Initial implementation using polling

Author: inukshuk

src/plugin.js

@@ -1,33 +1,158 @@
+const { Buffer } = require('node:buffer')
+const { setTimeout } = require('node:timers/promises')
+
+const TEXT_TITAN = 356425
+const PRINT_M1 = 39995
+
+const md5 = (data, encoding = 'hex') => {
+  let hash = createHash('md5')
+  hash.update(data)
+  return hash.digest(encoding)
+}
+
 export default class ArtimiPlugin {
   constructor (options, context) {
     this.options = Object.assign({}, ArtimiPlugin.defaults, options)
     this.context = context
+    this.controller = new AbortController()
   }
 
   async transcribe (transcription, { buffer, ...raw }) {
+    let { logger } = this.context
+    let { host, token, htr, model } = this.options
+
+    let images = [await this.upload(buffer)]
+    let config = {
+      model: model || (htr ? TEXT_TITAN : PRINT_M1)
+    }
+
+    logger.info('Submit transcription request...')
+    let res = await fetch(`${host}/transcription`, {
+      method: 'POST',
+      headers: {
+        'Authorization': `Bearer ${token}`,
+        'Content-Type': 'application/json'
+      },
+      body: JSON.stringify({ config, images })
+    })
+
+    if (!res.ok) {
+      throw new Error(await res.text())
+    }
+
+    let job = await res.json()
+    transcription.config.jobId = job.id
+
+    logger.info(`Fetch transcription #${job.id}...`)
+    await this.poll(job)
+
+    transcription.text = job.output.text
+    transcription.data = job.output.alto
+  }
+
+  async poll (job) {
+    let { logger } = this.context
+    let { host, interval, maxRetries } = this.options
+
+    let numRetries = 0
+
+    while (true) {
+      let next
+
+      try {
+        let res = await fetch(`${host}/transcription/${job.id}`, {
+          headers: {
+            Authorization: `Bearer ${token}`
+          }
+        })
+
+        if (!res.ok) {
+          throw new Error(await res.text())
+        }
+
+        next = await res.json()
+      } catch (err) {
+        logger.error({ stack: err.stack }, `Request failed: ${err.message}`)
+        if (++numRetries > maxRetries) {
+          throw err
+        }
+      }
+
+      switch (next.state) {
+        case 'completed':
+          return next
+        case 'created':
+        case 'active':
+        case 'paused':
+          break
+        default:
+          throw new Error(`transcription request state "${next.state}"`)
+      }
+
+      await new Promise(resolve => {
+        setTimeout(interval, null, {
+          signal: this.controller.signal
+        }).then(resolve, resolve)
+      })
+    }
+  }
+
+  async upload (buffer) {
     let { logger, sharp } = this.context
+    let { host, token } = this.options
 
     // TODO use PNG if image uses alpha
+    let type = 'jpeg'
     let image = await sharp.toBuffer('jpeg', buffer, { raw })
 
-    logger.info('Upload image to cache...')
-    logger.info('Submit transcription request...')
-    logger.info('Fetch transcription...')
+    let checksum = md5(image)
+    let contentMd5 = Buffer.from(checksum, 'hex').toString('base64')
+    let contentType = `image/${type}`
 
-    // logger.info(`transkribus process id set to ${proc.id}`)
-    // transcription.config.proc = proc.id
-    // await session.poll(proc)
+    let res = await fetch(`${host}/uploads/${checksum}.${type}`, {
+      method: 'PUT',
+      redirect: 'manual',
+      headers: {
+        'Authorization': `Bearer ${token}`,
+        'X-Content-Length': image.length,
+        'Content-MD5': contentMd5,
+        'Content-Type': contentType
+      }
+    })
 
-    // transcription.text = proc.content.text
-    // transcription.data = await session.alto(proc)
+    switch (res.status) {
+      case 204:
+        logger.info('Image already cached...')
+        return `${checksum}.${type}`
+      case 307:
+        logger.info('Upload image to cache...')
+        res = await fetch(res.headers.get('location'), {
+          method: 'PUT',
+          body: image,
+          headers: {
+            'Content-MD5': contentMd5,
+            'Content-Type': contentType
+          }
+        })
+        if (!res.ok) {
+          throw new Error(await res.text())
+        }
+        return `${checksum}.${type}`
+      default:
+        throw new Error(await res.text())
+    }
   }
 
   async unload () {
+    this.conroller.abort()
   }
 }
 
 ArtimiPlugin.defaults = {
   htr: true,
   model: null,
-  token: null
+  token: null,
+  interval: 10_000,
+  maxRetries: 3,
+  host: 'http://localhost:3000'
 }