docspack 0.1.0 → 0.2.0

package/src/style.ts

@@ -34,6 +34,11 @@ const SENTENCE_SPLIT = /(?<=[.!?])\s+/;
 /** A heading or list item starts a new unit, whatever the previous line ended with. */
 const BLOCK_START = /\n(?=\s*(?:[-*+]\s|\d+\.\s|#{1,6}\s|\|))/;
 const LONG_SENTENCE_WORDS = 40;
+/**
+ * A word carries a letter or a digit. Stripped code leaves its punctuation behind, so counting
+ * everything between spaces reports an enumeration of forty code spans as a forty-word sentence.
+ */
+const WORD = /[\p{L}\p{N}]/u;
 
 export interface FillerHit {
   readonly phrase: string;
@@ -68,7 +73,7 @@ export function findLongSentences(text: string, maxWords = LONG_SENTENCE_WORDS):
     .flatMap((block) => block.split(BLOCK_START))
     .flatMap((block) => block.split(SENTENCE_SPLIT))
     .map((sentence) => sentence.replace(/\s+/g, " ").trim())
-    .filter((sentence) => sentence.split(" ").filter(Boolean).length > maxWords);
+    .filter((sentence) => sentence.split(" ").filter((word) => WORD.test(word)).length > maxWords);
 }
 
 /**
@@ -94,11 +99,16 @@ export function contentFingerprint(text: string): string {
     .trim();
 }
 
+/** An authored directive, e.g. `<!-- docspack: tags=grid -->`. Metadata, not site boilerplate. */
+const DIRECTIVE_LINE = /^\s*<!--\s*docspack:/i;
+
 /** Removes lines a documentation site wraps around its content. */
 export function stripBoilerplate(text: string): string {
   return text
     .split("\n")
-    .filter((line) => !BOILERPLATE.some((pattern) => pattern.test(line)))
+    .filter(
+      (line) => DIRECTIVE_LINE.test(line) || !BOILERPLATE.some((pattern) => pattern.test(line)),
+    )
     .join("\n");
 }