OpenAI to Hashnode

[Surgical17]

Hi Everyone

So I am busy building a process in N8N with the intent of creating a draft blog post for myself using OpenAi. Trigger is a webhook where my draft data is sent, then gets sent to openai where it creates a draft for me (Best would be that the output is in markdown and ready for a blog post, then from there I want to send it to hashnode via GraphQl. But I am expoeriencing many issues from OpenAI Output being \n to quotes in it that messes with the variables json etc etc.

Has anyone made anything similar or any sugegstions how to build a working process. Been at it for weeks now

[n8n]

It looks like your topic is missing some important information. Could you provide the following if applicable.

• n8n version:
• Database (default: SQLite):
• n8n EXECUTIONS_PROCESS setting (default: own, main):
• Running n8n via (Docker, npm, n8n cloud, desktop app):
• Operating system:

[gualter]

hi @Surgical17

You might need to polish the response using the code node to prepare it for the next node. Can you share your workflow and an example response from OpenAI output?

Thanks!

[Surgical17]

Unfraternally I do not have the plan where I can share the workflow, but screenshots and code below:
Workflow:

Javascript code to cleanup Markdown:

return items.map(item => {
    let content = item.json.message.content;

    // 1. Normalize line breaks to \n
    content = content.replace(/\r\n/g, '\n').replace(/\r/g, '\n');

    // 2. Ensure there is a space after headers (e.g., '#Header' -> '# Header')
    content = content.replace(/^(#{1,6})(\S)/gm, '$1 $2');

    // 3. Ensure two line breaks between paragraphs
    content = content.replace(/([^\n])\n([^\n])/g, '$1\n\n$2');

    // 4. Trim trailing spaces on each line
    content = content.replace(/[ \t]+$/gm, '');

    // 5. Ensure code blocks have proper line breaks before and after
    content = content.replace(/```/g, '\n```\n');

    // 6. Remove any unnecessary multiple blank lines
    content = content.replace(/\n{3,}/g, '\n\n');

    // Trim the entire content
    content = content.trim();

    // Initialize title as null
    let title = null;

    // Split content into lines
    const lines = content.split('\n');

    // Check if the first line is a Markdown header
    const headerMatch = lines[0].match(/^(#{1,6})\s*(.*)$/);

    if (headerMatch) {
        // Extract the title without the '#' symbols and any leading/trailing spaces
        title = headerMatch[2].trim();

        // Remove the first line from the content
        lines.shift();

        // Rejoin the remaining lines into the cleaned content
        content = lines.join('\n').trim();
    }

    // Replace all double quotes with single quotes to avoid JSON issues
    content = content.replace(/"/g, "'");

    // Construct the desired JSON structure
    const output = {
        input: {
            title: title,
            contentMarkdown: content,
            publicationId: "my iD"
        }
    };

    return {
        output
    };
});

GraphQL Query in Json Format Request:

mutation CreateDraft($input: CreateDraftInput!) {
  createDraft(input: $input) {
    draft {
      id
      slug
      title
      subtitle
      author {
        username
        name
      }
      tagsV2 {
        __typename
      }
      canonicalUrl
      coverImage {
        url
      }
      readTimeInMinutes
      content {
        markdown
      }
      updatedAt
      seo {
        title
        description
      }
    }
  }
}

Json Variables for GraphQL:

{
  "input": {
    "title": "{{ $json.output.input.title }}",
    "contentMarkdown": "{{ $json.output.input.contentMarkdown }}",
    "publicationId": "my ID"
  }
}

A few Notes:

1. MY Javascript outputs the Variables in Json, this was just part of a test as I will explain below to try and just output the entire Json Variables
2. The main section causing issues is this section (“contentMarkdown”: “{{ $json.output.input.contentMarkdown }}”) Reason be the content in this section and output from OpenAI looks like This “#Header Title \n\n ##second title \n\n Text”

The issue is when there is characters such as " or \n in my json it cannot be send to Hashnode. But if I only have text with no special characters and add the markdown manually (“contentMarkdown”: “{{ $json.output.input.contentMarkdown1 }}\n\n{{ $json.output.input.contentMarkdown2 }}”) then it works fine.

[gualter]

Thanks for sharing this! What I meant by sharing the workflow was adding it inside markdown code:

paste workflow here

You can copy it directly from the canvas.

As for the issue, have you tried escaping the “\n” in the markdown directly by adding another backslash:

\\n