flag scientific script added

src/data/scripts/db.mjs

@@ -11,4 +11,10 @@ export const db = Knex({
     return db
     .select("word")
     .from("dictionary")
+}
+
+export const getBadWords = async (db) => {
+  return db
+    .select("word")
+    .from("bad_words")
 }

src/data/scripts/flagDerivatives.mjs

@@ -1,5 +1,4 @@
 import { db, getWords } from './db.mjs'
-import fs from 'fs'
 
 //const words = getWords(db)
 const allDefinitionsAreFormOf = (meanings) => {
@@ -43,7 +42,7 @@ const sampleMeanings =[
       }
     ]
 const words = await getWords(db)
-const deleted = []
+const flagged = []
 for (const word of words) {
     const res = 
     await db('dictionary')
@@ -51,11 +50,14 @@ for (const word of words) {
     .where('word', word.word)
     const meanings = JSON.parse(res[0].meanings)
     if(allDefinitionsAreFormOf(meanings)){
-        deleted.push(word)
+      await db('dictionary').
+      where('word', word.word)
+      .update('derivative',1)
+      flagged.push(word)
     }
 }
-console.log(`${deleted.length} entries deleted`)
+console.log(`${flagged.length} entries flagged`)
-console.dir(deleted)
+//console.dir(flagged)
 db.destroy()
 
 

src/data/scripts/flagScientific.mjs

@@ -0,0 +1,41 @@
+import { db, getWords } from './db.mjs'
+
+const words = await getWords(db)
+const allDefinitionsAreScientific = (meanings) => {
+    let scientific = 0
+    let totalDefs = 0
+    for (const meaning of meanings) {
+        for (const definition of meaning.definitions) {
+            totalDefs++
+            if (definition.topics) {
+                for (const topic of definition.topics) {
+                    if (topic === 'sciences') {
+                        scientific++
+                        break
+                    }
+                }
+            }
+        }
+    }
+    return scientific === totalDefs
+}
+
+let updated = []
+
+for (const word of words) {
+    const res = await db('dictionary')
+        .select('meanings')
+        .where('word', word.word)
+    const meanings = JSON.parse(res[0].meanings)
+    if (allDefinitionsAreScientific(meanings)) {
+        await db('dictionary')
+            .where('word', word.word)
+            .update('scientific', 1)
+        updated.push(word.word)
+    }
+}
+
+console.log(`${updated.length} words with only scientific definitions found.`)
+
+
+db.destroy()

src/data/scripts/pipeline.sh

@@ -1,7 +1,7 @@
 #!/bin/bash
 cd ../processing
 
-jq -r '. | select((.pos=="noun") or (.pos=="verb") or (.pos=="adj") or (.pos=="adv")) | select((.word | test("[^a-z]"))|not) | {word: .word, type: .pos, pronunciation: .sounds[0].ipa, definitions: (try .senses|map({definition: .glosses|join(" "), form_of: (.form_of!=null) }))}' \
+jq -r '. | select((.pos=="noun") or (.pos=="verb") or (.pos=="adj") or (.pos=="adv")) | select((.word | test("[^a-z]"))|not) | {word: .word, type: .pos, pronunciation: .sounds[0].ipa, definitions: (try .senses|map({definition: .glosses|join(" "), form_of: (.form_of!=null), topics:(try .topics) }))}' \
 wiktionary.json > wiktionary-p1.json
 
 jq --slurp '. | group_by(.word)[] | {word:.[0].word, pronunciation:.[0].pronunciation, meanings:[[([.[]|{type:.type, definitions:[try .definitions][]}] | to_entries[] | [{type: .value.type, index: .key, definitions: .value.definitions}][] )] | group_by(.type) | map({type: .[0].type, index: .[0].index, definitions: [.[].definitions|select(length > 0)[]]}) | sort_by(.index)[] | del(.index) ] }' \