我有一个 Rails 7.0.4 应用程序，其中有一个简单的 TranscriptionController，一旦提交了包含标题、描述和 audio_file 的表单，就会启动一个 Action Job 以使用 openai gem 处理音频转录。尽管生成的 URL 指向我要转码的 audio_file，但我仍然收到此错误：
Errno::ENOENT (No such file or directory @ rb_sysopen这是转录模型：
class Transcription < ApplicationRecord
has_one_attached :audio_file
def audio_file_url
Rails.application.routes.url_helpers.url_for(audio_file) if audio_file.attached?
end
end
这是 TranscriptionController 中的创建操作：
def create
@transcription = Transcription.new(transcription_params)
respond_to do |format|
if @transcription.save
# Process audio file using OpenAI's Whisper API
TranscriptionJob.perform_later(@transcription)
format.html { redirect_to transcription_url(@transcription), notice: "Transcription was successfully created. Check back later for transcription text." }
format.json { render :show, status: :created, location: @transcription }
else
format.html { render :new, status: :unprocessable_entity }
format.json { render json: @transcription.errors, status: :unprocessable_entity }
end
end
end
这里是最重要的部分，名为 transcription_job.rb 的 Active Job：
class TranscriptionJob < ApplicationJob
queue_as :default
require 'openai'
def perform(transcription)
# Do something later
client = OpenAI::Client.new(access_token: Rails.application.credentials.dig(:openai, :api_key))
logger.info "SMC DEBUG: OpenAI Key set"
file_path = transcription.audio_file_url
logger.info "SMC DEBUG: Audio file path is: #{file_path}"
response = client.transcribe(
parameters: {
model: "whisper-1",
file: File.open(file_path, "rb")
})
logger.info "SMC DEBUG: transcription sent"
# Update the transcription object with the returned text
transcription.transcriptionresult = response['text']
logger.info "SMC DEBUG: TranscriptionResult is: #{transcription.transcriptionresult}"
# Save the updated transcription object
transcription.save
end
end
以下是 Rails 日志中针对 Action Job 的一些相关反馈：
15:21:58 web.1 | [ActiveJob] Enqueued ActiveStorage::AnalyzeJob (Job ID: 7b3600ec-ee85-4be6-8faa-18f467fc719a) to Async(default) with arguments: #<GlobalID:0x00000001073c6880 @uri=#<URI::GID gid://transcriptionservice/ActiveStorage::Blob/59>>
15:21:58 web.1 | [ActiveJob] Enqueued TranscriptionJob (Job ID: 974566f7-7d83-48ee-b3ae-d6f02a006efb) to Async(default) with arguments: #<GlobalID:0x00000001073d7ba8 @uri=#<URI::GID gid://transcriptionservice/Transcription/61>>
15:21:58 web.1 | Redirected to http://localhost:3000/transcriptions/61
15:21:58 web.1 | Completed 302 Found in 60ms (ActiveRecord: 13.1ms | Allocations: 24625)
15:21:58 web.1 |
15:21:58 web.1 |
15:21:58 web.1 | Started GET "/transcriptions/61" for ::1 at 2023-05-14 15:21:58 -0400
15:21:58 web.1 | [ActiveJob] [ActiveStorage::AnalyzeJob] [7b3600ec-ee85-4be6-8faa-18f467fc719a] ActiveStorage::Blob Load (2.2ms) SELECT "active_storage_blobs".* FROM "active_storage_blobs" WHERE "active_storage_blobs"."id" = $1 LIMIT $2 [["id", 59], ["LIMIT", 1]]
15:21:58 web.1 | [ActiveJob] [ActiveStorage::AnalyzeJob] [7b3600ec-ee85-4be6-8faa-18f467fc719a] Performing ActiveStorage::AnalyzeJob (Job ID: 7b3600ec-ee85-4be6-8faa-18f467fc719a) from Async(default) enqueued at 2023-05-14T19:21:58Z with arguments: #<GlobalID:0x0000000107426ac8 @uri=#<URI::GID gid://transcriptionservice/ActiveStorage::Blob/59>>
15:21:58 web.1 | [ActiveJob] [ActiveStorage::AnalyzeJob] [7b3600ec-ee85-4be6-8faa-18f467fc719a] Disk Storage (0.6ms) Downloaded file from key: 49zqmrx6g54pk1pzuc3y2uyvtiyx
15:21:58 web.1 | [ActiveJob] [TranscriptionJob] [974566f7-7d83-48ee-b3ae-d6f02a006efb] Transcription Load (0.3ms) SELECT "transcriptions".* FROM "transcriptions" WHERE "transcriptions"."id" = $1 LIMIT $2 [["id", 61], ["LIMIT", 1]]
15:21:58 web.1 | [ActiveJob] [TranscriptionJob] [974566f7-7d83-48ee-b3ae-d6f02a006efb] Performing TranscriptionJob (Job ID: 974566f7-7d83-48ee-b3ae-d6f02a006efb) from Async(default) enqueued at 2023-05-14T19:21:58Z with arguments: #<GlobalID:0x000000010743ca80 @uri=#<URI::GID gid://transcriptionservice/Transcription/61>>
15:21:58 web.1 | [ActiveJob] [TranscriptionJob] [974566f7-7d83-48ee-b3ae-d6f02a006efb] SMC DEBUG: OpenAI Key set
15:21:58 web.1 | Processing by TranscriptionsController#show as TURBO_STREAM
15:21:58 web.1 | Parameters: {"id"=>"61"}
15:21:58 web.1 | [ActiveJob] [TranscriptionJob] [974566f7-7d83-48ee-b3ae-d6f02a006efb] ActiveStorage::Attachment Load (7.0ms) SELECT "active_storage_attachments".* FROM "active_storage_attachments" WHERE "active_storage_attachments"."record_id" = $1 AND "active_storage_attachments"."record_type" = $2 AND "active_storage_attachments"."name" = $3 LIMIT $4 [["record_id", 61], ["record_type", "Transcription"], ["name", "audio_file"], ["LIMIT", 1]]
15:21:58 web.1 | [ActiveJob] [TranscriptionJob] [974566f7-7d83-48ee-b3ae-d6f02a006efb] ↳ app/models/transcription.rb:5:in `audio_file_url'
15:21:58 web.1 | Transcription Load (1.7ms) SELECT "transcriptions".* FROM "transcriptions" WHERE "transcriptions"."id" = $1 LIMIT $2 [["id", 61], ["LIMIT", 1]]
15:21:58 web.1 | ↳ app/controllers/transcriptions_controller.rb:83:in `set_transcription'
15:21:58 web.1 | Rendering layout layouts/application.html.erb
15:21:58 web.1 | Rendering transcriptions/show.html.erb within layouts/application
15:21:58 web.1 | Rendered transcriptions/_transcription.html.erb (Duration: 0.0ms | Allocations: 25)
15:21:58 web.1 | [ActiveJob] [TranscriptionJob] [974566f7-7d83-48ee-b3ae-d6f02a006efb] ActiveStorage::Blob Load (1.4ms) SELECT "active_storage_blobs".* FROM "active_storage_blobs" WHERE "active_storage_blobs"."id" = $1 LIMIT $2 [["id", 59], ["LIMIT", 1]]
15:21:58 web.1 | Rendered transcriptions/show.html.erb within layouts/application (Duration: 1.1ms | Allocations: 843)
15:21:58 web.1 | [ActiveJob] [TranscriptionJob] [974566f7-7d83-48ee-b3ae-d6f02a006efb] ↳ app/models/transcription.rb:5:in `audio_file_url'
15:21:58 web.1 | [ActiveJob] [TranscriptionJob] [974566f7-7d83-48ee-b3ae-d6f02a006efb] SMC DEBUG: Audio file path is: http://localhost:3000/rails/active_storage/blobs/redirect/eyJfcmFpbHMiOnsibWVzc2FnZSI6IkJBaHBRQT09IiwiZXhwIjpudWxsLCJwdXIiOiJibG9iX2lkIn19--495ff1b4e7c2001a5ca50886bfd85e2ad6847c7b/test_audio.mp3
15:21:58 web.1 | [ActiveJob] [TranscriptionJob] [974566f7-7d83-48ee-b3ae-d6f02a006efb] Error performing TranscriptionJob (Job ID: 974566f7-7d83-48ee-b3ae-d6f02a006efb) from Async(default) in 25.03ms: Errno::ENOENT (No such file or directory @ rb_sysopen - http://localhost:3000/rails/active_storage/blobs/redirect/eyJfcmFpbHMiOnsibWVzc2FnZSI6IkJBaHBRQT09IiwiZXhwIjpudWxsLCJwdXIiOiJibG9iX2lkIn19--495ff1b4e7c2001a5ca50886bfd85e2ad6847c7b/test_audio.mp3):
当我尝试将该 url 放入我的浏览器时，它会下载正确的文件。
我尝试使用以下内容来引用路径而不是 URL：
file_path = Rails.application.routes.url_helpers.rails_blob_path(transcription.audio_file, only_path: true)
但这给了我一个类似的错误指向路径而不是 url:
15:38:35 web.1 | [ActiveJob] [TranscriptionJob] [854d3393-ed0b-41b7-8f5f-7d6469feb8a1] Error performing TranscriptionJob (Job ID: 854d3393-ed0b-41b7-8f5f-7d6469feb8a1) from Async(default) in 134.93ms: Errno::ENOENT (No such file or directory @ rb_sysopen - /rails/active_storage/blobs/redirect/eyJfcmFpbHMiOnsibWVzc2FnZSI6IkJBaHBRUT09IiwiZXhwIjpudWxsLCJwdXIiOiJibG9iX2lkIn19--f6270379699bf876802dd28de4cb4414cc7f96ba/test_audio.mp3):
我不知道如何解决这个问题。我很想从智囊团那里找到答案。
感谢@Chiperific 将我指向这个边缘指南：在这里输入链接描述这帮助我重写了我的代码如下。关键是将文件下载到系统上的临时文件位置，然后在 openai API 中指向该本地路径。
class TranscriptionJob < ApplicationJob
queue_as :default
require 'openai'
def perform(transcription)
# Provide the openai API key and initialize an instance of the openai client
client = OpenAI::Client.new(access_token: Rails.application.credentials.dig(:openai, :api_key))
# Initialize the transcriptionresult variable
transcription.transcriptionresult = ''
# It is necessary to download the file to temporary storage location first and then submit with the openai API.
transcription.audio_file.open do |file|
system '/tmp/transcribe', file.path
response = client.transcribe(
parameters: {
model: "whisper-1",
file: File.open(file.path, "rb")
})
# Make the transcriptionresult equal to the response['text'] produced by the openai transcription.
transcription.transcriptionresult = response['text']
# Save the updated transcription object
transcription.save
end
end
end