@scanmate/ocr 0.5.0 → 0.6.0

package/dist/index.esm.js

@@ -975,13 +975,21 @@ function verifyPrintedRun(original, scan, dpi, run, templates, options = {}) {
     minPrinted = MIN_PRINTED_SCORE,
     minRivals = scope === 'text' ? MIN_TEXT_RIVALS : MIN_RIVALS,
     minDigits = 2,
-    characters = scope === 'text' ? TEXT_CHARACTERS : FIGURE_CHARACTERS$1
+    characters = scope === 'figures' ? FIGURE_CHARACTERS$1 : TEXT_CHARACTERS,
+    claimed
   } = options;
   const printed = [...run.text].filter(character => character.trim() !== '');
-  const wanted = scope === 'text' ? textCells(printed, characters) : figureCells(printed, minDigits);
+  // What the reading says it is, lined up with what the original printed. The
+  // two must correspond character for character or the comparison is meaningless.
+  const proposed = scope === 'confirm' ? [...(claimed ?? '')].filter(character => character.trim() !== '') : null;
+  if (proposed !== null && (proposed.length === 0 || proposed.length !== printed.length)) return {
+    verified: false,
+    because: 'no-claim'
+  };
+  const wanted = proposed === null ? scope === 'text' ? textCells(printed, characters) : figureCells(printed, minDigits) : disputedCells(printed, proposed, characters);
   if (wanted.size === 0) return {
     verified: false,
-    because: 'no-figure'
+    because: proposed === null ? 'no-figure' : 'no-claim'
   };
   // A total on a shaded bar is printed light on dark; turned round, it is a figure like any other.
   const lightOnDark = printPolarity(original, dpi, run) === 'light-on-dark';
@@ -992,9 +1000,25 @@ function verifyPrintedRun(original, scan, dpi, run, templates, options = {}) {
     verified: false,
     because: 'unplaceable'
   };
-  // Rivals: every character the page prints in this face and size that a figure could use.
+  /**
+   * Who each cell is up against.
+   *
+   * Identifying an unknown glyph needs most of the alphabet: to assert "this is
+   * a 7" you must be able to rule the others out, and the bar of eight digits
+   * is what the false-call rate justified. A settlement is not asking that. The
+   * reading has already named its answer, so the question is which of two named
+   * characters this ink is - and two templates answer it. A wrong verdict then
+   * requires the ink to match the *other specific character* better, not merely
+   * to be hard to read.
+   */
   const rivals = [...characters].filter(character => templates.has(templateKey(run, character)));
-  if (rivals.length < minRivals) return {
+  const rivalsAt = index => {
+    if (proposed === null) return rivals;
+    const against = proposed[index];
+    return against !== undefined && templates.has(templateKey(run, against)) ? [against] : [];
+  };
+  const enough = proposed === null ? rivals.length >= minRivals : [...wanted].some(index => rivalsAt(index).length > 0);
+  if (!enough) return {
     verified: false,
     because: 'few-rivals'
   };
@@ -1042,7 +1066,7 @@ function verifyPrintedRun(original, scan, dpi, run, templates, options = {}) {
   let confidence = 1;
   let checked = 0;
   for (const [index, crop] of crops) {
-    const passes = recoveries.map(recovery => judgeCell(crop, printed[index], rivals, run, templates, recovery, limits));
+    const passes = recoveries.map(recovery => judgeCell(crop, printed[index], rivalsAt(index), run, templates, recovery, limits));
     const first = passes[0];
     if (first === undefined) continue;
     // Undecided unless every pass reads it the same way: a verdict that changes
@@ -1116,6 +1140,15 @@ function textCells(printed, characters) {
   for (const [index, character] of printed.entries()) if (characters.includes(character)) wanted.add(index);
   return wanted;
 }
+/** Where the print and the reading disagree, which is the whole of what a settlement asks about. */
+function disputedCells(printed, proposed, characters) {
+  const wanted = new Set();
+  for (const [index, character] of printed.entries()) {
+    const against = proposed[index];
+    if (against !== undefined && against !== character && characters.includes(character) && characters.includes(against)) wanted.add(index);
+  }
+  return wanted;
+}
 function figureCells(printed, minDigits) {
   const wanted = new Set();
   let group = [];
@@ -1740,7 +1773,8 @@ async function readPage(page, engine, options, templates) {
       'unplaceable': 0,
       'few-rivals': 0,
       'too-coarse': 0,
-      'undecided': 0
+      'undecided': 0,
+      'no-claim': 0
     }
   };
   const seenChanged = new Set();

package/dist/src/print-verification/verify-print.use-case.d.ts

@@ -70,10 +70,22 @@ export interface VerifyOptions {
     minRivals?: number;
     /**
      * What to check: the digits of a figure (`'figures'`, the default, cheap
-     * enough for a whole page), or every letter and digit of the run (`'text'`,
-     * for a single run under dispute).
+     * enough for a whole page), every letter and digit of the run (`'text'`, for
+     * a single run under dispute), or the one question a settlement actually
+     * asks (`'confirm'`).
+     *
+     * `'confirm'` needs {@link claimed} and answers "is this ink the character
+     * the original printed, or the one the reading says it is?" - which takes
+     * two templates rather than a near-complete alphabet, because the reading
+     * has already named the alternative. See {@link VerifyOptions.claimed}.
      */
-    scope?: 'figures' | 'text';
+    scope?: 'figures' | 'text' | 'confirm';
+    /**
+     * What the reading claims the run says, for `scope: 'confirm'`. Compared
+     * character by character against the original's own text, so it has to hold
+     * the same number of printed characters; anything else abstains.
+     */
+    claimed?: string;
     /** How badly the printed glyph must match for a character to count as changed. Default `0.85`. */
     maxPrinted?: number;
     /** Digits in a row before a group is treated as a figure. Default `2`. */
@@ -109,8 +121,10 @@ export interface CellVerification {
  * - `too-coarse` - the scan resolves the cell below the height it is matched
  *   at, so a verdict would rest on detail it never captured.
  * - `undecided` - measured, and nothing won by enough to say.
+ * - `no-claim` - `scope: 'confirm'` without a reading to test, or one that
+ *   does not line up with the print character for character.
  */
-export type PrintAbstention = 'no-figure' | 'unplaceable' | 'few-rivals' | 'too-coarse' | 'undecided';
+export type PrintAbstention = 'no-figure' | 'unplaceable' | 'few-rivals' | 'too-coarse' | 'undecided' | 'no-claim';
 /** What the ink said, or why it would not say. */
 export type PrintCheck = ({
     verified: true;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@scanmate/ocr",
-  "version": "0.5.0",
+  "version": "0.6.0",
   "description": "How closely a scan's text matches the original's: OCR matched run by run on the page, figures held exact, every difference located.",
   "license": "MIT",
   "author": "Eduardo Russo",
@@ -41,7 +41,7 @@
     "!**/*.js.map"
   ],
   "dependencies": {
-    "@scanmate/ink": "^0.5.0",
+    "@scanmate/ink": "^0.6.0",
     "@tesseract.js-data/eng": "^1.0.0",
     "fastest-levenshtein": "^1.0.16",
     "tesseract.js": "^7.0.0"
@@ -50,7 +50,7 @@
     "access": "public"
   },
   "devDependencies": {
-    "@scanmate/align": "^0.5.0",
-    "@scanmate/extract": "^0.5.0"
+    "@scanmate/align": "^0.6.0",
+    "@scanmate/extract": "^0.6.0"
   }
 }