Skip to content

Commit 1e5318a

Browse files
BalearicaBalearica
committed
Minor updates
1 parent 0a65142 commit 1e5318a

File tree

3 files changed

+48
-29
lines changed

3 files changed

+48
-29
lines changed

‎js/containers/imageContainer.js‎

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -276,6 +276,10 @@ export class ImageCache {
276276
* @param {boolean} [nativeOnly=true]
277277
*/
278278
static getImages = (n, props, nativeOnly = true) => {
279+
if (!ImageCache.inputModes.image && !ImageCache.inputModes.pdf) {
280+
return { native: undefined, binary: undefined };
281+
}
282+
279283
const significantRotation = Math.abs(pageMetricsArr[n].angle || 0) > 0.05;
280284

281285
const newNative = !ImageCache.native[n] || !imageUtils.compatible(ImageCache.nativeProps[n], props, significantRotation);

‎js/import/convertPageText.js‎

Lines changed: 37 additions & 24 deletions
Original file line numberDiff line numberDiff line change
@@ -4,11 +4,12 @@ import { calcWordCharMetrics } from '../utils/fontUtils.js';
44
import { FontCont } from '../containers/fontContainer.js';
55

66
const FONT_FAMILY = 'Times New Roman';
7-
const FONT_SIZE = 12;
8-
const CHAR_SPACING = 1;
7+
const FONT_SIZE = 14;
8+
const CHAR_SPACING = 0;
9+
const WORD_SPACING = 0;
910
const LINE_HEIGHT = 14.4;
10-
const ASCENDER_HEIGHT = 9.6;
11-
const DESCENDER_HEIGHT = 2.4;
11+
const MARGIN_VERTICAL = 30;
12+
const MARGIN_HORIZONTAL = 20;
1213

1314
/** @type {?opentype.Font} */
1415
let fontOpentype = null;
@@ -19,15 +20,24 @@ let fontOpentype = null;
1920
* @param {number} size
2021
* @param {opentype.Font} font
2122
*/
22-
function getTextAdvance(text, size, font) {
23+
function getTextWidth(text, size, font) {
2324
const { advanceArr, kerningArr } = calcWordCharMetrics(text, font);
2425

2526
const advanceTotal = advanceArr.reduce((a, b) => a + b, 0);
2627
const kerningTotal = kerningArr.reduce((a, b) => a + b, 0);
2728

28-
const wordWidth1 = (advanceTotal + kerningTotal) * (size / font.unitsPerEm);
29+
const wordLastGlyphMetrics = font.charToGlyph(text.at(-1)).getMetrics();
30+
const wordFirstGlyphMetrics = font.charToGlyph(text[0]).getMetrics();
31+
32+
// The `leftSideBearing`/`rightSideBearing`/ numbers reported by Opentype.js are not accurate for mono-spaced fonts, so `xMin`/`xMax` are used instead.
33+
const wordLeftBearing = wordFirstGlyphMetrics.xMin || 0;
34+
const lastGlyphMax = wordLastGlyphMetrics.xMax || 0;
35+
const wordRightBearing = advanceArr[advanceArr.length - 1] - lastGlyphMax;
36+
37+
const wordWidth1 = (advanceTotal + kerningTotal - wordLeftBearing - wordRightBearing);
38+
const wordWidth1Px = wordWidth1 * (size / font.unitsPerEm);
2939
const spacingTotalPx = (text.length - 1) * CHAR_SPACING;
30-
const wordWidth = wordWidth1 + spacingTotalPx;
40+
const wordWidth = wordWidth1Px + spacingTotalPx;
3141

3242
return wordWidth;
3343
}
@@ -76,6 +86,9 @@ export async function convertPageText({ textStr, pageDims = null }) {
7686
fontOpentype = (await FontCont.getFont({ font: FONT_FAMILY })).opentype;
7787
}
7888

89+
const ASCENDER_HEIGHT = fontOpentype.ascender * (FONT_SIZE / fontOpentype.unitsPerEm);
90+
const DESCENDER_HEIGHT = fontOpentype.descender * (FONT_SIZE / fontOpentype.unitsPerEm);
91+
7992
const lines = textStr.split(/\r?\n/);
8093

8194
if (!pageDims) {
@@ -97,39 +110,38 @@ export async function convertPageText({ textStr, pageDims = null }) {
97110

98111
let tablesPage = new LayoutDataTablePage(0);
99112
const pagesOut = [{ pageObj, dataTables: tablesPage }];
100-
const margin = 20;
101-
const availableWidth = pageDims.width - margin * 2;
113+
const availableWidth = pageDims.width - MARGIN_HORIZONTAL * 2;
102114

103-
let currentY = margin + ASCENDER_HEIGHT;
115+
let currentY = MARGIN_VERTICAL + LINE_HEIGHT / 2;
104116

105117
for (let lineIndex = 0; lineIndex < lines.length; lineIndex++) {
106118
const lineText = lines[lineIndex];
107119

108120
if (lineText.length === 0 || lineText.trim().length === 0) {
109121
currentY += LINE_HEIGHT;
110-
if (currentY + DESCENDER_HEIGHT > pageDims.height - margin) {
122+
if (currentY + FONT_SIZE > pageDims.height - MARGIN_VERTICAL) {
111123
pageIndex++;
112124
const newPage = new ocr.OcrPage(pageIndex, pageDims);
113125
newPage.textSource = 'text';
114126
const newTables = new LayoutDataTablePage(0);
115127
pagesOut.push({ pageObj: newPage, dataTables: newTables });
116128
pageObj = newPage;
117129
tablesPage = newTables;
118-
currentY = margin + ASCENDER_HEIGHT;
130+
currentY = MARGIN_VERTICAL + LINE_HEIGHT / 2;
119131
}
120132
continue;
121133
}
122134

123135
const wordTokens = splitIntoWords(lineText);
124136

125137
const parLines = [];
126-
let parRight = margin;
138+
let parRight = MARGIN_HORIZONTAL;
127139

128140
for (let idx = 0; idx < wordTokens.length;) {
129-
if (currentY + DESCENDER_HEIGHT > pageDims.height - margin) {
141+
if (currentY + FONT_SIZE > pageDims.height - MARGIN_VERTICAL) {
130142
if (parLines.length > 0) {
131143
const parBbox = {
132-
left: margin,
144+
left: MARGIN_HORIZONTAL,
133145
top: parLines[0].bbox.top,
134146
right: parRight,
135147
bottom: parLines[parLines.length - 1].bbox.bottom,
@@ -139,7 +151,7 @@ export async function convertPageText({ textStr, pageDims = null }) {
139151
for (const ln of parLines) ln.par = parObj;
140152
pageObj.pars.push(parObj);
141153
parLines.length = 0;
142-
parRight = margin;
154+
parRight = MARGIN_HORIZONTAL;
143155
}
144156
pageIndex++;
145157
const newPage = new ocr.OcrPage(pageIndex, pageDims);
@@ -148,34 +160,35 @@ export async function convertPageText({ textStr, pageDims = null }) {
148160
pagesOut.push({ pageObj: newPage, dataTables: newTables });
149161
pageObj = newPage;
150162
tablesPage = newTables;
151-
currentY = margin + ASCENDER_HEIGHT;
163+
currentY = MARGIN_VERTICAL + LINE_HEIGHT / 2;
152164
}
153165

154166
const baseline = [0, DESCENDER_HEIGHT];
155167
const lineTop = Math.round(currentY - ASCENDER_HEIGHT);
156168
const lineBottom = Math.round(currentY + DESCENDER_HEIGHT);
157169

158-
let currentX = margin;
170+
let currentX = MARGIN_HORIZONTAL;
159171
let widthSoFar = 0;
160172

161173
const lineBbox = {
162-
left: margin,
174+
left: MARGIN_HORIZONTAL,
163175
top: lineTop,
164-
right: margin,
176+
right: MARGIN_HORIZONTAL,
165177
bottom: lineBottom,
166178
};
167179
const lineObj = new ocr.OcrLine(
168180
pageObj,
169181
lineBbox,
170182
baseline,
171183
ASCENDER_HEIGHT,
172-
ASCENDER_HEIGHT - DESCENDER_HEIGHT,
184+
null,
173185
);
174186

175187
let lastConsumed = idx;
176188
for (let j = idx; j < wordTokens.length; j++) {
177189
const tok = wordTokens[j];
178-
const tokWidth = getTextAdvance(tok.text, FONT_SIZE, fontOpentype);
190+
let tokWidth = getTextWidth(tok.text, FONT_SIZE, fontOpentype);
191+
if (tok.isWhitespace) tokWidth += WORD_SPACING;
179192

180193
if (tok.isWhitespace) {
181194
if (lineObj.words.length === 0) {
@@ -218,7 +231,7 @@ export async function convertPageText({ textStr, pageDims = null }) {
218231
if (lineObj.words.length === 0) {
219232
const nextTok = wordTokens[idx];
220233
if (nextTok && !nextTok.isWhitespace) {
221-
const tokWidth = getTextAdvance(nextTok.text, FONT_SIZE, fontOpentype);
234+
const tokWidth = getTextWidth(nextTok.text, FONT_SIZE, fontOpentype);
222235
const wordBbox = {
223236
left: Math.round(currentX),
224237
top: lineTop,
@@ -258,7 +271,7 @@ export async function convertPageText({ textStr, pageDims = null }) {
258271

259272
if (parLines.length > 0) {
260273
const parBbox = {
261-
left: margin,
274+
left: MARGIN_HORIZONTAL,
262275
top: parLines[0].bbox.top,
263276
right: parRight,
264277
bottom: parLines[parLines.length - 1].bbox.bottom,

‎js/worker/compareOCRModule.js‎

Lines changed: 7 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -1454,7 +1454,7 @@ export async function nudgePageBaseline({
14541454
* This function is a WIP and not all options are implemented.
14551455
* @param {Object} args
14561456
* @param {OcrPage} args.page - Page to render.
1457-
* @param {import('../containers/imageContainer.js').ImageWrapper} args.image
1457+
* @param {import('../containers/imageContainer.js').ImageWrapper} [args.image]
14581458
* @param {dims} [args.pageDims] - Dimensions of page.
14591459
* @param {?number} [args.angle=0] - Angle of page.
14601460
* @param {("proof" | "invis" | "ebook" | "eval")} [args.displayMode='proof'] - Display mode.
@@ -1474,8 +1474,10 @@ export const renderPageStaticImp = async ({
14741474
const canvas = await ca.createCanvas(dims.width, dims.height);
14751475
const ctx = /** @type {OffscreenCanvasRenderingContext2D} */ (/** @type {unknown} */ (canvas.getContext('2d')));
14761476

1477-
const imageBit = await ca.getImageBitmap(image.src);
1478-
if (image) ctx.drawImage(imageBit, 0, 0);
1477+
if (image) {
1478+
const imageBit = await ca.getImageBitmap(image.src);
1479+
ctx.drawImage(imageBit, 0, 0);
1480+
}
14791481

14801482
angle = angle ?? 0;
14811483

@@ -1485,7 +1487,7 @@ export const renderPageStaticImp = async ({
14851487
const cosAngle = Math.cos(angle * (Math.PI / 180));
14861488

14871489
for (const lineObj of page.lines) {
1488-
const angleAdjLine = image.rotated ? ocr.calcLineStartAngleAdj(lineObj) : { x: 0, y: 0 };
1490+
const angleAdjLine = image?.rotated ? ocr.calcLineStartAngleAdj(lineObj) : { x: 0, y: 0 };
14891491

14901492
const baselineY = lineObj.bbox.bottom + lineObj.baseline[1] + angleAdjLine.y;
14911493
const lineLeftAdj = lineObj.bbox.left + angleAdjLine.x;
@@ -1505,7 +1507,7 @@ export const renderPageStaticImp = async ({
15051507

15061508
ctx.fillStyle = fill;
15071509

1508-
const angleAdjWord = wordObj.sup ? ocr.calcWordAngleAdj(wordObj) : { x: 0, y: 0 };
1510+
const angleAdjWord = wordObj.style.sup ? ocr.calcWordAngleAdj(wordObj) : { x: 0, y: 0 };
15091511

15101512
const wordMetrics = calcWordMetrics(wordObj);
15111513
const advanceArr = wordMetrics.advanceArr;

0 commit comments

Comments
 (0)