Previously, the confirmation dialog shown when leaving a livechat session was too generic and did not provide enough context to help agents ensure they were closing the correct conversation. As a result, agents could accidentally leave the wrong chat, especially when the livechat list is reordered due to new incoming messages.

This PR improves the dialog to make the action clearer and safer:
- Displays the visitor’s name in the confirmation title.
- Shows the most recent message from the conversation in the dialog body.
- Helps agents verify they are leaving the intended conversation.

task-5355106

Before this PR: 
<img width="593" height="138" alt="image" src="https://github.com/user-attachments/assets/363eeab9-0306-47cc-8f4f-4035862f7e29" />

After this PR:
<img width="1135" height="277" alt="image" src="https://github.com/user-attachments/assets/6a7b7aea-d3bf-4b52-be39-001448ba7afc" />

Before vs After (Chat Window)
<table> <tr> <td> <img width="256" height="422" src="https://github.com/user-attachments/assets/b895b831-e11d-4aee-8f35-cd79f5c77aaf" /> </td> <td> <img width="256" height="422" src="https://github.com/user-attachments/assets/d31c061d-6a6e-42cb-ae60-12a5329cd4d4" /> </td> </tr> </table>


---
I confirm I have signed the CLA and read the PR guidelines at www.odoo.com/submit-pr

**Issue:**
When invoices in a journal enabled to "Use Documents" for LATAM legal invoicing have a sequence prefix with numbers, the next invoice will restart the sequence and try to use already-used sequences.

**Steps to Reproduce:**
- Install l10n_pe_edi
- Change to PE Company
- Duplicate the Customer Invoice journal, give it shortcode "01A"
- Create an invoice in dupe journal, set customer to "Comercial Constructora los Patitos S.A.", set product and tax, Document Type = (01) Factura
- Confirm the invoice, it will have name "F F01-00000001"
- Duplicate the invoice

-> The new invoice has the same sequence number, and the document number is visible

Expected: Invoice is called "Draft" (/), document number is not visible, and upon confirming the invoice, it will have the name "F F01-00000002"

**Cause:**
- LATAM invoices do not follow the standard formats in sequence_mixin.py, so _deduce_sequence_number_reset will always return "never"
- However, the result of this method is used in _get_last_sequence_domain, and assumes that the sequence follows the corresponding format to "never" (_sequence_fixed_regex)
- This isn't the case if the prefix has numbers, it will be captured in the _sequence_yearly_regex
- anti-regex is used to filter sequences that aren't _sequence_fixed_regex, but this causes prefixes with numbers to never be found, and always restart the sequence

**Solution:**
- Add context value "no_anti_regex" to skip sequence exclusion
- If the invoice is LATAM, call _get_last_sequence_domain with context "no_anti_regex" = True so no sequences are excluded.
- Because LATAM invoices are always in a fixed format, we don't need to filter out sequences in other formats

opw-5111844

Forward-Port-Of: odoo/odoo#235000

**Issue:**
When invoices in a journal enabled to "Use Documents" for LATAM legal invoicing have a sequence prefix with numbers, the next invoice will restart the sequence and try to use already-used sequences.

**Steps to Reproduce:**
- Install l10n_pe_edi
- Change to PE Company
- Duplicate the Customer Invoice journal, give it shortcode "01A"
- Create an invoice in dupe journal, set customer to "Comercial Constructora los Patitos S.A.", set product and tax, Document Type = (01) Factura
- Confirm the invoice, it will have name "F F01-00000001"
- Duplicate the invoice

-> The new invoice has the same sequence number, and the document number is visible

Expected: Invoice is called "Draft" (/), document number is not visible, and upon confirming the invoice, it will have the name "F F01-00000002"

**Cause:**
- LATAM invoices do not follow the standard formats in sequence_mixin.py, so _deduce_sequence_number_reset will always return "never"
- However, the result of this method is used in _get_last_sequence_domain, and assumes that the sequence follows the corresponding format to "never" (_sequence_fixed_regex)
- This isn't the case if the prefix has numbers, it will be captured in the _sequence_yearly_regex
- anti-regex is used to filter sequences that aren't _sequence_fixed_regex, but this causes prefixes with numbers to never be found, and always restart the sequence

**Solution:**
- Add context value "no_anti_regex" to skip sequence exclusion
- If the invoice is LATAM, call _get_last_sequence_domain with context "no_anti_regex" = True so no sequences are excluded.
- Because LATAM invoices are always in a fixed format, we don't need to filter out sequences in other formats

opw-5111844

Forward-Port-Of: odoo/enterprise#101620

Issue before this commit:
=========================

The high-priority maintenance request count (todo_request_count_high_priority) 
was not calculated correctly.

Steps to Reproduce:
=========================
- Install the maintenance module.
- Create a maintenance request and set the Priority to High (3-starred) in the form view.
- Open the dashboard.
- Observe that the high-priority request count is not displayed.
- The count always remains 0, even when high-priority requests exist.

Cause of the issue:
=========================
In this [PR](https://github.com/odoo/odoo/pull/94866), the logic was mistakenly changed. 
The priority field is defined as a Selection field, but while computing the count, the comparison 
was done against an integer(3, not '3') instead of the actual string value. Since the stored 
value is '3' (string), the condition is always evaluated to False, resulting in a count of 0.

With This Commit:
=========================
Ensure that high-priority maintenance requests are correctly counted and displayed on 
the dashboard when they exist. This provides better visibility of critical requests and
helps users prioritise maintenance work effectively.

Forward-Port-Of: odoo/odoo#244987

**Steps to reproduce:**
- Install Accounting
- Go to Accounting dashboard

1) - On "Bank" journal card, select "Import File" option in dropdown menu
    - Upload a file containing a lot of statements (e.g. more than 170)
    - Test the import with "Test" button
2) - Open "Bank" journal
    - Select "Import records" in the cog menu
    - Upload a file containing a lot of statements (e.g. more than 170)
    - Test the import with "Test" button
3) - Open "Bank" journal
    - Click on "Upload" button
    - Upload a file containing a lot of statements (e.g. more than 170)
    - Test the import with "Test" button

**Issue:**
The issue happens when the test import fails.
Even if it was a test, all the records are created in the database.

**Cause:**
During a test import the records should not be reconciled after creation.
However, by default, "auto_statement_processing" property is True and trigger the reconciliation.
When an error is raised during the test import, a rollback is made on a previous savepoint.
But a commit can be done in "_cron_try_auto_reconcile_statement_lines" method, which results in the records being created for real.

**Solution:**
Set "auto_statement_processing" property to False for Test import.
For use case 1) and 2), we make sure that we follow the same flow than use case 3).

opw-5436937

Forward-Port-Of: odoo/enterprise#104954

Issue before this commit:
=========================

The high-priority maintenance request count (todo_request_count_high_priority) 
was not calculated correctly.

Steps to Reproduce:
=========================
- Install the maintenance module.
- Create a maintenance request and set the Priority to High (3-starred) in the form view.
- Open the dashboard.
- Observe that the high-priority request count is not displayed.
- The count always remains 0, even when high-priority requests exist.

Cause of the issue:
=========================
In this [PR](https://github.com/odoo/odoo/pull/94866), the logic was mistakenly changed. 
The priority field is defined as a Selection field, but while computing the count, the comparison 
was done against an integer(3, not '3') instead of the actual string value. Since the stored 
value is '3' (string), the condition is always evaluated to False, resulting in a count of 0.

With This Commit:
=========================
Ensure that high-priority maintenance requests are correctly counted and displayed on 
the dashboard when they exist. This provides better visibility of critical requests and
helps users prioritise maintenance work effectively.

Forward-Port-Of: odoo/odoo#244987

**Description of the issue/feature this PR addresses:**
In the `triggering_answer_ids` searchable dropdown, when no value is entered, the `_search_display_name` method of `chatbot_script_answer` is not called. Instead, the ORM falls back to the field’s default domain and returns all `chatbot.script.answer` records, including those from other scripts. When a value is entered, `_search_display_name` is triggered and the results are filtered correctly.

This behavior changed after PR #201587, where the `operator_optimization` step started executing before `determine_domain`. Since `determine_domain` is the step that triggers `_search_display_name`, it no longer gets called when the domain `('name', 'ilike', '')` is stripped by `operator_optimization`. Therefore, filtering only works when a non-empty filter value is provided.

**Current behavior before PR:**
All `chatbot.script.answer` records are shown in the `triggering_answer_ids` dropdown when no search value is entered, even if they don’t belong to the current chatbot script.

**Desired behavior after PR is merged:**
The `triggering_answer_ids` dropdown only shows answers belonging to the current chatbot script, regardless of whether a search value is entered.

task-[4968490](https://www.odoo.com/odoo/project/1519/tasks/4968490)

Forward-Port-Of: odoo/odoo#228192

Issue before this commit:
=========================

The high-priority maintenance request count (todo_request_count_high_priority) 
was not calculated correctly.

Steps to Reproduce:
=========================
- Install the maintenance module.
- Create a maintenance request and set the Priority to High (3-starred) in the form view.
- Open the dashboard.
- Observe that the high-priority request count is not displayed.
- The count always remains 0, even when high-priority requests exist.

Cause of the issue:
=========================
In this [PR](https://github.com/odoo/odoo/pull/94866), the logic was mistakenly changed. 
The priority field is defined as a Selection field, but while computing the count, the comparison 
was done against an integer(3, not '3') instead of the actual string value. Since the stored 
value is '3' (string), the condition is always evaluated to False, resulting in a count of 0.

With This Commit:
=========================
Ensure that high-priority maintenance requests are correctly counted and displayed on 
the dashboard when they exist. This provides better visibility of critical requests and
helps users prioritise maintenance work effectively.

Forward-Port-Of: odoo/odoo#244987

Issue before this commit:
=========================

The high-priority maintenance request count (todo_request_count_high_priority) 
was not calculated correctly.

Steps to Reproduce:
=========================
- Install the maintenance module.
- Create a maintenance request and set the Priority to High (3-starred) in the form view.
- Open the dashboard.
- Observe that the high-priority request count is not displayed.
- The count always remains 0, even when high-priority requests exist.

Cause of the issue:
=========================
In this [PR](https://github.com/odoo/odoo/pull/94866), the logic was mistakenly changed. 
The priority field is defined as a Selection field, but while computing the count, the comparison 
was done against an integer(3, not '3') instead of the actual string value. Since the stored 
value is '3' (string), the condition is always evaluated to False, resulting in a count of 0.

With This Commit:
=========================
Ensure that high-priority maintenance requests are correctly counted and displayed on 
the dashboard when they exist. This provides better visibility of critical requests and
helps users prioritise maintenance work effectively.

Forward-Port-Of: odoo/odoo#244987

### Issue:
The payment method `99 – Por Definir` was used as the default value for invoices, sale orders, and POS orders
This leads to fiscal inconsistencies, especially when invoices use the `PUE` payment policy, where this payment method is invalid

### Cause:
In the `_compute_l10n_mx_edi_payment_method_id` methods, the default value was always set to `Por Definir`

### Fix:
After discussion with the PO (MIAL), the chosen solution is to archive the payment method `99 – Por Definir`and remove it as a default value
All valid cases should already be handled explicitly, making it clear to the user that something is missing when the data is blank

### Steps to reproduce:
- Install `l10n_mx_edi` and switch to the MX company
- Create an invoice with today’s invoice date
- The payment policy is set to PUE
- Before the fix, the payment method is set to `Por Definir`

For Sale Order and POS Order tests, it's the default value as soon as you create an order

opw-5406038

Forward-Port-Of: odoo/enterprise#105058
Forward-Port-Of: odoo/enterprise#104164

In the appointment calendar, the "Month Year" (e.g. January 2026) is displayed at the top of the calendar selector as is standard for calendars. Unfortunately in some languages, the word month word used depends on grammatical situation.

Steps to reproduce:
- activate Ukrainian (or Polish) language
- Preview (i.e. the website view) of any appointment
- Change the preview into Ukrainian (or Polish)

Expected result:
Calendar month at top of calendar is shown in the nominative case:
e.g. January 2026 = січень 2026 (in Ukrainian)

Actual result:
Calendar month is shown in the genitive case (e.g. "of January", as in "the 12th of January):
e.g. January 2026 = січня 2026 (in Ukrainian)

Fix is to switch from the "MMMM Y" format (i.e. month based on grammar context) to "LLLL Y" (i.e. stand alone month) which will use the correct month case. Ref: https://www.unicode.org/reports/tr35/tr35-dates.html#dfst-month

Note that for most languages this won't make a difference since there is usually only 1 way of writing a month.

opw-5474705

picture diff (for January 2026):

before fix:
<img width="1168" height="468" alt="image" src="https://github.com/user-attachments/assets/36754439-e664-40f3-9c5b-7c9c76cd7cca" />


after fix:
<img width="1215" height="492" alt="image" src="https://github.com/user-attachments/assets/82f0665e-fe70-4b9f-b1d4-f420b52db38f" />

Forward-Port-Of: odoo/enterprise#104942

Steps to Reproduce:
Create an accrual plan that accrues 2 days monthly, credited at the start of the month. Create a Time Off Type with the Unit of Measure set to Hours. Create an allocation request using the above Time Off Type and Accrual Plan. Open the Time Off dashboard:
a. The current available balance is displayed in hours and is correctly computed based on the employee’s working schedule. b. However, when checking the balance for a future date, the system incorrectly adds 2 hours instead of 16 hours (i.e., 2 days converted to hours based on the working schedule).

Bug cause:
The system incorrectly identified the unit of measure because it was still referencing the request_unit field, which defaults to 'day'. Following a recent refactor where request_unit was replaced by unit_of_measure to define Time Off Type units, this specific logic flow was left unupdated, causing it to be calculated in days instead of hours.

Solution:
Updated the future leaves calculation logic to use unit_of_measure instead of request_unit. This ensures that when a Time Off Type is configured in hours, accruals defined in days are correctly converted based on the employee's working schedule.

Task: 5498292

---
I confirm I have signed the CLA and read the PR guidelines at www.odoo.com/submit-pr

enterprise: https://github.com/odoo/enterprise/pull/104332
upgrade: https://github.com/odoo/upgrade/pull/9269

* = im_livechat, test_discuss_full

This commit removes the fetched feature of Discuss messages, for the following reasons:
1. Redundancy: The original purpose of confirming server receipt is now better handled by the "pending" message state (spinner/opacity). Once a message is no longer pending, it is effectively "sent".
2. Reliability & Privacy: The previous implementation relied on a single bus notification, which when lost would lead to an inconsistent state. Fixing this reliability (by broadcasting status on connection) would negatively impact performance and, more critically, leak the presence of users who wish to appear "Offline".
3. Data Modeling: The model only tracks the last fetched ID, which is an inaccurate representation of message history if gaps exist.

This commit also simplifies the "Seen" indicator to reflect this change:
- single gray check: Some people have seen.
- double purple checks: Seen by all channel members.

task-5177299

community: https://github.com/odoo/odoo/pull/243853
upgrade: https://github.com/odoo/upgrade/pull/9269

This commit removes the fetched feature of Discuss messages, for the following reasons:
1. Redundancy: The original purpose of confirming server receipt is now better handled by the "pending" message state (spinner/opacity). Once a message is no longer pending, it is effectively "sent".
2. Reliability & Privacy: The previous implementation relied on a single bus notification, which when lost would lead to an inconsistent state. Fixing this reliability (by broadcasting status on connection) would negatively impact performance and, more critically, leak the presence of users who wish to appear "Offline".
3. Data Modeling: The model only tracks the last fetched ID, which is an inaccurate representation of message history if gaps exist.

This commit also simplifies the "Seen" indicator to reflect this change:
- single gray check: Some people have seen.
- double purple checks: Seen by all channel members.

task-5177299

**Issue**:
The displayed available quantity on the ecommerce product page is incorrect for rental products when "continue selling" is enabled.

**Steps to reproduce**:
- Create a rental product tracked in stock with a quantity of 5
- Enable "continue selling" and "show available quantity below 10"
- Go to the ecommerce page of this product
- Rent 3 units for a given period, confirm and pay
- Return to the ecommerce product page
-> Whatever the selected renting period, the displayed quantity is always 2

**Cause**:
The website displays `free_qty`:
https://github.com/odoo/enterprise/blob/41c729e22c5fd1abb690f8335e933f793be0b319/website_sale_stock_renting/static/src/xml/website_sale_stock_renting_product_availability.xml#L15

`free_qty` is computed in:
https://github.com/odoo-dev/odoo/blob/0935829ddaecd7b2b6eec9157f8f790b546d06ff/addons/website_sale_stock/models/product_template.py#L36

which leads to:
https://github.com/odoo/enterprise/blob/41c729e22c5fd1abb690f8335e933f793be0b319/website_sale_stock_renting/models/website.py#L10

and ultimately relies on:
https://github.com/odoo/odoo/blob/37bf1703c7478a3010b71cd60bbb43b3295a605b/addons/stock/models/product.py#L213

This computation does not take the selected renting period into account.

There is a period-aware computation here:
https://github.com/odoo/enterprise/blob/41c729e22c5fd1abb690f8335e933f793be0b319/website_sale_stock_renting/models/website.py#L15C17-L21C1

but it is only triggered when `product.allow_out_of_stock_order` is False (i.e. when "continue selling" is disabled).

opw-[5354163](https://www.odoo.com/web#id=5354163&view_type=form&model=project.task)

Forward-Port-Of: odoo/enterprise#104686
Forward-Port-Of: odoo/enterprise#103333

Modules containing valid Markdown in their description or README.md could cause Odoo to crash during startup or module updates if the content confused the reStructuredText (RST) parser.

### Steps to reproduce

1. Create a module with a manifest like this:

    ```py
    {
        'name': 'base',
        'description': """
        ....
    """,
    }
    ```

2. Start the Odoo server or attempt to install or update update the module.
3. The server crashes:

    ```
    docutils.utils.SystemMessage: (SEVERE/4) Unexpected section title or transition.
    ```

### Cause

The crash occurs during the execution of the `_get_desc` method in the `ir.module.module` model, which computes the `description_html` field.

When Odoo processes a module, it checks for a pre-rendered HTML description at `static/description/index.html`. If this file is missing, the `_get_desc` method attempts to generate HTML from the module's `description` field (often populated from the `README.md` file) by calling the `docutils.core.publish_string` function.

The **docutils** library is designed specifically for **reStructuredText (RST)**. If the input text contains structural patterns that violate RST rules—such as inconsistent header levels or "transitions" in invalid contexts—docutils flags a severe error and raises a `docutils.utils.SystemMessage` exception.

### Fix

This commit handles these exceptions and falls back to a raw text rendering. It also improves the logic by removing the restriction that prevented non-application modules from rendering their description via RST.

opw-5424131

Forward-Port-Of: odoo/odoo#243517

_
## Short functional explanation of the error
When an employee enters multiple attendances for a single day, the company tolerance time isn't computed correctly.

## Reproduction Steps
1. Go to attendances.
2. Click on configuration and scroll down to the Extra Hours section. Set a Tolerance Time in Favor of Company of 15 minutes.
3. Create 2 attendances for the same employee: one attendance from 8 to 15 for example, and a second one from 16 to 18:12.

### Expected behavior
As the overtime entered is 12 minutes, which is inferior to the company tolerance time of 15 minutes, no extra time should be computed.

### Unexpected behavior
12 minutes of overtime are computed.

## Origin of the issue
Let's say we enter 2 different shifts for the same day. Our work day should be 8 hours, and the sum of both shifts reaches 8 hours or more. We shouldn't have any overtime. However, in the code, the overtime is negative. This is compensated by, in our case, the post-work time: in our case, our overtime duration will be equal to -1, but our post-work time will be equal to 1.2. Both cancel each other, and in the end we obtain 0.2 of overtime, which corresponds to our 10 minutes overtime. However, in this code: https://github.com/odoo/odoo/blob/afcbd98594c9f7007f03a343ea40ea122b955459/addons/hr_attendance/models/hr_attendance.py#L374-L380 it isn't computed that way: because post-work time is 1.2, which is above our company tolerance time of 15 minutes (0.25 in the code), we will always be in the case where we exceed the tolerance time.

Hence, we have to "flatten" the overtime duration and the post-work time before reaching that piece of code.

note: the same bug exists for the employee tolerance time, which
is corrected in this commit.
note: the issue doesn't persist in 19.0, forwarding the tests.
__
opw-5136861

---

Forward-Port-Of: odoo/odoo#244186
Forward-Port-Of: odoo/odoo#242517

Currently, an error occurs when a user tries to post an invoice using a payment method that shares the same code as another payment method.

Steps to replicate:
- Install `l10n_mx_edi` and `accountant` with demo and switch to `ZAPATERIA URTADO ÑERI` (Mexican company).
- Go to `Accounting > Configuration > Payment Way Codes (MX)`.
- Open `Efectivo` and change its code to `02`.
- Create a new Invoice, select `Efectivo` in the Payment Way.
- Add a customer and a move line, then confirm the invoice and send it (make sure CFDI is checked).

Error:
```
  File '/home/odoo/src/enterprise/19.0/l10n_mx_edi/models/account_move.py', line 424, in _l10n_mx_edi_get_extra_invoice_report_values
    cfdi_infos['payment_way'] = f'{payment_way} - {payment_method.name}'
  File '/home/odoo/src/odoo/19.0/odoo/orm/fields.py', line 1659, in __get__
    record.ensure_one()
  File '/home/odoo/src/odoo/19.0/odoo/orm/models.py', line 5934, in ensure_one
    raise ValueError('Expected singleton: %s' % self)
ValueError: Expected singleton: l10n_mx_edi.payment.method(1, 22)
```

Cause:
- Issue originated through this [PR] that gave access to write on the model.
- As the user made the codes of two payment methods same, the [search] returned two records and while accessing `payment_method.name` on two records it results into this error.

Solution:
- Made the fields read-only via XML to prevent users from changing the payment method codes established by the Mexican government.
- Added limit to the search query to prevent multiple records. (for existing DBs that might have changed payment method codes).
- Removed unlink rights on the `l10n_mx_edi.payment.method` model.
- Added a SQL constraint to allow only unique values for the code.

[PR]: https://github.com/odoo/enterprise/pull/38046

[search]: https://github.com/odoo/enterprise/blob/18117c6a9fbf270ace1c551616828a85713d5225/l10n_mx_edi/models/account_move.py#L423

sentry-7171030995

Forward-Port-Of: odoo/enterprise#104914
Forward-Port-Of: odoo/enterprise#103944

**Steps to reproduce:**
- Create a portal user
- Go to the documents app
- Click on the marketing folder
- Share the marketing folder through a link (Anyone with a link = viewer + discoverable)
- Copy the share link
- Login with the portal user in an incognito window
- Paste the share link in an incognito browser
- Click on "brand 1" folder, result nothing is showing while there should be a folder and a picture
- Click on "brand 2"
- Click back on "brand 1" and now the folder and picture are visible
- If you click on a subfolder of "brand 1" you also get an error

**Issue:**
Discoverable subfolders accessed using `accessToken` are not available on
the first read of a user and this happens for each level of the
hierarchy (refresh is needed each time).

When using sharing link to display folders with a user,
the subfolder document access is created on `/documents/touch/` using
`_from_access_token`.

But on the js side the call is delayed (with debounce) and occurs after
the `web_search_read`. This means that subfolders are only accessible
after a refresh or by switching back and forth between folders.
Also, even after the folder is displayed, if there are other subfolders
in it, going deeper in the hierarchy won't work as well without a refresh
due to the `search_panel_select_range` missing the new folder.

**Fix:**
Not sure on the best way to fix this, the issue will always be related to
performance. Current fix checks if a reload is needed by sending a flag in the
`/documents/touch/<access_token>` request result when a new document
access was created.

opw-5156297

Forward-Port-Of: odoo/enterprise#104585
Forward-Port-Of: odoo/enterprise#99820

…records in hr_recruitment_reports demo data

Steps to reproduce:
1. Load demo data for hr_recruitment_reports module
2. Check applicant records in chatter
3. Observe that stage change messages don't show "Stage changed" label
4. Review mail.tracking.value records for applicant stage changes
5. Observe incorrect stage name "Initial Qualification" instead of "Qualification"

Bug cause:
mail.message records for applicant stage changes were missing the subtype_id field, which defaults to an incorrect subtype (mt_note instead of mt_applicant_stage_changed). This caused the messages to be treated as internal notes rather than stage change notifications.

Additionally, mail.tracking.value records used "Initial Qualification" as the old_value_char or new_value_char, which doesn't match the actual stage name defined in hr_recruitment module. The correct stage name is simply "Qualification" (ref: hr_recruitment.stage_job1).

Solution:
- Add subtype_id field with reference to hr_recruitment.mt_applicant_stage_changed to all mail.message records that track applicant stage transitions
- This ensures stage change events are properly identified and displayed in chatter with the correct "Stage changed" label
- Update old_value_char and new_value_char fields in mail.tracking.value records to use the correct stage name "Qualification" instead of "Initial Qualification"
- Ensures consistency with actual stage names and proper display in applicant chatter history

Affected records: 20 mail.message records in hr_recruitment_reports_demo.xml

Task Id:5454691

When trying to compute the document count
during the upgrade, we encountered a ValueError
because multiple records were found for a partner. The system expected a singleton

```
File "/home/odoo/src/enterprise/19.0/documents_hr/models/hr_employee.py", line 34, in _compute_document_count
    ('partner_id', '=', self.work_contact_id.id)
   File "/home/odoo/src/odoo/19.0/odoo/orm/fields_misc.py", line 112, in __get__
    raise ValueError("Expected singleton: %s" % record)
 ValueError: Expected singleton: res.partner(11393, 11612, 13026, 13085, 13066, 11674, 13007, 11240, 11420, 13086, 2687, 8998, 10309, 8195, 8468, 6439, 8151, 6580, 7928, 10301, 11058, 10515, 5274, 9243, 8141, 8435, 8889, 7761, 7733, 8443, 8545, 9252, 8457, 9980, 5495, 11424, 6458, 10558, 11070, 8924, 11731, 11528, 11615, 11766, 13021, 13080, 11758, 11742, 9306, 8826, 11004, 9393, 8879, 9317, 11652, 13075, 11744, 11160, 11644, 11763, 11416, 11618, 11732, 7931, 3846, 8442, 10299, 7916, 8429, 8057, 11061, 9342, 6435, 6553, 6512)
```

**Description of the issue/feature this PR addresses:**

Microsoft calendar synchronization may fail in environments with slower Microsoft Graph responses or large calendars because Graph API calls triggered after commit use a fixed 3-second timeout. This can lead to repeated synchronization failures even though the operation would succeed with slightly more time.

Additionally, when creating events, the Microsoft Graph request may time out after the event is successfully created on Microsoft’s side but before the response containing the event ID is returned. In this case, Odoo does not store the ID of the event and may create the same event again during the next sync, resulting in duplicate events.

**Current behavior before PR:**

Microsoft Graph requests (insert, update, delete) are executed with a hardcoded 3-second timeout. If the Graph API response takes longer:
	•	the synchronization fails,
	•	and in the case of event creation, Odoo may not receive the Microsoft event ID even though the event was created remotely, which can lead to duplicate events in Odoo.

**Desired behavior after PR is merged:**

The Microsoft Graph request timeout is configurable via the optional system parameter `microsoft_calendar.graph_timeout`.

If the parameter is not set, the behavior remains unchanged (default 3 seconds).
Administrators can increase the timeout to allow successful synchronization in slower environments or with large datasets, reducing synchronization failures and avoiding duplicate event creation caused by missing Microsoft IDs.


---
I confirm I have signed the CLA and read the PR guidelines at www.odoo.com/submit-pr

Forward-Port-Of: odoo/odoo#245022
Forward-Port-Of: odoo/odoo#241921

Scenario:

- Setup Unsplash and install website
- Set a user as "Website: Editor and Designer"
- Login as that user and go to any website page with a qweb view
- Go to edit mode and drop Text-Image in a view
- Replace the image with an unsplash image and then save

Result: the save fails without any message shown, and there is a
security access WARNING in the logs.

Note: a similar scenario can be done for a restricted editor that is
editing a HTML field it has write access to.

Issue: to save a model with res_id 0, we need to either be admin
(base.group_system) or the record creator. Since
9c9c58a5a10101532cbf046d21d4a63c2b7d2838 to bypass the mimetype
neutering of happening, we create the attachment as SUPERUSER.

Then when we modify the attachment url (for unsplash images), we have no
access right to the attachment since we are not the creator.

Fix: create the attachment with the current user, and only use
SUPERUSER to set the mimetype if it was neutered (ie. the user doesn't
have write access right to ir.ui.view, which in normal use case should
only happen for "Restricted Editor"). This way the image is created by
the user that uploaded it and not SUPERUSER.

opw-4850611
opw-5387258
opw-5489219

Forward-Port-Of: odoo/odoo#245085
Forward-Port-Of: odoo/odoo#219472

Currently, an error occurs when uploading `encrypted or incomplete` 
PDF files (missing EOF marker) while creating a quotation document 
header or footer.

**Steps to reproduce:**
- Install the `sale_pdf_quote_builder` module.
- Navigate to: Sales > Configuration > Headers/Footers.
- Upload encrypted file [1], or incomplete file [2].

**Error:**
`PyPDF2.errors.DependencyError: PyCryptodome is required for AES algorithm`
`PyPDF2.errors.PdfReadError: EOF marker not found`

**Root cause:**
At [3], `_get_form_fields_from_pdf` and `_ensure_document_not_encrypted` directly 
call `pdf.PdfFileReader`, when it fails to read or decrypt the file, Python raises 
an error.

**Fix:**
This commit prevents errors when users upload unreadable or encrypted PDF files.

[1]:
https://drive.google.com/file/d/1moSlwXHkqcV6_7zHBNhLMLi-9Ye_xDGJ/view?usp=sharing

[2]:
https://drive.google.com/file/d/16O4LLH8dL0RWmbOx4HrcooFUyesWaVi-/view?usp=sharing

[3]:
https://github.com/odoo/odoo/blob/694f1d0fb03b56dd41a59eb676e56622634cc91b/addons/sale_pdf_quote_builder/utils.py#L11

sentry-6928220164
opw-5227601

Forward-Port-Of: odoo/odoo#245167
Forward-Port-Of: odoo/odoo#230712

Issue before this commit:
=========================

The high-priority maintenance request count (todo_request_count_high_priority) 
was not calculated correctly.

Steps to Reproduce:
=========================
- Install the maintenance module.
- Create a maintenance request and set the Priority to High (3-starred) in the form view.
- Open the dashboard.
- Observe that the high-priority request count is not displayed.
- The count always remains 0, even when high-priority requests exist.

Cause of the issue:
=========================
In this [PR](https://github.com/odoo/odoo/pull/94866), the logic was mistakenly changed. 
The priority field is defined as a Selection field, but while computing the count, the comparison 
was done against an integer(3, not '3') instead of the actual string value. Since the stored 
value is '3' (string), the condition is always evaluated to False, resulting in a count of 0.

With This Commit:
=========================
Ensure that high-priority maintenance requests are correctly counted and displayed on 
the dashboard when they exist. This provides better visibility of critical requests and
helps users prioritise maintenance work effectively.

Forward-Port-Of: odoo/odoo#244987

Issue before this commit:
=========================

The high-priority maintenance request count (todo_request_count_high_priority) 
was not calculated correctly.

Steps to Reproduce:
=========================
- Install the maintenance module.
- Create a maintenance request and set the Priority to High (3-starred) in the form view.
- Open the dashboard.
- Observe that the high-priority request count is not displayed.
- The count always remains 0, even when high-priority requests exist.

Cause of the issue:
=========================
In this [PR](https://github.com/odoo/odoo/pull/94866), the logic was mistakenly changed. 
The priority field is defined as a Selection field, but while computing the count, the comparison 
was done against an integer(3, not '3') instead of the actual string value. Since the stored 
value is '3' (string), the condition is always evaluated to False, resulting in a count of 0.

With This Commit:
=========================
Ensure that high-priority maintenance requests are correctly counted and displayed on 
the dashboard when they exist. This provides better visibility of critical requests and
helps users prioritise maintenance work effectively.

Forward-Port-Of: odoo/odoo#244987

### Steps to reproduce:

- In the settings enable By-Products
- Create a product with an color attribute and 2 values: white, black
- Create a bom for that products and add 2 by product lines:
    - 1 x comp1 with a cost_share of 50% specific to the white att-value
    - 1 x comp2 with a cost_share of 70% specific to the Black att-value
> Try to save and you will raise a UserError: The total cost share for a BoM's by-products cannot exceed 100.

### Expected behavior:

The error should not be raised as the total cost_share is 50% for the white variant and 70% for the black one but none of them exceeds the 100% cost share.

### Cause of the Issue:

Currently the constraint does not take attribute values into accounts and simply sums the value of the cost share of all by-product lines: https://github.com/odoo/odoo/blob/bcc1397c7d694dbe61ecbd44d0320b9518df84cb/addons/mrp/models/mrp_bom.py#L201-L202

### Fix:

Just as for the total cost_share on kit products, we rely on the exclusion util and check for each existing product variant if the cost share set up is valid:
https://github.com/odoo/odoo/blob/7e81c528ae350aab4432207f5655dcfadf6ec627/addons/purchase_mrp/models/mrp_bom.py#L20-L23 see 3832793e3ce61aff0c7cf4673de84645a3469b3a

opw-5499773

---
I confirm I have signed the CLA and read the PR guidelines at www.odoo.com/submit-pr

Steps to reproduce the issue:
- Create a storable product “P1” with the following BoM:
    -Operation: OP1
- Create a MO to produce one unit of P1.
- Duplicate this MO.
- Confirm both MOs.
- Plan both MOs.
- Navigate to the Work Orders Planning page:
    - Move the work order of the second MO to the next day.
     - Block the work order of the first MO by the work order of the second MO.
- Go back to the MO list:
    - Unplan both MOs.
    - Attempt to replan the first MO.

Problem:
A validation error is triggered:
“The operation cannot be completed:
Create/update: a mandatory field is not set.
 Model: Production Order (mrp.production)
 Field: Start (date_start)"

When trying to plan the work order, we’ll checks if it is blocked by
another work order and skips planning if this is the case. Since the
work order is blocked by one from another MO, it remains unplanned:
https://github.com/odoo/odoo/blob/17.0/addons/mrp/models/mrp_production.py#L1510-L1512
As a result, the work order remains unplanned and does not have a
"leave_id" set. Consequently, the calculation for the "date_start" and
"date_finished" on the MO fails because the min function returns False.
This leads to an attempt to write a False value to the MO’s "date_start"
field:
https://github.com/odoo/odoo/blob/17.0/addons/mrp/models/mrp_production.py#L1518-L1521

opw-4393415