fix: custom regex

Author: seveibar

src/components/DocumentLabeler/index.js

@@ -12,9 +12,10 @@ export default function DocumentLabeler(props: LabelDocumentProps) {
   const [selectedLabels, changeSelectedLabels] = useState(
     props.initialLabels || (props.initialLabel ? [props.initialLabel] : [])
   )
-  const sequence = useMemo(() => stringToSequence(props.document), [
-    props.document
-  ])
+  const sequence = useMemo(
+    () => stringToSequence(props.document, props.separatorRegex),
+    [props.document]
+  )
   return (
     <div>
       <div>
@@ -58,7 +59,11 @@ export default function DocumentLabeler(props: LabelDocumentProps) {
             )
           })}
         </div>
-        <Document nothingHighlighted sequence={sequence} />
+        <Document
+          nothingHighlighted
+          sequence={sequence}
+          separatorRegex={props.separatorRegex}
+        />
       </div>
     </div>
   )

src/components/NLPAnnotator/index.story.js

@@ -32,6 +32,30 @@ storiesOf("NLPAnnotator", module)
       ]}
     />
   ))
+  .add("Sequence Labeler with Custom Regex", () => (
+    <NLPAnnotator
+      hotkeysEnabled
+      onChange={action("onChange")}
+      onFinish={action("onFinish")}
+      onNext={action("onNext")}
+      onPrev={action("onPrev")}
+      type="label-sequence"
+      document={`Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis pharetra ipsum tristique ligula venenatis placerat. Interdum et malesuada fames ac ante ipsum primis in faucibus. Fusce mollis velit nec tellus sollicitudin aliquam. In velit erat, iaculis id consectetur et, tincidunt sit amet mauris. Quisque ultricies, purus eleifend congue malesuada, ipsum erat molestie dolor, in pellentesque lacus purus vel nisl. Interdum et malesuada fames ac ante ipsum primis in faucibus. Nulla sed vestibulum magna. Quisque ut lorem imperdiet, aliquam velit nec, dictum felis.`}
+      labels={[
+        {
+          color: colors[0],
+          id: "noun",
+          displayName: "Noun"
+        },
+        {
+          color: colors[1],
+          id: "proper-noun",
+          displayName: "Proper Noun"
+        }
+      ]}
+      separatorRegex="."
+    />
+  ))
   .add("Document Labeler", () => (
     <NLPAnnotator
       hotkeysEnabled

src/components/RelationshipAnnotator/index.js

@@ -133,6 +133,7 @@ export default function RelationshipAnnotator(
       </LabelSelectorContainer>
       <div style={{ borderTop: "1px solid #ccc", marginTop: 8, paddingTop: 5 }}>
         <Document
+          separatorRegex={props.separatorRegex}
           colorLabelMap={colorLabelMap}
           nothingHighlighted={highlightedItems.length === 0}
           onCreateEmptyRelationship={([first, second]) => {

src/components/SequenceAnnotator/index.js

@@ -18,7 +18,7 @@ export default function SequenceAnnotator(props: SequenceAnnotatorProps) {
             ? [entity]
             : stringToSequence(entity.text, props.separatorRegex)
         )
-      : stringToSequence(props.document)
+      : stringToSequence(props.document, props.separatorRegex)
   )
   const colorLabelMap = useMemo(
     () =>
@@ -71,6 +71,7 @@ export default function SequenceAnnotator(props: SequenceAnnotatorProps) {
       </div>
       <div style={{ borderTop: "1px solid #ccc", marginTop: 8, paddingTop: 5 }}>
         <Document
+          separatorRegex={props.separatorRegex}
           colorLabelMap={colorLabelMap}
           nothingHighlighted={highlightedItems.length === 0}
           onHighlightedChanged={highlightedItems =>

src/string-to-sequence.js

@@ -1,8 +1,11 @@
 // @flow
 
-const stringToSequence = (doc: string, sepRe: RegExp = /[a-zA-ZÀ-ÿ]+/g) => {
+const stringToSequence = (
+  doc: string,
+  sepRe: RegExp | string = /[a-zA-ZÀ-ÿ]+/g
+) => {
   if (typeof sepRe === "string") {
-    sepRe = new RegExp(sepRe)
+    sepRe = new RegExp(sepRe, "g")
   }
   let m
   let indices = [0]
@@ -15,7 +18,6 @@ const stringToSequence = (doc: string, sepRe: RegExp = /[a-zA-ZÀ-ÿ]+/g) => {
   } while (m)
   indices = indices.concat([doc.length])
   return indices
-    .filter((_, i) => indices[i] !== indices[i + 1])
     .map((_, i) => ({
       text: doc.slice(indices[i], indices[i + 1]),
       textId: Math.random()

src/types.js

@@ -32,6 +32,7 @@ export type LabelDocumentProps = {
   multipleLabels?: boolean,
   document: string,
   initialLabels?: Array<string>,
+  separatorRegex?: string,
   onChange: (Array<string> | string | null) => any
 }