@rohal12/twee-ts 1.10.0 → 1.11.0

package/dist/bin/twee-ts.js

@@ -236,15 +236,32 @@ function passageToTiddler(p, pid, obfuscateRot13 = false) {
   const nameComment = obfuscateRot13 ? `<!-- ${commentSanitize(p.name)} -->` : "";
   return `<div tiddler=${quote(attrEscape(p.name))} tags=${quote(attrEscape(p.tags.join(" ")))} created=${quote(created)} modifier=${quote("twee")} twine-position=${quote(attrEscape(position))}>${nameComment}${content}</div>`;
 }
-function countWords(p) {
-  let text = p.text;
-  text = text.replace(/\n/g, "");
-  text = text.replace(/(?:\/%.+?%\/|\/\*.+?\*\/|<!--.+?-->)/gs, "");
-  const normalized = text.normalize("NFKD");
-  const count = [...normalized].length;
-  if (count === 0) return 0;
-  const words = Math.floor(count / 5);
-  return count % 5 > 0 ? words + 1 : words;
+function countWords(p, method = "tweego") {
+  switch (method) {
+    case "tweego": {
+      let text = p.text;
+      text = text.replace(/\n/g, "");
+      text = text.replace(/(?:\/%.+?%\/|\/\*.+?\*\/|<!--.+?-->)/gs, "");
+      const normalized = text.normalize("NFKD");
+      const count = [...normalized].length;
+      if (count === 0) return 0;
+      const words = Math.floor(count / 5);
+      return count % 5 > 0 ? words + 1 : words;
+    }
+    case "whitespace": {
+      let text = p.text;
+      text = text.replace(/(?:\/%.+?%\/|\/\*.+?\*\/|<!--.+?-->)/gs, "");
+      text = text.replace(/<<[^>]*>>/g, "");
+      text = text.replace(/\[\[([^\]|]*?)(?:\|[^\]]*?)?\]\]/g, "$1");
+      text = text.replace(/<[^>]+>/g, "");
+      const tokens = text.split(/\s+/).filter((t) => t.length > 0);
+      return tokens.length;
+    }
+    default: {
+      const _exhaustive = method;
+      throw new Error(`Unhandled word count method: ${_exhaustive}`);
+    }
+  }
 }
 function applyTagAliases(passages, aliases) {
   const entries = Object.entries(aliases);
@@ -462,13 +479,13 @@ function storyAdd(story, p, diagnostics) {
   }
   storyAppend(story, processed, diagnostics);
 }
-function getStoryStats(story) {
+function getStoryStats(story, wordCountMethod = "tweego") {
   let storyPassages = 0;
   let words = 0;
   for (const p of story.passages) {
     if (isStoryPassage(p)) {
       storyPassages++;
-      words += countWords(p);
+      words += countWords(p, wordCountMethod);
     }
   }
   return { passages: story.passages.length, storyPassages, words };
@@ -2293,7 +2310,7 @@ async function buildOutput(options, cache, changedFiles) {
     }
   }
   const stats = {
-    ...getStoryStats(story),
+    ...getStoryStats(story, options.wordCountMethod),
     files: [...processedFiles]
   };
   return { output, story, format, diagnostics, stats };
@@ -2514,6 +2531,7 @@ import { readFileSync as readFileSync4, existsSync as existsSync2 } from "fs";
 import { join as join5 } from "path";
 var CONFIG_FILENAME = "twee-ts.config.json";
 var VALID_OUTPUT_MODES = ["html", "twee3", "twee1", "twine2-archive", "twine1-archive", "json"];
+var VALID_WORD_COUNT_METHODS = ["tweego", "whitespace"];
 function loadConfig(dir) {
   const base = dir ?? process.cwd();
   const configPath = join5(base, CONFIG_FILENAME);
@@ -2600,6 +2618,11 @@ function validateConfig(data) {
       errors.push(`"outputMode" must be one of: ${VALID_OUTPUT_MODES.join(", ")}.`);
     }
   }
+  if ("wordCountMethod" in obj) {
+    if (typeof obj["wordCountMethod"] !== "string" || !VALID_WORD_COUNT_METHODS.includes(obj["wordCountMethod"])) {
+      errors.push(`"wordCountMethod" must be one of: ${VALID_WORD_COUNT_METHODS.join(", ")}.`);
+    }
+  }
   return errors;
 }
 function scaffoldConfig() {
@@ -2642,6 +2665,7 @@ var { values, positionals } = parseArgs({
     "no-remote": { type: "boolean" },
     "tag-alias": { type: "string", multiple: true },
     "source-info": { type: "boolean" },
+    "word-count-method": { type: "string" },
     config: { type: "string", short: "c" },
     "no-config": { type: "boolean" }
   }
@@ -2699,6 +2723,16 @@ async function main() {
       tagAliases[pair.slice(0, eq)] = pair.slice(eq + 1);
     }
   }
+  const VALID_WORD_COUNT_METHODS2 = ["tweego", "whitespace"];
+  const wordCountMethod = (() => {
+    const raw = values["word-count-method"] ?? config?.wordCountMethod;
+    if (raw === void 0) return void 0;
+    if (!VALID_WORD_COUNT_METHODS2.includes(raw)) {
+      console.error(`Error: Invalid --word-count-method "${raw}". Expected: ${VALID_WORD_COUNT_METHODS2.join(", ")}`);
+      process.exit(1);
+    }
+    return raw;
+  })();
   if (values.lint) {
     const lintResult = await lint({
       sources,
@@ -2714,7 +2748,8 @@ async function main() {
       formatUrls: values["format-url"] ?? config?.formatUrls,
       noRemote: values["no-remote"] ?? config?.noRemote ?? false,
       tagAliases,
-      sourceInfo: values["source-info"] ?? config?.sourceInfo ?? false
+      sourceInfo: values["source-info"] ?? config?.sourceInfo ?? false,
+      wordCountMethod
     });
     console.log(formatLintReport(lintResult));
     const hasErrors = lintResult.brokenLinks.length > 0 || lintResult.diagnostics.some((d) => d.level === "error");
@@ -2736,7 +2771,8 @@ async function main() {
     formatUrls: values["format-url"] ?? config?.formatUrls,
     noRemote: values["no-remote"] ?? config?.noRemote ?? false,
     tagAliases,
-    sourceInfo: values["source-info"] ?? config?.sourceInfo ?? false
+    sourceInfo: values["source-info"] ?? config?.sourceInfo ?? false,
+    wordCountMethod
   };
   const outFile = values.output ?? config?.output ?? "-";
   if (values.watch) {
@@ -2921,6 +2957,7 @@ Options:
   --format-url <url>        Direct format.js URL (repeatable)
   --tag-alias <alias=target> Map a tag to a special tag (repeatable)
   --source-info             Emit source file/line as data- attributes on passages
+  --word-count-method <m>   Word counting method: tweego (default), whitespace
   --no-remote               Disable remote format fetching
   -c, --config <file>       Config file path (default: ${CONFIG_FILENAME})
   --no-config               Skip config file loading