added html striped content for better search

raven/api/search.py

@@ -20,7 +20,7 @@ def get_search_result(filter_type, doctype, search_text=None, from_user=None, in
     }
 
     query = frappe.qb.from_(doctype).select(
-        doctype.name, doctype.file, doctype.owner, doctype.creation, doctype.message_type, doctype.channel_id, doctype.text).join(channel, JoinType.left).on(doctype.channel_id == channel.name).join(channel_member, JoinType.left).on(
+        doctype.name, doctype.file, doctype.owner, doctype.creation, doctype.message_type, doctype.channel_id, doctype.text, doctype.content).join(channel, JoinType.left).on(doctype.channel_id == channel.name).join(channel_member, JoinType.left).on(
             channel_member.channel_id == doctype.channel_id).where((channel.type != 'Private') | (channel_member.user_id == frappe.session.user))
 
     if filter_type == 'File':
@@ -40,7 +40,7 @@ def get_search_result(filter_type, doctype, search_text=None, from_user=None, in
             query = query.where(doctype.file.like(
                 "/private/files/%" + search_text + "%"))
         elif filter_type == 'Message':
-            query = query.where(doctype.text.like("%" + search_text + "%"))
+            query = query.where(doctype.content.like("%" + search_text + "%"))
         elif filter_type == 'Channel':
             query = query.where(
                 doctype.channel_name.like("%" + search_text + "%"))

raven/patches.txt

@@ -3,3 +3,4 @@
 [post_model_sync]
 raven.patches.v1_2.create_raven_users
 raven.patches.v1_3.create_raven_message_indexes #23
+raven.patches.v1_3.update_all_messages_to_include_message_content #1

raven/patches/v1_3/update_all_messages_to_include_message_content.py

@@ -0,0 +1,26 @@
+import frappe
+from frappe.utils import strip_html_tags
+
+
+def execute():
+    update_old_messages_to_include_message_content()
+
+
+def update_old_messages_to_include_message_content():
+    '''
+    Update all old messages to include message content
+    Message content is required for search
+    It is basically the message's text content but without any html tags
+    (this is done to improve search results)
+    This is a one-time operation, not required for new messages
+    '''
+    messages = frappe.db.get_all('Raven Message', fields=[
+                                 'name', 'text', 'message_type'])
+    for message in messages:
+        if message.text:
+            cleaned_text = strip_html_tags(message.text).replace(
+                '\ufeff', '').replace(' ', ' ')
+            content = cleaned_text
+            frappe.db.set_value(
+                'Raven Message', message.name, 'content', content)
+    frappe.db.commit()

raven/raven_messaging/doctype/raven_message/raven_message.json

@@ -22,7 +22,8 @@
   "is_reply",
   "linked_message",
   "link_doctype",
-  "link_document"
+  "link_document",
+  "content"
  ],
  "fields": [
   {
@@ -109,11 +110,17 @@
    "fieldtype": "Dynamic Link",
    "label": "Link Document",
    "options": "link_doctype"
+  },
+  {
+   "fieldname": "content",
+   "fieldtype": "Long Text",
+   "label": "Content",
+   "read_only": 1
   }
  ],
  "index_web_pages_for_search": 1,
  "links": [],
- "modified": "2023-12-22 18:25:02.237368",
+ "modified": "2024-01-19 14:56:44.534433",
  "modified_by": "Administrator",
  "module": "Raven Messaging",
  "name": "Raven Message",

raven/raven_messaging/doctype/raven_message/raven_message.py

@@ -3,9 +3,11 @@
 import frappe
 from frappe import _
 from frappe.model.document import Document
+from frappe.utils import strip_html_tags
 import json
 from raven.api.raven_message import track_visit
 
+
 class RavenMessage(Document):
     # begin: auto-generated types
     # This code is auto-generated. Do not modify anything in this block.
@@ -16,6 +18,7 @@ class RavenMessage(Document):
         from frappe.types import DF
 
         channel_id: DF.Link
+        content: DF.LongText | None
         file: DF.Attach | None
         file_thumbnail: DF.Attach | None
         image_height: DF.Data | None
@@ -91,3 +94,12 @@ def on_trash(self):
     def before_save(self):
         if frappe.db.get_value('Raven Channel', self.channel_id, 'type') != 'Private' or frappe.db.exists("Raven Channel Member", {"channel_id": self.channel_id, "user_id": frappe.session.user}):
             track_visit(self.channel_id)
+
+
+def on_doctype_update():
+    '''
+    Add indexes to Raven Message table
+    '''
+    # Index the selector (channel or message type) first for faster queries (less rows to sort in the next step)
+    frappe.db.add_index("Raven Message", ["channel_id", "creation"])
+    frappe.db.add_index("Raven Message", ["message_type", "creation"])