llamaparserb 0.1.1 → 0.2.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b0073944d6fc5235f1ee62b86656a396afaf63cdb92aecb2c65468f57767876d
-  data.tar.gz: 1cedf4c9db42955c7e9c3baac69b8a8a78213817b505e65033c00817fdf73163
+  metadata.gz: 7bc729e8371ed2f748eaef9a92b82380c2e7d674caf6c4fe1b17e7e53d8ea62a
+  data.tar.gz: 8f510a5be8efc2877b617bd7e4682a4cb0e92c4fd794e6e6ef036481f3d57284
 SHA512:
-  metadata.gz: fc4f606f909a118fed822fa47515c71b45daab1e26b8b059229dff9f9e90ad338be1fde4995ac0dd2a47ffe64378a6a7dda3c163c5cc598e7637f7d4126468f1
-  data.tar.gz: 502b1ffe2170259741d42f1bae218b725e7430de5166cdb58ffa4d03536e4220be0001ed04e07f4abe6c3e3a6158b70ca42dc887217efc61e49a6d382040f35c
+  metadata.gz: 5e6c3d6df9c69da63cf631a296c9618c074fc127c48cc23b398237dc015f4cb4d2bf4d2f4f50b7ed380ac053ff98c19ea9fc3530d890db34cf090e2084fb0821
+  data.tar.gz: 04b10a441f82670dc58d873d586cdd043b94f1a477b2d72b6494128d9ab76cd2f8656b1c5db0ad61487d4efe51dbc42e63b047c8101f6f74593d29e098965b52

data/CHANGELOG.md

@@ -6,6 +6,15 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.2.1] - 2024-11-28
+### Fixed
+- Fix parse_file to handle files that are not on the local filesystem
+
+## [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 = "0.2.1"
 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 +42,34 @@ 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
+        if file_type
+          job_id = create_job_from_io(file_input, file_type)
+          log "Started parsing binary data under job_id #{job_id}", :info
+        elsif File.exist?(file_input)
+          job_id = create_job_from_path(file_input)
+          log "Started parsing file under job_id #{job_id}", :info
+        else
+          raise Error, "file_type parameter is required for binary string input"
+        end
+      when IO, StringIO, Tempfile
+        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)
+      raise unless @options[:ignore_errors]
+      nil
     end
 
     private
@@ -97,15 +115,19 @@ module Llamaparserb
 
     def log(message, level = :debug)
       return unless @options[:verbose]
+
+      # Convert message to string and force UTF-8 encoding, replacing invalid characters
+      safe_message = message.to_s.encode("UTF-8", invalid: :replace, undef: :replace, replace: "?")
+
       case level
       when :info
-        logger.info(message)
+        logger.info(safe_message)
       when :warn
-        logger.warn(message)
+        logger.warn(safe_message)
       when :error
-        logger.error(message)
+        logger.error(safe_message)
       else
-        logger.debug(message)
+        logger.debug(safe_message)
       end
     end
 
@@ -144,9 +166,15 @@ 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
+        safe_message = if file_input.is_a?(String) && !File.exist?(file_input)
+          "binary data"
+        else
+          file_input.class.to_s
+        end
+
+        log "Error while parsing file (#{safe_message}): #{error.message}", :error
         nil
       else
         raise error
@@ -163,14 +191,43 @@ 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_or_string, file_type)
+      file_type = ".#{file_type}" unless file_type.start_with?(".")
+      validate_file_type!(file_type)
 
+      temp_file = Tempfile.new(["upload", file_type])
+      temp_file.binmode
+
+      case io_or_string
+      when String
+        temp_file.write(io_or_string.force_encoding("ASCII-8BIT"))
+      else
+        io_or_string.rewind if io_or_string.respond_to?(:rewind)
+        temp_file.write(io_or_string.read.force_encoding("ASCII-8BIT"))
+      end
+
+      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)
@@ -233,7 +290,12 @@ module Llamaparserb
     end
 
     def validate_file_type!(file_path)
-      extension = File.extname(file_path).downcase
+      extension = if file_path.start_with?(".")
+        file_path
+      else
+        File.extname(file_path).downcase
+      end
+
       unless SUPPORTED_FILE_TYPES.include?(extension)
         raise Error, "Unsupported file type: #{extension}. Supported types: #{SUPPORTED_FILE_TYPES.join(", ")}"
       end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: llamaparserb
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 0.2.1
 platform: ruby
 authors:
 - Heidar Bernhardsson