llamaparserb 0.1.1 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b0073944d6fc5235f1ee62b86656a396afaf63cdb92aecb2c65468f57767876d
-  data.tar.gz: 1cedf4c9db42955c7e9c3baac69b8a8a78213817b505e65033c00817fdf73163
+  metadata.gz: 14e4e5deea50bc3f5cb0e32c3c3029f6b26fda4cacf91be0f96e9f677f92d1e5
+  data.tar.gz: e3e6cb569456d1c649be22ad3e5af65e30484fce50582d7493f5e22b2ac03f61
 SHA512:
-  metadata.gz: fc4f606f909a118fed822fa47515c71b45daab1e26b8b059229dff9f9e90ad338be1fde4995ac0dd2a47ffe64378a6a7dda3c163c5cc598e7637f7d4126468f1
-  data.tar.gz: 502b1ffe2170259741d42f1bae218b725e7430de5166cdb58ffa4d03536e4220be0001ed04e07f4abe6c3e3a6158b70ca42dc887217efc61e49a6d382040f35c
+  metadata.gz: 72121ad4b70f95ddd2bfe23f129d5c9ee634543543ba43e4faa437da67bf3cbdf8bc1ea828aeb58d43073bed46c28d4010a6865b5bc006c32ccfc30a099d1d65
+  data.tar.gz: 5bea9e367b71d38fe8f6a1a80be0b4fd190e3e56b780e369d33c3319dfd91b483283f4b43ab8fc484468ab611c5e3efc639d8302ba611ccfb71258c459fdb783

data/CHANGELOG.md

@@ -6,6 +6,11 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.2.0] - 2024-11-28
+### Changed
+- Allow passing in a string or an IO object to `parse_file`
+- Add support for file type parameter to `parse_file`
+
 ## [0.1.1] - 2024-11-28
 ### Changed
 - Move gem ownership to Horizing

data/README.md

@@ -32,14 +32,43 @@ require 'llamaparserb'
 # Initialize client with API key
 client = Llamaparserb::Client.new(ENV['LLAMA_CLOUD_API_KEY'])
 
-# Parse a file to text (default)
+# Parse a file from disk (to text by default)
 text = client.parse_file('path/to/document.pdf')
 
+# Parse an in-memory file (requires file type)
+require 'open-uri'
+file_content = URI.open('https://example.com/document.pdf')
+text = client.parse_file(file_content, 'pdf')
+
 # Parse a file to markdown
 client = Llamaparserb::Client.new(ENV['LLAMA_CLOUD_API_KEY'], result_type: "markdown")
 markdown = client.parse_file('path/to/document.pdf')
 ```
 
+### File Input Options
+
+The `parse_file` method accepts two types of inputs:
+
+1. File path (String):
+```ruby
+client.parse_file('path/to/document.pdf')
+```
+
+2. IO object (requires file type parameter):
+```ruby
+# From a URL
+file_content = URI.open('https://example.com/document.pdf')
+client.parse_file(file_content, 'pdf')
+
+# From memory
+io = StringIO.new(file_content)
+client.parse_file(io, 'pdf')
+
+# From a Tempfile
+temp_file = Tempfile.new(['document', '.pdf'])
+client.parse_file(temp_file, 'pdf')
+```
+
 ### Advanced Options
 
 ```ruby

data/lib/llamaparserb/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Llamaparserb
-  VERSION = "0.1.1"
+  VERSION = "0.2.0"
 end

data/lib/llamaparserb.rb

@@ -8,6 +8,7 @@ require "mime/types"
 require "uri"
 require "async"
 require "logger"
+require "tempfile"
 
 module Llamaparserb
   class Error < StandardError; end
@@ -41,17 +42,27 @@ module Llamaparserb
       @connection = build_connection
     end
 
-    def parse_file(file_path)
-      job_id = create_job(file_path)
-      log "Started parsing file under job_id #{job_id}", :info
+    def parse_file(file_input, file_type = nil)
+      case file_input
+      when String
+        # Treat as file path
+        job_id = create_job_from_path(file_input)
+        log "Started parsing file under job_id #{job_id}", :info
+      when IO, StringIO, Tempfile
+        # Treat as file object
+        raise Error, "file_type parameter is required for IO objects" unless file_type
+        job_id = create_job_from_io(file_input, file_type)
+        log "Started parsing in-memory file under job_id #{job_id}", :info
+      else
+        raise Error, "Invalid input type. Expected String (file path) or IO object, got #{file_input.class}"
+      end
 
       wait_for_completion(job_id)
-
       result = get_result(job_id)
       log "Successfully retrieved result", :info
       result
     rescue => e
-      handle_error(e, file_path)
+      handle_error(e, file_input)
     end
 
     private
@@ -144,9 +155,9 @@ module Llamaparserb
       end
     end
 
-    def handle_error(error, file_path)
+    def handle_error(error, file_input)
       if @options[:ignore_errors]
-        log "Error while parsing file '#{file_path}': #{error.message}", :error
+        log "Error while parsing file '#{file_input}'", :error
         nil
       else
         raise error
@@ -163,14 +174,37 @@ module Llamaparserb
       end
     end
 
-    def create_job(file_path)
+    def create_job_from_path(file_path)
       validate_file_type!(file_path)
-
       file = Faraday::Multipart::FilePart.new(
         file_path,
         detect_content_type(file_path)
       )
+      create_job(file)
+    end
+
+    def create_job_from_io(io, file_type)
+      # Ensure file_type starts with a dot
+      file_type = ".#{file_type}" unless file_type.start_with?(".")
+      validate_file_type!(file_type)
+
+      temp_file = Tempfile.new(["upload", file_type])
+      temp_file.binmode
+      io.rewind
+      temp_file.write(io.read)
+      temp_file.rewind
+
+      file = Faraday::Multipart::FilePart.new(
+        temp_file,
+        detect_content_type(temp_file.path)
+      )
+      create_job(file)
+    ensure
+      temp_file&.close
+      temp_file&.unlink
+    end
 
+    def create_job(file)
       response = @connection.post("upload") do |req|
         req.headers["Authorization"] = "Bearer #{api_key}"
         req.body = upload_params(file)

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: llamaparserb
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 0.2.0
 platform: ruby
 authors:
 - Heidar Bernhardsson