data.extend(probs)
+                self._size += 1
+                return
+            if k == key:
+                # Update in place
+                off = self._value_indices[idx]
+                for i in range(len(probs)):
+                    self._prob_data[off + i] = probs[i]
+                return
+            idx = (idx + 1) & self._mask
+
+    def build(self) -> _BigramMap:
+        """Build an immutable _BigramMap."""
+        return _BigramMap(self._keys, self._value_indices,
+                          self._prob_data, self._mask)
+
+    def _resize(self):
+        new_capacity = (self._mask + 1) << 1
+        old_keys = self._keys
+        old_value_indices = self._value_indices
+        self._keys = array.array('I', [0]) * new_capacity
+        self._value_indices = array.array('I', [0]) * new_capacity
+        self._mask = new_capacity - 1
+        self._threshold = int(new_capacity * _MAX_LOAD_FACTOR)
+        self._size = 0
+        for i in range(len(old_keys)):
+            if old_keys[i] != _EMPTY:
+                self._rehash_put(old_keys[i], old_value_indices[i])
+
+    def _rehash_put(self, key: int, value_index: int):
+        idx = _mix(key) & self._mask
+        while True:
+            if self._keys[idx] == _EMPTY:
+                self._keys[idx] = key
+                self._value_indices[idx] = value_index
+                self._size += 1
+                return
+            idx = (idx + 1) & self._mask
+
+
+class Scores:
+    """Accumulates per-language scoring data during CJ text processing."""
+
+    __slots__ = (
+        "unigram_scores", "bigram_scores",
+        "unigram_hits_per_lang", "bigram_hits_per_lang",
+        "kana_count", "cj_char_count",
+    )
+
+    def __init__(self):
+        self.unigram_scores = [0.0] * _LANG_COUNT
+        self.bigram_scores = [0.0] * _LANG_COUNT
+        self.unigram_hits_per_lang = [0] * _LANG_COUNT
+        self.bigram_hits_per_lang = [0] * _LANG_COUNT
+        self.kana_count = 0
+        self.cj_char_count = 0
+
+    def clear(self):
+        """Reset all accumulated scores and counts to zero."""
+        for i in range(_LANG_COUNT):
+            self.unigram_scores[i] = 0.0
+            self.bigram_scores[i] = 0.0
+            self.unigram_hits_per_lang[i] = 0
+            self.bigram_hits_per_lang[i] = 0
+        self.kana_count = 0
+        self.cj_char_count = 0
+
+    def any_hits(self) -> bool:
+        """True if at least one unigram was matched in any language."""
+        return max(self.unigram_hits_per_lang) > 0
+
+
+class Results:
+    """Detection results, including accumulated Scores and the computed result."""
+
+    __slots__ = ("scores", "total_scores", "boosts", "result", "gap")
+
+    def __init__(self, scores: Optional[Scores] = None):
+        self.scores = scores if scores is not None else Scores()
+        self.total_scores = [0.0] * _LANG_COUNT
+        self.boosts = [0.0] * _LANG_COUNT
+        self.result: Optional[CJLanguage] = None
+        self.gap: float = 0.0
+
+    def clear(self):
+        """Reset all scores, boosts, and the computed result."""
+        self.scores.clear()
+        for i in range(_LANG_COUNT):
+            self.total_scores[i] = 0.0
+            self.boosts[i] = 0.0
+        self.result = None
+        self.gap = 0.0
+
+    def _compute_totals(self, placeholder_score: float):
+        max_unigram = max(self.scores.unigram_hits_per_lang)
+        max_bigram = max(self.scores.bigram_hits_per_lang)
+        for i in range(_LANG_COUNT):
+            self.total_scores[i] = (
+                self.scores.unigram_scores[i]
+                + (max_unigram - self.scores.unigram_hits_per_lang[i]) * placeholder_score
+                + self.scores.bigram_scores[i]
+                + (max_bigram - self.scores.bigram_hits_per_lang[i]) * placeholder_score
+            )
+            # Implement boosts: since logprob values are negative, a favorable boost
+            # is negative.
+            self.total_scores[i] -= self.boosts[i] * self.total_scores[i]
+
+    def to_short_string(self) -> str:
+        """Compact comma-separated representation of per-language relative scores.
+
+        Example: 'zh-hans:1.00,zh-hant:0.97,ja:0.85'
+        Returns '' if no result has been computed or the result is UNKNOWN.
+        """
+        if self.result is None or self.result is CJLanguage.UNKNOWN:
+            return ""
+        if self.scores.kana_count > 0 and self.result is CJLanguage.JAPANESE:
+            return "ja:1.0,zh-hans:0,zh-hant:0"
+
+        # Sort indices by total_scores descending
+        order = sorted(range(_LANG_COUNT), key=lambda i: self.total_scores[i], reverse=True)
+        best = self.total_scores[order[0]]
+        if best == 0:
+            return ""
+
+        parts = []
+        for li in order:
+            ratio = best / self.total_scores[li] if self.total_scores[li] != 0 else 0
+            parts.append(f"{CJ_LANGUAGES[li].iso_code}:{ratio:.2f}")
+        return ",".join(parts)
+
+    def __repr__(self) -> str:
+        code = self.result.iso_code if self.result else "None"
+        parts = [f"Results(result={code}"]
+        if self.scores.kana_count > 0:
+            total = self.scores.kana_count + self.scores.cj_char_count
+            parts.append(f" kana={self.scores.kana_count}/{total}")
+        for li in range(_LANG_COUNT):
+            parts.append(
+                f" | {CJ_LANGUAGES[li].iso_code}:"
+                f" uni={self.scores.unigram_scores[li]:.2f}"
+                f" bi={self.scores.bigram_scores[li]:.2f}"
+                f" total={self.total_scores[li]:.2f}"
+                f" biHits={self.scores.bigram_hits_per_lang[li]}"
+            )
+        parts.append(")")
+        return "".join(parts)
+
+
+class CJClassifier:
+    """Detects whether CJ text is Japanese, Chinese Simplified, or Chinese Traditional.
+
+    Load with ``CJClassifier.load()``. The model is cached as a singleton.
+    Call ``detect(text)`` to classify text.
+
+    Example::
+
+        cjc = CJClassifier.load()
+        cjc.detect("今天天气很好")  # => CJLanguage.CHINESE_SIMPLIFIED
+        cjc.detect("事務所")        # => CJLanguage.JAPANESE
+    """
+
+    # Class-level constants exposed for user code
+    CJ_LANGUAGES = CJ_LANGUAGES
+
+    def __init__(
+        self,
+        unigram_log_probs: List[float],
+        bigram_map: _BigramMap,
+        default_log_prob: float,
+    ):
+        self._unigram_log_probs = unigram_log_probs
+        self._bigram_map = bigram_map
+        self._default_log_prob = default_log_prob
+        self._tolerated_kana_threshold = 0.01
+
+    @property
+    def tolerated_kana_threshold(self) -> float:
+        """Kana fraction threshold above which text is classified as Japanese."""
+        return self._tolerated_kana_threshold
+
+    @tolerated_kana_threshold.setter
+    def tolerated_kana_threshold(self, value: float):
+        self._tolerated_kana_threshold = value
+
+    # ========================================================================
+    # Loading
+    # ========================================================================
+
+    @classmethod
+    def load(cls, path: Optional[str] = None, log_prob_floor: float = 0.0) -> "CJClassifier":
+        """Load a CJClassifier, using a cached singleton.
+
+        Args:
+            path: Filesystem path to a model file (may be gzipped).
+                  If None, loads the bundled model from the package.
+            log_prob_floor: Custom log-probability floor.
+                  If 0, uses the file's MinLogProb header value.
+
+        Returns:
+            A cached (or newly-loaded) CJClassifier instance.
+        """
+        cache_key = f"{path or 'bundled'}:{log_prob_floor}"
+        if cache_key in _cache:
+            return _cache[cache_key]
+
+        if path is not None:
+            instance = cls._load_from_file(path, log_prob_floor)
+        else:
+            instance = cls._load_bundled(log_prob_floor)
+
+        _cache[cache_key] = instance
+        return instance
+
+    @classmethod
+    def clear_cached_models(cls):
+        """Clear loaded model(s), primarily for testing."""
+        _cache.clear()
+
+    @classmethod
+    def _load_bundled(cls, log_prob_floor: float) -> "CJClassifier":
+        """Load the bundled model from package resources."""
+        # Compatible with Python 3.8+
+        try:
+            # Python 3.9+
+            ref = importlib.resources.files("cjclassifier").joinpath("cjlogprobs.gz")
+            data = ref.read_bytes()
+        except AttributeError:
+            # Python 3.8 fallback
+            with importlib.resources.open_binary("cjclassifier", "cjlogprobs.gz") as f:
+                data = f.read()
+
+        text = gzip.decompress(data).decode("utf-8")
+        return cls._parse_model(text, "bundled:cjlogprobs.gz", log_prob_floor)
+
+    @classmethod
+    def _load_from_file(cls, path: str, log_prob_floor: float) -> "CJClassifier":
+        """Load a model from a filesystem path."""
+        if path.endswith(".gz"):
+            with gzip.open(path, "rt", encoding="utf-8") as f:
+                text = f.read()
+        else:
+            with open(path, "r", encoding="utf-8") as f:
+                text = f.read()
+        return cls._parse_model(text, path, log_prob_floor)
+
+    @classmethod
+    def _parse_model(cls, text: str, label: str, log_prob_floor: float) -> "CJClassifier":
+        """Parse the model file format into arrays."""
+        lines = text.split("\n")
+
+        # Parse header
+        header = lines[0]
+        if not header.startswith("Languages: "):
+            raise ValueError(f"Invalid model file (bad header): {label}")
+
+        header_parts = header.split(" ")
+        lang_codes = header_parts[1].split(",")
+        lang_map = []
+        for code in lang_codes:
+            lang = CJLanguage.from_string(code)
+            if lang is CJLanguage.UNKNOWN:
+                raise ValueError(f"Unknown CJ language in header: {code} in {label}")
+            lang_map.append(int(lang))
+
+        # Parse MinLogProb from header
+        parsed_min_prob = None
+        for i in range(len(header_parts) - 1):
+            if header_parts[i] == "MinLogProb:":
+                parsed_min_prob = float(header_parts[i + 1])
+                break
+
+        if log_prob_floor == 0.0:
+            if parsed_min_prob is None:
+                raise ValueError(
+                    f"No MinLogProb in header and no explicit log_prob_floor: {label}"
+                )
+            default_log_prob = parsed_min_prob
+        else:
+            if parsed_min_prob is not None:
+                default_log_prob = max(log_prob_floor, parsed_min_prob)
+            else:
+                default_log_prob = log_prob_floor
+
+        # Parse unigram and bigram lines.
+        # Bigram data uses an open-addressing hash map backed by array.array
+        # to avoid the overhead of Python object boxing (~28 bytes per int/float
+        # vs 4 bytes in a C-level array).
+        unigram_log_probs = [0.0] * (_CJ_RANGE_SIZE * _LANG_COUNT)
+        bigram_builder = _BigramMapBuilder(16 * 1024)
+
+        unigram_count = 0
+        bigram_count = 0
+        probs = [0.0] * _LANG_COUNT  # reused across bigram lines
+
+        for line in lines[1:]:
+            if not line:
+                continue
+            parts = line.split(" ")
+            key = parts[0]
+            if len(parts) != len(lang_map) + 1:
+                raise ValueError(f"Column count mismatch on line: {line} in {label}")
+
+            if len(key) == 1:
+                # Unigram
+                c = ord(key)
+                idx = c - _CJ_RANGE_START
+                if 0 <= idx < _CJ_RANGE_SIZE:
+                    for col in range(len(lang_map)):
+                        val = max(float(parts[col + 1]), default_log_prob)
+                        unigram_log_probs[idx * _LANG_COUNT + lang_map[col]] = val
+                    unigram_count += 1
+
+            elif len(key) == 2:
+                # Bigram
+                any_higher = False
+                for col in range(len(lang_map)):
+                    val = float(parts[col + 1])
+                    if val < default_log_prob or val == 0.0:
+                        val = default_log_prob
+                    else:
+                        any_higher = True
+                    probs[lang_map[col]] = val
+                if any_higher:
+                    bigram_builder.put(ord(key[0]), ord(key[1]), probs)
+                    bigram_count += 1
+
+        logger.info(
+            "Loaded %d unigrams and %d bigrams, minLogProb: %.2f, from %s",
+            unigram_count, bigram_count, default_log_prob, label,
+        )
+
+        return cls(unigram_log_probs, bigram_builder.build(), default_log_prob)
+
+    # ========================================================================
+    # Detection
+    # ========================================================================
+
+    def detect(self, text: str, results: Optional[Results] = None) -> CJLanguage:
+        """Detect the language of the given CJ text.
+
+        Args:
+            text: The text to classify.
+            results: Optional Results instance to populate with scoring details.
+
+        Returns:
+            The detected CJLanguage, or CJLanguage.UNKNOWN if undetermined.
+        """
+        if results is None:
+            results = Results()
+        else:
+            results.clear()
+        self.add_text(text, results.scores)
+        return self.compute_result(results)
+
+    def add_text(self, text: str, scores: Scores):
+        """Add text to the scoring accumulator.
+
+        This is the lower-level API for incremental classification.
+        Call compute_result() when all text has been added.
+
+        Args:
+            text: A chunk of CJ text to process.
+            scores: The Scores instance to accumulate into.
+        """
+        prev = 0
+        prev_in_range = False
+
+        unigram_log_probs = self._unigram_log_probs
+        bigram_map = self._bigram_map
+        bigram_prob_data = bigram_map.prob_data
+        default_log_prob = self._default_log_prob
+
+        for ch in text:
+            c = ord(ch)
+            if _is_kana(c):
+                scores.kana_count += 1
+                continue
+            in_range = _in_main_cj_range(c)
+            if in_range:
+                scores.cj_char_count += 1
+                # Unigram scoring
+                idx = c - _CJ_RANGE_START
+                base = idx * _LANG_COUNT
+                for li in range(_LANG_COUNT):
+                    u_prob = unigram_log_probs[base + li]
+                    scores.unigram_scores[li] += max(u_prob, default_log_prob)
+                    if u_prob != 0:
+                        scores.unigram_hits_per_lang[li] += 1
+
+                # Bigram scoring
+                if prev_in_range:
+                    b_off = bigram_map.get_offset(prev, c)
+                    if b_off != 0:
+                        for li in range(_LANG_COUNT):
+                            bp = bigram_prob_data[b_off + li]
+                            scores.bigram_scores[li] += max(bp, default_log_prob)
+                            if bp != 0:
+                                scores.bigram_hits_per_lang[li] += 1
+
+            prev = c
+            prev_in_range = in_range
+
+    def compute_result(self, results: Results) -> CJLanguage:
+        """Compute the final language result from accumulated scores.
+
+        Args:
+            results: The Results instance containing accumulated scores.
+
+        Returns:
+            The detected CJLanguage, or CJLanguage.UNKNOWN if undetermined.
+        """
+        scores = results.scores
+
+        if scores.kana_count > 0:
+            kana_ratio = scores.kana_count / (scores.kana_count + scores.cj_char_count)
+            if kana_ratio > self._tolerated_kana_threshold:
+                results.result = CJLanguage.JAPANESE
+                results.gap = 1.0
+                results.total_scores[CJLanguage.JAPANESE] = 1.0
+                return results.result
+
+        if not scores.any_hits():
+            results.result = CJLanguage.UNKNOWN
+            results.gap = 0.0
+            return results.result
+
+        results._compute_totals(self._default_log_prob)
+
+        # Find best and second-best language
+        best_idx = 0
+        second_idx = -1
+        for li in range(1, _LANG_COUNT):
+            if results.total_scores[li] > results.total_scores[best_idx]:
+                second_idx = best_idx
+                best_idx = li
+            elif second_idx < 0 or results.total_scores[li] > results.total_scores[second_idx]:
+                second_idx = li
+
+        results.result = CJ_LANGUAGES[best_idx]
+
+        # Compute gap: 1 - (best / second). Scores are negative logprobs,
+        # so best is least negative.
+        best = results.total_scores[best_idx]
+        second = results.total_scores[second_idx] if second_idx >= 0 else best
+        results.gap = (1.0 - (best / second)) if second != 0.0 else 0.0
+
+        return results.result

cjclassifier/language.py

@@ -0,0 +1,90 @@
+# Copyright 2026 Jeremy Lilley (jeremy@jlilley.net)
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Language enum for CJClassifier."""
+
+from enum import IntEnum
+from typing import Dict, List, Tuple
+
+
+class CJLanguage(IntEnum):
+    """Represents the three CJ languages plus UNKNOWN.
+
+    The int value is the internal array index used by the classifier:
+    CHINESE_SIMPLIFIED=0, CHINESE_TRADITIONAL=1, JAPANESE=2, UNKNOWN=-1.
+    """
+
+    UNKNOWN = -1
+    CHINESE_SIMPLIFIED = 0
+    CHINESE_TRADITIONAL = 1
+    JAPANESE = 2
+
+    @property
+    def iso_code(self) -> str:
+        """ISO 639-1-style code (e.g. 'zh-hans', 'ja')."""
+        return _LANG_INFO[self][0]
+
+    @property
+    def iso_code3(self) -> str:
+        """ISO 639-3-style code (e.g. 'zho-hans', 'jpn')."""
+        return _LANG_INFO[self][1]
+
+    @property
+    def alt_names(self) -> List[str]:
+        """Alternate name strings recognized by from_string()."""
+        return list(_LANG_INFO[self][2])
+
+    def is_chinese(self) -> bool:
+        """True if this is CHINESE_SIMPLIFIED or CHINESE_TRADITIONAL."""
+        return self in (CJLanguage.CHINESE_SIMPLIFIED, CJLanguage.CHINESE_TRADITIONAL)
+
+    def is_japanese(self) -> bool:
+        """True if this is JAPANESE."""
+        return self is CJLanguage.JAPANESE
+
+    @classmethod
+    def from_string(cls, s: str) -> "CJLanguage":
+        """Look up a CJLanguage by name, ISO code, or alternate name (case-insensitive).
+
+        Returns UNKNOWN if not recognized.
+        """
+        if s is None:
+            return cls.UNKNOWN
+        return _BY_NAME.get(s.lower(), cls.UNKNOWN)
+
+
+# Per-language metadata: (iso_code, iso_code3, alt_names)
+_LANG_INFO: Dict[CJLanguage, Tuple[str, str, List[str]]] = {
+    CJLanguage.UNKNOWN: ("", "", []),
+    CJLanguage.CHINESE_SIMPLIFIED: (
+        "zh-hans", "zho-hans",
+        ["chinese", "zh", "zh-cn", "zh-hans-cn", "zh-hans-sg"],
+    ),
+    CJLanguage.CHINESE_TRADITIONAL: (
+        "zh-hant", "zho-hant",
+        ["zh-hant-hk", "zh-hk", "zh-hant-tw"],
+    ),
+    CJLanguage.JAPANESE: ("ja", "jpn", ["jp"]),
+}
+
+# Build the lookup table once at module load time.
+_BY_NAME: Dict[str, CJLanguage] = {}
+for _lang in CJLanguage:
+    _BY_NAME[_lang.name.lower()] = _lang
+    if _lang.iso_code:
+        _BY_NAME[_lang.iso_code] = _lang
+    if _lang.iso_code3:
+        _BY_NAME[_lang.iso_code3] = _lang
+    for _name in _lang.alt_names:
+        _BY_NAME[_name] = _lang

cjclassifier-1.0.2.dist-info/METADATA

@@ -0,0 +1,76 @@
+Metadata-Version: 2.4
+Name: cjclassifier
+Version: 1.0.2
+Summary: Classify text as Chinese Simplified, Chinese Traditional, or Japanese using a statistical model.
+Author-email: Jeremy Lilley <jeremy@jlilley.net>
+License-Expression: Apache-2.0
+Project-URL: Homepage, https://github.com/jlpka/cjclassifier
+Project-URL: Repository, https://github.com/jlpka/cjclassifier
+Keywords: chinese,japanese,language-detection,cjk,nlp
+Classifier: Development Status :: 5 - Production/Stable
+Classifier: Intended Audience :: Developers
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Text Processing :: Linguistic
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Dynamic: license-file
+
+# CJClassifier
+
+A focused, pure-Python library for distinguishing between **Japanese**,
+**Chinese Simplified**, and **Chinese Traditional** text using a statistical
+model of ideograph frequencies built from Japanese and Chinese language Wikipedia
+corpora.
+
+No external dependencies. The bundled model is ~7 MB on disk and ~29 MB in
+memory (loaded once and cached).
+
+## Install
+
+```bash
+pip install cjclassifier
+```
+
+## Usage
+
+```python
+from cjclassifier import CJClassifier, CJLanguage
+
+cjc = CJClassifier.load()
+
+cjc.detect("今天天气很好，我们去公园散步")   # => CJLanguage.CHINESE_SIMPLIFIED
+cjc.detect("今天天氣很好，我們去公園散步")   # => CJLanguage.CHINESE_TRADITIONAL
+cjc.detect("事務所")                         # => CJLanguage.JAPANESE  (all Kanji)
+cjc.detect("ひらがなとカタカナと")           # => CJLanguage.JAPANESE  (all kana)
+cjc.detect("hello")                          # => CJLanguage.UNKNOWN
+```
+
+## Detailed results
+
+```python
+from cjclassifier.classifier import Results
+
+results = Results()
+cjc.detect("今天天气很好", results)
+
+results.result             # CJLanguage.CHINESE_SIMPLIFIED
+results.gap                # confidence gap: 0 = dead heat, 1 = no contest
+results.total_scores       # per-language log-probability totals
+results.to_short_string()  # e.g. "zh-hans:1.00,zh-hant:0.97,ja:0.85"
+```
+
+## How it works
+
+CJClassifier uses a **unigram + bigram statistical model** trained on the
+Chinese and Japanese Wikipedia corpora. For every character and character-pair
+in the CJ range, the model stores per-language log-probabilities. At
+classification time the library sums these log-probabilities across the input
+and picks the language with the highest score.
+
+A Java implementation and the model-building tools are also available in the
+same repository: [github.com/jlpka/cjclassifier](https://github.com/jlpka/cjclassifier)
+
+## License
+
+Apache License 2.0

cjclassifier-1.0.2.dist-info/RECORD

@@ -0,0 +1,9 @@
+cjclassifier/__init__.py,sha256=HnqkDnoewd2ZCMR0EumFQ-FQ7SS5HafaQrsNQ9QaZnA,985
+cjclassifier/cjlogprobs.gz,sha256=sPyx6C2sEdLhFxABK1Y_exnuPpLOagHn3oBryq38AS8,7597074
+cjclassifier/classifier.py,sha256=fH4sdVA7uIr_CI8SUVqpdy8HDX0fSEKvY4P67lrs8wA,22152
+cjclassifier/language.py,sha256=nZPBeWAEnAIWnd5bmcCvYXHnlYJ4uy7PMHjZG4GGTO0,2923
+cjclassifier-1.0.2.dist-info/licenses/LICENSE,sha256=eL5ZFQlGwgXafnpu26n_FyrAToFQtQuKNecgNEiPiSM,11359
+cjclassifier-1.0.2.dist-info/METADATA,sha256=ika04ZOlxLx7LU5Tz-6ytDBGU1nAfYcIBlxfX5YAEn0,2657
+cjclassifier-1.0.2.dist-info/WHEEL,sha256=aeYiig01lYGDzBgS8HxWXOg3uV61G9ijOsup-k9o1sk,91
+cjclassifier-1.0.2.dist-info/top_level.txt,sha256=YgZwIODTloGw6uDvvgc7gEQqIkru_-c_XjrRmlnCbos,13
+cjclassifier-1.0.2.dist-info/RECORD,,

cjclassifier-1.0.2.dist-info/WHEEL

@@ -0,0 +1,5 @@
+Wheel-Version: 1.0
+Generator: setuptools (82.0.1)
+Root-Is-Purelib: true
+Tag: py3-none-any
+

cjclassifier-1.0.2.dist-info/licenses/LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either expressed or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

cjclassifier-1.0.2.dist-info/top_level.txt

@@ -0,0 +1 @@
+cjclassifier