[ADD] attachment_indexation_textract

attachment_indexation_textract/README.rst

@@ -0,0 +1,5 @@
+======================================================
+Attachments List and Document Indexation with Textract
+======================================================
+
+Module to index documents with Textract; install optional dependencies to support more file formats. Supported formats include .xls, .doc with antiword, etc.

attachment_indexation_textract/__init__.py

@@ -0,0 +1,4 @@
+# Copyright 2023 len-foss/Financial Way
+# License AGPL-3.0 or later (http://www.gnu.org/licenses/agpl.html).
+
+from . import models

attachment_indexation_textract/__manifest__.py

@@ -0,0 +1,20 @@
+# Copyright 2023 len-foss/Financial Way
+# License AGPL-3.0 or later (http://www.gnu.org/licenses/agpl.html).
+{
+    "name": "Attachments Indexation with Textract",
+    "category": "Hidden/Tools",
+    "version": "16.0.1.0.0",
+    "summary": "Attachments List and Document Indexation with PyMuPDF",
+    "author": "len-foss/FinancialWay,Odoo Community Association (OCA)",
+    "website": "https://github.com/OCA/knowledge",
+    "license": "AGPL-3",
+    "depends": ["attachment_indexation"],
+    "auto_install": True,
+    "installable": True,
+    "data": [],
+    "assets": {},
+    "external_dependencies": {
+        "python": ["textract"],
+        "bin": ["antiword"],
+    },
+}

attachment_indexation_textract/models/__init__.py

@@ -0,0 +1,4 @@
+# Copyright 2023 len-foss/Financial Way
+# License AGPL-3.0 or later (http://www.gnu.org/licenses/agpl.html).
+
+from . import ir_attachment

attachment_indexation_textract/models/ir_attachment.py

@@ -0,0 +1,36 @@
+# Copyright 2023 len-foss/Financial Way
+# License AGPL-3.0 or later (http://www.gnu.org/licenses/agpl.html).
+
+import logging
+import mimetypes
+import tempfile
+
+import textract
+
+from odoo import models
+
+_logger = logging.getLogger(__name__)
+
+
+class IrAttachment(models.Model):
+    _inherit = "ir.attachment"
+
+    def _index(self, bin_data, mimetype, checksum=None):
+        """Index documents with textract if available"""
+        if mimetype != "application/pdf":  # mupdf is better
+            buf = self.with_context(mimetype=mimetype)._index_textract(bin_data)
+        return buf or super()._index(bin_data, mimetype, checksum=checksum)
+
+    def _index_textract(self, bin_data):
+        """Index documents with textract if available"""
+        buf = ""
+        try:
+            mimetype = self.env.context.get("mimetype")
+            extension = mimetypes.guess_extension(mimetype)
+            with tempfile.NamedTemporaryFile(suffix=extension or "") as tmp_file:
+                tmp_file.write(bin_data)
+                file_path = tmp_file.name
+                buf = textract.process(file_path)
+        except Exception:
+            _logger.info(Exception, exc_info=True)
+        return buf

setup/attachment_indexation_textract/odoo/addons/attachment_indexation_textract

@@ -0,0 +1 @@
+../../../../attachment_indexation_textract

setup/attachment_indexation_textract/setup.py

@@ -0,0 +1,6 @@
+import setuptools
+
+setuptools.setup(
+    setup_requires=['setuptools-odoo'],
+    odoo_addon=True,
+)