Spaces:
Running
Running
Julien Chaumond
commited on
Commit
·
ee12ee4
1
Parent(s):
6556bb5
Checkpoint 3
Browse files- js-src/Displacy.ts +56 -15
- js-src/coref.ts +2 -2
js-src/Displacy.ts
CHANGED
@@ -10,14 +10,54 @@ interface Span {
|
|
10 |
end: number;
|
11 |
}
|
12 |
|
|
|
|
|
|
|
|
|
|
|
13 |
class Displacy {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
|
15 |
/**
|
16 |
* Render a text string and its entity spans
|
17 |
*
|
18 |
* *see displacy-ent.js*
|
|
|
19 |
*/
|
20 |
static render(text: string, spans: Span[]): string {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
21 |
let out = {
|
22 |
__content: "",
|
23 |
append(s: string) {
|
@@ -26,23 +66,24 @@ class Displacy {
|
|
26 |
};
|
27 |
let offset = 0;
|
28 |
|
29 |
-
|
30 |
-
|
31 |
-
const
|
32 |
-
|
33 |
-
|
34 |
-
out.append(
|
35 |
-
|
36 |
-
out.append('<br>');
|
37 |
-
}
|
38 |
-
});
|
39 |
|
40 |
-
|
41 |
-
// We do not filter any entity type out.
|
42 |
-
out.append(`<mark data-entity=${ type.toLowerCase() }>${ entity }</mark>`);
|
43 |
|
44 |
-
|
45 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
|
47 |
out.append(text.slice(offset, text.length));
|
48 |
return out.__content;
|
|
|
10 |
end: number;
|
11 |
}
|
12 |
|
13 |
+
interface SpanTag {
|
14 |
+
span: Span;
|
15 |
+
tag: "start" | "end";
|
16 |
+
}
|
17 |
+
|
18 |
class Displacy {
|
19 |
+
static sortSpans(spans: Span[]) {
|
20 |
+
spans.sort((a, b) => { /// `a` should come first when the result is < 0
|
21 |
+
if (a.start === b.start) {
|
22 |
+
return b.end - a.end; /// CAUTION.
|
23 |
+
}
|
24 |
+
return a.start - b.start;
|
25 |
+
});
|
26 |
+
|
27 |
+
// Check existence of **strict overlapping**
|
28 |
+
spans.forEach((s, i) => {
|
29 |
+
if (i < spans.length - 1) {
|
30 |
+
const sNext = spans[i+1];
|
31 |
+
if (s.start < sNext.start && s.end > sNext.start) {
|
32 |
+
alert("Spans: strict overlapping");
|
33 |
+
}
|
34 |
+
}
|
35 |
+
});
|
36 |
+
}
|
37 |
|
38 |
/**
|
39 |
* Render a text string and its entity spans
|
40 |
*
|
41 |
* *see displacy-ent.js*
|
42 |
+
* see https://github.com/explosion/displacy-ent/issues/2
|
43 |
*/
|
44 |
static render(text: string, spans: Span[]): string {
|
45 |
+
this.sortSpans(spans);
|
46 |
+
|
47 |
+
const tags: { [index: number]: SpanTag[] } = {};
|
48 |
+
const __addTag = (i: number, s: Span, tag: "start" | "end") => {
|
49 |
+
if (Array.isArray(tags[i])) {
|
50 |
+
tags[i].push({ span: s, tag: tag });
|
51 |
+
} else {
|
52 |
+
tags[i] = [{ span: s, tag: tag }];
|
53 |
+
}
|
54 |
+
};
|
55 |
+
for (const s of spans) {
|
56 |
+
__addTag(s.start, s, "start");
|
57 |
+
__addTag(s.end, s, "end");
|
58 |
+
}
|
59 |
+
// console.log(JSON.stringify(tags)); // todo remove
|
60 |
+
|
61 |
let out = {
|
62 |
__content: "",
|
63 |
append(s: string) {
|
|
|
66 |
};
|
67 |
let offset = 0;
|
68 |
|
69 |
+
const indexes = Object.keys(tags).map(k => parseInt(k, 10)).sort((a, b) => a - b); /// CAUTION
|
70 |
+
for (const i of indexes) {
|
71 |
+
const spanTags = tags[i];
|
72 |
+
// console.log(i, spanTags); // todo remove
|
73 |
+
if (i > offset) {
|
74 |
+
out.append(text.slice(offset, i));
|
75 |
+
}
|
|
|
|
|
|
|
76 |
|
77 |
+
offset = i;
|
|
|
|
|
78 |
|
79 |
+
for (const sT of spanTags) {
|
80 |
+
if (sT.tag === "start") {
|
81 |
+
out.append(`<mark data-entity="${ sT.span.type.toLowerCase() }">`);
|
82 |
+
} else {
|
83 |
+
out.append(`</mark>`);
|
84 |
+
}
|
85 |
+
}
|
86 |
+
}
|
87 |
|
88 |
out.append(text.slice(offset, text.length));
|
89 |
return out.__content;
|
js-src/coref.ts
CHANGED
@@ -63,9 +63,9 @@ class Coref {
|
|
63 |
}
|
64 |
|
65 |
render(res: Response) {
|
66 |
-
const markup = Displacy.render(res.cleanedText, (<any>res).entities);
|
67 |
if (!this.container) { return ; }
|
68 |
-
console.log(markup);
|
69 |
this.container.innerHTML = `<div class="text">${markup}</div>`;
|
70 |
}
|
71 |
}
|
|
|
63 |
}
|
64 |
|
65 |
render(res: Response) {
|
66 |
+
const markup = Displacy.render(res.cleanedText, (<any>res).entities); // We will sort the second param in Displacy
|
67 |
if (!this.container) { return ; }
|
68 |
+
console.log(markup); // todo remove
|
69 |
this.container.innerHTML = `<div class="text">${markup}</div>`;
|
70 |
}
|
71 |
}
|