fix: Merge pull request #275 from UniversalDataTool/fix/relations-fixes

add support for import text snippets with relations,
UniversalDataTool · Aug 26, 2020 · 4975550 · 4975550
2 parents f458e15 + 89cf6fe
commit 4975550
Show file tree

Hide file tree

Showing 6 changed files with 187 additions and 56 deletions.
diff --git a/package.json b/package.json
@@ -73,7 +73,7 @@
     "react": "^16.8.6",
     "react-dom": "^16.8.6",
     "react-markdown": "^4.1.0",
-    "react-nlp-annotate": "^0.3.13",
+    "react-nlp-annotate": "^0.3.14",
     "rimraf": "^3.0.2",
     "spelling": "^2.0.1"
   },

diff --git a/src/components/ImportPage/index.js b/src/components/ImportPage/index.js
@@ -222,7 +222,11 @@ export default ({
         <Button
           dialog="import-text-snippets"
           Icon={TextFieldsIcon}
-          onlySupportType={["text_entity_recognition", "text_classification"]}
+          onlySupportType={[
+            "text_entity_recognition",
+            "text_classification",
+            "text_entity_relations",
+          ]}
           type={dataset.interface.type}
         >
           {t("import-text-snippets")}

diff --git a/src/components/TextEntityRecognition/convert-react-nlp-annotate-types.js b/src/components/TextEntityRecognition/convert-react-nlp-annotate-types.js
@@ -32,7 +32,7 @@ export const simpleSequenceAndRelationsToEntitySequence = ({
         textId: seq.textId,
         label: seq.label,
         start: charsPassed,
-        end: charsPassed + seq.text.length,
+        end: charsPassed + seq.text.length - 1,
       })
     }
     charsPassed += seq.text.length
@@ -54,7 +54,8 @@ export const entitySequenceToSimpleSeq = (doc, entSeq) => {
         textId:
           entSeq[nextEntity].textId || Math.random().toString(36).slice(-6),
       })
-      i = entSeq[nextEntity].end
+      // i = entSeq[nextEntity].end
+      i += entSeq[nextEntity].text.length - 1
       simpleSeq.push({ text: "" })
       nextEntity += 1
     } else {
@@ -68,5 +69,13 @@ export const entitySequenceToSimpleSeq = (doc, entSeq) => {
       }
     }
   }
+  if (simpleSeq.map((s) => s.text).join("") !== doc) {
+    throw new Error(
+      "Combining annotation entities with document didn't match document.\n\n" +
+        simpleSeq.map((s) => s.text).join("") +
+        "\n\nvs\n\n" +
+        doc
+    )
+  }
   return simpleSeq.filter((item) => item.text.length > 0)
 }
diff --git a/src/components/TextEntityRelations/index.js b/src/components/TextEntityRelations/index.js
@@ -8,6 +8,7 @@ import {
 } from "../TextEntityRecognition/convert-react-nlp-annotate-types"
 
 export const TextEntityRelations = (props) => {
+  console.log(props)
   const [currentSampleIndex, setCurrentSampleIndex] = useClobberedState(
     props.sampleIndex,
     0
@@ -28,6 +29,8 @@ export const TextEntityRelations = (props) => {
     )
   }
 
+  console.log({ initialSequence, initialRelations })
+
   return (
     <NLPAnnotator
       key={currentSampleIndex}

diff --git a/src/components/TextEntityRelations/index.story.js b/src/components/TextEntityRelations/index.story.js
@@ -7,57 +7,172 @@ import { action } from "@storybook/addon-actions"
 
 import TextEntityRelations from "./"
 
-storiesOf("TextEntityRelations", module).add("Basic", () => (
-  <TextEntityRelations
-    onSaveTaskOutputItem={action("onSaveTaskOutputItem")}
-    onExit={action("onExit")}
-    interface={{
-      type: "text_entity_recognition",
-      description: "Label words or phrases as food or hat.",
-      overlapAllowed: false,
-      entityLabels: [
+storiesOf("TextEntityRelations", module)
+  .add("Basic", () => (
+    <TextEntityRelations
+      onSaveTaskOutputItem={action("onSaveTaskOutputItem")}
+      onExit={action("onExit")}
+      interface={{
+        type: "text_entity_recognition",
+        description: "Label words or phrases as food or hat.",
+        overlapAllowed: false,
+        entityLabels: [
+          {
+            id: "food",
+            displayName: "Food",
+          },
+          {
+            id: "hat",
+            displayName: "Hat",
+          },
+        ],
+        relationLabels: [
+          {
+            id: "subject-doing",
+            displayName: "Subject Doing",
+          },
+        ],
+      }}
+      samples={[
         {
-          id: "food",
-          displayName: "Food",
+          document:
+            "This strainer makes a great hat, I'll wear it while I serve spaghetti!",
+          annotation: {
+            entities: [
+              {
+                text: "strainer",
+                label: "hat",
+                start: 5,
+                end: 13,
+                textId: "id1",
+              },
+              {
+                text: "spaghetti",
+                label: "food",
+                start: 60,
+                end: 69,
+                textId: "id2",
+              },
+              { text: "I'll", start: 33, end: 36, textId: "id3" },
+              { text: "wear", start: 38, end: 41, textId: "id4" },
+            ],
+            relations: [{ from: "id3", to: "id4", label: "subject-doing" }],
+          },
         },
+      ]}
+    />
+  ))
+  .add("Saving bug with relations (should not error)", () => (
+    <TextEntityRelations
+      onSaveTaskOutputItem={action("onSaveTaskOutputItem")}
+      onExit={action("onExit")}
+      interface={{
+        type: "text_entity_recognition",
+        description: "Label words or phrases as food or hat.",
+        overlapAllowed: false,
+        entityLabels: [
+          {
+            id: "food",
+            displayName: "Food",
+          },
+          {
+            id: "hat",
+            displayName: "Hat",
+          },
+        ],
+        relationLabels: [
+          {
+            id: "subject",
+            displayName: "Subject",
+          },
+          {
+            id: "R1",
+          },
+        ],
+      }}
+      samples={[
         {
-          id: "hat",
-          displayName: "Hat",
+          document: "This is something i'd like to do",
+          annotation: {
+            entities: [
+              { text: "This", textId: "82tm4y", start: 0, end: 3 },
+              { text: "something", textId: "gejze6", start: 8, end: 16 },
+              { text: "d", textId: "ixwjg9", start: 20, end: 20 },
+              { text: "to", textId: "r8ayjq", start: 27, end: 28 },
+              { text: "do", textId: "wt3hbx", start: 30, end: 31 },
+            ],
+            relations: [
+              {
+                from: "82tm4y",
+                to: "gejze6",
+                label: "subject",
+                color: "#d32f2f",
+              },
+              { from: "ixwjg9", to: "gejze6", label: "R1", color: "#c2185b" },
+              {
+                from: "r8ayjq",
+                to: "gejze6",
+                label: "subject",
+                color: "#d32f2f",
+              },
+              { from: "ixwjg9", to: "wt3hbx", label: "R1", color: "#c2185b" },
+              {
+                from: "82tm4y",
+                to: "gejze6",
+                label: "subject",
+                color: "#d32f2f",
+              },
+            ],
+          },
         },
-      ],
-      relationLabels: [
+      ]}
+    />
+  ))
+  .add("Should not combine entities", () => (
+    <TextEntityRelations
+      onSaveTaskOutputItem={action("onSaveTaskOutputItem")}
+      onExit={action("onExit")}
+      interface={{
+        type: "text_entity_recognition",
+        description: "Label words or phrases as food or hat.",
+        overlapAllowed: false,
+        entityLabels: [
+          {
+            id: "food",
+            displayName: "Food",
+          },
+          {
+            id: "hat",
+            displayName: "Hat",
+          },
+        ],
+        relationLabels: [
+          {
+            id: "subject",
+            displayName: "Subject",
+          },
+          {
+            id: "R1",
+          },
+        ],
+      }}
+      samples={[
         {
-          id: "subject-doing",
-          displayName: "Subject Doing",
+          document: "This is something i'd like to do",
+          annotation: {
+            entities: [
+              {
+                text: "This",
+                textId: "this",
+                start: 0,
+                end: 3,
+                label: "subject",
+              },
+              { text: " is", textId: "is", start: 4, end: 6, label: "subject" },
+            ],
+            relations: [],
+          },
         },
-      ],
-    }}
-    samples={[
-      {
-        document:
-          "This strainer makes a great hat, I'll wear it while I serve spaghetti!",
-        annotation: {
-          entities: [
-            {
-              text: "strainer",
-              label: "hat",
-              start: 5,
-              end: 13,
-              textId: "id1",
-            },
-            {
-              text: "spaghetti",
-              label: "food",
-              start: 60,
-              end: 69,
-              textId: "id2",
-            },
-            { text: "I'll", start: 32, end: 36, textId: "id3" },
-            { text: "wear", start: 37, end: 41, textId: "id4" },
-          ],
-          relations: [{ from: "id3", to: "id4", label: "subject-doing" }],
-        },
-      },
-    ]}
-  />
-))
+      ]}
+    />
+  ))
diff --git a/yarn.lock b/yarn.lock
@@ -16303,10 +16303,10 @@ react-native-get-random-values@^1.4.0:
   dependencies:
     fast-base64-decode "^1.0.0"
 
-react-nlp-annotate@^0.3.13:
-  version "0.3.13"
-  resolved "https://registry.yarnpkg.com/react-nlp-annotate/-/react-nlp-annotate-0.3.13.tgz#8652ebf35a5585d4ab921f1966470f5cda35d71c"
-  integrity sha512-JJjCAE8vzZxPP+u2BPAmCHvWDduBTPqOh13N+YGMUs26aqQYDrOOnjZQN9VLnySXtwcGkeJbufH6q3k+A8Hlbw==
+react-nlp-annotate@^0.3.14:
+  version "0.3.14"
+  resolved "https://registry.yarnpkg.com/react-nlp-annotate/-/react-nlp-annotate-0.3.14.tgz#d96c495d0e5da259a5b02de11b72df3d44157267"
+  integrity sha512-SSg1jsD7rh6WA4OSCcs4VbaTfIibGlRd0qLcADDc8HdioALiodJcCvkPQG5A4nt1NfAZaBEn/QOs9FUoei57rg==
   dependencies:
     "@material-ui/lab" "^4.0.0-alpha.56"
     chroma-js "^2.0.3"