extract-pdf 0.1.21 → 0.1.23

package/src/models/stashing-stream.ts

@@ -1,86 +1,86 @@
-/**
- * @description Abstract stream processor that buffers ("stashes") consecutive
- * items matching a predicate before flushing them as a group. Subclasses
- * implement `shouldStash`, `doMatchesStash`, and `doFlushStash` to define what
- * constitutes a run. Used by `WordDetectionStream` in `LineConverter` (grouping
- * same-format text spans) and `VerticalsStream` in `VerticalToHorizontal`
- * (merging single-character vertical lines into one horizontal line).
- */
-// Abstract stream which allows stash items temporarily
-export default class StashingStream {
-  results: any[];
-  stash: any[];
-
-  constructor() {
-    if (this.constructor === StashingStream) {
-      throw new TypeError("Can not construct abstract class.");
-    }
-    this.results = [];
-    this.stash = [];
-  }
-
-  consumeAll(items: any[]): void {
-    items.forEach((item) => this.consume(item));
-  }
-
-  consume(item: any): void {
-    if (this.shouldStash(item)) {
-      if (!this.matchesStash(item)) {
-        this.flushStash();
-      }
-      this.pushOnStash(item);
-    } else {
-      if (this.stash.length > 0) {
-        this.flushStash();
-      }
-      this.results.push(item);
-    }
-  }
-
-  pushOnStash(item: any): void {
-    this.onPushOnStash(item);
-    this.stash.push(item);
-  }
-
-  complete(): any[] {
-    if (this.stash.length > 0) {
-      this.flushStash();
-    }
-    return this.results;
-  }
-
-  matchesStash(item: any): boolean {
-    if (this.stash.length === 0) {
-      return true;
-    }
-    const lastItem = this.stash[this.stash.length - 1];
-    return this.doMatchesStash(lastItem, item);
-  }
-
-  flushStash(): void {
-    if (this.stash.length > 0) {
-      this.doFlushStash(this.stash, this.results);
-      this.stash = [];
-    }
-  }
-
-  onPushOnStash(item: any): void {
-    // sub-classes may override
-  }
-
-  shouldStash(item: any): boolean {
-    throw new TypeError(" Do not call abstract method foo from child." + item);
-  }
-
-  doMatchesStash(lastItem: any, item: any): boolean {
-    throw new TypeError(
-      " Do not call abstract method foo from child." + lastItem + item,
-    );
-  }
-
-  doFlushStash(stash: any[], results: any[]): void {
-    throw new TypeError(
-      " Do not call abstract method foo from child." + stash + results,
-    );
-  }
-}
+/**
+ * @description Abstract stream processor that buffers ("stashes") consecutive
+ * items matching a predicate before flushing them as a group. Subclasses
+ * implement `shouldStash`, `doMatchesStash`, and `doFlushStash` to define what
+ * constitutes a run. Used by `WordDetectionStream` in `LineConverter` (grouping
+ * same-format text spans) and `VerticalsStream` in `VerticalToHorizontal`
+ * (merging single-character vertical lines into one horizontal line).
+ */
+// Abstract stream which allows stash items temporarily
+export default class StashingStream {
+  results: any[];
+  stash: any[];
+
+  constructor() {
+    if (this.constructor === StashingStream) {
+      throw new TypeError("Can not construct abstract class.");
+    }
+    this.results = [];
+    this.stash = [];
+  }
+
+  consumeAll(items: any[]): void {
+    items.forEach((item) => this.consume(item));
+  }
+
+  consume(item: any): void {
+    if (this.shouldStash(item)) {
+      if (!this.matchesStash(item)) {
+        this.flushStash();
+      }
+      this.pushOnStash(item);
+    } else {
+      if (this.stash.length > 0) {
+        this.flushStash();
+      }
+      this.results.push(item);
+    }
+  }
+
+  pushOnStash(item: any): void {
+    this.onPushOnStash(item);
+    this.stash.push(item);
+  }
+
+  complete(): any[] {
+    if (this.stash.length > 0) {
+      this.flushStash();
+    }
+    return this.results;
+  }
+
+  matchesStash(item: any): boolean {
+    if (this.stash.length === 0) {
+      return true;
+    }
+    const lastItem = this.stash[this.stash.length - 1];
+    return this.doMatchesStash(lastItem, item);
+  }
+
+  flushStash(): void {
+    if (this.stash.length > 0) {
+      this.doFlushStash(this.stash, this.results);
+      this.stash = [];
+    }
+  }
+
+  onPushOnStash(item: any): void {
+    // sub-classes may override
+  }
+
+  shouldStash(item: any): boolean {
+    throw new TypeError(" Do not call abstract method foo from child." + item);
+  }
+
+  doMatchesStash(lastItem: any, item: any): boolean {
+    throw new TypeError(
+      " Do not call abstract method foo from child." + lastItem + item,
+    );
+  }
+
+  doFlushStash(stash: any[], results: any[]): void {
+    throw new TypeError(
+      " Do not call abstract method foo from child." + stash + results,
+    );
+  }
+}

package/src/models/text-item-line-grouper.ts

@@ -1,41 +1,41 @@
-/**
- * @description Groups a flat array of `TextItem`s into per-line buckets by
- * clustering items whose y-coordinates differ by less than half of
- * `mostUsedDistance`. Each bucket is then sorted by x so items read left-to-right,
- * which is critical for footnote detection and word ordering in `CompactLines`.
- */
-
-import TextItem from "./text-item";
-import { sortByX } from "../utils/page-item-functions";
-
-// Groups all text items which are on the same y line
-export default class TextItemLineGrouper {
-  mostUsedDistance: number;
-
-  constructor(options: { mostUsedDistance?: number }) {
-    this.mostUsedDistance = options.mostUsedDistance || 12;
-  }
-
-  group(textItems: TextItem[]): TextItem[][] {
-    const lines: TextItem[][] = [];
-    var currentLine: TextItem[] = [];
-    textItems.forEach((item) => {
-      if (
-        currentLine.length > 0 &&
-        Math.abs(currentLine[0].y - item.y) >= this.mostUsedDistance / 2
-      ) {
-        lines.push(currentLine);
-        currentLine = [];
-      }
-      currentLine.push(item);
-    });
-    lines.push(currentLine);
-
-    lines.forEach((lineItems) => {
-      sortByX(lineItems);
-    });
-    return lines;
-  }
-}
-
-
+/**
+ * @description Groups a flat array of `TextItem`s into per-line buckets by
+ * clustering items whose y-coordinates differ by less than half of
+ * `mostUsedDistance`. Each bucket is then sorted by x so items read left-to-right,
+ * which is critical for footnote detection and word ordering in `CompactLines`.
+ */
+
+import TextItem from "./text-item";
+import { sortByX } from "../utils/page-item-functions";
+
+// Groups all text items which are on the same y line
+export default class TextItemLineGrouper {
+  mostUsedDistance: number;
+
+  constructor(options: { mostUsedDistance?: number }) {
+    this.mostUsedDistance = options.mostUsedDistance || 12;
+  }
+
+  group(textItems: TextItem[]): TextItem[][] {
+    const lines: TextItem[][] = [];
+    var currentLine: TextItem[] = [];
+    textItems.forEach((item) => {
+      if (
+        currentLine.length > 0 &&
+        Math.abs(currentLine[0].y - item.y) >= this.mostUsedDistance / 2
+      ) {
+        lines.push(currentLine);
+        currentLine = [];
+      }
+      currentLine.push(item);
+    });
+    lines.push(currentLine);
+
+    lines.forEach((lineItems) => {
+      sortByX(lineItems);
+    });
+    return lines;
+  }
+}
+
+

package/src/models/word.ts

@@ -1,31 +1,31 @@
-/**
- * @description Minimal token model for a single word within a `LineItem`.
- * Carries the text `string`, an optional `type` (`WordType`: LINK,
- * FOOTNOTE_LINK, FOOTNOTE) for semantic rendering, and an optional `format`
- * (`WordFormat`: BOLD, OBLIQUE, BOLD_OBLIQUE) for inline HTML styling.
- */
-
-export interface WordFormatEntry {
-  name: string;
-  startSymbol: string;
-  endSymbol: string;
-}
-
-export interface WordTypeEntry {
-  name: string;
-  attachWithoutWhitespace?: boolean;
-  plainTextFormat?: boolean;
-  toText(string: string): string;
-}
-
-export default class Word {
-  string: string;
-  type: WordTypeEntry | null;
-  format: WordFormatEntry | null;
-
-  constructor(options: { string: string; type?: WordTypeEntry | null; format?: WordFormatEntry | null }) {
-    this.string = options.string;
-    this.type = options.type ?? null;
-    this.format = options.format ?? null;
-  }
-}
+/**
+ * @description Minimal token model for a single word within a `LineItem`.
+ * Carries the text `string`, an optional `type` (`WordType`: LINK,
+ * FOOTNOTE_LINK, FOOTNOTE) for semantic rendering, and an optional `format`
+ * (`WordFormat`: BOLD, OBLIQUE, BOLD_OBLIQUE) for inline HTML styling.
+ */
+
+export interface WordFormatEntry {
+  name: string;
+  startSymbol: string;
+  endSymbol: string;
+}
+
+export interface WordTypeEntry {
+  name: string;
+  attachWithoutWhitespace?: boolean;
+  plainTextFormat?: boolean;
+  toText(string: string): string;
+}
+
+export default class Word {
+  string: string;
+  type: WordTypeEntry | null;
+  format: WordFormatEntry | null;
+
+  constructor(options: { string: string; type?: WordTypeEntry | null; format?: WordFormatEntry | null }) {
+    this.string = options.string;
+    this.type = options.type ?? null;
+    this.format = options.format ?? null;
+  }
+}