chatgpt-to-markdown 1.4.0 → 1.5.1

package/README.md

@@ -97,6 +97,8 @@ git push --follow-tags
 
 ## Release notes
 
+- 1.5.1: 22 Mar 2024. Handle unicode filenames
+- 1.5.0: 28 Nov 2023. Handle `tether_browsing_display`, `tether_quote` and `system_error`
 - 1.4.0: 29 Oct 2023. Handle multi-modal text from Dall-E
 - 1.3.0: 29 Sep 2023. Set create and update dates from chat
 - 1.2.0: 28 Sep 2023. Added test cases

package/index.js

@@ -9,6 +9,7 @@ import path from "path";
 function sanitizeFileName(title) {
   return title
     .replace(/[<>:"\/\\|?*\n]/g, " ")
+    .replace(/[^\w\s]/gi, " ")
     .replace(/\s+/g, " ")
     .trim();
 }
@@ -33,8 +34,8 @@ function wrapHtmlTagsInBackticks(text) {
 function indent(str) {
   return str
     .split("\n")
-    .map((v) => `    ${v}`)
-    .join("\n");
+    .map((v) => `    ${v}\n`)
+    .join("");
 }
 
 const dateFormat = new Intl.DateTimeFormat("en-US", {
@@ -85,18 +86,24 @@ async function chatgptToMarkdown(json, sourceDir, { dateFormat } = { dateFormat:
           content.content_type == "text"
             ? content.parts.join("\n")
             : content.content_type == "code"
-            ? "```" + content.language.replace("unknown", "") + "\n" + content.text + "\n```"
-            : content.content_type == "execution_output"
-            ? "```\n" + content.text + "\n```"
-            : content.content_type == "multimodal_text"
-            ? content.parts
-                .map((part) =>
-                  part.content_type === "image_asset_pointer"
-                    ? `Image (${part.width}x${part.height}): ${part?.metadata?.dalle?.prompt}\n\n`
-                    : `${part.content_type}\n\n`,
-                )
-                .join("")
-            : content;
+              ? "```" + content.language.replace("unknown", "") + "\n" + content.text + "\n```"
+              : content.content_type == "execution_output"
+                ? "```\n" + content.text + "\n```"
+                : content.content_type == "multimodal_text"
+                  ? content.parts
+                      .map((part) =>
+                        part.content_type === "image_asset_pointer"
+                          ? `Image (${part.width}x${part.height}): ${part?.metadata?.dalle?.prompt}\n\n`
+                          : `${part.content_type}\n\n`,
+                      )
+                      .join("")
+                  : content.content_type == "tether_browsing_display"
+                    ? "```\n" + content.result + "\n```"
+                    : content.content_type == "tether_quote"
+                      ? "```\n" + `${content.title} (${content.url})\n\n${content.text}` + "\n```"
+                      : content.content_type == "system_error"
+                        ? `${content.name}\n\n${content.text}\n\n`
+                        : content;
         // Ignore empty content
         if (!body.trim()) return "";
         // Indent user / tool messages. The sometimes contain code and whitespaces are relevant

package/index.test.js

@@ -47,6 +47,7 @@ describe("chatgptToMarkdown", () => {
 
     Hello
 
+
 `);
   });
 
@@ -129,6 +130,48 @@ describe("chatgptToMarkdown", () => {
     expect(fileContent).not.toContain("## user (John)");
   });
 
+  it("should handle tether_browsing_display content", async () => {
+    const json = [
+      {
+        title: "Test Conversation",
+        create_time: 1630454400,
+        update_time: 1630458000,
+        mapping: {
+          0: {
+            message: {
+              author: { role: "tool", name: "browser" },
+              content: { content_type: "tether_browsing_display", result: "L0: x" },
+            },
+          },
+        },
+      },
+    ];
+    await chatgptToMarkdown(json, tempDir);
+    const fileContent = await fs.readFile(path.join(tempDir, "Test Conversation.md"), "utf8");
+    expect(fileContent).toContain("```\nL0: x\n```");
+  });
+
+  it("should handle tether_quote content", async () => {
+    const json = [
+      {
+        title: "Test Conversation",
+        create_time: 1630454400,
+        update_time: 1630458000,
+        mapping: {
+          0: {
+            message: {
+              author: { role: "tool", name: "browser" },
+              content: { content_type: "tether_quote", url: "x.com", domain: "x.com", title: "T", text: "X" },
+            },
+          },
+        },
+      },
+    ];
+    await chatgptToMarkdown(json, tempDir);
+    const fileContent = await fs.readFile(path.join(tempDir, "Test Conversation.md"), "utf8");
+    expect(fileContent).toContain("```\nT (x.com)\n\nX\n```");
+  });
+
   it("should handle multimodal_text content", async () => {
     const json = [
       {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "chatgpt-to-markdown",
-  "version": "1.4.0",
+  "version": "1.5.1",
   "description": "Convert ChatGPT exported conversations.json to Markdown",
   "main": "index.js",
   "type": "module",
@@ -23,6 +23,6 @@
   },
   "devDependencies": {
     "jest": "^29.7.0",
-    "prettier": "^3.0.3"
+    "prettier": "^3.2.5"
   }
 }