Merge pull request #24 from ConnerWithAnE/levenshtein

feat: use levenshtein distance for long strings

Author: liminalities

web/package-lock.json

@@ -16,15 +16,17 @@
     "@nextui-org/system": "^2.2.6",
     "@nextui-org/theme": "^2.2.11",
     "@types/react-icons": "^3.0.0",
+    "fastest-levenshtein": "^1.0.16",
     "framer-motion": "^11.13.1",
     "react": "^18.3.1",
     "react-dom": "^18.3.1",
+    "react-icons": "^5.4.0",
     "react-router-dom": "^7.0.1",
-    "reacticons": "^0.0.1",
-    "react-icons": "^5.4.0"
+    "reacticons": "^0.0.1"
   },
   "devDependencies": {
     "@eslint/js": "^9.15.0",
+    "@types/node": "^22.13.13",
     "@types/react": "^18.3.12",
     "@types/react-dom": "^18.3.1",
     "@vitejs/plugin-react": "^4.3.4",

web/package.json

@@ -24,6 +24,7 @@ import { useForm } from "../../DataContext";
 import { HiCheckCircle } from "react-icons/hi2";
 import { HiExclamationTriangle } from "react-icons/hi2";
 import { HiExclamationCircle } from "react-icons/hi2";
+import { compareStringPasses } from "../../types/levenshtein-distance";
 
 type EntrySliverProp = {
   gptPass: GPTResponse;
@@ -158,13 +159,23 @@ export default function EntrySliver({
           const tests_2 = pass_2[i][typesafeKey];
           const tests_3 = pass_3[i][typesafeKey];
 
-          // if the key is special_notes, just fill in one value and return
-          // this is a compromise, as special_notes is often one long string and hard to compare
+          // the 'special_notes' field is a long string and thus should probably be compared using a different method
           if (typesafeKey === "special_notes") {
-            updatedTest = {
-              ...updatedTest,
-              [typesafeKey]: tests_1,
-            };
+            const result = compareStringPasses(tests_1, tests_2, tests_3);
+            if (result.pass) {
+              updatedTest = {
+                ...updatedTest,
+                [typesafeKey]: result.pass,
+              };
+            }
+            if (result.severity) {
+              addConflict2(
+                updatedConflicts,
+                `parts-${testIndex}-sees-${i}-${key}`,
+                result.severity
+              );
+            }
+
             return;
           }
 
@@ -238,13 +249,22 @@ export default function EntrySliver({
           const tests_2 = pass_2[i][typesafeKey];
           const tests_3 = pass_3[i][typesafeKey];
 
-          // if the key is special_notes, just fill in one value and return
-          // this is a compromise, as special_notes is often one long string and hard to compare
+          // the 'special_notes' field is a long string and thus should probably be compared using a different method
           if (typesafeKey === "special_notes") {
-            updatedTest = {
-              ...updatedTest,
-              [typesafeKey]: tests_1,
-            };
+            const result = compareStringPasses(tests_1, tests_2, tests_3);
+            if (result.pass) {
+              updatedTest = {
+                ...updatedTest,
+                [typesafeKey]: result.pass,
+              };
+            }
+            if (result.severity) {
+              addConflict2(
+                updatedConflicts,
+                `parts-${testIndex}-sees-${i}-${key}`,
+                result.severity
+              );
+            }
             return;
           }
 
@@ -318,16 +338,25 @@ export default function EntrySliver({
           const tests_2 = pass_2[i][typesafeKey];
           const tests_3 = pass_3[i][typesafeKey];
 
-          // if the key is special_notes, just fill in one value and return
-          // this is a compromise, as special_notes is often one long string and hard to compare
+          // the 'special_notes' field is a long string and thus should probably be compared using a different method
           if (typesafeKey === "special_notes") {
-            updatedTest = {
-              ...updatedTest,
-              [typesafeKey]: tests_1,
-            };
+            const result = compareStringPasses(tests_1, tests_2, tests_3);
+            if (result.pass) {
+              updatedTest = {
+                ...updatedTest,
+                [typesafeKey]: result.pass,
+              };
+            }
+            if (result.severity) {
+              addConflict2(
+                updatedConflicts,
+                `parts-${testIndex}-sees-${i}-${key}`,
+                result.severity
+              );
+            }
+
             return;
           }
-
           if (
             tests_1 === tests_2 &&
             tests_1 === tests_3 &&
@@ -447,13 +476,24 @@ export default function EntrySliver({
             };
             return;
           }
-          // if the key is other_details, just fill in one value and return
-          // this is a compromise, as other_details is often one long string and hard to compare
+
+          // the 'other_details' field is a long string and thus should probably be compared using a different method
           if (typesafeKey === "other_details") {
-            updatedPart = {
-              ...updatedPart,
-              [typesafeKey]: parts_1,
-            };
+            const result = compareStringPasses(parts_1, parts_2, parts_3);
+            if (result.pass) {
+              updatedPart = {
+                ...updatedPart,
+                [typesafeKey]: result.pass,
+              };
+            }
+            if (result.severity) {
+              addConflict2(
+                updatedConflicts,
+                `parts-${i}-${key}`,
+                result.severity
+              );
+            }
+
             return;
           }
 
@@ -544,12 +584,20 @@ export default function EntrySliver({
       if (editedEntry[typesafeKey] !== undefined) {
         return;
       }
-      // the 'objective' field is a long string and thus cannot be compared via string comparison
+      // the 'objective' field is a long string and thus should probably be compared using a different method
       if (typesafeKey === "objective") {
-        updatedEntry = {
-          ...updatedEntry,
-          objective: pass_1,
-        };
+        const result = compareStringPasses(pass_1, pass_2, pass_3);
+
+        if (result.pass) {
+          updatedEntry = {
+            ...updatedEntry,
+            objective: result.pass,
+          };
+        }
+        if (result.severity) {
+          addConflict2(updatedConflicts, key, result.severity);
+        }
+
         return;
       }
       // Compare the part information in the passes

web/src/components/database-entries/entry-sliver.tsx

@@ -818,7 +818,7 @@ export const mockFullDataTypePasses5: FullDataType[] = [
             increased_power_usage: false,
             power_usage_description: "string",
             special_notes:
-              "This is a test to make sure that the rendering works",
+              "Testing included latchup protection circuitry to prevent destructive latchup.",
           },
           {
             source: "Co60",
@@ -829,7 +829,7 @@ export const mockFullDataTypePasses5: FullDataType[] = [
             increased_power_usage: false,
             power_usage_description: "string",
             special_notes:
-              "This is a test to make sure that the rendering works",
+              "Rated as moderate risk due to low predicted event rates and/or non-destructive nature of observed latchups",
           },
         ],
         sees: [
@@ -959,7 +959,7 @@ export const mockFullDataTypePasses5: FullDataType[] = [
             increased_power_usage: false,
             power_usage_description: "string",
             special_notes:
-              "This is a test to make sure that the rendering works",
+              "Testing included latchup protection circuitry to prevent destructive latchup.",
           },
           {
             source: "Co60",
@@ -970,7 +970,7 @@ export const mockFullDataTypePasses5: FullDataType[] = [
             increased_power_usage: false,
             power_usage_description: "string",
             special_notes:
-              "This is a test to make sure that the rendering works",
+              "Rated as moderate risk due to non-destructive nature of observed latchups and/or low predicted event rates",
           },
         ],
         sees: [
@@ -1100,7 +1100,7 @@ export const mockFullDataTypePasses5: FullDataType[] = [
             increased_power_usage: false,
             power_usage_description: "string",
             special_notes:
-              "This is a test to make sure that the rendering works",
+              "Testing included latchup protection circuitry to prevent destructive latchup.",
           },
           {
             source: "Co60",
@@ -1111,7 +1111,7 @@ export const mockFullDataTypePasses5: FullDataType[] = [
             increased_power_usage: false,
             power_usage_description: "string",
             special_notes:
-              "This is a test to make sure that the rendering works",
+              "Determined to have moderate risk because of non-destructive nature of observed latchups and/or low predicted event rates",
           },
         ],
         sees: [

web/src/mockfulldatatype.tsx

@@ -0,0 +1,25 @@
+import { distance } from "fastest-levenshtein";
+import { Severity } from "./types";
+
+export function compareStringPasses(
+  a: string,
+  b: string,
+  c: string
+): { pass?: string; severity?: Severity } {
+  const distance1 = distance(a, b);
+  const distance2 = distance(b, c);
+  const distance3 = distance(a, c);
+
+  // all passes are similar enough
+  if (distance1 < 5 && distance2 < 5 && distance3 < 5) {
+    return { pass: a };
+  }
+  // if two passes are similar enough
+  if ((distance1 < 5 && distance3 < 5) || (distance1 < 5 && distance2 < 5)) {
+    return { pass: a, severity: 1 };
+  }
+  if (distance2 < 5 && distance3 < 5) {
+    return { pass: b, severity: 1 };
+  }
+  return { severity: 2 };
+}