feat: include parsed value in details for invalid text fields (#51)

When a value doesn't match the expected regex, it still has a clear value with start and end indexes. BREAKING CHANGE: The result's `details` may now include a non-null `value` even when it's invalid. The `start` and `end` fields will refer to it instead of the entire range when that happens. The `fields` object is unchanged and still contains `null` for invalid fields. Closes: #50
cheminfo · Feb 23, 2024 · b6e2652 · b6e2652
1 parent 5be3275
commit b6e2652
Show file tree

Hide file tree

Showing 9 changed files with 115 additions and 19 deletions.
diff --git a/src/parse/__tests__/swissDrivingLicense.test.ts b/src/parse/__tests__/swissDrivingLicense.test.ts
@@ -25,7 +25,7 @@ describe('parse Swiss Driving License', () => {
       valid: true,
       autocorrect: [],
     });
-    expect(result.details[result.details.length - 1]).toStrictEqual({
+    expect(result.details.at(-1)).toStrictEqual({
       label: 'First name',
       field: 'firstName',
       value: 'FABIENNE',
@@ -55,6 +55,68 @@ describe('parse Swiss Driving License', () => {
       lastName: 'MARCHAND',
     });
   });
+  it('invalid text', () => {
+    const MRZ = [
+      'AAA001D<<',
+      'FACHE305142128097<<800126<<<<<',
+      'M4RCHAND<<FABI3NNE<<<<<<<<<<<<',
+    ];
+
+    const result = parse(MRZ);
+    expect(result.format).toBe('SWISS_DRIVING_LICENSE');
+    expect(result.valid).toBe(false);
+    expect(result.details.filter((a) => !a.valid)).toHaveLength(2);
+    expect(result.details[0]).toStrictEqual({
+      label: 'Document number',
+      field: 'documentNumber',
+      ranges: [{ line: 0, start: 0, end: 9, raw: 'AAA001D<<' }],
+      line: 0,
+      start: 0,
+      end: 7,
+      value: 'AAA001D',
+      valid: true,
+      autocorrect: [],
+    });
+    expect(result.details.at(-1)).toStrictEqual({
+      label: 'First name',
+      field: 'firstName',
+      value: 'FABI3NNE',
+      valid: false,
+      ranges: [
+        {
+          line: 2,
+          start: 0,
+          end: 30,
+          raw: 'M4RCHAND<<FABI3NNE<<<<<<<<<<<<',
+        },
+      ],
+      line: 2,
+      start: 10,
+      end: 18,
+      error:
+        'invalid text: FABI3NNE<<<<<<<<<<<<. Must match the following regular expression: /^[A-Z<]+<*$/',
+      autocorrect: [],
+    });
+    expect(result.details.at(-2)).toMatchObject({
+      field: 'lastName',
+      value: 'M4RCHAND',
+      valid: false,
+      line: 2,
+      start: 0,
+      end: 8,
+    });
+    expect(result.fields).toStrictEqual({
+      documentNumber: 'AAA001D',
+      languageCode: 'D',
+      documentCode: 'FA',
+      issuingState: 'CHE',
+      pinCode: '305142128',
+      versionNumber: '097',
+      birthDate: '800126',
+      firstName: null,
+      lastName: null,
+    });
+  });
   it('Use autocorrect', () => {
     const MRZ = [
       'AAA001D<<',

diff --git a/src/parse/createFieldParser.ts b/src/parse/createFieldParser.ts
@@ -1,5 +1,6 @@
 'use strict';
 
+import { ParseTextError } from '../parsers/parseText';
 import { Autocorrect, Details, FieldName, Range } from '../types';
 
 import { autoCorrection } from './autoCorrection';
@@ -100,8 +101,13 @@ export default function createFieldParser(
         result.start = range.start + parsed.start;
         result.end = range.start + parsed.end;
       }
-    } catch (e) {
-      result.error = e.message;
+    } catch (err) {
+      result.error = err.message;
+      if (err instanceof ParseTextError) {
+        result.value = err.value;
+        result.start = range.start + err.start;
+        result.end = range.start + err.end;
+      }
     }
 
     for (const autocorrectElement of autocorrect) {

diff --git a/src/parse/getResult.ts b/src/parse/getResult.ts
@@ -20,14 +20,16 @@ function getDetails(
 
 function getFields(details: Details[]) {
   const fields: FieldRecords = {};
-  let valid = true;
+  let allValid = true;
   for (const detail of details) {
-    if (!detail.valid) valid = false;
+    if (!detail.valid) {
+      allValid = false;
+    }
     if (detail.field) {
-      fields[detail.field] = detail.value;
+      fields[detail.field] = detail.valid ? detail.value : null;
     }
   }
-  return { fields, valid };
+  return { fields, allValid };
 }
 
 function getCorrection(
@@ -71,6 +73,6 @@ export function getResult(
     format,
     details,
     fields: fields.fields,
-    valid: fields.valid,
+    valid: fields.allValid,
   };
 }
diff --git a/src/parsers/parseDocumentNumberOptional.ts b/src/parsers/parseDocumentNumberOptional.ts
@@ -8,14 +8,17 @@ export function parseDocumentNumberOptional(
 ) {
   if (checkDigit === '<') {
     const firstFiller = optional.indexOf('<');
-    const value = parseText(optional.substring(firstFiller + 1));
+    const value = parseText(
+      optional.substring(firstFiller + 1),
+      firstFiller + 1,
+    );
     return {
       value,
       start: firstFiller + 1,
       end: firstFiller + 1 + value.length,
     };
   } else {
-    const value = parseText(optional);
+    const value = parseText(optional, 0);
     return {
       value,
       start: 0,

diff --git a/src/parsers/parseFirstName.ts b/src/parsers/parseFirstName.ts
@@ -4,7 +4,11 @@ import { parseText } from './parseText';
 
 export default function parseFirstName(source: string) {
   const withoutStart = source.replace(/.*?<{2}/, '');
-  const value = parseText(withoutStart, /^[A-Z<]+<*$/);
+  const value = parseText(
+    withoutStart,
+    source.length - withoutStart.length,
+    /^[A-Z<]+<*$/,
+  );
   const start = source.length - withoutStart.length;
   return {
     value,

diff --git a/src/parsers/parseLastName.ts b/src/parsers/parseLastName.ts
@@ -3,7 +3,7 @@
 import { parseText } from './parseText';
 
 export default function parseLastName(source: string) {
-  const parsed = parseText(source.replace(/<{2}.*/, ''), /^[A-Z<]*<*$/);
+  const parsed = parseText(source.replace(/<{2}.*$/, ''), 0, /^[A-Z<]*<*$/);
   return {
     value: parsed,
     start: 0,

diff --git a/src/parsers/parseOptional.ts b/src/parsers/parseOptional.ts
@@ -3,11 +3,11 @@
 import { parseText } from './parseText';
 
 export function parseOptional(source: string) {
-  const value = parseText(source);
+  const value = parseText(source, 0);
 
   return {
     value,
     start: 0,
-    end: 0 + value.length,
+    end: value.length,
   };
 }
diff --git a/src/parsers/parsePersonalNumber.ts b/src/parsers/parsePersonalNumber.ts
@@ -3,7 +3,7 @@
 import { parseText } from './parseText';
 
 export function parsePersonalNumber(source: string) {
-  const value = parseText(source, /^[A-Z0-9<]+<*$/);
+  const value = parseText(source, 0, /^[A-Z0-9<]+<*$/);
   return {
     value,
     start: 0,

diff --git a/src/parsers/parseText.ts b/src/parsers/parseText.ts
@@ -2,11 +2,30 @@
 
 import { cleanText } from './cleanText';
 
-export function parseText(source: string, regexp = /^[0-9A-Z<]+$/) {
-  if (!source.match(regexp)) {
-    throw new Error(
+export function parseText(
+  source: string,
+  initialStart: number,
+  regexp = /^[0-9A-Z<]+$/,
+) {
+  const cleaned = cleanText(source);
+  if (!regexp.test(source)) {
+    throw new ParseTextError(
       `invalid text: ${source}. Must match the following regular expression: ${regexp.toString()}`,
+      cleaned,
+      initialStart,
+      initialStart + cleaned.length,
     );
   }
-  return cleanText(source);
+  return cleaned;
+}
+
+export class ParseTextError extends Error {
+  constructor(
+    message: string,
+    public readonly value: string,
+    public readonly start: number,
+    public readonly end: number,
+  ) {
+    super(message);
+  }
 }