Merge pull request #1072 from mendableai/nsc/json-format

(feat/formats) Extract format renamed to json format

README.md

@@ -250,8 +250,8 @@ curl -X POST https://api.firecrawl.dev/v1/scrape \
     -H 'Authorization: Bearer YOUR_API_KEY' \
     -d '{
       "url": "https://www.mendable.ai/",
-      "formats": ["extract"],
+      "formats": ["json"],
-      "extract": {
+      "jsonOptions": {
         "schema": {
           "type": "object",
           "properties": {
@@ -296,7 +296,7 @@ curl -X POST https://api.firecrawl.dev/v1/scrape \
       "ogSiteName": "Mendable",
       "sourceURL": "https://mendable.ai/"
     },
-    "llm_extraction": {
+    "json": {
       "company_mission": "Train a secure AI on your technical resources that answers customer and employee questions so your team doesn't have to",
       "supports_sso": true,
       "is_open_source": false,
@@ -316,8 +316,8 @@ curl -X POST https://api.firecrawl.dev/v1/scrape \
     -H 'Authorization: Bearer YOUR_API_KEY' \
     -d '{
       "url": "https://docs.firecrawl.dev/",
-      "formats": ["extract"],
+      "formats": ["json"],
-      "extract": {
+      "jsonOptions": {
         "prompt": "Extract the company mission from the page."
       }
     }'
@@ -447,12 +447,12 @@ class TopArticlesSchema(BaseModel):
     top: List[ArticleSchema] = Field(..., max_items=5, description="Top 5 stories")
 
 data = app.scrape_url('https://news.ycombinator.com', {
-    'formats': ['extract'],
+    'formats': ['json'],
-    'extract': {
+    'jsonOptions': {
         'schema': TopArticlesSchema.model_json_schema()
     }
 })
-print(data["extract"])
+print(data["json"])
 ```
 
 ## Using the Node SDK
@@ -526,10 +526,10 @@ const schema = z.object({
 });
 
 const scrapeResult = await app.scrapeUrl("https://news.ycombinator.com", {
-  extractorOptions: { extractionSchema: schema },
+  jsonOptions: { extractionSchema: schema },
 });
 
-console.log(scrapeResult.data["llm_extraction"]);
+console.log(scrapeResult.data["json"]);
 ```
 
 ## Open Source vs Cloud Offering

apps/api/src/controllers/v1/scrape.ts

@@ -33,6 +33,7 @@ export async function scrapeController(
     basePriority: 10,
   });
 
+
   await addScrapeJob(
     {
       url: req.body.url,
@@ -96,7 +97,7 @@ export async function scrapeController(
     // Don't bill if we're early returning
     return;
   }
-  if (req.body.extract && req.body.formats.includes("extract")) {
+  if (req.body.extract && req.body.formats.includes("extract") ) {
     creditsToBeBilled = 5;
   }
 

apps/api/src/controllers/v1/types.ts

@@ -125,6 +125,7 @@ export const scrapeOptions = z
         "screenshot",
         "screenshot@fullPage",
         "extract",
+        "json"
       ])
       .array()
       .optional()
@@ -139,7 +140,10 @@ export const scrapeOptions = z
     onlyMainContent: z.boolean().default(true),
     timeout: z.number().int().positive().finite().safe().optional(),
     waitFor: z.number().int().nonnegative().finite().safe().default(0),
+    // Deprecate this to jsonOptions
     extract: extractOptions.optional(),
+    // New
+    jsonOptions: extractOptions.optional(),
     mobile: z.boolean().default(false),
     parsePDF: z.boolean().default(true),
     actions: actionsSchema.optional(),
@@ -242,20 +246,43 @@ export const scrapeRequestSchema = scrapeOptions
     (obj) => {
       const hasExtractFormat = obj.formats?.includes("extract");
       const hasExtractOptions = obj.extract !== undefined;
+      const hasJsonFormat = obj.formats?.includes("json");
+      const hasJsonOptions = obj.jsonOptions !== undefined;
       return (
         (hasExtractFormat && hasExtractOptions) ||
-        (!hasExtractFormat && !hasExtractOptions)
+        (!hasExtractFormat && !hasExtractOptions) ||
+        (hasJsonFormat && hasJsonOptions) ||
+        (!hasJsonFormat && !hasJsonOptions)
       );
     },
     {
       message:
-        "When 'extract' format is specified, 'extract' options must be provided, and vice versa",
+        "When 'extract' or 'json' format is specified, corresponding options must be provided, and vice versa",
     },
   )
   .transform((obj) => {
-    if ((obj.formats?.includes("extract") || obj.extract) && !obj.timeout) {
+    // Handle timeout
-      return { ...obj, timeout: 60000 };
+    if ((obj.formats?.includes("extract") || obj.extract || obj.formats?.includes("json") || obj.jsonOptions) && !obj.timeout) {
+      obj = { ...obj, timeout: 60000 };
     }
+
+    if(obj.formats?.includes("json")) {
+      obj.formats.push("extract");
+    }
+
+    // Convert JSON options to extract options if needed
+    if (obj.jsonOptions && !obj.extract) {
+      obj = {
+        ...obj,
+        extract: {
+          prompt: obj.jsonOptions.prompt,
+          systemPrompt: obj.jsonOptions.systemPrompt,
+          schema: obj.jsonOptions.schema,
+          mode: "llm"
+        }
+      };
+    }
+
     return obj;
   });
 
@@ -410,6 +437,7 @@ export type Document = {
   links?: string[];
   screenshot?: string;
   extract?: any;
+  json?: any;
   warning?: string;
   actions?: {
     screenshots?: string[];

apps/api/src/scraper/scrapeURL/transformers/llmExtract.ts

@@ -233,7 +233,12 @@ export async function performLLMExtract(
       document.markdown,
       document.warning,
     );
+    
+    if (meta.options.formats.includes("json")) {
+      document.json = extract;
+    } else {
       document.extract = extract;
+    }
     document.warning = warning;
   }
 

apps/js-sdk/firecrawl/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mendable/firecrawl-js",
-  "version": "1.14.1",
+  "version": "1.15.0",
   "description": "JavaScript SDK for Firecrawl API",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",

apps/js-sdk/firecrawl/src/index.ts

@@ -78,7 +78,7 @@ export interface FirecrawlDocument<T = any, ActionsSchema extends (ActionsResult
  * Defines the options and configurations available for scraping web content.
  */
 export interface CrawlScrapeOptions {
-  formats: ("markdown" | "html" | "rawHtml" | "content" | "links" | "screenshot" | "screenshot@fullPage" | "extract")[];
+  formats: ("markdown" | "html" | "rawHtml" | "content" | "links" | "screenshot" | "screenshot@fullPage" | "extract" | "json")[];
   headers?: Record<string, string>;
   includeTags?: string[];
   excludeTags?: string[];
@@ -127,6 +127,11 @@ export interface ScrapeParams<LLMSchema extends zt.ZodSchema = any, ActionsSchem
     schema?: LLMSchema;
     systemPrompt?: string;
   };
+  json?:{
+    prompt?: string;
+    schema?: LLMSchema;
+    systemPrompt?: string;
+  }
   actions?: ActionsSchema;
 }
 
@@ -393,6 +398,23 @@ export default class FirecrawlApp {
         },
       };
     }
+
+    if (jsonData?.jsonOptions?.schema) {
+      let schema = jsonData.jsonOptions.schema;
+      // Try parsing the schema as a Zod schema
+      try {
+        schema = zodToJsonSchema(schema);
+      } catch (error) {
+        
+      }
+      jsonData = {
+        ...jsonData,
+        jsonOptions: {
+          ...jsonData.jsonOptions,
+          schema: schema,
+        },
+      };
+    }
     try {
       const response: AxiosResponse = await axios.post(
         this.apiUrl + `/v1/scrape`,
@@ -772,6 +794,23 @@ export default class FirecrawlApp {
         },
       };
     }
+    if (jsonData?.jsonOptions?.schema) {
+      let schema = jsonData.jsonOptions.schema;
+
+      // Try parsing the schema as a Zod schema
+      try {
+        schema = zodToJsonSchema(schema);
+      } catch (error) {
+        
+      }
+      jsonData = {
+        ...jsonData,
+        jsonOptions: {
+          ...jsonData.jsonOptions,
+          schema: schema,
+        },
+      };
+    }
     try {
       const response: AxiosResponse = await this.postRequest(
         this.apiUrl + `/v1/batch/scrape`,

apps/python-sdk/firecrawl/__init__.py

@@ -13,7 +13,7 @@ import os
 
 from .firecrawl import FirecrawlApp # noqa
 
-__version__ = "1.9.0"
+__version__ = "1.10.0"
 
 # Define the logger for the Firecrawl project
 logger: logging.Logger = logging.getLogger("firecrawl")

apps/python-sdk/firecrawl/firecrawl.py

@@ -112,6 +112,18 @@ class FirecrawlApp:
                 if key not in ['extract']:
                     scrape_params[key] = value
 
+            json = params.get("jsonOptions", {})
+            if json:
+                if 'schema' in json and hasattr(json['schema'], 'schema'):
+                    json['schema'] = json['schema'].schema()
+                scrape_params['jsonOptions'] = json
+
+            # Include any other params directly at the top level of scrape_params
+            for key, value in params.items():
+                if key not in ['jsonOptions']:
+                    scrape_params[key] = value
+
+
         endpoint = f'/v1/scrape'
         # Make the POST request with the prepared headers and JSON data
         response = requests.post(

apps/python-sdk/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 dynamic = ["version"]
-name = "firecrawl-py"
+name = "firecrawl"
 description = "Python SDK for Firecrawl API"
 readme = {file="README.md", content-type = "text/markdown"}
 requires-python = ">=3.8"

apps/python-sdk/setup.py

@@ -17,7 +17,7 @@ def get_version():
 
 
 setup(
-    name="firecrawl-py",
+    name="firecrawl",
     version=get_version(),
     url="https://github.com/mendableai/firecrawl",
     author="Mendable.ai",