introduce entrypoint to get the status of async process

Inist-CNRS · Dec 20, 2024 · 5d3ab2c · 5d3ab2c
1 parent fa71402
commit 5d3ab2c
Show file tree

Hide file tree

Showing 9 changed files with 90 additions and 19 deletions.
diff --git a/services/data-computer/examples.http b/services/data-computer/examples.http
@@ -17,6 +17,17 @@ Content-Type: application/json
 	}
 ]
 
+###
+# @name v1RetrieveStatus
+POST {{host}}/v1/retrieve-status HTTP/1.1
+Content-Type: application/json
+
+[
+	{
+		"value":"36ji5xEk9"
+	}
+]
+
 ###
 # @name v1RetrieveJSON
 POST {{host}}/v1/retrieve-json?indent=true HTTP/1.1

diff --git a/services/data-computer/v1/buffer-pdf.cfg b/services/data-computer/v1/buffer-pdf.cfg
@@ -16,8 +16,9 @@ content = get('value')
 identifier = get('id')
 location = /tmp/upload
 
-[debug]
-text = fix('Data received by', env('generator'), 'for', env('identifier')).join(' ')
+[combine]
+primer = PDF received
+file = timestamp.cfg
 
 [exchange]
 value = get('filename')

diff --git a/services/data-computer/v1/buffer.cfg b/services/data-computer/v1/buffer.cfg
@@ -13,8 +13,9 @@ identifier = env('identifier')
 location = /tmp/upload
 compress = true
 
-[debug]
-text = fix('Data received by', env('generator'), 'for', env('identifier')).join(' ')
+[combine]
+primer = Data received
+file = timestamp.cfg
 
 [exchange]
 value = get('filename')

diff --git a/services/data-computer/v1/charger.cfg b/services/data-computer/v1/charger.cfg
@@ -28,3 +28,8 @@ value = self().omit([env('pid')])
 
 [metrics]
 bucket = charger
+
+[combine]
+primer = Data charged
+file = timestamp.cfg
+
diff --git a/services/data-computer/v1/logger.cfg b/services/data-computer/v1/logger.cfg
@@ -8,8 +8,9 @@ bucket = logger
 # On ne garde que la première erreur déclénchée
 [shift]
 
-[debug]
-text = Error trapped
+[combine]
+primer = Error trapped
+file = timestamp.cfg
 
 [assign]
 path = body.identifier
@@ -51,5 +52,6 @@ identifier = env('identifier')
 jsonl = true
 compress = false
 
-[debug]
-text = Error was saved
+[combine]
+primer = Error was saved
+file = timestamp.cfg
diff --git a/services/data-computer/v1/recorder.cfg b/services/data-computer/v1/recorder.cfg
@@ -2,9 +2,9 @@
 plugin = basics
 plugin = analytics
 
-[singleton]
-[singleton/debug]
-text = fix('One first result received by', env('generator'), 'for', env('identifier')).join(' ')
+[combine]
+primer = One first result received
+file = timestamp.cfg
 
 [metrics]
 bucket = recorder
@@ -34,8 +34,9 @@ value = env('headers.x-webhook-success')
 path = body
 value = self().pick(['size', 'atime', 'mtime', 'ctime']).set('identifier', env('identifier')).set('generator', env('generator')).set('state', 'ready')
 
-[swing/debug]
-text = fix('Result generated by', env('generator'), 'for', env('identifier')).join(' ')
+[swing/combine]
+primer = Result generated
+file = timestamp.cfg
 
 # Step 2.4.2 (générique): Envoyer la requète HTTP
 [swing/URLFetch]
@@ -46,11 +47,13 @@ retries = 5
 timeout = 30000
 
 # Step 2.4.3 (faculatif) : Ajouter une trace dans log
-[swing/debug]
-text = fix('WebHook triggered by', env('generator'), 'for', env('identifier')).join(' ')
+[swing/combine]
+primer = WebHook triggered
+file = timestamp.cfg
 
 # Step 2.5 (faculatif) : Ajouter une trace dans log
-[debug]
-text = fix('Process completed by', env('generator'), 'for', env('identifier')).join(' ')
+[swing/combine]
+primer = Process completed
+file = timestamp.cfg
 
 
diff --git a/services/data-computer/v1/retrieve-status.ini b/services/data-computer/v1/retrieve-status.ini
@@ -0,0 +1,27 @@
+# Entrypoint output format
+mimeType = application/jsonl
+extension = jsonl
+
+# OpenAPI Documentation - JSON format (dot notation)
+post.operationId = post-v1-retrieve
+post.summary = Récupération du status d'un traitement
+post.description = Les traitements étant asynchrones le status du traitement peut-être récupér par cette route
+post.tags.0 = data-computer
+post.responses.default.description = Fichier corpus au format jsonl
+post.responses.default.content.application/jsonl.schema.type = string
+post.requestBody.content.application/json.example.0.value = xMkWJX7GU
+post.requestBody.content.application/json.schema.$ref = #/components/schemas/JSONStream
+post.requestBody.required = true
+
+[use]
+plugin = basics
+
+[JSONParse]
+separator = *
+
+[exchange]
+value = get('value')
+
+[FILELoad]
+location = /tmp/logs
+
diff --git a/services/data-computer/v1/statistics.ini b/services/data-computer/v1/statistics.ini
@@ -57,8 +57,6 @@ file = buffer.cfg
 path = value
 
 # Step 2.1.2 (spécifique): Créer des tableaux de paires des segments (ou Bigramme)
-[fork/delegate/debug]
-
 [fork/delegate/assign]
 path = _statistics.value.input
 value = get('value')

diff --git a/services/data-computer/v1/timestamp.cfg b/services/data-computer/v1/timestamp.cfg
@@ -0,0 +1,23 @@
+[debug]
+text = timestamp
+
+[fork]
+standalone = true
+logger = logger.cfg
+
+[fork/replace]
+path = timestamp
+value = fix(Date.now())
+path = generator
+value = env('generator')
+path = identifier
+value = env('identifier')
+path = message
+value = self()
+
+[fork/FILESave]
+location = /tmp/logs
+identifier = env('identifier')
+jsonl = true
+compress = false
+append = true