tree-fs 0.1.3 → 0.1.5

package/README.md

@@ -18,7 +18,7 @@ It is designed to be the **standard "Paste & Go" receiver for AI-generated code*
 LLMs (ChatGPT, Claude, DeepSeek) are great at planning architectures but bad at executing them.
 They often output this:
 
-```text
+```bash
 my-app
 ├── src
 │   ├── index.js
@@ -87,13 +87,13 @@ generateFS(tree, path.resolve(__dirname, "./output"))
 console.log("Structure created!")
 ```
 
-## 💡 Syntax Guide & Robustness
+## { } Syntax Guide & Robustness
 
 tree-fs is built to handle the "messy reality" of text inputs.
 
 ### 1. Comments are ignored
 Great for annotated AI outputs.
-```text
+```bash
 src
 ├── auth.js  # Handles JWT tokens
 └── db.js    # Connection logic
@@ -102,7 +102,7 @@ src
 
 ### 2. Explicit Folders
 If a name looks like a file but is actually a folder (e.g., version numbers), end it with a slash `/`.
-```text
+```bash
 api
 ├── v1.5/     <-- Created as a folder
 └── v2.0/     <-- Created as a folder
@@ -110,7 +110,7 @@ api
 
 ### 3. Smart Nesting
 If an item has children indented below it, it is **automatically treated as a folder**, even if it has a dot.
-```text
+```bash
 app
 └── v2.5            <-- Treated as folder because it has a child
     └── migrator.js
@@ -118,7 +118,7 @@ app
 
 ### 4. Markdown & Symbols
 We handle standard tree characters, ASCII art, and bullets.
-```text
+```bash
 project
 - src
   + components
@@ -133,11 +133,33 @@ Known files without extensions are correctly identified as files.
 
 ### 6. Indicators & Comments
 We strip out common markers used to highlight specific files in documentation.
-```text
+```bash
 project
 ├── src/  <-- Working directory
 ├── utils.js // Deprecated
 └── .env # Do not commit
+```
+
+*Result: Creates folder src and files utils.js, .env. All comments are ignored.*
+
+### 7. Rich Text & Emojis
+We automatically clean up "decorative" trees often generated by newer AI models (like Claude or GPT-4o), regardless of where the decoration is placed.
+
+**It handles:**
+*   **Leading/Trailing Emojis:** `📁 src`, `index.js 🚀`, `✨ components ✨`
+*   **Explanations:** `index.js (Core Logic)`, `main.py (Entry)`
+
+```bash
+my-project
+├── 📁 src 🚀
+│   ├── main.js       (The Brain) 🧠
+│   └── theme.css     (Dark Mode)
+└── 📄 package.json 📦
+```
+
+*Result: Creates folder src and files main.js, theme.css, package.json.*
+
+**Note:** Internal emojis (logo_🔥.png) and filenames with parentheses (image(1).png) are preserved. We only strip "detached" decorations separated by spaces.
 
 ## 📦 CI/CD Integration
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tree-fs",
-  "version": "0.1.3",
+  "version": "0.1.5",
   "description": "Generate file system structures from text-based directory trees. The standard receiver for AI-generated project scaffolding.",
   "bin": {
     "tree-fs": "bin/tree-fs.js"

package/src/normaliser.js

@@ -1,52 +1,58 @@
 // src/normaliser.js
 
-// Catch standard tree characters, markdown bullets, and ASCII symbols (+, >, -)
+// 1. Tree Characters & Bullets
 const STRIP_REGEX = /^[\s│├└─•*|\-+>]+/
 
+// 2. Leading Emojis (e.g. "📁 src")
+const LEADING_EMOJI_REGEX = /^[\p{Emoji}\u200d\ufe0f]+\s*/u
+
+// 3. Trailing Emojis (e.g. "index.js 🚀")
+// Must have a space before it to preserve files like "logo_🔥.png"
+const TRAILING_EMOJI_REGEX = /\s+[\p{Emoji}\u200d\ufe0f]+$/u
+
+// 4. Trailing Parenthesis (e.g. "index.js (Logic)")
+const PAREN_COMMENT_REGEX = /\s+\([^)]+\)$/
+
 function normaliseLines(input) {
   return input
     .split("\n")
-    .map(line => line.replace(/\r/g, "")) // Handle Windows CRLF
+    .map(line => line.replace(/\r/g, ""))
     .filter(Boolean)
     .map(raw => {
-      // 1. Windows Fix: Normalize backslashes to forward slashes immediately
+      // A. Normalize slashes
       const normalizedRaw = raw.replace(/\\/g, "/")
 
-      // 2. Calculate indent based on the full prefix (spaces + tree chars)
-      const match = normalizedRaw.match(STRIP_REGEX)
-      const prefixLength = match ? match[0].length : 0
+      // B. Calculate Indent
+      const treeMatch = normalizedRaw.match(STRIP_REGEX)
+      const prefixLength = treeMatch ? treeMatch[0].length : 0
 
-      // 3. Strip comments
-      // Look for multiple comment styles. We use "space + marker" to avoid false positives.
+      // C. Strip Explicit Comments (#, //, <--)
       const commentMarkers = [" #", " <--", " //"]
       let splitIndex = -1
-
       for (const marker of commentMarkers) {
         const idx = normalizedRaw.indexOf(marker)
         if (idx !== -1) {
-          // Keep the earliest marker found
-          if (splitIndex === -1 || idx < splitIndex) {
-            splitIndex = idx
-          }
+          if (splitIndex === -1 || idx < splitIndex) splitIndex = idx
         }
       }
-
       let cleaned = splitIndex !== -1 ? normalizedRaw.substring(0, splitIndex) : normalizedRaw
 
-      // 4. Check for explicit trailing slash
-      const endsWithSlash = cleaned.trim().endsWith("/")
-
-      // 5. Clean up the name
+      // D. Deep Cleaning Chain
+      // We repeat the trailing checks to handle mixed cases like "file.js (Logic) 🚀"
       cleaned = cleaned
-        .replace(STRIP_REGEX, "") // Remove tree characters
-        .replace(/\/$/, "")       // Remove trailing slash for name storage
+        .replace(STRIP_REGEX, "")           // Remove tree chars
+        .replace(LEADING_EMOJI_REGEX, "")   // Remove leading emojis
+        .replace(TRAILING_EMOJI_REGEX, "")  // Remove trailing emojis (Pass 1)
+        .replace(PAREN_COMMENT_REGEX, "")   // Remove trailing parens
+        .replace(TRAILING_EMOJI_REGEX, "")  // Remove trailing emojis (Pass 2 - catches leftovers)
+        .replace(/\/$/, "")                 // Remove trailing slash
         .trim()
 
       return {
         raw: normalizedRaw,
         indent: prefixLength,
         name: cleaned,
-        explicitFolder: endsWithSlash
+        explicitFolder: normalizedRaw.trim().endsWith("/")
       }
     })
     .filter(line => line.name.length > 0)