From 20df0d9671143f019b77230587b3fd92523026c1 Mon Sep 17 00:00:00 2001
From: Ruben Bimberg <ruben.bimberg@mni.thm.de>
Date: Sat, 19 Jun 2021 00:07:42 +0200
Subject: [PATCH] Change endpoint to spacy
Related to https://git.thm.de/rbmb29/spacy-api-docker/-/commit/a9ec007db16dac56956b4fb77dd078a221641da9
---
proxy.conf.json | 2 +-
.../spacy-dialog/spacy-dialog.component.ts | 5 +-
src/app/services/http/spacy.service.ts | 63 +++++--------------
3 files changed, 20 insertions(+), 50 deletions(-)
diff --git a/proxy.conf.json b/proxy.conf.json
index 603e23b9e..9e32ed540 100644
--- a/proxy.conf.json
+++ b/proxy.conf.json
@@ -9,7 +9,7 @@
"logLevel": "debug"
},
"/spacy": {
- "target": "https://spacy.frag.jetzt/noun",
+ "target": "https://spacy.frag.jetzt/spacy",
"secure": true,
"changeOrigin": true,
"pathRewrite": {
diff --git a/src/app/components/shared/_dialogs/spacy-dialog/spacy-dialog.component.ts b/src/app/components/shared/_dialogs/spacy-dialog/spacy-dialog.component.ts
index c77fddc5e..37233b4fd 100644
--- a/src/app/components/shared/_dialogs/spacy-dialog/spacy-dialog.component.ts
+++ b/src/app/components/shared/_dialogs/spacy-dialog/spacy-dialog.component.ts
@@ -72,10 +72,9 @@ export class SpacyDialogComponent implements OnInit, AfterContentInit {
.subscribe(words => {
const keywords: Keyword[] = [];
for (const word of words) {
- const newWord = word.trim();
- if (keywords.findIndex(item => item.word === newWord) < 0) {
+ if (keywords.findIndex(item => item.word === word) < 0) {
keywords.push({
- word: newWord,
+ word,
completed: false,
editing: false,
selected: false
diff --git a/src/app/services/http/spacy.service.ts b/src/app/services/http/spacy.service.ts
index f8965a97a..5e04619a0 100644
--- a/src/app/services/http/spacy.service.ts
+++ b/src/app/services/http/spacy.service.ts
@@ -6,22 +6,24 @@ import { catchError, map, tap } from 'rxjs/operators';
export type Model = 'de' | 'en' | 'fr' | 'es' | 'it' | 'nl' | 'pt' | 'auto';
-//[B]egin, [I]nside, [O]utside or unset
-type EntityPosition = 'B' | 'I' | 'O' | '';
+type KeywordType = 'entity' | 'noun';
+
+interface NounKeyword {
+ type: KeywordType;
+ lemma: string;
+ text: string;
+ dep: string;
+ tag: string;
+ pos: string;
+}
-interface NounToken {
- dep: string; // dependency inside the sentence
- // eslint-disable-next-line @typescript-eslint/naming-convention
- entity_pos: EntityPosition; // entity position
- // eslint-disable-next-line @typescript-eslint/naming-convention
- entity_type: string; // entity type
- lemma: string; // lemma of token
- tag: string; // tag of token
- text: string; // text of token
+interface EntityKeyword extends NounKeyword {
+ entityType: string;
}
-type NounCompound = NounToken[];
-type NounCompoundList = NounCompound[];
+type AbstractKeyword = NounKeyword | EntityKeyword;
+
+type KeywordList = AbstractKeyword[];
const httpOptions = {
// eslint-disable-next-line @typescript-eslint/naming-convention
@@ -37,44 +39,13 @@ export class SpacyService extends BaseHttpService {
super();
}
- private static processCompound(result: string[], data: NounCompound) {
- let isInEntity = false;
- let start = 0;
- const pushNew = (i: number) => {
- if (start < i) {
- result.push(data.slice(start, i).reduce((acc, current) => acc + ' ' + current.lemma, ''));
- start = i;
- }
- };
- data.forEach((noun, i) => {
- if (noun.entity_pos === 'B' || (noun.entity_pos === 'I' && !isInEntity)) {
- // entity begins
- pushNew(i);
- isInEntity = true;
- } else if (isInEntity) {
- if (noun.entity_pos === '' || noun.entity_pos === 'O') {
- // entity ends
- pushNew(i);
- isInEntity = false;
- }
- }
- });
- pushNew(data.length);
- }
-
getKeywords(text: string, model: Model): Observable<string[]> {
const url = '/spacy';
- return this.http.post<NounCompoundList>(url, {text, model}, httpOptions)
+ return this.http.post<KeywordList>(url, {text, model}, httpOptions)
.pipe(
tap(_ => ''),
catchError(this.handleError<any>('getKeywords')),
- map((result: NounCompoundList) => {
- const filteredNouns: string[] = [];
- result.forEach(compound => {
- SpacyService.processCompound(filteredNouns, compound);
- });
- return filteredNouns;
- })
+ map((result: KeywordList) => result.map(e => e.type === 'entity' ? e.text.trim() : e.lemma.trim()))
);
}
}
--
GitLab