Add DOCX export worker and feature

arkindex/process/models.py

@@ -70,6 +70,7 @@ class ProcessMode(Enum):
 class ExportFormat(Enum):
     PDF = "pdf"
     PageXML = "page_xml"
+    DOCX = "docx"
 
 
 class Process(IndexableModel):
@@ -703,6 +704,7 @@ class ArkindexFeature(Enum):
     # FEATURE_FORMAT_MAP dictionary which maps export formats to arkindex features
     ExportPDF = "pdf_export"
     ExportPageXML = "pagexml_export"
+    ExportDOCX = "docx_export"
 
 
 class WorkerVersion(models.Model):
@@ -1222,5 +1224,6 @@ class CorpusWorkerVersion(models.Model):
 # A mapping of export formats with their corresponding arkindex feature
 FEATURE_FORMAT_MAP = {
     ExportFormat.PDF: ArkindexFeature.ExportPDF,
-    ExportFormat.PageXML: ArkindexFeature.ExportPageXML
+    ExportFormat.PageXML: ArkindexFeature.ExportPageXML,
+    ExportFormat.DOCX: ArkindexFeature.ExportDOCX
 }

arkindex/system_workers.yml

@@ -17,3 +17,6 @@ features:
     pagexml_export:
         image: registry.gitlab.teklia.com/arkindex/workers/export:0.2.0
         command: worker-export-pagexml
+    docx_export:
+        image: registry.gitlab.teklia.com/arkindex/workers/export:0.2.0
+        command: worker-export-docx