fix: generate two level json schema

Author: MichalKalita

src/utils/input-schema-to-markdown.ts

@@ -4,41 +4,71 @@ function visibleEmpty(value: string) {
     return value === '' ? '<empty>' : value;
 }
 
-export function inputSchemaToMarkdown(inputSchema: IActorInputSchema) {
-    const requiredFields = new Set(inputSchema.required || []);
+function formatProperty(key: string, value: any, requiredFields: Set<string>, level = 2): string {
+    const isRequired = requiredFields.has(key);
+    const requiredText = isRequired ? 'required' : 'optional';
 
-    let markdown = '# Input Schema';
-    if (inputSchema.description) {
-        markdown += '\n\n';
-        markdown += inputSchema.description;
-    }
+    let result = `${'#'.repeat(level)} \`${key}\` ${requiredText} ${value.type}`;
 
-    for (const [key, value] of Object.entries(inputSchema.properties)) {
-        const isRequired = requiredFields.has(key);
-        const requiredText = isRequired ? 'required' : 'optional';
+    if (value.format) {
+        result += ` format:${value.format}`;
+    }
 
-        let line = `## \`${key}\` ${requiredText} ${value.type}`;
+    if (value.prefill !== undefined && !Array.isArray(value.prefill)) {
+        result += ' prefill:';
+        result += visibleEmpty(String(value.prefill));
+    } else if (value.default !== undefined) {
+        result += ' default:';
+        result += visibleEmpty(String(value.default));
+    }
 
-        if (value.prefill !== undefined && !Array.isArray(value.prefill)) {
-            line += ' prefill:';
-            line += visibleEmpty(String(value.prefill));
-        } else if (value.default !== undefined) {
-            line += ' default:';
-            line += visibleEmpty(String(value.default));
+    // Handle nested properties for objects
+    if (value.type === 'object' && value.properties) {
+        result += '\n';
+        const nestedEntries = Object.entries(value.properties);
+        for (let i = 0; i < nestedEntries.length; i++) {
+            const [nestedKey, nestedValue] = nestedEntries[i];
+            result += formatProperty(nestedKey, nestedValue, requiredFields, level + 1);
+            if (i < nestedEntries.length - 1) {
+                result += '\n';
+            }
         }
+        return result;
+    }
 
-        markdown += '\n\n';
-        markdown += line;
-        markdown += '\n';
-
+    if (value.enum || value.description) {
+        result += '\n';
         if (value.enum) {
             let enumLine = 'options: ';
             enumLine += value.enum.map(visibleEmpty).join(', ');
-            markdown += enumLine;
-            markdown += '\n';
+            result += enumLine;
+            result += '\n';
+        }
+        if (value.description) {
+            result += value.description;
         }
+    }
+
+    return result;
+}
+
+export function inputSchemaToMarkdown(inputSchema: IActorInputSchema) {
+    const requiredFields = new Set(inputSchema.required || []);
 
-        markdown += value.description;
+    let markdown = '# JSON Schema';
+    if (inputSchema.description) {
+        markdown += '\n\n';
+        markdown += inputSchema.description;
+    }
+    markdown += '\n\n'; // Add blank line after title/description
+
+    const properties = Object.entries(inputSchema.properties);
+    for (let i = 0; i < properties.length; i++) {
+        const [key, value] = properties[i];
+        markdown += formatProperty(key, value, requiredFields);
+        if (i < properties.length - 1) {
+            markdown += '\n\n'; // Add blank line between properties
+        }
     }
 
     return markdown;

tests/unit/input-schema-to-markdown.test.ts

@@ -43,7 +43,7 @@ describe('inputSchemaToMarkdown', () => {
 
         const result = inputSchemaToMarkdown(schema);
         expect(result).toMatchInlineSnapshot(`
-          "# Input Schema
+          "# JSON Schema
 
           This scraper will get post and page details from Facebook pages of your choice. To try it out, just paste a Facebook Page URL and click ▷ Start. If you need any guidance, just <a href='https://blog.apify.com/scrape-facebook-posts-data/' target='_blank' rel='noopener'>follow this tutorial</a>.
 
@@ -176,7 +176,7 @@ describe('inputSchemaToMarkdown', () => {
 
         const result = inputSchemaToMarkdown(schema);
         expect(result).toMatchInlineSnapshot(`
-          "# Input Schema
+          "# JSON Schema
 
           To extract contact details from Google places, simply enter 🔍 <b>Search term</b>, add 📍 <b>Location</b>, and 💯 <b>Number of places</b> to extract. Section 🎯 <b>Filters</b> contains various extra features, filters, and sorting options. <br><br> Sections <b>with asterisk*</b> are just alternative ways to start the input (📡 Geolocation parameters, 🛰 Polygons, 🔗 URLs). They can be combined with any of the features and sorting options from the <b>Filters</b> section.
 
@@ -234,4 +234,66 @@ describe('inputSchemaToMarkdown', () => {
           Max 300 results per search URL. Valid format for URLs contains <code>google.com/maps/</code>. This feature also supports uncommon URL formats such as: <code>google.com?cid=***</code>, <code>goo.gl/maps</code>, and custom place list URL."
         `);
     });
+
+    it('should format schema for rag web browser results', () => {
+        const schema = {
+            type: 'object',
+            properties: {
+                crawl: {
+                    type: 'object',
+                    properties: {
+                        httpStatusCode: {
+                            type: 'integer',
+                        },
+                        httpStatusMessage: {
+                            type: 'string',
+                        },
+                        loadedAt: {
+                            type: 'string',
+                            format: 'date-time',
+                        },
+                        uniqueKey: {
+                            type: 'string',
+                        },
+                        requestStatus: {
+                            type: 'string',
+                        },
+                    },
+                },
+                searchResult: {
+                    type: 'object',
+                },
+                metadata: {
+                    type: 'object',
+                },
+                query: {
+                    type: 'string',
+                },
+                markdown: {
+                    type: 'string',
+                    format: 'style',
+                },
+            },
+        };
+
+        const result = inputSchemaToMarkdown(schema);
+        expect(result).toMatchInlineSnapshot(`
+          "# JSON Schema
+
+          ## \`crawl\` optional object
+          ### \`httpStatusCode\` optional integer
+          ### \`httpStatusMessage\` optional string
+          ### \`loadedAt\` optional string format:date-time
+          ### \`uniqueKey\` optional string
+          ### \`requestStatus\` optional string
+
+          ## \`searchResult\` optional object
+
+          ## \`metadata\` optional object
+
+          ## \`query\` optional string
+
+          ## \`markdown\` optional string format:style"
+        `);
+    });
 });