@see-ms/converter 0.1.2 → 0.1.4

package/dist/index.mjs

@@ -1,7 +1,7 @@
 // src/converter.ts
 import pc3 from "picocolors";
-import path9 from "path";
-import fs8 from "fs-extra";
+import path11 from "path";
+import fs9 from "fs-extra";
 
 // src/filesystem.ts
 import fs from "fs-extra";
@@ -472,13 +472,20 @@ import * as cheerio2 from "cheerio";
 import fs5 from "fs-extra";
 import path6 from "path";
 function cleanClassName(className) {
-  return className.split(" ").filter((cls) => !cls.startsWith("c-") && !cls.startsWith("w-")).filter((cls) => cls.length > 0).map((cls) => cls.replace(/-/g, "_")).join(" ");
+  return className.split(" ").filter((cls) => !cls.startsWith("c-") && !cls.startsWith("w-")).filter((cls) => cls.length > 0).join(" ");
 }
 function getPrimaryClass(classAttr) {
   if (!classAttr) return null;
   const cleaned = cleanClassName(classAttr);
   const classes = cleaned.split(" ").filter((c) => c.length > 0);
-  return classes[0] || null;
+  if (classes.length === 0) return null;
+  const original = classes[0];
+  return {
+    selector: original,
+    // Keep original with dashes for CSS selector
+    fieldName: original.replace(/-/g, "_")
+    // Normalize for field name
+  };
 }
 function getContextModifier(_$, $el) {
   let $current = $el.parent();
@@ -535,11 +542,11 @@ function detectEditableFields(templateHtml) {
   const potentialCollections = /* @__PURE__ */ new Map();
   $("[class]").each((_, el) => {
     const primaryClass = getPrimaryClass($(el).attr("class"));
-    if (primaryClass && (primaryClass.includes("card") || primaryClass.includes("item") || primaryClass.includes("post") || primaryClass.includes("feature")) && !primaryClass.includes("image") && !primaryClass.includes("inner")) {
-      if (!potentialCollections.has(primaryClass)) {
-        potentialCollections.set(primaryClass, []);
+    if (primaryClass && (primaryClass.fieldName.includes("card") || primaryClass.fieldName.includes("item") || primaryClass.fieldName.includes("post") || primaryClass.fieldName.includes("feature")) && !primaryClass.fieldName.includes("image") && !primaryClass.fieldName.includes("inner")) {
+      if (!potentialCollections.has(primaryClass.fieldName)) {
+        potentialCollections.set(primaryClass.fieldName, []);
       }
-      potentialCollections.get(primaryClass)?.push(el);
+      potentialCollections.get(primaryClass.fieldName)?.push(el);
     }
   });
   potentialCollections.forEach((elements, className) => {
@@ -553,40 +560,43 @@ function detectEditableFields(templateHtml) {
           collectionElements.add(child);
         });
       });
+      const collectionClassInfo = getPrimaryClass($(elements[0]).attr("class"));
+      const collectionSelector = collectionClassInfo ? `.${collectionClassInfo.selector}` : `.${className}`;
       $first.find("img").each((_, img) => {
         if (isInsideButton($, img)) return;
         const $img = $(img);
         const $parent = $img.parent();
-        const parentClass = getPrimaryClass($parent.attr("class"));
-        if (parentClass && parentClass.includes("image")) {
-          collectionFields.image = `.${parentClass}`;
+        const parentClassInfo = getPrimaryClass($parent.attr("class"));
+        if (parentClassInfo && parentClassInfo.fieldName.includes("image")) {
+          collectionFields.image = `.${parentClassInfo.selector}`;
           return false;
         }
       });
       $first.find("div").each((_, el) => {
-        const primaryClass = getPrimaryClass($(el).attr("class"));
-        if (primaryClass && primaryClass.includes("tag") && !primaryClass.includes("container")) {
-          collectionFields.tag = `.${primaryClass}`;
+        const classInfo = getPrimaryClass($(el).attr("class"));
+        if (classInfo && classInfo.fieldName.includes("tag") && !classInfo.fieldName.includes("container")) {
+          collectionFields.tag = `.${classInfo.selector}`;
           return false;
         }
       });
       $first.find("h1, h2, h3, h4, h5, h6").first().each((_, el) => {
-        const primaryClass = getPrimaryClass($(el).attr("class"));
-        if (primaryClass) {
-          collectionFields.title = `.${primaryClass}`;
+        const classInfo = getPrimaryClass($(el).attr("class"));
+        if (classInfo) {
+          collectionFields.title = `.${classInfo.selector}`;
         }
       });
       $first.find("p").first().each((_, el) => {
-        const primaryClass = getPrimaryClass($(el).attr("class"));
-        if (primaryClass) {
-          collectionFields.description = `.${primaryClass}`;
+        const classInfo = getPrimaryClass($(el).attr("class"));
+        if (classInfo) {
+          collectionFields.description = `.${classInfo.selector}`;
         }
       });
       $first.find("a, NuxtLink").not(".c_button, .c_icon_button").each((_, el) => {
         const $link = $(el);
         const linkText = $link.text().trim();
         if (linkText) {
-          collectionFields.link = `.${getPrimaryClass($link.attr("class")) || "a"}`;
+          const classInfo = getPrimaryClass($link.attr("class"));
+          collectionFields.link = classInfo ? `.${classInfo.selector}` : "a";
           return false;
         }
       });
@@ -596,7 +606,7 @@ function detectEditableFields(templateHtml) {
           collectionName += "s";
         }
         detectedCollections[collectionName] = {
-          selector: `.${className}`,
+          selector: collectionSelector,
           fields: collectionFields
         };
       }
@@ -607,25 +617,30 @@ function detectEditableFields(templateHtml) {
     if (collectionElements.has(el)) return;
     const $el = $(el);
     const text = $el.text().trim();
-    const primaryClass = getPrimaryClass($el.attr("class"));
+    const classInfo = getPrimaryClass($el.attr("class"));
     if (text) {
       let fieldName;
-      if (primaryClass && !primaryClass.startsWith("heading_")) {
-        fieldName = primaryClass;
+      let selector;
+      if (classInfo && !classInfo.fieldName.startsWith("heading_")) {
+        fieldName = classInfo.fieldName;
+        selector = `.${classInfo.selector}`;
       } else {
         const $parent = $el.closest('[class*="header"], [class*="hero"], [class*="cta"]').first();
-        const parentClass = getPrimaryClass($parent.attr("class"));
+        const parentClassInfo = getPrimaryClass($parent.attr("class"));
         const modifier = getContextModifier($, $el);
-        if (parentClass) {
-          fieldName = modifier ? `${modifier}_${parentClass}` : parentClass;
+        if (parentClassInfo) {
+          fieldName = modifier ? `${modifier}_${parentClassInfo.fieldName}` : parentClassInfo.fieldName;
+          selector = classInfo ? `.${classInfo.selector}` : `.${parentClassInfo.selector}`;
         } else if (modifier) {
           fieldName = `${modifier}_heading`;
+          selector = classInfo ? `.${classInfo.selector}` : el.tagName.toLowerCase();
         } else {
           fieldName = `heading_${index}`;
+          selector = classInfo ? `.${classInfo.selector}` : el.tagName.toLowerCase();
         }
       }
       detectedFields[fieldName] = {
-        selector: primaryClass ? `.${primaryClass}` : el.tagName.toLowerCase(),
+        selector,
         type: "plain",
         editable: true
       };
@@ -635,11 +650,11 @@ function detectEditableFields(templateHtml) {
     if (collectionElements.has(el)) return;
     const $el = $(el);
     const text = $el.text().trim();
-    const primaryClass = getPrimaryClass($el.attr("class"));
-    if (text && text.length > 20 && primaryClass) {
+    const classInfo = getPrimaryClass($el.attr("class"));
+    if (text && text.length > 20 && classInfo) {
       const hasFormatting = $el.find("strong, em, b, i, a, NuxtLink").length > 0;
-      detectedFields[primaryClass] = {
-        selector: `.${primaryClass}`,
+      detectedFields[classInfo.fieldName] = {
+        selector: `.${classInfo.selector}`,
         type: hasFormatting ? "rich" : "plain",
         editable: true
       };
@@ -651,11 +666,11 @@ function detectEditableFields(templateHtml) {
     const $el = $(el);
     if (isDecorativeImage($, $el)) return;
     const $parent = $el.parent();
-    const parentClass = getPrimaryClass($parent.attr("class"));
-    if (parentClass) {
-      const fieldName = parentClass.includes("image") ? parentClass : `${parentClass}_image`;
+    const parentClassInfo = getPrimaryClass($parent.attr("class"));
+    if (parentClassInfo) {
+      const fieldName = parentClassInfo.fieldName.includes("image") ? parentClassInfo.fieldName : `${parentClassInfo.fieldName}_image`;
       detectedFields[fieldName] = {
-        selector: `.${parentClass}`,
+        selector: `.${parentClassInfo.selector}`,
         type: "image",
         editable: true
       };
@@ -669,8 +684,9 @@ function detectEditableFields(templateHtml) {
     }).first().text().trim();
     if (text && text.length > 2) {
       const $parent = $el.closest('[class*="cta"]').first();
-      const parentClass = getPrimaryClass($parent.attr("class")) || "button";
-      detectedFields[`${parentClass}_button_text`] = {
+      const parentClassInfo = getPrimaryClass($parent.attr("class"));
+      const fieldName = parentClassInfo ? `${parentClassInfo.fieldName}_button_text` : "button_text";
+      detectedFields[fieldName] = {
         selector: `.c_button`,
         type: "plain",
         editable: true
@@ -736,6 +752,18 @@ function mapFieldTypeToStrapi(fieldType) {
   };
   return typeMap[fieldType] || "string";
 }
+function pluralize(word) {
+  if (word.endsWith("s") || word.endsWith("x") || word.endsWith("z") || word.endsWith("ch") || word.endsWith("sh")) {
+    return word + "es";
+  }
+  if (word.endsWith("y") && word.length > 1) {
+    const secondLast = word[word.length - 2];
+    if (!"aeiou".includes(secondLast.toLowerCase())) {
+      return word.slice(0, -1) + "ies";
+    }
+  }
+  return word + "s";
+}
 function pageToStrapiSchema(pageName, fields) {
   const attributes = {};
   for (const [fieldName, field] of Object.entries(fields)) {
@@ -748,12 +776,14 @@ function pageToStrapiSchema(pageName, fields) {
     }
   }
   const displayName = pageName.split("-").map((word) => word.charAt(0).toUpperCase() + word.slice(1)).join(" ");
+  const kebabCaseName = pageName;
+  const pluralName = pluralize(kebabCaseName);
   return {
     kind: "singleType",
-    collectionName: pageName.replace(/-/g, "_"),
+    collectionName: kebabCaseName,
     info: {
-      singularName: pageName.replace(/-/g, "_"),
-      pluralName: pageName.replace(/-/g, "_"),
+      singularName: kebabCaseName,
+      pluralName,
       displayName
     },
     options: {
@@ -780,13 +810,14 @@ function collectionToStrapiSchema(collectionName, collection) {
     };
   }
   const displayName = collectionName.split(/[-_]/).map((word) => word.charAt(0).toUpperCase() + word.slice(1)).join(" ");
-  const singularName = collectionName.endsWith("s") ? collectionName.slice(0, -1) : collectionName;
+  const kebabCaseName = collectionName.replace(/_/g, "-");
+  const singularName = kebabCaseName.endsWith("s") ? kebabCaseName.slice(0, -1) : kebabCaseName;
   return {
     kind: "collectionType",
-    collectionName: collectionName.replace(/[-_]/g, "_"),
+    collectionName: kebabCaseName,
     info: {
-      singularName: singularName.replace(/[-_]/g, "_"),
-      pluralName: collectionName.replace(/[-_]/g, "_"),
+      singularName,
+      pluralName: kebabCaseName,
       displayName
     },
     options: {
@@ -898,6 +929,186 @@ const { data } = await $fetch('http://localhost:1337/api/portfolio-cards')
   await fs7.writeFile(readmePath, content, "utf-8");
 }
 
+// src/content-extractor.ts
+import * as cheerio3 from "cheerio";
+import path9 from "path";
+function extractContentFromHTML(html, _pageName, pageManifest) {
+  const $ = cheerio3.load(html);
+  const content = {
+    fields: {},
+    collections: {}
+  };
+  if (pageManifest.fields) {
+    for (const [fieldName, field] of Object.entries(pageManifest.fields)) {
+      const selector = field.selector;
+      const element = $(selector).first();
+      if (element.length > 0) {
+        if (field.type === "image") {
+          const src = element.attr("src") || element.find("img").attr("src") || "";
+          content.fields[fieldName] = src;
+        } else {
+          const text = element.text().trim();
+          content.fields[fieldName] = text;
+        }
+      }
+    }
+  }
+  if (pageManifest.collections) {
+    for (const [collectionName, collection] of Object.entries(pageManifest.collections)) {
+      const items = [];
+      const collectionElements = $(collection.selector);
+      collectionElements.each((_, elem) => {
+        const item = {};
+        const $elem = $(elem);
+        for (const [fieldName, fieldSelector] of Object.entries(collection.fields)) {
+          const fieldElement = $elem.find(fieldSelector).first();
+          if (fieldElement.length > 0) {
+            if (fieldName === "image" || fieldName.includes("image")) {
+              const src = fieldElement.attr("src") || fieldElement.find("img").attr("src") || "";
+              item[fieldName] = src;
+            } else if (fieldName === "link" || fieldName === "url") {
+              const href = fieldElement.attr("href") || "";
+              item[fieldName] = href;
+            } else {
+              const text = fieldElement.text().trim();
+              item[fieldName] = text;
+            }
+          }
+        }
+        if (Object.keys(item).length > 0) {
+          items.push(item);
+        }
+      });
+      if (items.length > 0) {
+        content.collections[collectionName] = items;
+      }
+    }
+  }
+  return content;
+}
+function extractAllContent(htmlFiles, manifest) {
+  const extractedContent = {
+    pages: {}
+  };
+  for (const [pageName, pageManifest] of Object.entries(manifest.pages)) {
+    const html = htmlFiles.get(pageName);
+    if (html) {
+      const content = extractContentFromHTML(html, pageName, pageManifest);
+      extractedContent.pages[pageName] = content;
+    }
+  }
+  return extractedContent;
+}
+function normalizeImagePath(imageSrc) {
+  if (!imageSrc) return "";
+  if (imageSrc.startsWith("/")) return imageSrc;
+  const filename = path9.basename(imageSrc);
+  if (imageSrc.includes("images/")) {
+    return `/images/${filename}`;
+  }
+  return `/${filename}`;
+}
+function formatForStrapi(extracted) {
+  const seedData = {};
+  for (const [pageName, content] of Object.entries(extracted.pages)) {
+    if (Object.keys(content.fields).length > 0) {
+      const formattedFields = {};
+      for (const [fieldName, value] of Object.entries(content.fields)) {
+        if (fieldName.includes("image") || fieldName.includes("bg")) {
+          formattedFields[fieldName] = normalizeImagePath(value);
+        } else {
+          formattedFields[fieldName] = value;
+        }
+      }
+      seedData[pageName] = formattedFields;
+    }
+    for (const [collectionName, items] of Object.entries(content.collections)) {
+      const formattedItems = items.map((item) => {
+        const formattedItem = {};
+        for (const [fieldName, value] of Object.entries(item)) {
+          if (fieldName === "image" || fieldName.includes("image")) {
+            formattedItem[fieldName] = normalizeImagePath(value);
+          } else {
+            formattedItem[fieldName] = value;
+          }
+        }
+        return formattedItem;
+      });
+      seedData[collectionName] = formattedItems;
+    }
+  }
+  return seedData;
+}
+
+// src/seed-writer.ts
+import fs8 from "fs-extra";
+import path10 from "path";
+async function writeSeedData(outputDir, seedData) {
+  const seedDir = path10.join(outputDir, "cms-seed");
+  await fs8.ensureDir(seedDir);
+  const seedPath = path10.join(seedDir, "seed-data.json");
+  await fs8.writeJson(seedPath, seedData, { spaces: 2 });
+}
+async function createSeedReadme(outputDir) {
+  const readmePath = path10.join(outputDir, "cms-seed", "README.md");
+  const content = `# CMS Seed Data
+
+Auto-extracted content from your Webflow export, ready to seed into Strapi.
+
+## What's in this folder?
+
+\`seed-data.json\` contains the actual content extracted from your HTML:
+- **Single types** - Page-specific content (homepage, about page, etc.)
+- **Collection types** - Repeating items (portfolio cards, team members, etc.)
+
+## Structure
+
+\`\`\`json
+{
+  "index": {
+    "hero_heading_container": "Actual heading from HTML",
+    "hero_bg_image": "/images/hero.jpg",
+    ...
+  },
+  "portfolio_cards": [
+    {
+      "image": "/images/card1.jpg",
+      "tag": "Technology",
+      "description": "Card description"
+    }
+  ]
+}
+\`\`\`
+
+## How to Seed Strapi
+
+### Option 1: Manual Entry
+1. Open Strapi admin panel
+2. Go to Content Manager
+3. Create entries using the data from \`seed-data.json\`
+
+### Option 2: Automated Seeding (Coming Soon)
+We'll provide a seeding script that:
+1. Uploads images to Strapi media library
+2. Creates content entries via Strapi API
+3. Handles relationships between content types
+
+## Image Paths
+
+Image paths in the seed data reference files in your Nuxt \`public/\` directory:
+- \`/images/hero.jpg\` \u2192 \`public/images/hero.jpg\`
+
+When seeding Strapi, these images will be uploaded to Strapi's media library.
+
+## Next Steps
+
+1. Review the extracted data for accuracy
+2. Set up your Strapi instance with the schemas from \`cms-schemas/\`
+3. Use this seed data to populate your CMS
+`;
+  await fs8.writeFile(readmePath, content, "utf-8");
+}
+
 // src/converter.ts
 async function convertWebflowExport(options) {
   const { inputDir, outputDir, boilerplate } = options;
@@ -906,7 +1117,7 @@ async function convertWebflowExport(options) {
   console.log(pc3.dim(`Output: ${outputDir}`));
   try {
     await setupBoilerplate(boilerplate, outputDir);
-    const inputExists = await fs8.pathExists(inputDir);
+    const inputExists = await fs9.pathExists(inputDir);
     if (!inputExists) {
       throw new Error(`Input directory not found: ${inputDir}`);
     }
@@ -922,10 +1133,17 @@ async function convertWebflowExport(options) {
     console.log(pc3.blue("\n\u{1F50D} Finding HTML files..."));
     const htmlFiles = await findHTMLFiles(inputDir);
     console.log(pc3.green(`  \u2713 Found ${htmlFiles.length} HTML files`));
+    const htmlContentMap = /* @__PURE__ */ new Map();
+    for (const htmlFile of htmlFiles) {
+      const html = await readHTMLFile(inputDir, htmlFile);
+      const pageName = htmlFile.replace(".html", "").replace(/\//g, "-");
+      htmlContentMap.set(pageName, html);
+      console.log(pc3.dim(`  Stored: ${pageName} from ${htmlFile}`));
+    }
     console.log(pc3.blue("\n\u2699\uFE0F  Converting HTML to Vue components..."));
     let allEmbeddedStyles = "";
     for (const htmlFile of htmlFiles) {
-      const html = await readHTMLFile(inputDir, htmlFile);
+      const html = htmlContentMap.get(htmlFile.replace(".html", "").replace(/\//g, "-"));
       const parsed = parseHTML(html, htmlFile);
       if (parsed.embeddedStyles) {
         allEmbeddedStyles += `
@@ -941,7 +1159,7 @@ ${parsed.embeddedStyles}
     }
     await formatVueFiles(outputDir);
     console.log(pc3.blue("\n\u{1F50D} Analyzing pages for CMS fields..."));
-    const pagesDir = path9.join(outputDir, "pages");
+    const pagesDir = path11.join(outputDir, "pages");
     const manifest = await generateManifest(pagesDir);
     await writeManifest(outputDir, manifest);
     const totalFields = Object.values(manifest.pages).reduce(
@@ -955,12 +1173,26 @@ ${parsed.embeddedStyles}
     console.log(pc3.green(`  \u2713 Detected ${totalFields} fields across ${Object.keys(manifest.pages).length} pages`));
     console.log(pc3.green(`  \u2713 Detected ${totalCollections} collections`));
     console.log(pc3.green("  \u2713 Generated cms-manifest.json"));
+    console.log(pc3.blue("\n\u{1F4DD} Extracting content from HTML..."));
+    console.log(pc3.dim(`  HTML map has ${htmlContentMap.size} entries`));
+    console.log(pc3.dim(`  Manifest has ${Object.keys(manifest.pages).length} pages`));
+    const extractedContent = extractAllContent(htmlContentMap, manifest);
+    const seedData = formatForStrapi(extractedContent);
+    await writeSeedData(outputDir, seedData);
+    await createSeedReadme(outputDir);
+    const pagesWithContent = Object.keys(seedData).filter((key) => {
+      const data = seedData[key];
+      if (Array.isArray(data)) return data.length > 0;
+      return Object.keys(data).length > 0;
+    }).length;
+    console.log(pc3.green(`  \u2713 Extracted content from ${pagesWithContent} pages`));
+    console.log(pc3.green(`  \u2713 Generated cms-seed/seed-data.json`));
     console.log(pc3.blue("\n\u{1F4CB} Generating Strapi schemas..."));
     const schemas = manifestToSchemas(manifest);
     await writeAllSchemas(outputDir, schemas);
     await createStrapiReadme(outputDir);
     console.log(pc3.green(`  \u2713 Generated ${Object.keys(schemas).length} Strapi content types`));
-    console.log(pc3.dim("    View schemas in: strapi/src/api/"));
+    console.log(pc3.dim("    View schemas in: cms-schemas/"));
     if (allEmbeddedStyles.trim()) {
       console.log(pc3.blue("\n\u2728 Writing embedded styles..."));
       const dedupedStyles = deduplicateStyles(allEmbeddedStyles);
@@ -975,7 +1207,7 @@ ${parsed.embeddedStyles}
       await updateNuxtConfig(outputDir, assets.css);
       console.log(pc3.green("  \u2713 Config updated"));
     } catch (error) {
-      console.log(pc3.yellow("  \u26A0 Could not update nuxt.config.ts automatically"));
+      console.log(pc3.yellow("  \u26A0  Could not update nuxt.config.ts automatically"));
       console.log(pc3.dim("    Please add CSS files manually"));
     }
     console.log(pc3.blue("\n\u{1F3A8} Setting up editor overlay..."));
@@ -988,10 +1220,11 @@ ${parsed.embeddedStyles}
     console.log(pc3.green("\n\u2705 Conversion completed successfully!"));
     console.log(pc3.cyan("\n\u{1F4CB} Next steps:"));
     console.log(pc3.dim(`  1. cd ${outputDir}`));
-    console.log(pc3.dim("  2. Review cms-manifest.json"));
-    console.log(pc3.dim("  3. Copy strapi/ schemas to your Strapi project"));
-    console.log(pc3.dim("  4. pnpm install && pnpm dev"));
-    console.log(pc3.dim("  5. Visit http://localhost:3000?preview=true to edit inline!"));
+    console.log(pc3.dim("  2. Review cms-manifest.json and cms-seed/seed-data.json"));
+    console.log(pc3.dim("  3. Set up Strapi and install schemas from cms-schemas/"));
+    console.log(pc3.dim("  4. Seed Strapi with data from cms-seed/"));
+    console.log(pc3.dim("  5. pnpm install && pnpm dev"));
+    console.log(pc3.dim("  6. Visit http://localhost:3000?preview=true to edit inline!"));
   } catch (error) {
     console.error(pc3.red("\n\u274C Conversion failed:"));
     console.error(pc3.red(error instanceof Error ? error.message : String(error)));