1 files changed, 79 insertions, 0 deletions
diff --git a/src/nlp/spacy.ts b/src/nlp/spacy.ts
new file mode 100644
index 0000000..d79de55
--- /dev/null
+++ b/src/nlp/spacy.ts
@@ -0,0 +1,79 @@
+import type { AsyncRes, Result } from "sortug";
+import { detectLang } from "./iso";
+const ENDPOINT = "http://localhost:8102";
+
+export async function run(text: string, langg?: string): AsyncRes<SpacyRes> {
+  try {
+    const lang = langg ? langg : detectLang(text);
+    const body = JSON.stringify({ string: text, lang });
+    const opts = {
+      headers: {
+        "Content-type": "application/json",
+        "X-API-KEY": Bun.env.SORTUG_NLP_API_KEY!,
+      },
+      method: "POST",
+      body,
+    };
+    const res = await fetch(ENDPOINT + "/spacy", opts);
+    const j = await res.json();
+    console.log("spacy", j);
+    return { ok: j };
+  } catch (e) {
+    return { error: `${e}` };
+  }
+}
+
+export type SpacyResBig = {
+  doc: {
+    text: string;
+    ents: any[];
+    sents: Array<{ start: number; end: number }>;
+    tokens: Token[];
+  };
+  segs: Sentence[];
+};
+export type SpacyRes = {
+  input: string;
+  segments: Sentence[];
+};
+export type Sentence = {
+  text: string;
+  start: number;
+  end: number;
+  root: Token;
+  subj: Token;
+  arcs: Arc[];
+  words: Word[];
+};
+export type Arc = {
+  start: number;
+  end: number;
+  label: string; // deprel label
+  dir: string;
+};
+export type Token = {
+  id: number;
+  head: number;
+  start: number;
+  end: number;
+  dep: string;
+  lemma: string;
+  morph: string;
+  pos: string;
+  tag: string;
+  text: string;
+};
+
+export interface Word extends Token {
+  ancestors: number[];
+  children: [];
+  n_lefts: number;
+  n_rights: number;
+  left_edge: number;
+  right_edge: number;
+  morph_map: Record<string, string>;
+}
+
+export function isChild(w: Word, topId: number): boolean {
+  return w.id === topId || w.ancestors.includes(topId);
+}