新版本
This commit is contained in:
commit
680ed62b5b
509 files changed
+54763
No files matched your search
@@ -0,0 +1,580 @@
|
||||
import { PFiles } from "../file.ts";
|
||||
import { PDFParse } from "pdf-parse";
|
||||
import {
|
||||
PDFArray,
|
||||
PDFDocument,
|
||||
PDFHexString,
|
||||
PDFName,
|
||||
PDFNumber,
|
||||
} from "pdf-lib";
|
||||
import "openpgp";
|
||||
import { resolveChromiumPath } from "./utils.ts";
|
||||
import { DocuBuilder } from "./docu_builder.ts";
|
||||
|
||||
type IndexItemLayout = {
|
||||
pageIndex: number;
|
||||
left: number;
|
||||
top: number;
|
||||
width: number;
|
||||
height: number;
|
||||
fontSize: number;
|
||||
};
|
||||
|
||||
const PDF_PAGE_HEIGHT_PT = 842;
|
||||
const PDF_VIEWPORT_WIDTH_PX = 794;
|
||||
const CSS_PX_TO_PT = 72 / 96;
|
||||
|
||||
export class IndexBuilder {
|
||||
indexed_map = new Map<
|
||||
string,
|
||||
{
|
||||
page: number;
|
||||
passage_index: number;
|
||||
column_name: string;
|
||||
}
|
||||
>();
|
||||
total_pages: number = 0;
|
||||
template_dir: string = "";
|
||||
passage_dir: string = "";
|
||||
file_path: string = "";
|
||||
cfg: any;
|
||||
constructor(file_path: string, cfg: any) {
|
||||
if (cfg.journal.build_mode == "flat") {
|
||||
this.template_dir = cfg.journal.from_directory;
|
||||
this.passage_dir = cfg.journal.from_directory;
|
||||
} else {
|
||||
this.template_dir = cfg.journal.template_directory;
|
||||
this.passage_dir = cfg.journal.from_directory;
|
||||
}
|
||||
this.file_path = file_path;
|
||||
this.cfg = cfg;
|
||||
}
|
||||
|
||||
get_index_titles_from_file_name(mode: "detached" | "flat"): {
|
||||
titles: Array<string>;
|
||||
passages_index: Map<string, number>;
|
||||
passages_column_name: Map<string, string>;
|
||||
} {
|
||||
const pfiles = new PFiles(mode, this.passage_dir, this.cfg);
|
||||
const titles: Array<string> = [];
|
||||
const passages_index: Map<string, number> = new Map();
|
||||
const passages_column_name: Map<string, string> = new Map();
|
||||
pfiles.pcolumes.forEach((colume) => {
|
||||
colume.pfiles.forEach((pfile) => {
|
||||
titles.push(pfile.meta.title);
|
||||
passages_index.set(pfile.meta.title, pfile.meta.index);
|
||||
passages_column_name.set(pfile.meta.title, pfile.meta.colume);
|
||||
});
|
||||
});
|
||||
return { titles, passages_index, passages_column_name };
|
||||
}
|
||||
|
||||
async read_pdf() {
|
||||
const parser = new PDFParse({ url: this.file_path });
|
||||
try {
|
||||
const pdfText = await parser.getText();
|
||||
this.total_pages = pdfText.total;
|
||||
return pdfText;
|
||||
} finally {
|
||||
await parser.destroy();
|
||||
}
|
||||
}
|
||||
//修改实现为在每页文本中查找目录标题,支持目录标题跨行的情况。通过维护一个标题游标,逐页扫描PDF文本,尝试匹配当前标题。每页文本被拆分成行,并进行规范化(去除空格和换行)。对于每个标题,尝试将连续的行合并起来进行匹配,如果找到匹配则记录该标题对应的页码和章节索引,并移动到下一个标题。这样可以更准确地处理目录标题在PDF中被拆分成多行的情况。
|
||||
async build_index() {
|
||||
const pdf_obj = await this.read_pdf();
|
||||
const { titles, passages_index, passages_column_name } = this
|
||||
.get_index_titles_from_file_name(
|
||||
this.cfg.journal.build_mode ?? "flat",
|
||||
);
|
||||
// Normalize titles (remove spaces/newlines) and require exact equality.
|
||||
// Matching supports title text split across multiple PDF lines.
|
||||
const normalizedTitles = titles.map((t) => ({
|
||||
orig: t,
|
||||
norm: t.replaceAll("\n", "").replaceAll(" ", ""),
|
||||
}));
|
||||
|
||||
let titleCursor = 0;
|
||||
|
||||
for (let i = 1; i <= this.total_pages; i++) {
|
||||
if (titleCursor >= normalizedTitles.length) {
|
||||
break;
|
||||
}
|
||||
|
||||
const pageTextRaw = pdf_obj.getPageText(i);
|
||||
const lines = pageTextRaw.split(/\r?\n/).map((l: string) => l.trim())
|
||||
.filter(
|
||||
(l: string) => l.length > 0,
|
||||
);
|
||||
|
||||
const normalizedLines = lines.map((line: string) =>
|
||||
line.replaceAll(" ", "")
|
||||
);
|
||||
const tt = normalizedTitles[titleCursor];
|
||||
|
||||
for (let start = 0; start < normalizedLines.length; start++) {
|
||||
let merged = "";
|
||||
|
||||
for (let end = start; end < normalizedLines.length; end++) {
|
||||
merged += normalizedLines[end];
|
||||
|
||||
if (merged.length > tt.norm.length) {
|
||||
break;
|
||||
}
|
||||
|
||||
if (merged === tt.norm) {
|
||||
if (!this.indexed_map.has(tt.orig)) {
|
||||
this.indexed_map.set(tt.orig, {
|
||||
page: i,
|
||||
passage_index: passages_index.get(tt.orig)?.valueOf() ?? 0,
|
||||
column_name: passages_column_name.get(tt.orig) ?? "",
|
||||
});
|
||||
}
|
||||
titleCursor++;
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
if (titleCursor >= normalizedTitles.length) {
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private get_sorted_index_entries() {
|
||||
return [...this.indexed_map.entries()].sort((a, b) => {
|
||||
const pageA = a[1].page;
|
||||
const pageB = b[1].page;
|
||||
if (pageA !== pageB) {
|
||||
return pageA - pageB;
|
||||
}
|
||||
|
||||
const indexA = a[1].passage_index;
|
||||
const indexB = b[1].passage_index;
|
||||
if (indexA !== indexB) {
|
||||
return indexA - indexB;
|
||||
}
|
||||
|
||||
return a[0].localeCompare(b[0], "zh-Hans-CN");
|
||||
});
|
||||
}
|
||||
|
||||
as_html_page(): string {
|
||||
const styles = Deno.readTextFileSync(`${this.template_dir}/style.html`);
|
||||
const fonts = Deno.readTextFileSync(`${this.template_dir}/fonts.html`);
|
||||
const debugIndexLayout = Deno.env.get("INDEX_DEBUG_LAYOUT") === "1";
|
||||
const debugStyles = debugIndexLayout
|
||||
? `<style>
|
||||
index li { outline: 1px solid rgba(255, 0, 0, 0.45); }
|
||||
index li a { outline: 1px solid rgba(0, 128, 255, 0.45); }
|
||||
index li page_title { outline: 1px solid rgba(0, 180, 0, 0.45); background: rgba(0, 180, 0, 0.06); }
|
||||
index li page_column { outline: 1px solid rgba(255, 140, 0, 0.45); background: rgba(255, 140, 0, 0.06); }
|
||||
index li page_num { outline: 1px solid rgba(128, 0, 255, 0.45); background: rgba(128, 0, 255, 0.06); }
|
||||
</style>`
|
||||
: "";
|
||||
const pageStyles = `<style>
|
||||
@page {
|
||||
size: A4;
|
||||
margin: 20mm 0 20mm 0;
|
||||
}
|
||||
html, body {
|
||||
width: 100%;
|
||||
height: 100%;
|
||||
}
|
||||
index {
|
||||
display: block;
|
||||
width: 100%;
|
||||
}
|
||||
</style>`;
|
||||
const html =
|
||||
`<html><head>${styles}${fonts}${pageStyles}${debugStyles}</head><body><index><h1>目录</h1><ul>${
|
||||
this.get_sorted_index_entries().map((
|
||||
[title, { page, column_name }],
|
||||
) =>
|
||||
`<li class="index-item"><a href="#page-${page}"><page_column>${column_name}</page_column> <page_title>${title}</page_title> <page_num>${page}</page_num></a></li>`
|
||||
).join("\n")
|
||||
}</ul></index></body></html>`;
|
||||
|
||||
return html;
|
||||
}
|
||||
|
||||
private getPdfMarginTopPt(): number {
|
||||
const topMargin = this.cfg?.playwright?.margin?.top ?? "20mm";
|
||||
const match = /^([0-9.]+)\s*mm$/i.exec(String(topMargin).trim());
|
||||
if (match) {
|
||||
return Number(match[1]) / 25.4 * 72;
|
||||
}
|
||||
return (20 / 25.4) * 72;
|
||||
}
|
||||
|
||||
private getPdfMarginBottomPt(): number {
|
||||
const bottomMargin = this.cfg?.playwright?.margin?.bottom ?? "20mm";
|
||||
const match = /^([0-9.]+)\s*mm$/i.exec(String(bottomMargin).trim());
|
||||
if (match) {
|
||||
return Number(match[1]) / 25.4 * 72;
|
||||
}
|
||||
return (20 / 25.4) * 72;
|
||||
}
|
||||
|
||||
// deno-lint-ignore no-explicit-any
|
||||
async html2pdf(cfg: any) {
|
||||
const builder = new DocuBuilder(cfg);
|
||||
try {
|
||||
await builder.with_browser();
|
||||
builder.pages = [await this.as_html_page()];
|
||||
await builder.build_as_pdf(cfg);
|
||||
} finally {
|
||||
await builder.close_browser();
|
||||
}
|
||||
}
|
||||
|
||||
private async get_index_item_layouts(index_html: string): Promise<
|
||||
Array<IndexItemLayout>
|
||||
> {
|
||||
const { chromium } = await import("playwright");
|
||||
const browser = await chromium.launch({
|
||||
executablePath: resolveChromiumPath(),
|
||||
headless: true,
|
||||
args: ["--no-sandbox", "--disable-dev-shm-usage"],
|
||||
});
|
||||
const page = await browser.newPage({
|
||||
viewport: {
|
||||
width: PDF_VIEWPORT_WIDTH_PX,
|
||||
height: 1123,
|
||||
},
|
||||
});
|
||||
try {
|
||||
await page.setContent(index_html, { timeout: 120000 });
|
||||
await page.emulateMedia({ media: "print" });
|
||||
return await page.evaluate(
|
||||
({ printableHeightPx }: { printableHeightPx: number }) => {
|
||||
// Browser-side DOM types are not available in Deno's type checker.
|
||||
const doc = (globalThis as any).document as any;
|
||||
const browserGlobal = globalThis as any;
|
||||
const items = Array.from(
|
||||
doc.querySelectorAll("li.index-item"),
|
||||
) as any[];
|
||||
return items.map((item) => {
|
||||
const titleEl = item.querySelector("page_title") as any;
|
||||
const itemRect = item.getBoundingClientRect();
|
||||
const titleClientRects = titleEl?.getClientRects?.();
|
||||
const itemClientRects = item.getClientRects?.();
|
||||
const titleRect = titleClientRects?.length
|
||||
? titleClientRects[0]
|
||||
: titleEl?.getBoundingClientRect() ?? itemRect;
|
||||
const anchorRect = itemClientRects?.length
|
||||
? itemClientRects[0]
|
||||
: titleRect;
|
||||
const computedStyle = titleEl
|
||||
? browserGlobal.getComputedStyle(titleEl)
|
||||
: null;
|
||||
const fontSizePx = computedStyle
|
||||
? Number.parseFloat(computedStyle.fontSize || "0")
|
||||
: 0;
|
||||
const pageIndex = Math.floor(
|
||||
Math.max(anchorRect.top, 0) / printableHeightPx,
|
||||
);
|
||||
const pageTopPx = pageIndex * printableHeightPx;
|
||||
return {
|
||||
pageIndex,
|
||||
left: titleRect.left,
|
||||
width: titleRect.width,
|
||||
top: anchorRect.top - pageTopPx,
|
||||
height: titleRect.height,
|
||||
fontSize: fontSizePx,
|
||||
};
|
||||
});
|
||||
},
|
||||
{
|
||||
printableHeightPx: (PDF_PAGE_HEIGHT_PT - this.getPdfMarginTopPt() -
|
||||
this.getPdfMarginBottomPt()) / CSS_PX_TO_PT,
|
||||
},
|
||||
);
|
||||
} finally {
|
||||
await page.close();
|
||||
await browser.close();
|
||||
}
|
||||
}
|
||||
|
||||
async combine_index_pdf_to_main_pdf(
|
||||
index_pdf_path: string = "./tmps/index.pdf",
|
||||
insert_index_after_page = 1,
|
||||
) {
|
||||
const combinedPdf = await PDFDocument.create();
|
||||
const index_html = this.as_html_page();
|
||||
const indexPdfBytes = await Deno.readFile(index_pdf_path);
|
||||
const mainPdfBytes = await Deno.readFile(this.file_path);
|
||||
|
||||
const indexPdf = await PDFDocument.load(indexPdfBytes);
|
||||
const mainPdf = await PDFDocument.load(mainPdfBytes);
|
||||
|
||||
const indexPages = await combinedPdf.copyPages(
|
||||
indexPdf,
|
||||
indexPdf.getPageIndices(),
|
||||
);
|
||||
const mainPages = await combinedPdf.copyPages(
|
||||
mainPdf,
|
||||
mainPdf.getPageIndices(),
|
||||
);
|
||||
//插入index页到指定位置,默认插入到第一页后面
|
||||
if (mainPages.length > 0) {
|
||||
mainPages.slice(0, insert_index_after_page).forEach((page: any) =>
|
||||
combinedPdf.addPage(page)
|
||||
);
|
||||
indexPages.forEach((page: any) => combinedPdf.addPage(page));
|
||||
mainPages.slice(insert_index_after_page).forEach((page: any) =>
|
||||
combinedPdf.addPage(page)
|
||||
);
|
||||
} else {
|
||||
indexPages.forEach((page: any) => combinedPdf.addPage(page));
|
||||
}
|
||||
|
||||
await this.add_index_link_annotations_to_pdf(
|
||||
combinedPdf,
|
||||
this.get_sorted_index_entries(),
|
||||
indexPages.length,
|
||||
insert_index_after_page,
|
||||
index_html,
|
||||
);
|
||||
|
||||
const combinedPdfBytes = await combinedPdf.save();
|
||||
await Deno.writeFile(this.file_path, combinedPdfBytes);
|
||||
return indexPages.length;
|
||||
}
|
||||
|
||||
private remap_page_after_index_insert(
|
||||
page: number,
|
||||
inserted_index_pages: number,
|
||||
insert_index_after_page: number,
|
||||
): number {
|
||||
if (inserted_index_pages <= 0) {
|
||||
return page;
|
||||
}
|
||||
if (page <= insert_index_after_page) {
|
||||
return page;
|
||||
}
|
||||
return page + inserted_index_pages;
|
||||
}
|
||||
|
||||
private add_link_annotation(
|
||||
page: any,
|
||||
ctx: any,
|
||||
rect: [number, number, number, number],
|
||||
destPageRef: any,
|
||||
destPageHeight: number,
|
||||
) {
|
||||
const annotation = ctx.obj({
|
||||
Type: PDFName.of("Annot"),
|
||||
Subtype: PDFName.of("Link"),
|
||||
Rect: PDFArray.withContext(ctx),
|
||||
Border: PDFArray.withContext(ctx),
|
||||
Dest: PDFArray.withContext(ctx),
|
||||
});
|
||||
|
||||
const rectArray = annotation.lookup(PDFName.of("Rect"), PDFArray);
|
||||
rect.forEach((value) => rectArray.push(PDFNumber.of(value)));
|
||||
|
||||
const borderArray = annotation.lookup(PDFName.of("Border"), PDFArray);
|
||||
borderArray.push(PDFNumber.of(0));
|
||||
borderArray.push(PDFNumber.of(0));
|
||||
borderArray.push(PDFNumber.of(0));
|
||||
|
||||
const destArray = annotation.lookup(PDFName.of("Dest"), PDFArray);
|
||||
destArray.push(destPageRef);
|
||||
destArray.push(PDFName.of("FitH"));
|
||||
destArray.push(PDFNumber.of(destPageHeight));
|
||||
|
||||
const annotationRef = ctx.register(annotation);
|
||||
const annots = page.node.Annots();
|
||||
if (annots) {
|
||||
annots.push(annotationRef);
|
||||
} else {
|
||||
page.node.set(PDFName.of("Annots"), ctx.obj([annotationRef]));
|
||||
}
|
||||
}
|
||||
|
||||
private add_index_link_annotations(
|
||||
pdfDoc: PDFDocument,
|
||||
sorted_entries: Array<
|
||||
[
|
||||
string,
|
||||
{ page: number; column_name: string },
|
||||
]
|
||||
>,
|
||||
index_item_layouts: Array<{
|
||||
pageIndex: number;
|
||||
top: number;
|
||||
left: number;
|
||||
width: number;
|
||||
height: number;
|
||||
fontSize: number;
|
||||
}>,
|
||||
inserted_index_pages: number,
|
||||
insert_index_after_page: number,
|
||||
) {
|
||||
if (
|
||||
inserted_index_pages <= 0 || sorted_entries.length === 0 ||
|
||||
index_item_layouts.length === 0
|
||||
) {
|
||||
return;
|
||||
}
|
||||
|
||||
const pages = pdfDoc.getPages();
|
||||
const indexPages = pages.slice(
|
||||
insert_index_after_page,
|
||||
insert_index_after_page + inserted_index_pages,
|
||||
);
|
||||
if (indexPages.length === 0) {
|
||||
return;
|
||||
}
|
||||
|
||||
const ctx = pdfDoc.context;
|
||||
const pageHeight = indexPages[0].getSize().height;
|
||||
const pageWidth = indexPages[0].getSize().width;
|
||||
const marginTop = this.getPdfMarginTopPt();
|
||||
const minRectHeight = 12;
|
||||
|
||||
sorted_entries.forEach((entry, index) => {
|
||||
const layout = index_item_layouts[index];
|
||||
if (!layout) {
|
||||
return;
|
||||
}
|
||||
const layoutLeftPt = layout.left * CSS_PX_TO_PT;
|
||||
const layoutWidthPt = layout.width * CSS_PX_TO_PT;
|
||||
const layoutHeightPt = layout.height * CSS_PX_TO_PT;
|
||||
const xLeft = Math.max(layoutLeftPt, 0);
|
||||
const xRight = Math.min(layoutLeftPt + layoutWidthPt, pageWidth);
|
||||
const mappedPage = this.remap_page_after_index_insert(
|
||||
entry[1].page,
|
||||
inserted_index_pages,
|
||||
insert_index_after_page,
|
||||
);
|
||||
const destPageIndex = Math.min(
|
||||
Math.max(mappedPage - 1, 0),
|
||||
pages.length - 1,
|
||||
);
|
||||
const destPage = pages[destPageIndex];
|
||||
const destPageRef = (destPage as any).ref;
|
||||
const destPageHeight = destPage.getSize().height;
|
||||
const totalHeight = Math.max(layoutHeightPt, minRectHeight);
|
||||
const targetPage = indexPages[
|
||||
Math.min(
|
||||
Math.max(layout.pageIndex, 0),
|
||||
indexPages.length - 1,
|
||||
)
|
||||
];
|
||||
if (!targetPage) {
|
||||
return;
|
||||
}
|
||||
const localTopPt = layout.top * CSS_PX_TO_PT;
|
||||
const yTop = pageHeight - marginTop - localTopPt;
|
||||
const yBottom = yTop - totalHeight;
|
||||
this.add_link_annotation(
|
||||
targetPage as any,
|
||||
ctx,
|
||||
[xLeft, yBottom, xRight, yTop],
|
||||
destPageRef,
|
||||
destPageHeight,
|
||||
);
|
||||
});
|
||||
}
|
||||
|
||||
private async add_index_link_annotations_to_pdf(
|
||||
pdfDoc: PDFDocument,
|
||||
sorted_entries: Array<
|
||||
[
|
||||
string,
|
||||
{ page: number; column_name: string },
|
||||
]
|
||||
>,
|
||||
inserted_index_pages: number,
|
||||
insert_index_after_page: number,
|
||||
index_html: string,
|
||||
) {
|
||||
const index_item_layouts = await this.get_index_item_layouts(index_html);
|
||||
this.add_index_link_annotations(
|
||||
pdfDoc,
|
||||
sorted_entries,
|
||||
index_item_layouts,
|
||||
inserted_index_pages,
|
||||
insert_index_after_page,
|
||||
);
|
||||
}
|
||||
|
||||
async add_pdf_outline(options?: {
|
||||
inserted_index_pages?: number;
|
||||
insert_index_after_page?: number;
|
||||
open_outline_by_default?: boolean;
|
||||
}) {
|
||||
const inserted_index_pages = options?.inserted_index_pages ?? 0;
|
||||
const insert_index_after_page = options?.insert_index_after_page ?? 1;
|
||||
const open_outline_by_default = options?.open_outline_by_default ?? true;
|
||||
|
||||
const pdfBytes = await Deno.readFile(this.file_path);
|
||||
const pdfDoc = await PDFDocument.load(pdfBytes);
|
||||
const pages = pdfDoc.getPages();
|
||||
|
||||
if (pages.length === 0 || this.indexed_map.size === 0) {
|
||||
return;
|
||||
}
|
||||
|
||||
const sorted_entries = this.get_sorted_index_entries();
|
||||
|
||||
const ctx = pdfDoc.context;
|
||||
// deno-lint-ignore no-explicit-any
|
||||
const itemNodes: Array<{ ref: any; dict: any }> = [];
|
||||
|
||||
for (const [title, { page }] of sorted_entries) {
|
||||
const mappedPage = this.remap_page_after_index_insert(
|
||||
page,
|
||||
inserted_index_pages,
|
||||
insert_index_after_page,
|
||||
);
|
||||
const targetPageIndex = Math.min(
|
||||
Math.max(mappedPage - 1, 0),
|
||||
pages.length - 1,
|
||||
);
|
||||
// deno-lint-ignore no-explicit-any
|
||||
const pageRef = (pages[targetPageIndex] as any).ref;
|
||||
const targetPageHeight = pages[targetPageIndex].getSize().height;
|
||||
const dict = ctx.obj({
|
||||
Title: PDFHexString.fromText(title),
|
||||
Dest: [
|
||||
pageRef,
|
||||
PDFName.of("FitH"),
|
||||
PDFNumber.of(targetPageHeight),
|
||||
],
|
||||
});
|
||||
const ref = ctx.register(dict);
|
||||
itemNodes.push({ ref, dict });
|
||||
}
|
||||
|
||||
const outlinesDict = ctx.obj({
|
||||
Type: PDFName.of("Outlines"),
|
||||
Count: itemNodes.length,
|
||||
First: itemNodes[0].ref,
|
||||
Last: itemNodes[itemNodes.length - 1].ref,
|
||||
});
|
||||
const outlinesRef = ctx.register(outlinesDict);
|
||||
|
||||
itemNodes.forEach((node, index) => {
|
||||
node.dict.set(PDFName.of("Parent"), outlinesRef);
|
||||
if (index > 0) {
|
||||
node.dict.set(PDFName.of("Prev"), itemNodes[index - 1].ref);
|
||||
}
|
||||
if (index < itemNodes.length - 1) {
|
||||
node.dict.set(PDFName.of("Next"), itemNodes[index + 1].ref);
|
||||
}
|
||||
});
|
||||
|
||||
// deno-lint-ignore no-explicit-any
|
||||
const catalog = (pdfDoc.catalog as any).dict;
|
||||
catalog.set(PDFName.of("Outlines"), outlinesRef);
|
||||
if (open_outline_by_default) {
|
||||
catalog.set(PDFName.of("PageMode"), PDFName.of("UseOutlines"));
|
||||
}
|
||||
|
||||
const outlinedPdfBytes = await pdfDoc.save();
|
||||
await Deno.writeFile(this.file_path, outlinedPdfBytes);
|
||||
}
|
||||
}
|
||||
Reference in new issue
Block a user