o vOiG@sdZddlmZddlmZdZdZdZdZdZ d Z eeeee e Z e Z d Ze Ze e ZeeZee Zee Zd Zd Zd ZeeZeeeddeeZeeeee e dZdddddddZd ddZddZddZeeddZZZ eeddZ!Z"Z#eeddZ$Z%eeeZ&Z'Z(ee!eZ)Z*Z+ee$eZ,ee&Z-Z.Z/ee)Z0Z1Z2ee,Z3Z4dS)!z(Constants for processing Pinyin strings.) whitespace)escapeu aāàáǎu eēéěèu iīíǐìu oōóǒòu uūúǔùu vüǖǘǚǜbpmfdtnlgkhjqxzcsrwyu ·012345:-'z"#$%&'()*+,-/\:;<=>@[]^_`{|}~z.!?Naeiouvrrr r r uvüFc Csdjttttttd}|}|D]\}}t |dkr$d|||<qd|r*dnd||d|d |d |d |d |d dS)aBuilds a Pinyin syllable re pattern. Syllables can be preceded by a middle dot (tone mark). Syllables that end in a consonant are only valid if they aren't followed directly by a vowel with no apostrophe in between. The rough approach used to validate a Pinyin syllable is: 1. Get the longest valid syllable. 2. If it ends in a consonant make sure it's not followed directly by a vowel (hyphens and apostrophes don't count). 3. If the above didn't match, repeat for the next longest valid match. Lookahead assertions are used to ensure that hyphens and apostrophes are only considered valid if used correctly. This helps to weed out non-Pinyin strings. z(?![{a}{e}{i}{o}{u}{v}]|u:)rz[{}]u(?:·|‧)?(?:(?:(?:[zcs]h|[gkh])u%(a)sng%(consonant_end)s)|(?:[jqx]i%(o)sng%(consonant_end)s)|(?:[nljqx]i%(a)sng%(consonant_end)s)|(?:(?:[zcs]h?|[dtnlgkhrjqxy])u%(a)sn%(consonant_end)s)|(?:(?:[zcs]h|[gkh])u%(a)si)|(?:(?:[zc]h?|[rdtnlgkhsy])%(o)sng%(consonant_end)s)|(?:(?:[zcs]h?|[rbpmfdtnlgkhw])?%(e)sng%(consonant_end)s)|(?:(?:[zcs]h?|[rbpmfdtnlgkhwy])?%(a)sng%(consonant_end)s)|(?:[bpmdtnljqxy]%(i)sng%(consonant_end)s)|(?:[bpmdtnljqx]i%(a)sn%(consonant_end)s)|(?:[bpmdtnljqx]i%(a)so)|(?:[nl](?:v|u:|ü)%(e)s)|(?:[nl](?:%(v)s|u:))|(?:[jqxy]u%(e)s)|(?:[bpmnljqxy]%(i)sn%(consonant_end)s)|(?:[mdnljqx]i%(u)s)|(?:[bpmdtnljqx]i%(e)s)|(?:[dljqx]i%(a)s)|(?:(?:[zcs]h?|[rdtnlgkhxqjy])%(u)sn%(consonant_end)s)|(?:(?:[zcs]h?|[rdtgkh])u%(i)s)|(?:(?:[zcs]h?|[rdtnlgkh])u%(o)s)|(?:(?:[zcs]h|[rgkh])u%(a)s)|(?:(?:[zcs]h?|[rbpmfdngkhw])?%(e)sn%(consonant_end)s)|(?:(?:[zcs]h?|[rbpmfdtnlgkhwy])?%(a)sn%(consonant_end)s)|(?:(?:[zcs]h?|[rpmfdtnlgkhy])?%(o)su)|(?:(?:[zcs]h?|[rbpmdtnlgkhy])?%(a)so)|(?:(?:[zs]h|[bpmfdtnlgkhwz])?%(e)si)|(?:(?:[zcs]h?|[bpmdtnlgkhw])?%(a)si)|(?:(?:[zcs]h?|[rjqxybpmdtnl])%(i)s)|(?:(?:[zcs]h?|[rwbpmfdtnlgkhjqxwy])%(u)s)|(?:%(e)s(?:r%(consonant_end)s)?)|(?:(?:[zcs]h?|[rmdtnlgkhy])%(e)s)|(?:[bpmfwyl]?%(o)s)|(?:(?:[zcs]h|[bpmfdtnlgkhzcswy])?%(a)s)|(?:r%(consonant_end)s))z[0-5]?rrr r r r ) consonant_endrrr r r r ) format_a_e_i_o_u_vcopyitemslen)vowels tone_numbersr_vowelsr sr?/home/ubuntu/.local/lib/python3.10/site-packages/zhon/pinyin.py _build_syl8s(   %'r cCsdj||d|d|ddS)aBuilds a Pinyin word re pattern from a Pinyin syllable re pattern. A word is defined as a series of consecutive valid Pinyin syllables with optional hyphens and apostrophes interspersed. Hyphens must be followed immediately by another valid Pinyin syllable. Apostrophes must be followed by another valid Pinyin syllable that starts with an 'a', 'e', or 'o'. z3(?:{syl}(?:-(?={syl})|'(?=[{a}{e}{o}])(?={syl}))?)+rrr )sylrrr )r)r!rrrr _build_words r"cCsdj|ttttdS)a!Builds a Pinyin sentence re pattern from a Pinyin word re pattern. A sentence is defined as a series of valid Pinyin words, punctuation (non-stops), and spaces followed by a single stop and zero or more container-closing punctuation marks (e.g. apostrophe and brackets). z?(?:{word}|[{non_stops}]|(?sH   L