feat: 完成 DOCX 视觉门禁与字体兼容映射

2026-07-31 20:13:19 +08:00
parent 10bc62cee4
commit e5da699ea3
32 changed files with 3015 additions and 17 deletions
@@ -0,0 +1,158 @@
+import type {
+  PdfPointBounds,
+  PdfTextItemSnapshot,
+  PdfTextLineSnapshot,
+} from "./types.js";
+
+const ZERO_WIDTH_AND_CONTROL = /[\u0000-\u0008\u000b\u000c\u000e-\u001f\u007f\u200b-\u200d\u2060\ufeff]/gu;
+const PAGE_NUMBER_PATTERNS = [
+  /^(?:[-—–]\s*)?\d+(?:\s*[/／]\s*\d+)?(?:\s*[-—–])?$/u,
+  /^第\s*\d+\s*页(?:\s*(?:[/／]|共)\s*\d+\s*页)?$/u,
+  /^\d+\s*页(?:\s*(?:[/／]|共)\s*\d+\s*页)?$/u,
+];
+
+export function normalizePdfText(value: string): string {
+  return value
+    .normalize("NFKC")
+    .replace(ZERO_WIDTH_AND_CONTROL, "")
+    .replace(/\u00a0/gu, " ")
+    .replace(/[ \t]+/gu, " ")
+    .trim();
+}
+
+export function normalizePdfContentText(value: string): string {
+  return normalizePdfText(value).replace(/\s+/gu, "");
+}
+
+function unionBounds(items: readonly PdfTextItemSnapshot[]): PdfPointBounds {
+  const left = Math.min(...items.map((item) => item.bounds.x));
+  const top = Math.min(...items.map((item) => item.bounds.y));
+  const right = Math.max(
+    ...items.map((item) => item.bounds.x + item.bounds.width),
+  );
+  const bottom = Math.max(
+    ...items.map((item) => item.bounds.y + item.bounds.height),
+  );
+  return {
+    x: left,
+    y: top,
+    width: Math.max(0, right - left),
+    height: Math.max(0, bottom - top),
+  };
+}
+
+function median(values: readonly number[]): number {
+  if (values.length === 0) {
+    return 0;
+  }
+  const sorted = [...values].sort((left, right) => left - right);
+  const middle = Math.floor(sorted.length / 2);
+  const value = sorted[middle] ?? 0;
+  if (sorted.length % 2 === 1) {
+    return value;
+  }
+  return ((sorted[middle - 1] ?? value) + value) / 2;
+}
+
+function joinLineItems(items: readonly PdfTextItemSnapshot[]): string {
+  let result = "";
+  let previous: PdfTextItemSnapshot | undefined;
+  for (const item of items) {
+    const text = item.normalizedText;
+    if (!text) {
+      continue;
+    }
+    if (previous && result) {
+      const gap =
+        item.bounds.x - (previous.bounds.x + previous.bounds.width);
+      const referenceHeight = Math.max(
+        1,
+        Math.min(previous.bounds.height, item.bounds.height),
+      );
+      if (
+        gap > referenceHeight * 0.24 &&
+        !result.endsWith(" ") &&
+        !text.startsWith(" ")
+      ) {
+        result += " ";
+      }
+    }
+    result += text;
+    previous = item;
+  }
+  return normalizePdfText(result);
+}
+
+export function isPageNumberLine(
+  lineText: string,
+  bounds: PdfPointBounds,
+  pageHeightPt: number,
+): boolean {
+  const normalized = normalizePdfText(lineText);
+  if (!PAGE_NUMBER_PATTERNS.some((pattern) => pattern.test(normalized))) {
+    return false;
+  }
+  const lineMiddle = bounds.y + bounds.height / 2;
+  const topBand = pageHeightPt * 0.08;
+  const bottomBand = pageHeightPt * 0.85;
+  return lineMiddle <= topBand || lineMiddle >= bottomBand;
+}
+
+export function aggregatePdfTextLines(
+  sourceItems: readonly PdfTextItemSnapshot[],
+  pageHeightPt: number,
+): PdfTextLineSnapshot[] {
+  const items = sourceItems
+    .filter((item) => item.normalizedText.length > 0)
+    .sort(
+      (left, right) =>
+        left.bounds.y - right.bounds.y || left.bounds.x - right.bounds.x,
+    );
+  const typicalHeight = median(items.map((item) => item.bounds.height));
+  const baselineTolerance = Math.max(1.5, typicalHeight * 0.15);
+  const groups: PdfTextItemSnapshot[][] = [];
+
+  for (const item of items) {
+    const lastGroup = groups.at(-1);
+    const lastBaseline = lastGroup
+      ? median(lastGroup.map((entry) => entry.baselineY))
+      : undefined;
+    if (
+      lastGroup &&
+      lastBaseline !== undefined &&
+      Math.abs(item.baselineY - lastBaseline) <= baselineTolerance
+    ) {
+      lastGroup.push(item);
+    } else {
+      groups.push([item]);
+    }
+  }
+
+  return groups.map((group) => {
+    const sortedItems = [...group].sort(
+      (left, right) => left.bounds.x - right.bounds.x,
+    );
+    const bounds = unionBounds(sortedItems);
+    const normalizedText = joinLineItems(sortedItems);
+    return {
+      text: normalizedText,
+      normalizedText,
+      bounds,
+      baselineY: median(sortedItems.map((item) => item.baselineY)),
+      role: isPageNumberLine(normalizedText, bounds, pageHeightPt)
+        ? "page-number"
+        : "content",
+      items: sortedItems,
+    };
+  });
+}
+
+export function buildPageContentText(
+  lines: readonly PdfTextLineSnapshot[],
+): string {
+  return lines
+    .filter((line) => line.role === "content")
+    .map((line) => line.normalizedText)
+    .filter(Boolean)
+    .join("\n");
+}