@scrabble-solver/word-definitions 2.9.1 → 2.9.3

package/src/parse/parse.test.ts

@@ -10,6 +10,10 @@ export const readTestFile = (filepath: string): string => {
 };
 
 const tests = [
+  { locale: Locale.DE_DE, word: 'hm' },
+  { locale: Locale.DE_DE, word: 'ho' },
+  { locale: Locale.DE_DE, word: 'kolla' },
+  { locale: Locale.DE_DE, word: 'vom' },
   { locale: Locale.EN_US, word: 'awe' },
   { locale: Locale.EN_US, word: 'pawn' },
   { locale: Locale.EN_US, word: 'pawnee' },

package/src/parse/parse.ts

@@ -19,11 +19,11 @@ const parsePerLocale: Record<Locale, (html: string) => ParseResult> = {
 };
 
 const parse = (locale: Locale, html: string): ParseResult => {
-  const { definitions, isAllowed } = parsePerLocale[locale](html);
+  const { definitions, exists } = parsePerLocale[locale](html);
 
   return {
     definitions: unique(definitions.map(normalizeDefinition).filter(Boolean)),
-    isAllowed,
+    exists,
   };
 };
 

package/src/parse/parseEnglish.ts

@@ -1,17 +1,21 @@
-import cheerio from 'cheerio';
+import { load } from 'cheerio';
 
 import { ParseResult } from '../types';
 
+const DOES_NOT_EXIST_MESSAGE =
+  // eslint-disable-next-line max-len
+  "The word you've entered isn't in the dictionary. Click on a spelling suggestion below or try again using the search bar above.";
+
 const parseEnglish = (html: string): ParseResult => {
-  const $ = cheerio.load(html);
+  const $ = load(html);
   $('strong.mw_t_bc').replaceWith(', ');
   $('.text-lowercase').remove();
   $('[id^=dictionary-entry]').find('.dtText > *:not(a)').remove();
   const $definitions = $('[id^=dictionary-entry]').find('.dtText, .cxl-ref');
 
   return {
-    definitions: Array.from($definitions).map((definition) => $(definition).text()),
-    isAllowed: $definitions.length > 0,
+    definitions: Array.from($definitions).map((definition) => $(definition).text().replace(/\n/g, '')),
+    exists: $('.spelling-suggestion-text').text().trim() !== DOES_NOT_EXIST_MESSAGE,
   };
 };
 

package/src/parse/parseFrench.ts

@@ -1,14 +1,14 @@
-import cheerio from 'cheerio';
+import { load } from 'cheerio';
 
 import { ParseResult } from '../types';
 
 const parseFrench = (html: string): ParseResult => {
-  const $ = cheerio.load(html);
+  const $ = load(html);
   const $definitions = $('.tlf_cdefinition');
 
   return {
     definitions: Array.from($definitions).map((definition) => $(definition).text()),
-    isAllowed: $('#vitemselected span').length > 0,
+    exists: $('#vitemselected span').length > 0,
   };
 };
 

package/src/parse/parseGerman.ts

@@ -1,21 +1,80 @@
-import cheerio from 'cheerio';
+import { Cheerio, CheerioAPI, Element, load } from 'cheerio';
 
 import { ParseResult } from '../types';
 
 const parseGerman = (html: string): ParseResult => {
-  const $ = cheerio.load(html);
-  const $meaningOverview = $('.bedeutungsuebersicht');
-  let $definitions;
-  if ($meaningOverview.length > 0) {
-    $definitions = $('.bedeutungsuebersicht > ol > li > a');
-  } else {
-    $definitions = $('.dwdswb-lesart .dwdswb-definition');
+  const $ = load(html);
+
+  const definitions = [parseBedeutungsubersicht, parseBedeutungen, parseBedeutung].reduce<string[]>(
+    (results, parse) => (results.length === 0 ? parse($) : results),
+    [],
+  );
+  const exists = Array.from($('.label-danger')).every((label) => $(label).text() !== 'Hinweis');
+
+  return { definitions, exists };
+};
+
+const parseBedeutungsubersicht = ($: CheerioAPI): string[] => {
+  Array.from($('.bedeutungsuebersicht ol > li > a')).forEach((item) => {
+    $(item).text($(item).text().replace(/\n/g, ''));
+  });
+
+  Array.from($('.bedeutungsuebersicht ol > li > ol > li')).forEach((item) => {
+    const text = `\n${$(item).text().replace(/\n/g, '')}`;
+    const $text = $(`<div>${text}</div>`);
+    $(item).replaceWith($text);
+  });
+
+  Array.from($('.bedeutungsuebersicht ol > li > ol')).forEach((list) => {
+    const $list = $(list);
+    const html = $list.html() || '';
+    const $html = $(`<div>${html}</div>`);
+    const $prev = $list.prev('a');
+
+    if ($prev) {
+      $prev.append($html);
+      $(list).remove();
+    } else {
+      $(list).replaceWith($html);
+    }
+  });
+
+  return parseDefinitions($, $('.bedeutungsuebersicht ol > li'));
+};
+
+const parseBedeutung = ($: CheerioAPI): string[] => {
+  return parseDefinitions($, $('.dwdswb-lesart .dwdswb-definition-spezifizierung'));
+};
+
+const parseBedeutungen = ($: CheerioAPI): string[] => {
+  const definitions = parseDefinitions($, $('.dwdswb-lesart .dwdswb-definition'));
+
+  if (definitions.length > 0) {
+    return definitions;
   }
 
-  return {
-    definitions: Array.from($definitions).map((definition) => $(definition).text().replace('/\n/', '')),
-    isAllowed: $definitions.length > 0,
-  };
+  const $references = $('.dwdswb-lesart .dwdswb-verweis');
+  const references = Array.from($references).reduce<string[]>((result, reference) => {
+    const html = reference.attribs['data-content'] || '<span />';
+    const values = $(html)
+      .text()
+      .split(';')
+      .map((value) => value.trim());
+
+    return result.concat(values);
+  }, []);
+
+  return references;
+};
+
+const parseDefinitions = ($: CheerioAPI, $definitions: Cheerio<Element>) => {
+  return Array.from($definitions).map((definition) =>
+    $(definition)
+      .text()
+      .replace(/[ ]+/g, ' ')
+      .replace(/[ ]\n/g, '\n')
+      .replace(/^[0-9]+\.\s/g, ''),
+  );
 };
 
 export default parseGerman;

package/src/parse/parsePolish.ts

@@ -1,16 +1,16 @@
-import cheerio from 'cheerio';
+import { load } from 'cheerio';
 
 import { ParseResult } from '../types';
 
 const parsePolish = (html: string): ParseResult => {
-  const $ = cheerio.load(html);
+  const $ = load(html);
   const $header = $($('h1')[0]);
   const $isAllowed = $header.next();
   const $definitions = $header.next().next().next().next();
 
   return {
     definitions: $definitions.text().trim().split(/\d+\./),
-    isAllowed: $isAllowed.text().trim().indexOf('dopuszczalne w grach') >= 0,
+    exists: $isAllowed.text().trim().indexOf('dopuszczalne w grach') >= 0,
   };
 };
 

package/src/parse/parseSpanish.ts

@@ -1,9 +1,9 @@
-import cheerio from 'cheerio';
+import { load } from 'cheerio';
 
 import { ParseResult } from '../types';
 
 const parseSpanish = (html: string): ParseResult => {
-  const $ = cheerio.load(html);
+  const $ = load(html);
   $('.verdBold14 + .gris11 + .gris13').remove();
   $('br + .gris13').remove();
   $('.grisItalic13 + .gris13').remove();
@@ -28,7 +28,7 @@ const parseSpanish = (html: string): ParseResult => {
       .filter(Boolean)
       .map((definition) => definition.replace(/\s+\.$/g, ''))
       .map((definition) => (definition.endsWith('.') ? definition : `${definition}.`)),
-    isAllowed: definitions.length > 0,
+    exists: $('.wrapper > p > strong').text() !== 'No se ha encontrado la palabra exacta',
   };
 };
 

package/src/types.ts

@@ -1,4 +1,4 @@
 export interface ParseResult {
   definitions: string[];
-  isAllowed: boolean;
+  exists: boolean;
 }