@inshapardaz/likhari-converters 0.1.46 → 0.1.47

package/dist/convert/index.d.ts

@@ -0,0 +1,26 @@
+import type { SerializedEditorState } from 'lexical';
+import { type FormatRegistry } from '../registry.js';
+import type { ConverterContext, FormatId } from '../types.js';
+export interface ConvertContentOptions {
+    /** The format `content` is in. */
+    inputFormat: FormatId;
+    /** The format to write. */
+    outputFormat: FormatId;
+    /** Passed to the parse and the serialize (image URL resolver and the like). */
+    context?: ConverterContext;
+    /** Defaults to the registry with all four built-in formats. */
+    registry?: FormatRegistry;
+}
+/** Reads `content` and returns it as Lexical state. */
+export declare function convertContentToState(content: string, options: Omit<ConvertContentOptions, 'outputFormat'>): SerializedEditorState;
+/**
+ * Converts a document from one format to another through the editor's own parsers and
+ * serializers, so the result is exactly what the editor would save in that format. Needs no
+ * editor or DOM for Markdown, Lexical JSON and plain text (HTML input uses `DOMParser`).
+ *
+ * Going to a format that holds less than the input (rich text to plain text, a table to
+ * Markdown, ...) degrades the way the format's fidelity matrix says; see
+ * `docs/lexical-editor-spec.md` and `docs/markdown-dialect.md`.
+ */
+export declare function convertContent(content: string, options: ConvertContentOptions): string;
+//# sourceMappingURL=index.d.ts.map

package/dist/convert/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/convert/index.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,qBAAqB,EAAE,MAAM,SAAS,CAAC;AACrD,OAAO,EAAyB,KAAK,cAAc,EAAE,MAAM,aAAa,CAAC;AACzE,OAAO,KAAK,EAAE,gBAAgB,EAAE,QAAQ,EAAE,MAAM,UAAU,CAAC;AAE3D,MAAM,WAAW,qBAAqB;IACpC,kCAAkC;IAClC,WAAW,EAAE,QAAQ,CAAC;IACtB,2BAA2B;IAC3B,YAAY,EAAE,QAAQ,CAAC;IACvB,+EAA+E;IAC/E,OAAO,CAAC,EAAE,gBAAgB,CAAC;IAC3B,+DAA+D;IAC/D,QAAQ,CAAC,EAAE,cAAc,CAAC;CAC3B;AAED,uDAAuD;AACvD,wBAAgB,qBAAqB,CAAC,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,IAAI,CAAC,qBAAqB,EAAE,cAAc,CAAC,GAAG,qBAAqB,CAQlI;AAED;;;;;;;;GAQG;AACH,wBAAgB,cAAc,CAAC,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,qBAAqB,GAAG,MAAM,CAGtF"}

package/dist/convert/index.js

@@ -0,0 +1,25 @@
+import { defaultFormatRegistry } from '../registry.js';
+/** Reads `content` and returns it as Lexical state. */
+export function convertContentToState(content, options) {
+    const { inputFormat, context, registry = defaultFormatRegistry } = options;
+    try {
+        return registry.parse(inputFormat, content, context);
+    }
+    catch (error) {
+        const reason = error instanceof Error ? error.message : String(error);
+        throw new Error(`convertContent: the input is not valid ${inputFormat}: ${reason}`);
+    }
+}
+/**
+ * Converts a document from one format to another through the editor's own parsers and
+ * serializers, so the result is exactly what the editor would save in that format. Needs no
+ * editor or DOM for Markdown, Lexical JSON and plain text (HTML input uses `DOMParser`).
+ *
+ * Going to a format that holds less than the input (rich text to plain text, a table to
+ * Markdown, ...) degrades the way the format's fidelity matrix says; see
+ * `docs/lexical-editor-spec.md` and `docs/markdown-dialect.md`.
+ */
+export function convertContent(content, options) {
+    const registry = options.registry ?? defaultFormatRegistry;
+    return registry.serialize(options.outputFormat, convertContentToState(content, options), options.context);
+}

package/dist/index.d.ts

@@ -4,4 +4,6 @@ export { plainTextConverter } from './plain-text/index.js';
 export { lexicalJsonConverter } from './lexical-json/index.js';
 export { markdownConverter } from './markdown/index.js';
 export { htmlConverter } from './html/index.js';
+export { mergePages, mergePagesToState, type MergePagesOptions, type PageInput } from './merge/index.js';
+export { convertContent, convertContentToState, type ConvertContentOptions } from './convert/index.js';
 //# sourceMappingURL=index.d.ts.map

package/dist/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,cAAc,SAAS,CAAC;AACxB,cAAc,YAAY,CAAC;AAC3B,OAAO,EAAE,kBAAkB,EAAE,MAAM,cAAc,CAAC;AAClD,OAAO,EAAE,oBAAoB,EAAE,MAAM,gBAAgB,CAAC;AACtD,OAAO,EAAE,iBAAiB,EAAE,MAAM,YAAY,CAAC;AAC/C,OAAO,EAAE,aAAa,EAAE,MAAM,QAAQ,CAAC"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,cAAc,SAAS,CAAC;AACxB,cAAc,YAAY,CAAC;AAC3B,OAAO,EAAE,kBAAkB,EAAE,MAAM,cAAc,CAAC;AAClD,OAAO,EAAE,oBAAoB,EAAE,MAAM,gBAAgB,CAAC;AACtD,OAAO,EAAE,iBAAiB,EAAE,MAAM,YAAY,CAAC;AAC/C,OAAO,EAAE,aAAa,EAAE,MAAM,QAAQ,CAAC;AACvC,OAAO,EAAE,UAAU,EAAE,iBAAiB,EAAE,KAAK,iBAAiB,EAAE,KAAK,SAAS,EAAE,MAAM,SAAS,CAAC;AAChG,OAAO,EAAE,cAAc,EAAE,qBAAqB,EAAE,KAAK,qBAAqB,EAAE,MAAM,WAAW,CAAC"}

package/dist/index.js

@@ -4,3 +4,5 @@ export { plainTextConverter } from './plain-text/index.js';
 export { lexicalJsonConverter } from './lexical-json/index.js';
 export { markdownConverter } from './markdown/index.js';
 export { htmlConverter } from './html/index.js';
+export { mergePages, mergePagesToState } from './merge/index.js';
+export { convertContent, convertContentToState } from './convert/index.js';

package/dist/markdown/parse.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"parse.d.ts","sourceRoot":"","sources":["../../src/markdown/parse.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAE,qBAAqB,EAAE,MAAM,SAAS,CAAC;AACrD,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,UAAU,CAAC;AAmUjD,wBAAgB,aAAa,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,CAAC,EAAE,gBAAgB,GAAG,qBAAqB,CAO3F"}
+{"version":3,"file":"parse.d.ts","sourceRoot":"","sources":["../../src/markdown/parse.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAE,qBAAqB,EAAE,MAAM,SAAS,CAAC;AACrD,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,UAAU,CAAC;AAoUjD,wBAAgB,aAAa,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,CAAC,EAAE,gBAAgB,GAAG,qBAAqB,CAO3F"}

package/dist/markdown/parse.js

@@ -264,6 +264,8 @@ function blocks(nodes, extra = {}) {
             case 'leafDirective': {
                 if (node.name === 'pagebreak')
                     out.push({ type: 'page-break', version: 1 });
+                else if (node.name === 'blank')
+                    out.push(paragraph());
                 else if (node.name === 'figure') {
                     const attrs = node.attributes ?? {};
                     const src = safeImageSrc(attrs.src ?? '');
@@ -302,7 +304,7 @@ function restoreUnknownDirectives(node, source) {
     node.children = node.children.map((child) => {
         const c = child;
         const unknownText = c.type === 'textDirective' && !KNOWN_TEXT_DIRECTIVES.has(c.name ?? '');
-        const unknownLeaf = c.type === 'leafDirective' && c.name !== 'pagebreak' && c.name !== 'figure';
+        const unknownLeaf = c.type === 'leafDirective' && c.name !== 'pagebreak' && c.name !== 'blank' && c.name !== 'figure';
         if ((unknownText || unknownLeaf) && c.position?.start.offset !== undefined && c.position.end.offset !== undefined) {
             const raw = source.slice(c.position.start.offset, c.position.end.offset);
             return unknownLeaf ? { type: 'paragraph', children: [{ type: 'text', value: raw }] } : { type: 'text', value: raw };

package/dist/markdown/serialize.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"serialize.d.ts","sourceRoot":"","sources":["../../src/markdown/serialize.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAE,qBAAqB,EAAE,MAAM,SAAS,CAAC;AACrD,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,UAAU,CAAC;AAKjD,OAAO,EAAa,KAAK,KAAK,EAAe,MAAM,SAAS,CAAC;AA6C7D,2GAA2G;AAC3G,wBAAgB,YAAY,CAAC,KAAK,EAAE,MAAM,GAAG,KAAK,CAWjD;AAqHD,wFAAwF;AACxF,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,KAAK,GAAG,KAAK,GAAG,IAAI,CAMhE;AAuJD,wBAAgB,iBAAiB,CAAC,KAAK,EAAE,qBAAqB,EAAE,GAAG,CAAC,EAAE,gBAAgB,GAAG,MAAM,CAM9F"}
+{"version":3,"file":"serialize.d.ts","sourceRoot":"","sources":["../../src/markdown/serialize.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAE,qBAAqB,EAAE,MAAM,SAAS,CAAC;AACrD,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,UAAU,CAAC;AAKjD,OAAO,EAAa,KAAK,KAAK,EAAe,MAAM,SAAS,CAAC;AA6C7D,2GAA2G;AAC3G,wBAAgB,YAAY,CAAC,KAAK,EAAE,MAAM,GAAG,KAAK,CAWjD;AAqHD,wFAAwF;AACxF,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,KAAK,GAAG,KAAK,GAAG,IAAI,CAMhE;AAuJD,wBAAgB,iBAAiB,CAAC,KAAK,EAAE,qBAAqB,EAAE,GAAG,CAAC,EAAE,gBAAgB,GAAG,MAAM,CAc9F"}

package/dist/markdown/serialize.js

@@ -2,7 +2,7 @@ import { unified } from 'unified';
 import remarkStringify from 'remark-stringify';
 import remarkGfm from 'remark-gfm';
 import remarkDirective from 'remark-directive';
-import { FORMAT_BOLD, FORMAT_CODE, FORMAT_HIGHLIGHT, FORMAT_ITALIC, FORMAT_STRIKETHROUGH, FORMAT_SUBSCRIPT, FORMAT_SUPERSCRIPT, FORMAT_UNDERLINE, rootChildren, } from '../shared/serialized.js';
+import { FORMAT_BOLD, FORMAT_CODE, FORMAT_HIGHLIGHT, FORMAT_ITALIC, FORMAT_STRIKETHROUGH, FORMAT_SUBSCRIPT, FORMAT_SUPERSCRIPT, FORMAT_UNDERLINE, isEmptyParagraph, rootChildren, } from '../shared/serialized.js';
 import { directive } from './mdast.js';
 import { layoutOrDefault, scaleOrDefault } from '../shared/poetry.js';
 import { nodeToHtml, setFootnoteOrder } from '../html/serialize.js';
@@ -333,7 +333,14 @@ function blockToMd(node, ctx) {
 export function serializeMarkdown(state, ctx) {
     const blocks = rootChildren(state);
     setFootnoteOrder(blocks);
-    const tree = { type: 'root', children: blocks.flatMap((n) => blockToMd(n, ctx)) };
+    // Markdown has no empty paragraph, so a run of them at the end of the document (which marks where a
+    // page's last paragraph ends, see mergePages) is written as `::blank` lines to survive a save.
+    let end = blocks.length;
+    while (end > 0 && (isEmptyParagraph(blocks[end - 1]) || blocks[end - 1].type === 'footnote-list'))
+        end -= 1;
+    // A document that is nothing but empty paragraphs is still just empty (end === 0).
+    const children = blocks.flatMap((n, i) => end > 0 && i >= end && isEmptyParagraph(n) ? [directive('leafDirective', 'blank', {})] : blockToMd(n, ctx));
+    const tree = { type: 'root', children };
     // remark-stringify consumes mdast; our MdNode is a structural subset of it.
     return String(processor.stringify(tree)).replace(/\n+$/, '\n').replace(/^\n+$/, '');
 }

package/dist/merge/continuation.d.ts

@@ -0,0 +1,23 @@
+import { type SNode } from '../shared/serialized.js';
+/**
+ * Whether a block's text stops mid-sentence: it does not end in a sentence mark (once any
+ * closing quote or bracket is set aside). A comma, colon, semicolon, dash or a plain word all
+ * count as unfinished. A block with no text at all counts as finished.
+ */
+export declare function endsMidSentence(block: SNode): boolean;
+/**
+ * Whether two blocks could be one paragraph: both are plain paragraphs, the second is not
+ * indented (an indent starts a new paragraph) and has text, and they agree on text direction
+ * and alignment. A direction left unset on either side is compatible with anything.
+ */
+export declare function canJoinParagraphs(previous: SNode | undefined, next: SNode | undefined): boolean;
+/**
+ * Whether a page's text ends with an empty line, which marks the end of a paragraph: a trailing
+ * empty paragraph in the parsed page (Markdown writes one as `::blank`). Markdown written by hand
+ * may instead just end with a blank line (`\n\n`, not only a final newline), which the parser
+ * would drop, so for Markdown the raw text is checked too.
+ */
+export declare function endsWithBlankLine(rawFormat: string, rawText: string, blocks: SNode[]): boolean;
+/** Appends `next`'s content to `previous` (which is changed), separated by one space unless one is already there. */
+export declare function joinParagraphs(previous: SNode, next: SNode): void;
+//# sourceMappingURL=continuation.d.ts.map

package/dist/merge/continuation.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"continuation.d.ts","sourceRoot":"","sources":["../../src/merge/continuation.ts"],"names":[],"mappings":"AAAA,OAAO,EAA8B,KAAK,KAAK,EAAE,MAAM,sBAAsB,CAAC;AAc9E;;;;GAIG;AACH,wBAAgB,eAAe,CAAC,KAAK,EAAE,KAAK,GAAG,OAAO,CAGrD;AAKD;;;;GAIG;AACH,wBAAgB,iBAAiB,CAAC,QAAQ,EAAE,KAAK,GAAG,SAAS,EAAE,IAAI,EAAE,KAAK,GAAG,SAAS,GAAG,OAAO,CAM/F;AAED;;;;;GAKG;AACH,wBAAgB,iBAAiB,CAAC,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,GAAG,OAAO,CAG9F;AAKD,qHAAqH;AACrH,wBAAgB,cAAc,CAAC,QAAQ,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,GAAG,IAAI,CAOjE"}

package/dist/merge/continuation.js

@@ -0,0 +1,63 @@
+import { isEmptyParagraph, textNode } from '../shared/serialized.js';
+/** Marks that end a sentence, in Urdu and Latin script: ۔ ؟ ! ? . … and the CJK full stop. */
+const SENTENCE_END = /[۔؟!?.…。]/;
+/** Closing quotes and brackets that may follow the mark and still end the sentence: ” ’ " ' ) ] » */
+const CLOSERS = /[”’"')\]»]+$/;
+/** The text of a block, from its text nodes only. */
+function plainText(node) {
+    if (node.type === 'text')
+        return String(node.text ?? '');
+    if (node.type === 'linebreak')
+        return '\n';
+    return (node.children ?? []).map(plainText).join('');
+}
+/**
+ * Whether a block's text stops mid-sentence: it does not end in a sentence mark (once any
+ * closing quote or bracket is set aside). A comma, colon, semicolon, dash or a plain word all
+ * count as unfinished. A block with no text at all counts as finished.
+ */
+export function endsMidSentence(block) {
+    const text = plainText(block).trimEnd().replace(CLOSERS, '').trimEnd();
+    return text !== '' && !SENTENCE_END.test(text.charAt(text.length - 1));
+}
+/** The alignment a paragraph effectively has: '' and 'start' are the same thing. */
+const alignment = (block) => (block.format === '' || block.format == null ? 'start' : String(block.format));
+/**
+ * Whether two blocks could be one paragraph: both are plain paragraphs, the second is not
+ * indented (an indent starts a new paragraph) and has text, and they agree on text direction
+ * and alignment. A direction left unset on either side is compatible with anything.
+ */
+export function canJoinParagraphs(previous, next) {
+    if (!previous || !next || previous.type !== 'paragraph' || next.type !== 'paragraph')
+        return false;
+    if (plainText(next).trim() === '' || plainText(previous).trim() === '')
+        return false;
+    if (Number(next.indent ?? 0) > 0)
+        return false;
+    if (previous.direction && next.direction && previous.direction !== next.direction)
+        return false;
+    return alignment(previous) === alignment(next);
+}
+/**
+ * Whether a page's text ends with an empty line, which marks the end of a paragraph: a trailing
+ * empty paragraph in the parsed page (Markdown writes one as `::blank`). Markdown written by hand
+ * may instead just end with a blank line (`\n\n`, not only a final newline), which the parser
+ * would drop, so for Markdown the raw text is checked too.
+ */
+export function endsWithBlankLine(rawFormat, rawText, blocks) {
+    if (isEmptyParagraph(blocks[blocks.length - 1]))
+        return true;
+    return rawFormat === 'markdown' && /(\r?\n[ \t]*){2,}$/.test(rawText);
+}
+const startsWithSpace = (children) => /^\s/.test(plainText({ type: 'x', children: children.slice(0, 1) }));
+const endsWithSpace = (children) => /\s$/.test(plainText({ type: 'x', children: children.slice(-1) }));
+/** Appends `next`'s content to `previous` (which is changed), separated by one space unless one is already there. */
+export function joinParagraphs(previous, next) {
+    const head = previous.children ?? [];
+    const tail = next.children ?? [];
+    const needsSpace = head.length > 0 && tail.length > 0 && !endsWithSpace(head) && !startsWithSpace(tail);
+    previous.children = [...head, ...(needsSpace ? [textNode(' ')] : []), ...tail];
+    // The merged paragraph keeps the first one's direction unless it had none.
+    if (!previous.direction && next.direction)
+        previous.direction = next.direction;
+}

package/dist/merge/index.d.ts

@@ -0,0 +1,51 @@
+import type { SerializedEditorState } from 'lexical';
+import { type FormatRegistry } from '../registry.js';
+import type { ConverterContext, FormatId } from '../types.js';
+/** One page: its text, plus its format when that differs from `MergePagesOptions.inputFormat`. */
+export type PageInput = string | {
+    content: string;
+    format?: FormatId;
+};
+export interface MergePagesOptions {
+    /** The format of every page given as a bare string. Defaults to 'markdown'. */
+    inputFormat?: FormatId;
+    /** The format of the combined document. */
+    outputFormat: FormatId;
+    /** Leave out pages with no content: nothing, whitespace, or only empty paragraphs. Off by default. */
+    skipBlankPages?: boolean;
+    /**
+     * Join pages that carry on the previous page's last paragraph into one paragraph, instead
+     * of breaking the page. Off by default. Both modes only join when the next page opens with an
+     * unindented paragraph of the same direction and alignment after a plain paragraph; headings,
+     * lists, quotes, tables and blank pages never join.
+     *
+     * - `'blank-line'`: a page that ends with an empty line or empty paragraph ends its paragraph,
+     *   so the next page starts a new one; a page that does not, continues. The marker is consumed,
+     *   so it does not appear in the output. Markdown is read from the raw text (it must end with a
+     *   blank line, `\n\n`), because Markdown has no empty paragraph; see the README.
+     * - `'punctuation'`: a guess from the text. The page continues when its last paragraph does
+     *   not end with `۔ ؟ ! ? . …` (closing quotes and brackets aside).
+     *
+     * A page break cannot sit inside a paragraph, so none is inserted where pages are joined.
+     */
+    joinParagraphs?: 'blank-line' | 'punctuation';
+    /** Passed to every parse and to the final serialize (image URL resolver and the like). */
+    context?: ConverterContext;
+    /** Defaults to the registry with all four built-in formats. */
+    registry?: FormatRegistry;
+}
+/** Parses each page and returns the combined document as Lexical state, without serializing it. */
+export declare function mergePagesToState(pages: PageInput[], options?: Omit<MergePagesOptions, 'outputFormat'>): SerializedEditorState;
+/**
+ * Merges pages (Markdown, Lexical JSON, or any other supported format) into one document,
+ * with a page break between each pair of pages, in the requested output format. Needs no
+ * editor or DOM, so it can run in a server, a worker or a build script.
+ *
+ * With `skipBlankPages`, blank pages are left out and no break is added for them.
+ *
+ * Footnote ids are only unique within a page, so a page whose footnote ids clash with an
+ * earlier page's gets them renamed (`1` becomes `1-2`), in references and notes alike; the
+ * notes of all pages end up in one list at the end of the document.
+ */
+export declare function mergePages(pages: PageInput[], options: MergePagesOptions): string;
+//# sourceMappingURL=index.d.ts.map

package/dist/merge/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/merge/index.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,qBAAqB,EAAE,MAAM,SAAS,CAAC;AACrD,OAAO,EAAyB,KAAK,cAAc,EAAE,MAAM,aAAa,CAAC;AACzE,OAAO,KAAK,EAAE,gBAAgB,EAAE,QAAQ,EAAE,MAAM,UAAU,CAAC;AAI3D,kGAAkG;AAClG,MAAM,MAAM,SAAS,GAAG,MAAM,GAAG;IAAE,OAAO,EAAE,MAAM,CAAC;IAAC,MAAM,CAAC,EAAE,QAAQ,CAAA;CAAE,CAAC;AAExE,MAAM,WAAW,iBAAiB;IAChC,+EAA+E;IAC/E,WAAW,CAAC,EAAE,QAAQ,CAAC;IACvB,2CAA2C;IAC3C,YAAY,EAAE,QAAQ,CAAC;IACvB,sGAAsG;IACtG,cAAc,CAAC,EAAE,OAAO,CAAC;IACzB;;;;;;;;;;;;;;OAcG;IACH,cAAc,CAAC,EAAE,YAAY,GAAG,aAAa,CAAC;IAC9C,0FAA0F;IAC1F,OAAO,CAAC,EAAE,gBAAgB,CAAC;IAC3B,+DAA+D;IAC/D,QAAQ,CAAC,EAAE,cAAc,CAAC;CAC3B;AAED,mGAAmG;AACnG,wBAAgB,iBAAiB,CAAC,KAAK,EAAE,SAAS,EAAE,EAAE,OAAO,GAAE,IAAI,CAAC,iBAAiB,EAAE,cAAc,CAAM,GAAG,qBAAqB,CA8ClI;AAED;;;;;;;;;;GAUG;AACH,wBAAgB,UAAU,CAAC,KAAK,EAAE,SAAS,EAAE,EAAE,OAAO,EAAE,iBAAiB,GAAG,MAAM,CAGjF"}

package/dist/merge/index.js

@@ -0,0 +1,95 @@
+import { defaultFormatRegistry } from '../registry.js';
+import { elementBase, isEmptyParagraph, makeState, rootChildren } from '../shared/serialized.js';
+import { canJoinParagraphs, endsMidSentence, endsWithBlankLine, joinParagraphs } from './continuation.js';
+/** Parses each page and returns the combined document as Lexical state, without serializing it. */
+export function mergePagesToState(pages, options = {}) {
+    const { inputFormat = 'markdown', skipBlankPages = false, joinParagraphs: joinMode, context, registry = defaultFormatRegistry } = options;
+    const body = [];
+    const footnotes = [];
+    const takenFootnoteIds = new Set();
+    let kept = 0;
+    let previousEndedParagraph = false;
+    pages.forEach((page, index) => {
+        const { content, format = inputFormat } = typeof page === 'string' ? { content: page } : page;
+        let blocks;
+        try {
+            blocks = structuredClone(rootChildren(registry.parse(format, content, context)));
+        }
+        catch (error) {
+            const reason = error instanceof Error ? error.message : String(error);
+            throw new Error(`mergePages: page ${index + 1} is not valid ${format}: ${reason}`);
+        }
+        if (skipBlankPages && blocks.every(isBlank))
+            return;
+        // Each page brings its own footnote list; they are combined into one at the end.
+        const pageLists = blocks.filter((block) => block.type === 'footnote-list');
+        const pageBody = blocks.filter((block) => block.type !== 'footnote-list');
+        uniquifyFootnoteIds([...pageBody, ...pageLists], takenFootnoteIds);
+        for (const list of pageLists)
+            footnotes.push(...(list.children ?? []));
+        const previous = body[body.length - 1];
+        let continues = false;
+        if (joinMode === 'punctuation')
+            continues = !!previous && endsMidSentence(previous);
+        if (joinMode === 'blank-line') {
+            continues = !previousEndedParagraph;
+            previousEndedParagraph = endsWithBlankLine(format, content, [...pageBody]);
+            // The empty line only marks the end of the paragraph; it is not content to keep.
+            if (pageBody.some((block) => !isEmptyParagraph(block))) {
+                while (isEmptyParagraph(pageBody[pageBody.length - 1]))
+                    pageBody.pop();
+            }
+        }
+        const joins = kept > 0 && continues && canJoinParagraphs(previous, pageBody[0]);
+        if (joins)
+            joinParagraphs(previous, pageBody.shift());
+        else if (kept > 0)
+            body.push({ type: 'page-break', version: 1 });
+        kept += 1;
+        body.push(...pageBody);
+    });
+    return makeState(footnotes.length > 0 ? [...body, elementBase('footnote-list', footnotes)] : body);
+}
+/**
+ * Merges pages (Markdown, Lexical JSON, or any other supported format) into one document,
+ * with a page break between each pair of pages, in the requested output format. Needs no
+ * editor or DOM, so it can run in a server, a worker or a build script.
+ *
+ * With `skipBlankPages`, blank pages are left out and no break is added for them.
+ *
+ * Footnote ids are only unique within a page, so a page whose footnote ids clash with an
+ * earlier page's gets them renamed (`1` becomes `1-2`), in references and notes alike; the
+ * notes of all pages end up in one list at the end of the document.
+ */
+export function mergePages(pages, options) {
+    const registry = options.registry ?? defaultFormatRegistry;
+    return registry.serialize(options.outputFormat, mergePagesToState(pages, options), options.context);
+}
+/** A block with nothing in it: a paragraph holding only whitespace and line breaks. Anything else (an image, a rule, a table, a break) is content. */
+function isBlank(block) {
+    if (block.type !== 'paragraph')
+        return false;
+    return (block.children ?? []).every((child) => (child.type === 'text' ? String(child.text ?? '').trim() === '' : child.type === 'linebreak'));
+}
+/** Renames this page's footnote ids that an earlier page already uses, then claims all its ids. */
+function uniquifyFootnoteIds(blocks, taken) {
+    const nodes = [];
+    const collect = (node) => {
+        if (typeof node.footnoteId === 'string')
+            nodes.push(node);
+        (node.children ?? []).forEach(collect);
+    };
+    blocks.forEach(collect);
+    const renamed = new Map();
+    for (const node of nodes) {
+        const id = node.footnoteId;
+        if (!renamed.has(id)) {
+            let next = id;
+            for (let n = 2; taken.has(next); n += 1)
+                next = `${id}-${n}`;
+            renamed.set(id, next);
+        }
+        node.footnoteId = renamed.get(id);
+    }
+    renamed.forEach((id) => taken.add(id));
+}

package/dist/shared/serialized.d.ts

@@ -33,4 +33,6 @@ export declare function isInlineNode(node: SNode): boolean;
  * order, among all such nodes. Walks the whole tree once so every converter
  * numbers references and their matching footnote-list entries identically. */
 export declare function collectFootnoteOrder(blocks: SNode[]): Map<string, number>;
+/** A paragraph with nothing in it: no children, or only whitespace and line breaks. */
+export declare function isEmptyParagraph(node: SNode | undefined): boolean;
 //# sourceMappingURL=serialized.d.ts.map

package/dist/shared/serialized.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"serialized.d.ts","sourceRoot":"","sources":["../../src/shared/serialized.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,qBAAqB,EAAE,MAAM,SAAS,CAAC;AAErD;;;;GAIG;AACH,MAAM,WAAW,KAAK;IACpB,IAAI,EAAE,MAAM,CAAC;IACb,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,QAAQ,CAAC,EAAE,KAAK,EAAE,CAAC;IACnB,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAAC;CACxB;AAED,eAAO,MAAM,WAAW,IAAI,CAAC;AAC7B,eAAO,MAAM,aAAa,IAAI,CAAC;AAC/B,eAAO,MAAM,oBAAoB,IAAI,CAAC;AACtC,eAAO,MAAM,gBAAgB,IAAI,CAAC;AAClC,eAAO,MAAM,WAAW,KAAK,CAAC;AAC9B,eAAO,MAAM,gBAAgB,KAAK,CAAC;AACnC,eAAO,MAAM,kBAAkB,KAAK,CAAC;AACrC,eAAO,MAAM,gBAAgB,MAAM,CAAC;AAEpC,6DAA6D;AAC7D,eAAO,MAAM,SAAS,KAAK,CAAC;AAE5B,MAAM,MAAM,SAAS,GAAG,KAAK,GAAG,KAAK,GAAG,IAAI,CAAC;AAE7C,wBAAgB,YAAY,CAAC,KAAK,EAAE,qBAAqB,GAAG,KAAK,EAAE,CAElE;AAED,wBAAgB,QAAQ,CAAC,IAAI,EAAE,MAAM,EAAE,MAAM,SAAI,EAAE,KAAK,SAAK,GAAG,KAAK,CAEpE;AAED,wBAAgB,WAAW,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,KAAK,EAAE,EAAE,KAAK,GAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAM,GAAG,KAAK,CAEvG;AAED,wBAAgB,SAAS,CAAC,QAAQ,GAAE,KAAK,EAAO,EAAE,KAAK,GAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAM,GAAG,KAAK,CAE5F;AAED,wBAAgB,SAAS,CAAC,MAAM,EAAE,KAAK,EAAE,GAAG,qBAAqB,CAKhE;AAED,qDAAqD;AACrD,wBAAgB,YAAY,CAAC,IAAI,EAAE,KAAK,GAAG,OAAO,CAEjD;AAED;;;8EAG8E;AAC9E,wBAAgB,oBAAoB,CAAC,MAAM,EAAE,KAAK,EAAE,GAAG,GAAG,CAAC,MAAM,EAAE,MAAM,CAAC,CAUzE"}
+{"version":3,"file":"serialized.d.ts","sourceRoot":"","sources":["../../src/shared/serialized.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,qBAAqB,EAAE,MAAM,SAAS,CAAC;AAErD;;;;GAIG;AACH,MAAM,WAAW,KAAK;IACpB,IAAI,EAAE,MAAM,CAAC;IACb,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,QAAQ,CAAC,EAAE,KAAK,EAAE,CAAC;IACnB,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAAC;CACxB;AAED,eAAO,MAAM,WAAW,IAAI,CAAC;AAC7B,eAAO,MAAM,aAAa,IAAI,CAAC;AAC/B,eAAO,MAAM,oBAAoB,IAAI,CAAC;AACtC,eAAO,MAAM,gBAAgB,IAAI,CAAC;AAClC,eAAO,MAAM,WAAW,KAAK,CAAC;AAC9B,eAAO,MAAM,gBAAgB,KAAK,CAAC;AACnC,eAAO,MAAM,kBAAkB,KAAK,CAAC;AACrC,eAAO,MAAM,gBAAgB,MAAM,CAAC;AAEpC,6DAA6D;AAC7D,eAAO,MAAM,SAAS,KAAK,CAAC;AAE5B,MAAM,MAAM,SAAS,GAAG,KAAK,GAAG,KAAK,GAAG,IAAI,CAAC;AAE7C,wBAAgB,YAAY,CAAC,KAAK,EAAE,qBAAqB,GAAG,KAAK,EAAE,CAElE;AAED,wBAAgB,QAAQ,CAAC,IAAI,EAAE,MAAM,EAAE,MAAM,SAAI,EAAE,KAAK,SAAK,GAAG,KAAK,CAEpE;AAED,wBAAgB,WAAW,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,KAAK,EAAE,EAAE,KAAK,GAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAM,GAAG,KAAK,CAEvG;AAED,wBAAgB,SAAS,CAAC,QAAQ,GAAE,KAAK,EAAO,EAAE,KAAK,GAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAM,GAAG,KAAK,CAE5F;AAED,wBAAgB,SAAS,CAAC,MAAM,EAAE,KAAK,EAAE,GAAG,qBAAqB,CAKhE;AAED,qDAAqD;AACrD,wBAAgB,YAAY,CAAC,IAAI,EAAE,KAAK,GAAG,OAAO,CAEjD;AAED;;;8EAG8E;AAC9E,wBAAgB,oBAAoB,CAAC,MAAM,EAAE,KAAK,EAAE,GAAG,GAAG,CAAC,MAAM,EAAE,MAAM,CAAC,CAUzE;AAED,uFAAuF;AACvF,wBAAgB,gBAAgB,CAAC,IAAI,EAAE,KAAK,GAAG,SAAS,GAAG,OAAO,CAMjE"}

package/dist/shared/serialized.js

@@ -45,3 +45,9 @@ export function collectFootnoteOrder(blocks) {
     blocks.forEach(visit);
     return order;
 }
+/** A paragraph with nothing in it: no children, or only whitespace and line breaks. */
+export function isEmptyParagraph(node) {
+    return (!!node &&
+        node.type === 'paragraph' &&
+        (node.children ?? []).every((child) => (child.type === 'text' ? String(child.text ?? '').trim() === '' : child.type === 'linebreak')));
+}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@inshapardaz/likhari-converters",
-  "version": "0.1.46",
+  "version": "0.1.47",
   "description": "Format transformers (Markdown/HTML/plain-text/Lexical JSON) against the canonical Lexical EditorState, isolated from UI.",
   "license": "MIT",
   "repository": {