bnc-lookup 1.0.6__py3-none-any.whl → 1.1.0__py3-none-any.whl

bnc_lookup/__init__.py

@@ -1,5 +1,18 @@
 from bnc_lookup.find_bnc import FindBnc
+from bnc_lookup.find_freq import FindFreq
 
 
-def is_bnc_term(input_text: str) -> bool:
+def exists(input_text: str) -> bool:
+    """Check if a word exists in the BNC corpus."""
     return FindBnc().exists(input_text)
+
+
+def bucket(input_text: str) -> int | None:
+    """
+    Get frequency bucket for a word.
+
+    Returns:
+        1-100: Bucket number (1=most frequent, 100=least frequent)
+        None: Word not found in BNC
+    """
+    return FindFreq().bucket(input_text)

bnc_lookup/find_freq.py

@@ -0,0 +1,59 @@
+# !/usr/bin/env python
+# -*- coding: UTF-8 -*-
+""" Lookup frequency bucket for BNC words """
+
+import hashlib
+import importlib
+
+_cache = {}
+
+
+def _get_bucket_dict(prefix: str) -> dict:
+    if prefix not in _cache:
+        module = importlib.import_module(f'bnc_lookup.freq.f_{prefix}')
+        _cache[prefix] = getattr(module, f'buckets_{prefix}')
+    return _cache[prefix]
+
+
+def _calculate_md5(input_text: str) -> str:
+    return hashlib.md5(input_text.lower().strip().encode()).hexdigest()
+
+
+def _lookup_bucket(input_text: str) -> int | None:
+    if not input_text:
+        return None
+    h = _calculate_md5(input_text)
+    prefix, suffix = h[:2], h[2:]
+    try:
+        return _get_bucket_dict(prefix).get(suffix)
+    except ModuleNotFoundError:
+        return None
+
+
+class FindFreq:
+    """ Lookup frequency bucket for BNC words """
+
+    def __init__(self):
+        pass
+
+    def bucket(self, input_text: str) -> int | None:
+        """
+        Get frequency bucket for a word.
+
+        Returns:
+            1-100: Bucket number (1=most frequent, 100=least frequent)
+            None: Word not found in BNC
+        """
+        input_text = input_text.lower().strip()
+
+        result = _lookup_bucket(input_text)
+        if result is not None:
+            return result
+
+        # Try singular form if plural
+        if input_text.endswith('s') and len(input_text) > 3:
+            result = _lookup_bucket(input_text[:-1])
+            if result is not None:
+                return result
+
+        return None

bnc_lookup/freq/__init__.py

@@ -0,0 +1,519 @@
+#!/usr/bin/env python
+# -*- coding: UTF-8 -*-
+# AUTO-GENERATED - DO NOT MODIFY
+
+from .f_00 import buckets_00
+from .f_01 import buckets_01
+from .f_02 import buckets_02
+from .f_03 import buckets_03
+from .f_04 import buckets_04
+from .f_05 import buckets_05
+from .f_06 import buckets_06
+from .f_07 import buckets_07
+from .f_08 import buckets_08
+from .f_09 import buckets_09
+from .f_0a import buckets_0a
+from .f_0b import buckets_0b
+from .f_0c import buckets_0c
+from .f_0d import buckets_0d
+from .f_0e import buckets_0e
+from .f_0f import buckets_0f
+from .f_10 import buckets_10
+from .f_11 import buckets_11
+from .f_12 import buckets_12
+from .f_13 import buckets_13
+from .f_14 import buckets_14
+from .f_15 import buckets_15
+from .f_16 import buckets_16
+from .f_17 import buckets_17
+from .f_18 import buckets_18
+from .f_19 import buckets_19
+from .f_1a import buckets_1a
+from .f_1b import buckets_1b
+from .f_1c import buckets_1c
+from .f_1d import buckets_1d
+from .f_1e import buckets_1e
+from .f_1f import buckets_1f
+from .f_20 import buckets_20
+from .f_21 import buckets_21
+from .f_22 import buckets_22
+from .f_23 import buckets_23
+from .f_24 import buckets_24
+from .f_25 import buckets_25
+from .f_26 import buckets_26
+from .f_27 import buckets_27
+from .f_28 import buckets_28
+from .f_29 import buckets_29
+from .f_2a import buckets_2a
+from .f_2b import buckets_2b
+from .f_2c import buckets_2c
+from .f_2d import buckets_2d
+from .f_2e import buckets_2e
+from .f_2f import buckets_2f
+from .f_30 import buckets_30
+from .f_31 import buckets_31
+from .f_32 import buckets_32
+from .f_33 import buckets_33
+from .f_34 import buckets_34
+from .f_35 import buckets_35
+from .f_36 import buckets_36
+from .f_37 import buckets_37
+from .f_38 import buckets_38
+from .f_39 import buckets_39
+from .f_3a import buckets_3a
+from .f_3b import buckets_3b
+from .f_3c import buckets_3c
+from .f_3d import buckets_3d
+from .f_3e import buckets_3e
+from .f_3f import buckets_3f
+from .f_40 import buckets_40
+from .f_41 import buckets_41
+from .f_42 import buckets_42
+from .f_43 import buckets_43
+from .f_44 import buckets_44
+from .f_45 import buckets_45
+from .f_46 import buckets_46
+from .f_47 import buckets_47
+from .f_48 import buckets_48
+from .f_49 import buckets_49
+from .f_4a import buckets_4a
+from .f_4b import buckets_4b
+from .f_4c import buckets_4c
+from .f_4d import buckets_4d
+from .f_4e import buckets_4e
+from .f_4f import buckets_4f
+from .f_50 import buckets_50
+from .f_51 import buckets_51
+from .f_52 import buckets_52
+from .f_53 import buckets_53
+from .f_54 import buckets_54
+from .f_55 import buckets_55
+from .f_56 import buckets_56
+from .f_57 import buckets_57
+from .f_58 import buckets_58
+from .f_59 import buckets_59
+from .f_5a import buckets_5a
+from .f_5b import buckets_5b
+from .f_5c import buckets_5c
+from .f_5d import buckets_5d
+from .f_5e import buckets_5e
+from .f_5f import buckets_5f
+from .f_60 import buckets_60
+from .f_61 import buckets_61
+from .f_62 import buckets_62
+from .f_63 import buckets_63
+from .f_64 import buckets_64
+from .f_65 import buckets_65
+from .f_66 import buckets_66
+from .f_67 import buckets_67
+from .f_68 import buckets_68
+from .f_69 import buckets_69
+from .f_6a import buckets_6a
+from .f_6b import buckets_6b
+from .f_6c import buckets_6c
+from .f_6d import buckets_6d
+from .f_6e import buckets_6e
+from .f_6f import buckets_6f
+from .f_70 import buckets_70
+from .f_71 import buckets_71
+from .f_72 import buckets_72
+from .f_73 import buckets_73
+from .f_74 import buckets_74
+from .f_75 import buckets_75
+from .f_76 import buckets_76
+from .f_77 import buckets_77
+from .f_78 import buckets_78
+from .f_79 import buckets_79
+from .f_7a import buckets_7a
+from .f_7b import buckets_7b
+from .f_7c import buckets_7c
+from .f_7d import buckets_7d
+from .f_7e import buckets_7e
+from .f_7f import buckets_7f
+from .f_80 import buckets_80
+from .f_81 import buckets_81
+from .f_82 import buckets_82
+from .f_83 import buckets_83
+from .f_84 import buckets_84
+from .f_85 import buckets_85
+from .f_86 import buckets_86
+from .f_87 import buckets_87
+from .f_88 import buckets_88
+from .f_89 import buckets_89
+from .f_8a import buckets_8a
+from .f_8b import buckets_8b
+from .f_8c import buckets_8c
+from .f_8d import buckets_8d
+from .f_8e import buckets_8e
+from .f_8f import buckets_8f
+from .f_90 import buckets_90
+from .f_91 import buckets_91
+from .f_92 import buckets_92
+from .f_93 import buckets_93
+from .f_94 import buckets_94
+from .f_95 import buckets_95
+from .f_96 import buckets_96
+from .f_97 import buckets_97
+from .f_98 import buckets_98
+from .f_99 import buckets_99
+from .f_9a import buckets_9a
+from .f_9b import buckets_9b
+from .f_9c import buckets_9c
+from .f_9d import buckets_9d
+from .f_9e import buckets_9e
+from .f_9f import buckets_9f
+from .f_a0 import buckets_a0
+from .f_a1 import buckets_a1
+from .f_a2 import buckets_a2
+from .f_a3 import buckets_a3
+from .f_a4 import buckets_a4
+from .f_a5 import buckets_a5
+from .f_a6 import buckets_a6
+from .f_a7 import buckets_a7
+from .f_a8 import buckets_a8
+from .f_a9 import buckets_a9
+from .f_aa import buckets_aa
+from .f_ab import buckets_ab
+from .f_ac import buckets_ac
+from .f_ad import buckets_ad
+from .f_ae import buckets_ae
+from .f_af import buckets_af
+from .f_b0 import buckets_b0
+from .f_b1 import buckets_b1
+from .f_b2 import buckets_b2
+from .f_b3 import buckets_b3
+from .f_b4 import buckets_b4
+from .f_b5 import buckets_b5
+from .f_b6 import buckets_b6
+from .f_b7 import buckets_b7
+from .f_b8 import buckets_b8
+from .f_b9 import buckets_b9
+from .f_ba import buckets_ba
+from .f_bb import buckets_bb
+from .f_bc import buckets_bc
+from .f_bd import buckets_bd
+from .f_be import buckets_be
+from .f_bf import buckets_bf
+from .f_c0 import buckets_c0
+from .f_c1 import buckets_c1
+from .f_c2 import buckets_c2
+from .f_c3 import buckets_c3
+from .f_c4 import buckets_c4
+from .f_c5 import buckets_c5
+from .f_c6 import buckets_c6
+from .f_c7 import buckets_c7
+from .f_c8 import buckets_c8
+from .f_c9 import buckets_c9
+from .f_ca import buckets_ca
+from .f_cb import buckets_cb
+from .f_cc import buckets_cc
+from .f_cd import buckets_cd
+from .f_ce import buckets_ce
+from .f_cf import buckets_cf
+from .f_d0 import buckets_d0
+from .f_d1 import buckets_d1
+from .f_d2 import buckets_d2
+from .f_d3 import buckets_d3
+from .f_d4 import buckets_d4
+from .f_d5 import buckets_d5
+from .f_d6 import buckets_d6
+from .f_d7 import buckets_d7
+from .f_d8 import buckets_d8
+from .f_d9 import buckets_d9
+from .f_da import buckets_da
+from .f_db import buckets_db
+from .f_dc import buckets_dc
+from .f_dd import buckets_dd
+from .f_de import buckets_de
+from .f_df import buckets_df
+from .f_e0 import buckets_e0
+from .f_e1 import buckets_e1
+from .f_e2 import buckets_e2
+from .f_e3 import buckets_e3
+from .f_e4 import buckets_e4
+from .f_e5 import buckets_e5
+from .f_e6 import buckets_e6
+from .f_e7 import buckets_e7
+from .f_e8 import buckets_e8
+from .f_e9 import buckets_e9
+from .f_ea import buckets_ea
+from .f_eb import buckets_eb
+from .f_ec import buckets_ec
+from .f_ed import buckets_ed
+from .f_ee import buckets_ee
+from .f_ef import buckets_ef
+from .f_f0 import buckets_f0
+from .f_f1 import buckets_f1
+from .f_f2 import buckets_f2
+from .f_f3 import buckets_f3
+from .f_f4 import buckets_f4
+from .f_f5 import buckets_f5
+from .f_f6 import buckets_f6
+from .f_f7 import buckets_f7
+from .f_f8 import buckets_f8
+from .f_f9 import buckets_f9
+from .f_fa import buckets_fa
+from .f_fb import buckets_fb
+from .f_fc import buckets_fc
+from .f_fd import buckets_fd
+from .f_fe import buckets_fe
+from .f_ff import buckets_ff
+
+__all__ = [
+    'buckets_00',
+    'buckets_01',
+    'buckets_02',
+    'buckets_03',
+    'buckets_04',
+    'buckets_05',
+    'buckets_06',
+    'buckets_07',
+    'buckets_08',
+    'buckets_09',
+    'buckets_0a',
+    'buckets_0b',
+    'buckets_0c',
+    'buckets_0d',
+    'buckets_0e',
+    'buckets_0f',
+    'buckets_10',
+    'buckets_11',
+    'buckets_12',
+    'buckets_13',
+    'buckets_14',
+    'buckets_15',
+    'buckets_16',
+    'buckets_17',
+    'buckets_18',
+    'buckets_19',
+    'buckets_1a',
+    'buckets_1b',
+    'buckets_1c',
+    'buckets_1d',
+    'buckets_1e',
+    'buckets_1f',
+    'buckets_20',
+    'buckets_21',
+    'buckets_22',
+    'buckets_23',
+    'buckets_24',
+    'buckets_25',
+    'buckets_26',
+    'buckets_27',
+    'buckets_28',
+    'buckets_29',
+    'buckets_2a',
+    'buckets_2b',
+    'buckets_2c',
+    'buckets_2d',
+    'buckets_2e',
+    'buckets_2f',
+    'buckets_30',
+    'buckets_31',
+    'buckets_32',
+    'buckets_33',
+    'buckets_34',
+    'buckets_35',
+    'buckets_36',
+    'buckets_37',
+    'buckets_38',
+    'buckets_39',
+    'buckets_3a',
+    'buckets_3b',
+    'buckets_3c',
+    'buckets_3d',
+    'buckets_3e',
+    'buckets_3f',
+    'buckets_40',
+    'buckets_41',
+    'buckets_42',
+    'buckets_43',
+    'buckets_44',
+    'buckets_45',
+    'buckets_46',
+    'buckets_47',
+    'buckets_48',
+    'buckets_49',
+    'buckets_4a',
+    'buckets_4b',
+    'buckets_4c',
+    'buckets_4d',
+    'buckets_4e',
+    'buckets_4f',
+    'buckets_50',
+    'buckets_51',
+    'buckets_52',
+    'buckets_53',
+    'buckets_54',
+    'buckets_55',
+    'buckets_56',
+    'buckets_57',
+    'buckets_58',
+    'buckets_59',
+    'buckets_5a',
+    'buckets_5b',
+    'buckets_5c',
+    'buckets_5d',
+    'buckets_5e',
+    'buckets_5f',
+    'buckets_60',
+    'buckets_61',
+    'buckets_62',
+    'buckets_63',
+    'buckets_64',
+    'buckets_65',
+    'buckets_66',
+    'buckets_67',
+    'buckets_68',
+    'buckets_69',
+    'buckets_6a',
+    'buckets_6b',
+    'buckets_6c',
+    'buckets_6d',
+    'buckets_6e',
+    'buckets_6f',
+    'buckets_70',
+    'buckets_71',
+    'buckets_72',
+    'buckets_73',
+    'buckets_74',
+    'buckets_75',
+    'buckets_76',
+    'buckets_77',
+    'buckets_78',
+    'buckets_79',
+    'buckets_7a',
+    'buckets_7b',
+    'buckets_7c',
+    'buckets_7d',
+    'buckets_7e',
+    'buckets_7f',
+    'buckets_80',
+    'buckets_81',
+    'buckets_82',
+    'buckets_83',
+    'buckets_84',
+    'buckets_85',
+    'buckets_86',
+    'buckets_87',
+    'buckets_88',
+    'buckets_89',
+    'buckets_8a',
+    'buckets_8b',
+    'buckets_8c',
+    'buckets_8d',
+    'buckets_8e',
+    'buckets_8f',
+    'buckets_90',
+    'buckets_91',
+    'buckets_92',
+    'buckets_93',
+    'buckets_94',
+    'buckets_95',
+    'buckets_96',
+    'buckets_97',
+    'buckets_98',
+    'buckets_99',
+    'buckets_9a',
+    'buckets_9b',
+    'buckets_9c',
+    'buckets_9d',
+    'buckets_9e',
+    'buckets_9f',
+    'buckets_a0',
+    'buckets_a1',
+    'buckets_a2',
+    'buckets_a3',
+    'buckets_a4',
+    'buckets_a5',
+    'buckets_a6',
+    'buckets_a7',
+    'buckets_a8',
+    'buckets_a9',
+    'buckets_aa',
+    'buckets_ab',
+    'buckets_ac',
+    'buckets_ad',
+    'buckets_ae',
+    'buckets_af',
+    'buckets_b0',
+    'buckets_b1',
+    'buckets_b2',
+    'buckets_b3',
+    'buckets_b4',
+    'buckets_b5',
+    'buckets_b6',
+    'buckets_b7',
+    'buckets_b8',
+    'buckets_b9',
+    'buckets_ba',
+    'buckets_bb',
+    'buckets_bc',
+    'buckets_bd',
+    'buckets_be',
+    'buckets_bf',
+    'buckets_c0',
+    'buckets_c1',
+    'buckets_c2',
+    'buckets_c3',
+    'buckets_c4',
+    'buckets_c5',
+    'buckets_c6',
+    'buckets_c7',
+    'buckets_c8',
+    'buckets_c9',
+    'buckets_ca',
+    'buckets_cb',
+    'buckets_cc',
+    'buckets_cd',
+    'buckets_ce',
+    'buckets_cf',
+    'buckets_d0',
+    'buckets_d1',
+    'buckets_d2',
+    'buckets_d3',
+    'buckets_d4',
+    'buckets_d5',
+    'buckets_d6',
+    'buckets_d7',
+    'buckets_d8',
+    'buckets_d9',
+    'buckets_da',
+    'buckets_db',
+    'buckets_dc',
+    'buckets_dd',
+    'buckets_de',
+    'buckets_df',
+    'buckets_e0',
+    'buckets_e1',
+    'buckets_e2',
+    'buckets_e3',
+    'buckets_e4',
+    'buckets_e5',
+    'buckets_e6',
+    'buckets_e7',
+    'buckets_e8',
+    'buckets_e9',
+    'buckets_ea',
+    'buckets_eb',
+    'buckets_ec',
+    'buckets_ed',
+    'buckets_ee',
+    'buckets_ef',
+    'buckets_f0',
+    'buckets_f1',
+    'buckets_f2',
+    'buckets_f3',
+    'buckets_f4',
+    'buckets_f5',
+    'buckets_f6',
+    'buckets_f7',
+    'buckets_f8',
+    'buckets_f9',
+    'buckets_fa',
+    'buckets_fb',
+    'buckets_fc',
+    'buckets_fd',
+    'buckets_fe',
+    'buckets_ff',
+]