[ADD] attachment_indexation_mupdf

attachment_indexation_mupdf/README.rst

@@ -0,0 +1,5 @@
+=====================================================
+Attachments List and Document Indexation with PyMuPDF
+=====================================================
+
+Module to index pdf document using state-of-the-art library.

attachment_indexation_mupdf/__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_mupdf/__manifest__.py

@@ -0,0 +1,17 @@
+# Copyright 2023 len-foss/Financial Way
+# License AGPL-3.0 or later (http://www.gnu.org/licenses/agpl.html).
+{
+    "name": "Attachments List and Document Indexation with PyMuPDF",
+    "category": "Hidden/Tools",
+    "version": "16.0.0.0.0",
+    "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": ["PyMuPDF"]},
+}

attachment_indexation_mupdf/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_mupdf/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 io
+import logging
+
+from odoo import models
+
+_logger = logging.getLogger(__name__)
+
+try:
+    import fitz
+except ImportError:
+    fitz = None
+    _logger.warning(
+        "Attachment indexation of PDF documents is unavailable"
+        "because PyMuPDF cannot be loaded."
+    )
+
+
+class IrAttachment(models.Model):
+    _inherit = "ir.attachment"
+
+    def _index_pdf(self, bin_data):
+        """Index PDF documents with MuPDF if available"""
+        if fitz is None:
+            return super()._index_pdf(bin_data)
+        buf = ""
+        try:
+            f = io.BytesIO(bin_data)
+            doc = fitz.open(stream=f, filetype="pdf")
+            for page in doc:
+                buf += page.get_text()
+        except Exception:  # pylint: disable=except-pass
+            pass
+        return buf

attachment_indexation_mupdf/tests/__init__.py

@@ -0,0 +1,2 @@
+# License AGPL-3.0 or later (http://www.gnu.org/licenses/agpl.html).
+from . import test_indexation

attachment_indexation_mupdf/tests/test_indexation.py

@@ -0,0 +1,27 @@
+# Copyright 2023 len-foss/Financial Way
+# License AGPL-3.0 or later (http://www.gnu.org/licenses/agpl.html).
+
+import os
+from unittest import skipIf
+
+from odoo.tests.common import TransactionCase, tagged
+
+directory = os.path.dirname(__file__)
+
+try:
+    import fitz
+except ImportError:
+    fitz = None
+
+
+@tagged("post_install", "-at_install")
+class TestCaseIndexation(TransactionCase):
+    @skipIf(fitz is None, "PyMyPDF is not installed")
+    def test_attachment_pdf_indexation(self):
+        with open(os.path.join(directory, "files", "test_content.pdf"), "rb") as file:
+            pdf = file.read()
+            text = self.env["ir.attachment"]._index(pdf, "application/pdf")
+            # note that the whitespace character is not the same as with pdfminer
+            self.assertEqual(
+                text, "TestContent!!\n", "the index content should be correct"
+            )

setup/attachment_indexation_mupdf/odoo/addons/attachment_indexation_mupdf

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

setup/attachment_indexation_mupdf/setup.py

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