@getguru/import-engine 0.1.2 → 0.1.4

package/docx-import-cli.js

@@ -98755,6 +98755,34 @@ var importType = typeArgIndex !== -1 ? process.argv[typeArgIndex + 1] : "";
 var dom = new import_jsdom.JSDOM("", {
   url: `http://localhost:8000/docx-import/index.html?type=${importType}&embed=`
 });
+var UNUSED_INPUT_EVENTS = /* @__PURE__ */ new Set([
+  "focus",
+  "focusin",
+  "keydown",
+  "keyup",
+  "mouseover",
+  "mousedown",
+  "mouseup",
+  "click",
+  "mouseout"
+]);
+var addWindowListener = dom.window.addEventListener;
+dom.window.addEventListener = function(type2, ...rest2) {
+  if (UNUSED_INPUT_EVENTS.has(type2)) {
+    return void 0;
+  }
+  return addWindowListener.call(this, type2, ...rest2);
+};
+var windowProperties = require("jsdom/lib/jsdom/living/window-properties.js");
+for (const hook of ["elementAttached", "elementDetached", "elementAttributeModified"]) {
+  const track = windowProperties[hook];
+  windowProperties[hook] = function(element, ...rest2) {
+    if (!element._ownerDocument._defaultView) {
+      return void 0;
+    }
+    return track.call(this, element, ...rest2);
+  };
+}
 var ALLOWED_GLOBALS = [
   "window",
   "document",
@@ -102636,6 +102664,52 @@ function markLinkRecorded(doc, link2) {
 function isLinkRecorded(doc, link2) {
   return !!doc.recordedLinks && doc.recordedLinks.has(link2);
 }
+var RECORDED_LINK_ATTRIBUTE = "data-guru-import-recorded-link";
+function serializeDoctype(doctype) {
+  if (!doctype) {
+    return "";
+  }
+  const quote = (id) => id.includes('"') ? `'${id}'` : `"${id}"`;
+  if (doctype.publicId) {
+    return `<!DOCTYPE ${doctype.name} PUBLIC ${quote(doctype.publicId)}${doctype.systemId ? " " + quote(doctype.systemId) : ""}>`;
+  }
+  if (doctype.systemId) {
+    return `<!DOCTYPE ${doctype.name} SYSTEM ${quote(doctype.systemId)}>`;
+  }
+  return `<!DOCTYPE ${doctype.name}>`;
+}
+function serializeDocTree(div, doc) {
+  for (const element of div.querySelectorAll("pre, textarea, listing")) {
+    const first2 = element.firstChild;
+    if (first2 && first2.nodeType === 3 && first2.data.startsWith("\n")) {
+      first2.data = "\n" + first2.data;
+    }
+  }
+  if (doc.recordedLinks) {
+    for (const element of div.querySelectorAll("*")) {
+      if (doc.recordedLinks.has(element)) {
+        element.setAttribute(RECORDED_LINK_ATTRIBUTE, "");
+      }
+    }
+  }
+  return flattenString(serializeDoctype(div.ownerDocument.doctype) + div.outerHTML);
+}
+function flattenString(string) {
+  string.charCodeAt(0);
+  return string;
+}
+function parseDocTree(source, doc) {
+  const div = new DOMParser().parseFromString(source, "text/html").documentElement;
+  const hadRecordedLinks = !!doc.recordedLinks;
+  delete doc.recordedLinks;
+  if (hadRecordedLinks) {
+    for (const element of div.querySelectorAll(`[${RECORDED_LINK_ATTRIBUTE}]`)) {
+      element.removeAttribute(RECORDED_LINK_ATTRIBUTE);
+      markLinkRecorded(doc, element);
+    }
+  }
+  return div;
+}
 function safeDecode(string) {
   try {
     return decodeURIComponent(string);
@@ -108438,6 +108512,12 @@ var card_schema_default = {
     strikethrough: {
       type: "boolean"
     },
+    superscript: {
+      type: "boolean"
+    },
+    subscript: {
+      type: "boolean"
+    },
     underline: {
       type: "boolean"
     },
@@ -111340,11 +111420,17 @@ function buildSlabPostSlugIndex(allDocs, zip) {
   }
   return index;
 }
+function getSlabPostSlugIndex(zip, allDocs) {
+  if (!zip.slabPostSlugIndex) {
+    zip.slabPostSlugIndex = buildSlabPostSlugIndex(allDocs, zip);
+  }
+  return zip.slabPostSlugIndex;
+}
 function resolveSlabPostLinks(div, doc, allDocs) {
   if (!doc.zip?.isSlabExport) {
     return;
   }
-  const index = buildSlabPostSlugIndex(allDocs, doc.zip);
+  const index = getSlabPostSlugIndex(doc.zip, allDocs);
   Array.from(div.querySelectorAll("a[href]")).forEach((link2) => {
     const href = link2.getAttribute("href");
     if (!href || SLAB_TOPIC_URL_PATTERN.test(href)) {
@@ -111445,6 +111531,32 @@ var STYLE_ATTRIBUTES_TO_KEEP = [
   "text-align",
   "text-decoration"
 ];
+var VERTICAL_ALIGN_BLOCK_SELECTOR = "p, div, h1, h2, h3, h4, h5, h6, ul, ol, li, table, blockquote, pre";
+function convertVerticalAlignToTags(root2) {
+  Array.from(root2.querySelectorAll("[style]")).forEach(function(el) {
+    if (el.closest(".ghq-card-content__markdown")) {
+      return;
+    }
+    const values2 = parseStyleString(el.getAttribute("style"));
+    const key = Object.keys(values2).find((k) => k.toLowerCase() === "vertical-align");
+    if (!key) {
+      return;
+    }
+    const value = values2[key].replace(/!\s*important\s*$/i, "").trim().toLowerCase();
+    if (value !== "super" && value !== "sub") {
+      return;
+    }
+    const tagName = value === "super" ? "sup" : "sub";
+    if (el.closest(tagName) || !el.textContent.trim() && !el.firstElementChild || el.querySelector(VERTICAL_ALIGN_BLOCK_SELECTOR)) {
+      return;
+    }
+    const tag = el.ownerDocument.createElement(tagName);
+    while (el.firstChild) {
+      tag.appendChild(el.firstChild);
+    }
+    el.appendChild(tag);
+  });
+}
 function parseStyleString(text2) {
   const result3 = {};
   const pairs3 = text2.split(";");
@@ -112030,6 +112142,7 @@ function cleanUpHtml(element, doc = {}) {
       );
     }
   });
+  convertVerticalAlignToTags(element);
   Array.from(element.querySelectorAll("[style]")).forEach(function(element2) {
     if (element2.closest(".ghq-card-content__markdown")) {
       return;
@@ -114320,7 +114433,7 @@ var import_jszip5 = __toESM(require_lib3());
 var import_url_join6 = __toESM(require_url_join());
 
 // packages/import-engine/package.json
-var version = "0.1.2";
+var version = "0.1.4";
 
 // src/docx-import/detectors.js
 function isGoogleSitesDoc(div) {
@@ -114344,7 +114457,7 @@ var showError;
 var updateLoader;
 var showReport;
 var docLinkIndex = /* @__PURE__ */ new Map();
-var toolVersion = true ? `${version}+${"4d61d22"}` : version;
+var toolVersion = true ? `${version}+${"b8fcb63"}` : version;
 var pendingUpload = null;
 function confirmPendingUpload() {
   const upload = pendingUpload;
@@ -114508,6 +114621,21 @@ async function processImage(doc, image2, imageIndex, newZip, resourceLookup = {}
     }
   });
 }
+var resourceLookupSizes = /* @__PURE__ */ new WeakMap();
+function getResourceLookupSize(resourceLookup) {
+  if (!resourceLookupSizes.has(resourceLookup)) {
+    resourceLookupSizes.set(resourceLookup, Object.keys(resourceLookup).length);
+  }
+  return resourceLookupSizes.get(resourceLookup);
+}
+function addToResourceLookup(resourceLookup, unsafeResourceName, safeResourceName) {
+  const size2 = getResourceLookupSize(resourceLookup);
+  const isNewKey = !Object.hasOwn(resourceLookup, unsafeResourceName);
+  resourceLookup[unsafeResourceName] = safeResourceName;
+  if (isNewKey && Object.hasOwn(resourceLookup, unsafeResourceName)) {
+    resourceLookupSizes.set(resourceLookup, size2 + 1);
+  }
+}
 async function resolveResource(element, doc, newZip, resourceLookup = {}) {
   return new Promise(async function(resolve, reject2) {
     try {
@@ -114536,10 +114664,10 @@ async function resolveResource(element, doc, newZip, resourceLookup = {}) {
         } else {
           safeResourceName = sanitizeFilename(unsafeResourceName);
           if (safeResourceName !== unsafeResourceName) {
-            const nextIndex = Object.keys(resourceLookup).length;
+            const nextIndex = getResourceLookupSize(resourceLookup);
             safeResourceName = `file${nextIndex}_${safeResourceName}`;
           }
-          resourceLookup[unsafeResourceName] = safeResourceName;
+          addToResourceLookup(resourceLookup, unsafeResourceName, safeResourceName);
         }
         const resourcePath = "resources/" + safeResourceName;
         const fileData = await (import_jszip5.default.support.uint8array ? file.async("uint8array") : file.async("base64"));
@@ -114774,7 +114902,7 @@ async function preProcessDoc(doc, docTitleLookup = {}) {
       doc.zip.source = source;
     }
   }
-  doc.htmlContent = div;
+  doc.htmlSource = serializeDocTree(div, doc);
   updateLoader({
     step: 1,
     docIndex: doc.index,
@@ -114805,7 +114933,8 @@ async function postProcessDoc(newZip, doc, resourceLookup = {}, externalLookup =
     writeCardToZip(newZip, doc, "");
     return;
   }
-  const div = doc.htmlContent;
+  const div = parseDocTree(doc.htmlSource, doc);
+  delete doc.htmlSource;
   updateLoader({
     step: 1,
     docIndex: doc.index,
@@ -114874,6 +115003,14 @@ async function postProcessDoc(newZip, doc, resourceLookup = {}, externalLookup =
   }
   writeCardToZip(newZip, doc, cleanUpHtml(div, doc));
 }
+function releaseDocContent(doc, buffersReadFromZip) {
+  delete doc.htmlSource;
+  delete doc.originalContent;
+  delete doc.recordedLinks;
+  if (buffersReadFromZip.has(doc)) {
+    delete doc.buffer;
+  }
+}
 function makeErrorHandler(progress) {
   return function handleError(event) {
     const message = event.reason?.message || event.message || event.error?.message || "Unknown error";
@@ -114957,6 +115094,7 @@ async function runProcessDocs(params, progress) {
       zipToCheck.docxAnchorIndex = buildDocxAnchorIndex(docs, zipToCheck);
     }
   }
+  const buffersReadFromZip = /* @__PURE__ */ new WeakSet();
   await doAll(docs, async function(doc, docIndex, nextDoc) {
     doc.index = docIndex;
     try {
@@ -114965,6 +115103,7 @@ async function runProcessDocs(params, progress) {
       } else if (doc.zip) {
         const buffer = await doc.zip.file(doc.filename).async("arraybuffer");
         doc.buffer = buffer;
+        buffersReadFromZip.add(doc);
         await preProcessDoc(doc, docTitleLookup);
       } else {
         await preProcessDoc(doc);
@@ -114972,6 +115111,7 @@ async function runProcessDocs(params, progress) {
     } catch (err) {
       recordFailure(doc, err);
     } finally {
+      delete doc.div;
       nextDoc();
     }
   });
@@ -114981,6 +115121,9 @@ async function runProcessDocs(params, progress) {
     if (zipToCheck.source === "confluence") {
       zipToCheck.confluencePageIndex = buildConfluencePageIndex(docs, zipToCheck);
     }
+    if (zipToCheck.isSlabExport) {
+      zipToCheck.slabPostSlugIndex = buildSlabPostSlugIndex(docs, zipToCheck);
+    }
   }
   await doAll(docs, async function(doc, docIndex, nextDoc) {
     try {
@@ -114988,6 +115131,7 @@ async function runProcessDocs(params, progress) {
     } catch (err) {
       recordFailure(doc, err);
     } finally {
+      releaseDocContent(doc, buffersReadFromZip);
       nextDoc();
     }
   });

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@getguru/import-engine",
-  "version": "0.1.2",
+  "version": "0.1.4",
   "description": "Command-line engine that turns Notion, Confluence, Google, Word, PDF, Markdown and spreadsheet exports into Guru import zips.",
   "bin": {
     "guru-import-engine": "docx-import-cli.js"