You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
35 lines
58 KiB
35 lines
58 KiB
4 years ago
|
"""
|
||
|
|
sphinx.search.tr
|
||
|
|
~~~~~~~~~~~~~~~~
|
||
|
|
|
||
|
|
Turkish search language: includes the JS Turkish stemmer.
|
||
|
|
|
||
|
|
:copyright: Copyright 2007-2013 by the Sphinx team, see AUTHORS.
|
||
|
|
:license: BSD, see LICENSE for details.
|
||
|
|
"""
|
||
|
|
|
||
|
|
from typing import Dict, Set
|
||
|
|
|
||
|
|
import snowballstemmer
|
||
|
|
|
||
|
|
from sphinx.search import SearchLanguage
|
||
|
|
|
||
|
|
js_stemmer = """
|
||
|
|
var JSX={};(function(q){function r(b,e){var a=function(){};a.prototype=e.prototype;var c=new a;for(var d in b){b[d].prototype=c}}function Q(c,b){for(var a in b.prototype)if(b.prototype.hasOwnProperty(a))c.prototype[a]=b.prototype[a]}function j(a,b,d){function c(a,b,c){delete a[b];a[b]=c;return c}Object.defineProperty(a,b,{get:function(){return c(a,b,d())},set:function(d){c(a,b,d)},enumerable:true,configurable:true})}function R(a,b,c){return a[b]=a[b]/c|0}var M=parseInt;var K=parseFloat;function P(a){return a!==a}var A=isFinite;var G=encodeURIComponent;var F=decodeURIComponent;var E=encodeURI;var D=decodeURI;var C=Object.prototype.toString;var H=Object.prototype.hasOwnProperty;function p(){}q.require=function(b){var a=y[b];return a!==undefined?a:null};q.profilerIsRunning=function(){return p.getResults!=null};q.getProfileResults=function(){return(p.getResults||function(){return{}})()};q.postProfileResults=function(a,b){if(p.postResults==null)throw new Error('profiler has not been turned on');return p.postResults(a,b)};q.resetProfileResults=function(){if(p.resetResults==null)throw new Error('profiler has not been turned on');return p.resetResults()};q.DEBUG=false;function I(){};r([I],Error);function d(a,b,c){this.G=a.length;this.A_=a;this.D_=b;this.J=c;this.I=null;this.E_=null};r([d],Object);function u(){};r([u],Object);function m(){var a;var b;var c;this.F={};a=this.E='';b=this._=0;c=this.A=a.length;this.D=0;this.B=b;this.C=c};r([m],u);function B(a,b){a.E=b.E;a._=b._;a.A=b.A;a.D=b.D;a.B=b.B;a.C=b.C};function v(b,d,c,e){var a;if(b._>=b.A){return false}a=b.E.charCodeAt(b._);if(a>e||a<c){return false}a-=c;if((d[a>>>3]&1<<(a&7))===0){return false}b._++;return true};function f(b,d,c,e){var a;if(b._<=b.D){return false}a=b.E.charCodeAt(b._-1);if(a>e||a<c){return false}a-=c;if((d[a>>>3]&1<<(a&7))===0){return false}b._--;return true};function t(a,d,c,e){var b;if(a._<=a.D){return false}b=a.E.charCodeAt(a._-1);if(b>e||b<c){a._--;return true}b-=c;if((d[b>>>3]&1<<(b&7))===0){a._--;return true}return false};function s(a,b,d){var c;if(a.A-a._<b){return false}if(a.E.slice(c=a._,c+b)!==d){return false}a._+=b;return true};function g(a,b,d){var c;if(a._-a.D<b){return false}if(a.E.slice((c=a._)-b,c)!==d){return false}a._-=b;return true};function b(d,m,p){var b;var g;var e;var n;var f;var k;var l;var i;var h;var c;var a;var j;var o;b=0;g=p;e=d._;n=d.D;f=0;k=0;l=false;while(true){i=b+(g-b>>1);h=0;c=f<k?f:k;a=m[i];for(j=a.G-1-c;j>=0;j--){if(e-c===n){h=-1;break}h=d.E.charCodeAt(e-1-c)-a.A_.charCodeAt(j);if(h!==0){break}c++}if(h<0){g=i;k=c}else{b=i;f=c}if(g-b<=1){if(b>0){break}if(g===b){break}if(l){break}l=true}}while(true){a=m[b];if(f>=a.G){d._=e-a.G|0;if(a.I==null){return a.J}o=a.I(d);d._=e-a.G|0;if(o){return a.J}}b=a.D_;if(b<0){return 0}}return-1};function n(a,b,d,e){var c;c=e.length-(d-b);a.E=a.E.slice(0,b)+e+a.E.slice(d);a.A+=c|0;if(a._>=d){a._+=c|0}else if(a._>b){a._=b}return c|0};function e(a,f){var b;var c;var d;var e;b=false;if((c=a.B)<0||c>(d=a.C)||d>(e=a.A)||e>a.E.length?false:true){n(a,a.B,a.C,f);b=true}return b};m.prototype.H=function(){return false};m.prototype.B_=function(b){var a;var c;var d;var e;a=this.F['.'+b];if(a==null){c=this.E=b;d=this._=0;e=this.A=c.length;this.D=0;this.B=d;this.C=e;this.H();a=this.E;this.F['.'+b]=a}return a};m.prototype.stemWord=m.prototype.B_;m.prototype.C_=function(e){var d;var b;var c;var a;var f;var g;var h;d=[];for(b=0;b<e.length;b++){c=e[b];a=this.F['.'+c];if(a==null){f=this.E=c;g=this._=0;h=this.A=f.length;this.D=0;this.B=g;this.C=h;this.H();a=this.E;this.F['.'+c]=a}d.push(a)}return d};m.prototype.stemWords=m.prototype.C_;function a(){m.call(this);this.B_continue_stemming_noun_suffixes=false;this.I_strlen=0};r([a],m);a.prototype.K=function(a){this.B_continue_stemming_noun_suffixes=a.B_continue_stemming_noun_suffixes;this.I_strlen=a.I_strlen;B(this,a)};a.prototype.copy_from=a.prototype.K;a.prototype.O=function(){var E;var q;var b;var e;var h;var i;var j;var k;var l;var m;var n;var o;var p;var c;var r;var s;var t;var u;var d;var v;var w;var x;var y;var z;var A;var B;var C;var D;var G;var H;var I;var
|
||
|
|
var Stemmer = JSX.require("src/turkish-stemmer.jsx").TurkishStemmer;
|
||
|
|
"""
|
||
|
|
|
||
|
|
|
||
|
|
class SearchTurkish(SearchLanguage):
|
||
|
|
lang = 'tr'
|
||
|
|
language_name = 'Turkish'
|
||
|
|
js_stemmer_rawcode = 'turkish-stemmer.js'
|
||
|
|
js_stemmer_code = js_stemmer
|
||
|
|
stopwords = set() # type: Set[str]
|
||
|
|
|
||
|
|
def init(self, options: Dict) -> None:
|
||
|
|
self.stemmer = snowballstemmer.stemmer('turkish')
|
||
|
|
|
||
|
|
def stem(self, word: str) -> str:
|
||
|
|
return self.stemmer.stemWord(word.lower())
|