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
21 KiB
35 lines
21 KiB
4 years ago
|
"""
|
||
|
|
sphinx.search.ro
|
||
|
|
~~~~~~~~~~~~~~~~
|
||
|
|
|
||
|
|
Romanian search language: includes the JS Romanian 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(j){function l(b,e){var a=function(){};a.prototype=e.prototype;var c=new a;for(var d in b){b[d].prototype=c}}function L(c,b){for(var a in b.prototype)if(b.prototype.hasOwnProperty(a))c.prototype[a]=b.prototype[a]}function h(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 M(a,b,c){return a[b]=a[b]/c|0}var E=parseInt;var C=parseFloat;function N(a){return a!==a}var A=isFinite;var z=encodeURIComponent;var y=decodeURIComponent;var x=encodeURI;var w=decodeURI;var u=Object.prototype.toString;var D=Object.prototype.hasOwnProperty;function k(){}j.require=function(b){var a=r[b];return a!==undefined?a:null};j.profilerIsRunning=function(){return k.getResults!=null};j.getProfileResults=function(){return(k.getResults||function(){return{}})()};j.postProfileResults=function(a,b){if(k.postResults==null)throw new Error('profiler has not been turned on');return k.postResults(a,b)};j.resetProfileResults=function(){if(k.resetResults==null)throw new Error('profiler has not been turned on');return k.resetResults()};j.DEBUG=false;function t(){};l([t],Error);function a(a,b,c){this.F=a.length;this.K=a;this.L=b;this.I=c;this.H=null;this.P=null};l([a],Object);function n(){};l([n],Object);function g(){var a;var b;var c;this.G={};a=this.E='';b=this._=0;c=this.A=a.length;this.D=0;this.B=b;this.C=c};l([g],n);function v(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 d(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 e(a,d,c,e){var b;if(a._>=a.A){return false}b=a.E.charCodeAt(a._);if(b>e||b<c){a._++;return true}b-=c;if((d[b>>>3]&1<<(b&7))===0){a._++;return true}return false};function p(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 m(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 i(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 q(f,m,p){var b;var d;var e;var n;var g;var k;var l;var i;var h;var c;var a;var j;var o;b=0;d=p;e=f._;n=f.A;g=0;k=0;l=false;while(true){i=b+(d-b>>>1);h=0;c=g<k?g:k;a=m[i];for(j=c;j<a.F;j++){if(e+c===n){h=-1;break}h=f.E.charCodeAt(e+c)-a.K.charCodeAt(j);if(h!==0){break}c++}if(h<0){d=i;k=c}else{b=i;g=c}if(d-b<=1){if(b>0){break}if(d===b){break}if(l){break}l=true}}while(true){a=m[b];if(g>=a.F){f._=e+a.F|0;if(a.H==null){return a.I}o=a.H(a.P);f._=e+a.F|0;if(o){return a.I}}b=a.L;if(b<0){return 0}}return-1};function f(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.F-1-c;j>=0;j--){if(e-c===n){h=-1;break}h=d.E.charCodeAt(e-1-c)-a.K.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.F){d._=e-a.F|0;if(a.H==null){return a.I}o=a.H(d);d._=e-a.F|0;if(o){return a.I}}b=a.L;if(b<0){return 0}}return-1};function s(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 c(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){s(a,a.B,a.C,f);b=true}return b};g.prototype.J=function(){return false};g.prototype.b=function(b){var a;var c;var d;var e;a=this.G['.'+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.J();a=this.E;this.G['.'+b]=a}return a};g.prototype.stemWord=g.prototype.b;g.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.G['.'+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.J();a=this.E;this.G['.'+c]=a}d.push(a)}return d};g.prototype.st
|
||
|
|
var Stemmer = JSX.require("src/romanian-stemmer.jsx").RomanianStemmer;
|
||
|
|
"""
|
||
|
|
|
||
|
|
|
||
|
|
class SearchRomanian(SearchLanguage):
|
||
|
|
lang = 'ro'
|
||
|
|
language_name = 'Romanian'
|
||
|
|
js_stemmer_rawcode = 'romanian-stemmer.js'
|
||
|
|
js_stemmer_code = js_stemmer
|
||
|
|
stopwords = set() # type: Set[str]
|
||
|
|
|
||
|
|
def init(self, options: Dict) -> None:
|
||
|
|
self.stemmer = snowballstemmer.stemmer('romanian')
|
||
|
|
|
||
|
|
def stem(self, word: str) -> str:
|
||
|
|
return self.stemmer.stemWord(word.lower())
|