postext 1.28.5 → 1.29.0

package/dist/measure/cjkCompose.js

@@ -35,7 +35,8 @@
 import { createBoundingBox } from '../vdt';
 import { isJapaneseLanguage } from '../locale';
 import { lineIndentAt, lineInsetsAt, lineMeasure, markInsetLines } from './types';
-import { measureInkBox, measureInkExtent, measureTextWidth, normalSpaceWidthFor } from './canvas';
+import { getMeasureLanguage, measureInkBox, measureInkExtent, measureTextWidth, normalSpaceWidthFor, setMeasureLanguage } from './canvas';
+import { spanShapingLanguage } from './spanLanguage';
 import { atomicSpanToken, refLabelTokens, setLabelTabs, emergencySplit, expandSmallCaps, pickSpanFont, scriptMetrics, setsObject, spanScriptFields, stackedScriptPairs, textWidth, tokenSegment, URL_LIKE_RE, urlBreakIndices, } from './rich';
 import { cjkBreakAllowed, cjkClassOf, getCjkLineBreak, isCjkGrapheme, isFullwidthAlnum, isFullwidthDigit, isJapaneseLineBreak, getCjkCircledNumbers, isLabelEndProhibited, isLineEndProhibited, isLineStartProhibited, isWordInnerMark, setCjkCircledNumbers, } from './cjkClasses';
 import { hasCJK } from './cjk';
@@ -169,6 +170,10 @@ function spanLanguageHere(span) {
         lang = d.lang;
     if (lang === undefined)
         return undefined;
+    // In a document shaped in its own language every run that names one
+    // carries it (#764): it was measured in it, and is painted in it.
+    if (getMeasureLanguage() !== undefined)
+        return lang;
     return getMeasureRegion() === 'japan' || isJapaneseLanguage(lang) ? lang : undefined;
 }
 function styleOf(span, fonts, vertical = false) {
@@ -639,230 +644,241 @@ function buildUnits(spans, fonts, letterSpacingPx, vertical = false, route = fal
     // Where each span's units start, for the kanbun marks (#430).
     const spanStarts = [];
     for (let si = 0; si < spans.length; si++) {
-        const span = spans[si];
-        spanStarts.push(units.length);
-        // A ruby base: one unit, sized with its reading once its neighbours
-        // are known (`sizeRubies`).
-        if (span.ruby && span.text.length > 0) {
-            units.push({ ...rubyBaseUnit(span, fonts, letterSpacingPx, vertical), ...(zwsp ? { zwspBefore: true } : {}), ...glue() });
-            zwsp = false;
-            continue;
-        }
-        if (vertical && (span.combineUpright || span.orientation) && span.text.length > 0 && !setsObject(span)) {
-            const first = units.length;
-            orientedUnits(span, styleOf(span, fonts, vertical), letterSpacingPx, zwsp, units);
-            if (units.length > first)
-                Object.assign(units[first], glue());
-            zwsp = false;
-            continue;
-        }
-        // A reference whose label is words (#702): one unit per word and per
-        // space, each carrying the reference; only the first is glued to the
-        // text before it, so a line may break after a space of the label.
-        const whole = refLabelTokens(span, fonts.normal, fonts.bold, fonts.italic, fonts.boldItalic, letterSpacingPx)
-            ?? atomicSpanToken(span, fonts.normal, fonts.bold, fonts.italic, fonts.boldItalic, letterSpacingPx);
-        const atomics = whole === undefined ? [] : Array.isArray(whole) ? whole : [whole];
-        for (const [k, atomic] of atomics.entries()) {
-            const style = styleOf(span, fonts, vertical);
-            if (atomic.kind === 'space') {
+        // A run that names its language is measured in it (#764).
+        const ownLanguage = spanShapingLanguage(spans[si]);
+        const languageBefore = getMeasureLanguage();
+        if (ownLanguage !== undefined)
+            setMeasureLanguage(ownLanguage);
+        try {
+            const span = spans[si];
+            spanStarts.push(units.length);
+            // A ruby base: one unit, sized with its reading once its neighbours
+            // are known (`sizeRubies`).
+            if (span.ruby && span.text.length > 0) {
+                units.push({ ...rubyBaseUnit(span, fonts, letterSpacingPx, vertical), ...(zwsp ? { zwspBefore: true } : {}), ...glue() });
+                zwsp = false;
+                continue;
+            }
+            if (vertical && (span.combineUpright || span.orientation) && span.text.length > 0 && !setsObject(span)) {
+                const first = units.length;
+                orientedUnits(span, styleOf(span, fonts, vertical), letterSpacingPx, zwsp, units);
+                if (units.length > first)
+                    Object.assign(units[first], glue());
+                zwsp = false;
+                continue;
+            }
+            // A reference whose label is words (#702): one unit per word and per
+            // space, each carrying the reference; only the first is glued to the
+            // text before it, so a line may break after a space of the label.
+            const whole = refLabelTokens(span, fonts.normal, fonts.bold, fonts.italic, fonts.boldItalic, letterSpacingPx)
+                ?? atomicSpanToken(span, fonts.normal, fonts.bold, fonts.italic, fonts.boldItalic, letterSpacingPx);
+            const atomics = whole === undefined ? [] : Array.isArray(whole) ? whole : [whole];
+            for (const [k, atomic] of atomics.entries()) {
+                const style = styleOf(span, fonts, vertical);
+                if (atomic.kind === 'space') {
+                    units.push({
+                        kind: 'space',
+                        text: atomic.text,
+                        width: atomic.width,
+                        graphemes: graphemeCount(atomic.text),
+                        first: 'western',
+                        last: 'western',
+                        firstCjk: false,
+                        lastCjk: false,
+                        style,
+                        at: 0,
+                        token: atomic,
+                    });
+                    continue;
+                }
+                const graphemes = graphemesOf(atomic.text);
+                const object = atomic.chip !== undefined || atomic.mathRender !== undefined || atomic.swatch !== undefined;
+                const firstG = graphemes[0] ?? '';
+                const lastG = graphemes[graphemes.length - 1] ?? '';
                 units.push({
-                    kind: 'space',
+                    kind: 'atomic',
                     text: atomic.text,
                     width: atomic.width,
-                    graphemes: graphemeCount(atomic.text),
-                    first: 'western',
-                    last: 'western',
-                    firstCjk: false,
-                    lastCjk: false,
+                    graphemes: graphemes.length,
+                    first: object ? 'ideograph' : cjkClassOf(firstG),
+                    last: object ? 'ideograph' : cjkClassOf(lastG),
+                    firstCjk: object || isCjkGrapheme(firstG),
+                    lastCjk: object || isCjkGrapheme(lastG),
                     style,
                     at: 0,
                     token: atomic,
+                    ...((atomic.refResourceId !== undefined || atomic.footnoteId !== undefined) && k === 0 ? { glueBefore: true } : {}),
+                    ...(zwsp ? { zwspBefore: true } : {}),
+                    ...glue(),
                 });
-                continue;
+                zwsp = false;
             }
-            const graphemes = graphemesOf(atomic.text);
-            const object = atomic.chip !== undefined || atomic.mathRender !== undefined || atomic.swatch !== undefined;
-            const firstG = graphemes[0] ?? '';
-            const lastG = graphemes[graphemes.length - 1] ?? '';
-            units.push({
-                kind: 'atomic',
-                text: atomic.text,
-                width: atomic.width,
-                graphemes: graphemes.length,
-                first: object ? 'ideograph' : cjkClassOf(firstG),
-                last: object ? 'ideograph' : cjkClassOf(lastG),
-                firstCjk: object || isCjkGrapheme(firstG),
-                lastCjk: object || isCjkGrapheme(lastG),
-                style,
-                at: 0,
-                token: atomic,
-                ...((atomic.refResourceId !== undefined || atomic.footnoteId !== undefined) && k === 0 ? { glueBefore: true } : {}),
-                ...(zwsp ? { zwspBefore: true } : {}),
-                ...glue(),
-            });
-            zwsp = false;
-        }
-        if (atomics.length > 0)
-            continue;
-        if (span.text.length === 0)
-            continue;
-        const style = styleOf(span, fonts, vertical);
-        const graphemes = graphemesOf(span.text);
-        const spanFirst = units.length;
-        let run = [];
-        let runAt = 0;
-        let runLink;
-        let space = '';
-        let spaceAt = 0;
-        // A lone mark that may pair with the next one (—, …), by unit index.
-        let pairOpen = -1;
-        let at = 0;
-        // The link a character at `offset` of the span is part of: its span and
-        // range, so two links to one target stay apart.
-        const links = span.links;
-        const linkAt = (offset) => {
-            if (!links || links.length === 0)
-                return undefined;
-            for (let li = 0; li < links.length; li++) {
-                const l = links[li];
-                if (offset >= l.start && offset < l.end)
-                    return `${si}:${li}`;
-            }
-            return undefined;
-        };
-        const push = (u) => {
-            units.push({
-                ...u,
-                style,
-                ...(zwsp ? { zwspBefore: true } : {}),
-                ...(style.script && units.length === spanFirst ? { glueBefore: true } : {}),
-                ...glue(),
-            });
-            zwsp = false;
-        };
-        const flushRun = () => {
-            if (run.length === 0)
-                return;
-            const text = run.join('');
-            const firstG = run[0];
-            const lastG = run[run.length - 1];
-            // Vertical text: a short number at either end stands in one upright
-            // cell (`cjk.uprightDigits`), and the run is Chinese on that side;
-            // tracking follows the cell once.
-            const cells = vertical ? numberCellEdges(run) : undefined;
-            push({
-                kind: 'text',
-                text,
-                width: textWidth(text, style.font, style.smallCaps) + track(vertical && letterSpacingPx !== 0 ? verticalTrackCount(text) : run.length),
-                graphemes: run.length,
-                first: cells?.start ? 'ideograph' : cjkClassOf(firstG),
-                last: cells?.end ? 'ideograph' : cjkClassOf(lastG),
-                firstCjk: cells?.start === true,
-                lastCjk: cells?.end === true,
-                at: runAt,
-                run: true,
-                ...(cells?.start ? { cellStart: true } : {}),
-                ...(cells?.end ? { cellEnd: true } : {}),
-                ...(URL_LIKE_RE.test(text) ? { url: true } : {}),
-                ...(runLink !== undefined ? { link: runLink } : {}),
-            });
-            run = [];
-            pairOpen = -1;
-        };
-        const flushSpace = () => {
-            if (space === '')
-                return;
-            units.push({
-                kind: 'space',
-                text: space,
-                width: measureTextWidth(space, style.font) + track(graphemeCount(space)),
-                graphemes: graphemeCount(space),
-                first: 'western',
-                last: 'western',
-                firstCjk: false,
-                lastCjk: false,
-                style,
-                at: spaceAt,
-            });
-            space = '';
-            pairOpen = -1;
-        };
-        for (let i = 0; i < graphemes.length; i++) {
-            const g = graphemes[i];
-            const gAt = at;
-            at += g.length;
-            if (isBreakingSpace(g[0])) {
-                flushRun();
-                wj = false;
-                if (space === '')
-                    spaceAt = gAt;
-                space += g;
+            if (atomics.length > 0)
                 continue;
-            }
-            flushSpace();
-            if (g === '\u200B') {
-                flushRun();
-                zwsp = true;
-                pairOpen = -1;
+            if (span.text.length === 0)
                 continue;
-            }
-            if (g === '\u2060') {
-                flushRun();
-                wj = true;
+            const style = styleOf(span, fonts, vertical);
+            const graphemes = graphemesOf(span.text);
+            const spanFirst = units.length;
+            let run = [];
+            let runAt = 0;
+            let runLink;
+            let space = '';
+            let spaceAt = 0;
+            // A lone mark that may pair with the next one (—, …), by unit index.
+            let pairOpen = -1;
+            let at = 0;
+            // The link a character at `offset` of the span is part of: its span and
+            // range, so two links to one target stay apart.
+            const links = span.links;
+            const linkAt = (offset) => {
+                if (!links || links.length === 0)
+                    return undefined;
+                for (let li = 0; li < links.length; li++) {
+                    const l = links[li];
+                    if (offset >= l.start && offset < l.end)
+                        return `${si}:${li}`;
+                }
+                return undefined;
+            };
+            const push = (u) => {
+                units.push({
+                    ...u,
+                    style,
+                    ...(zwsp ? { zwspBefore: true } : {}),
+                    ...(style.script && units.length === spanFirst ? { glueBefore: true } : {}),
+                    ...glue(),
+                });
+                zwsp = false;
+            };
+            const flushRun = () => {
+                if (run.length === 0)
+                    return;
+                const text = run.join('');
+                const firstG = run[0];
+                const lastG = run[run.length - 1];
+                // Vertical text: a short number at either end stands in one upright
+                // cell (`cjk.uprightDigits`), and the run is Chinese on that side;
+                // tracking follows the cell once.
+                const cells = vertical ? numberCellEdges(run) : undefined;
+                push({
+                    kind: 'text',
+                    text,
+                    width: textWidth(text, style.font, style.smallCaps) + track(vertical && letterSpacingPx !== 0 ? verticalTrackCount(text) : run.length),
+                    graphemes: run.length,
+                    first: cells?.start ? 'ideograph' : cjkClassOf(firstG),
+                    last: cells?.end ? 'ideograph' : cjkClassOf(lastG),
+                    firstCjk: cells?.start === true,
+                    lastCjk: cells?.end === true,
+                    at: runAt,
+                    run: true,
+                    ...(cells?.start ? { cellStart: true } : {}),
+                    ...(cells?.end ? { cellEnd: true } : {}),
+                    ...(URL_LIKE_RE.test(text) ? { url: true } : {}),
+                    ...(runLink !== undefined ? { link: runLink } : {}),
+                });
+                run = [];
                 pairOpen = -1;
-                continue;
-            }
-            // A soft hyphen: the composer breaks Western words only when they are
-            // wider than the line, so it is left out, and the word stays whole.
-            if (g === '\u00AD')
-                continue;
-            const link = linkAt(gAt);
-            if (!isCjkHere(g, graphemes[i - 1], graphemes[i + 1])) {
-                // A link that starts or ends inside a run cuts it: the two parts
-                // still never part (neither is CJK) and take no space between them.
-                if (run.length > 0 && link !== runLink)
+            };
+            const flushSpace = () => {
+                if (space === '')
+                    return;
+                units.push({
+                    kind: 'space',
+                    text: space,
+                    width: measureTextWidth(space, style.font) + track(graphemeCount(space)),
+                    graphemes: graphemeCount(space),
+                    first: 'western',
+                    last: 'western',
+                    firstCjk: false,
+                    lastCjk: false,
+                    style,
+                    at: spaceAt,
+                });
+                space = '';
+                pairOpen = -1;
+            };
+            for (let i = 0; i < graphemes.length; i++) {
+                const g = graphemes[i];
+                const gAt = at;
+                at += g.length;
+                if (isBreakingSpace(g[0])) {
                     flushRun();
-                if (run.length === 0) {
-                    runAt = gAt;
-                    runLink = link;
+                    wj = false;
+                    if (space === '')
+                        spaceAt = gAt;
+                    space += g;
+                    continue;
                 }
-                run.push(g);
-                continue;
+                flushSpace();
+                if (g === '\u200B') {
+                    flushRun();
+                    zwsp = true;
+                    pairOpen = -1;
+                    continue;
+                }
+                if (g === '\u2060') {
+                    flushRun();
+                    wj = true;
+                    pairOpen = -1;
+                    continue;
+                }
+                // A soft hyphen: the composer breaks Western words only when they are
+                // wider than the line, so it is left out, and the word stays whole.
+                if (g === '\u00AD')
+                    continue;
+                const link = linkAt(gAt);
+                if (!isCjkHere(g, graphemes[i - 1], graphemes[i + 1])) {
+                    // A link that starts or ends inside a run cuts it: the two parts
+                    // still never part (neither is CJK) and take no space between them.
+                    if (run.length > 0 && link !== runLink)
+                        flushRun();
+                    if (run.length === 0) {
+                        runAt = gAt;
+                        runLink = link;
+                    }
+                    run.push(g);
+                    continue;
+                }
+                flushRun();
+                // —— and …… are one unit of two ems; a third mark opens another.
+                if (pairOpen >= 0 && units[pairOpen].text === g && !zwsp && units[pairOpen].link === link) {
+                    const u = units[pairOpen];
+                    u.text += g;
+                    u.width += cellAdvance(g, style.font, vertical, u.first) + track(1);
+                    u.graphemes = 2;
+                    u.first = u.last = g === '\u2026' || g === '\u22EF' ? 'ellipsis' : 'dash';
+                    pairOpen = -1;
+                    continue;
+                }
+                let cls = cjkClassOf(g);
+                // A single em dash (or horizontal bar) joins two words as a connector
+                // (clreq §6.1.1): never at a line start.
+                if (g === '\u2014' || g === '\u2015')
+                    cls = 'connector';
+                push({
+                    kind: 'text',
+                    text: g,
+                    width: (style.smallCaps && !vertical ? textWidth(g, style.font, true) : cellAdvance(g, style.font, vertical, cls)) + track(1),
+                    graphemes: 1,
+                    first: cls,
+                    last: cls,
+                    firstCjk: true,
+                    lastCjk: true,
+                    at: gAt,
+                    ...(link !== undefined ? { link } : {}),
+                });
+                pairOpen = pairable(g) ? units.length - 1 : -1;
             }
             flushRun();
-            // —— and …… are one unit of two ems; a third mark opens another.
-            if (pairOpen >= 0 && units[pairOpen].text === g && !zwsp && units[pairOpen].link === link) {
-                const u = units[pairOpen];
-                u.text += g;
-                u.width += cellAdvance(g, style.font, vertical, u.first) + track(1);
-                u.graphemes = 2;
-                u.first = u.last = g === '\u2026' || g === '\u22EF' ? 'ellipsis' : 'dash';
-                pairOpen = -1;
-                continue;
-            }
-            let cls = cjkClassOf(g);
-            // A single em dash (or horizontal bar) joins two words as a connector
-            // (clreq §6.1.1): never at a line start.
-            if (g === '\u2014' || g === '\u2015')
-                cls = 'connector';
-            push({
-                kind: 'text',
-                text: g,
-                width: (style.smallCaps && !vertical ? textWidth(g, style.font, true) : cellAdvance(g, style.font, vertical, cls)) + track(1),
-                graphemes: 1,
-                first: cls,
-                last: cls,
-                firstCjk: true,
-                lastCjk: true,
-                at: gAt,
-                ...(link !== undefined ? { link } : {}),
-            });
-            pairOpen = pairable(g) ? units.length - 1 : -1;
+            flushSpace();
+            if (units.length === spanFirst + 1 && units[spanFirst].kind === 'text')
+                wholeSpan.add(spanFirst);
+        }
+        finally {
+            if (ownLanguage !== undefined)
+                setMeasureLanguage(languageBefore);
         }
-        flushRun();
-        flushSpace();
-        if (units.length === spanFirst + 1 && units[spanFirst].kind === 'text')
-            wholeSpan.add(spanFirst);
     }
     if (spans.some((s) => s.kunten))
         attachKunten(spans, spanStarts, units);