simplify a bit

Author: yostashiro

web_form_banner/README.rst

@@ -43,8 +43,9 @@ Usage
 =====
 
 #. Go to *Settings > Tachnical > User Interface > Form Banner Rules* and create a rule.
-#. Choose Model, (optionally) restrict Form Views, set Default Severity, Target XPath
-   (insertion point), Position, and configure the message.
+#. Choose Model, select Trigger Fields (optional), set Default Severity, select Views
+   (optional), update Target XPath (insertion point) and Position, and configure the
+   message.
 #. Save. Open any matching form record—the banner will appear and auto-refresh after
    load/save/reload.
 
@@ -68,6 +69,13 @@ Evaluation context variables available in Message Value Code:
 * `user`: Current user (`env.user`).
 * `ctx`: Copy of the current context (`dict(env.context)`).
 * `record`: Current record (the form's record).
+* `draft`: The persisted field values of the ORM record (before applying the current
+  form's unsaved changes) + the current unsaved changes on trigger fields.
+  Should be used instead of `record` when your rule is triggered dynamically by an
+  update to a trigger field. It doesn't include any values from complex fields
+  (x2many/reference, etc).
+* `record_id`: Integer id of the record being edited, or `False` if the form
+  is creating a new record.
 * `model`: Shortcut to the current model (`env[record._name]`).
 * `url_for(obj)`: Helper that returns a backend form URL for `obj`.
 * `context_today(ts=None)`: User-timezone “today” (date) for reliable date comparisons.
@@ -79,6 +87,19 @@ Evaluation context variables available in Message Value Code:
 
 All of the above are injected by the module to the safe_eval locals.
 
+Trigger Fields
+~~~~~~~~~~~~~~
+
+*Trigger Fields* is an optional list of model fields that, when changed in the open
+form, cause the banner to **recompute live**. If left empty, the banner does **not**
+auto-refresh as the user edits the form.
+
+When a trigger fires, the module sends the current draft values to the server, sanitizes
+them, builds an evaluation record, and re-runs your `message_value_code`.
+
+You should use `draft` instead of `record` to access the current form values if your
+rule is triggered based on an update to a trigger field.
+
 Message setting examples:
 ~~~~~~~~~~~~~~~~~~~~~~~~~
 
@@ -123,7 +144,7 @@ It is also possible to use "convenience placeholders" without an explicit `value
 .. code-block:: python
 
   {
-    "visible": record.amount_total > 30000,
+    "visible": record.amount_total >= 30000,
     "severity": "danger" if record.amount_total >= 100000 else "warning",
     "values": {"amount_total": record.amount_total},
   }
@@ -141,7 +162,7 @@ It is also possible to use "convenience placeholders" without an explicit `value
     "values": {"validity_date": record.validity_date},
   }
 
-**E) Pending activities on a task (uses env)**
+**E) Pending activities on a task (uses `env`)**
 
 * Model: `project.task`
 * Message: `There are ${cnt} pending activities.`
@@ -152,19 +173,30 @@ It is also possible to use "convenience placeholders" without an explicit `value
   cnt = env["mail.activity"].search_count([("res_model","=",record._name),("res_id","=",record.id)])
   result = {"visible": cnt > 0, "values": {"cnt": cnt}}
 
-**F) HTML banner linking to the customer's last sales order**
+**F) Product is missing internal reference (uses trigger fields)**
+
+* Model: `product.template`
+* Trigger Fields: `default_code`
+* Message: `Make sure to set an internal reference!`
+* Message Value Code:
+
+.. code-block:: python
+
+  {"visible": not bool(draft.default_code)}
+
+**G) HTML banner linking to the customer's last sales order (uses trigger fields)**
 
 * Model: `sale.order`
+* Trigger Fields: `partner_id`
 * Message: (leave blank; `html` provided by Message Value Code)
 * Message Value Code (multi-line with `result`):
 
 .. code-block:: python
 
-  last = model.search(
-    [("partner_id", "=", record.partner_id.id), ("id", "<", record.id)],
-    order="date_order desc, id desc",
-    limit=1,
-  )
+  domain = [("partner_id", "=", draft.partner_id.id)]
+  if record_id:
+    domain += [("id", "<", record_id)]
+  last = model.search(domain, order="date_order desc, id desc", limit=1)
   if last:
     html = "<strong>Previous order:</strong> <a href='%s'>%s</a>" % (url_for(last), last.name)
     result = {"visible": True, "html": html}
@@ -174,9 +206,23 @@ It is also possible to use "convenience placeholders" without an explicit `value
 Known issues / Roadmap
 ======================
 
+Banner presentation inside `<group>`
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
 Placing a full-width inline banner inside `<group>` is currently not supported. The
-banner will be limited to 50% of the group's width, and its label/value ratio will be
-forced to 1:1.
+presentation of the banner and the child fields will be distorted.
+
+Limitations of `draft` eval context variable
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+* `draft` is always available in the eval context, but for new records (`record_id` =
+  `False`) it only contains the trigger fields from the banner rules.
+* For existing records, `draft` overlays the trigger field values on top of the
+  persisted record; all other fields come from `Model.new` defaults rather than the
+  database.
+* Only simple field types are included: `char`, `text`, `html`, `selection`, `boolean`,
+  `integer`, `float`, `monetary`, `date`, `datetime`, and `many2one` (normalized to an
+  integer ID). **x2many/reference/other types are omitted.**
 
 Bug Tracker
 ===========
@@ -202,6 +248,7 @@ Contributors
 * `Quartile <https://www.quartile.co>`_:
 
   * Yoshi Tashiro
+  * Aung Ko Ko Lin
 
 Maintainers
 ~~~~~~~~~~~

web_form_banner/models/ir_model.py

@@ -41,11 +41,7 @@ def fields_view_get(
                 continue
             # Lightweight placeholder; JS will fill and toggle visibility
             css = "o_form_banner alert alert-%s" % (rule.severity or "danger")
-            triggers_fields = (
-                ",".join(rule.trigger_field_ids.mapped("name"))
-                if getattr(rule, "trigger_field_ids", False)
-                else ""
-            )
+            trigger_fields = ",".join(rule.trigger_field_ids.mapped("name"))
             node = etree.Element(
                 "div",
                 {
@@ -54,7 +50,7 @@ def fields_view_get(
                     "data-rule-id": str(rule.id),
                     "data-model": self._name,
                     "data-default-severity": (rule.severity or "danger"),
-                    "data-trigger-fields": triggers_fields,
+                    "data-trigger-fields": trigger_fields,
                     "style": "display:none;",
                 },
             )

web_form_banner/models/web_form_banner_rule.py

@@ -21,29 +21,6 @@
 _logger = logging.getLogger(__name__)
 
 
-class _EvalRecordProxy:
-    """Read-only-ish view of an existing record with field overrides."""
-    __slots__ = ("_b", "_o")
-
-    def __init__(self, base, overrides):
-        self._b = base
-        self._o = overrides
-
-    def __getattr__(self, name):
-        # Prefer explicit overrides; otherwise fall back to the base record
-        if name in self._o:
-            return self._o[name]
-        return getattr(self._b, name)
-
-    def __repr__(self):
-        return f"<_EvalRecordProxy base={self._b} overrides={list(self._o.keys())}>"
-
-    @property
-    def id(self):
-        # Keep the real DB id for RPCs / URL building / permissions
-        return self._b.id
-
-
 _SIMPLE_FIELD_TYPES = frozenset(
     {
         "char", "text", "html", "selection", "boolean",
@@ -191,7 +168,7 @@ def _get_eval_context(self, record):
         return eval_ctx
 
     @api.model
-    def _sanitize_draft(self, model, form_vals):
+    def _sanitize_values(self, model, form_vals):
         """Return a sanitized dict of simple field values safe for new()/eval."""
         flds = self.env[model]._fields
         out = {}
@@ -203,26 +180,16 @@ def _sanitize_draft(self, model, form_vals):
 
     @api.model
     def _build_eval_record(self, model, res_id, vals):
-        """Build the record used for evaluation.
-        - existing record: wrap with overrides but keep real id
-        - new record: new(vals)
-        """
-        if not res_id:
-            return self.env[model].new(vals) if vals else self.env[model]
-        base = self.env[model].browse(int(res_id))
-        if not vals:
-            return base
-        flds = self.env[model]._fields
-        ovr = {}
-        for n, v in vals.items():
-            f = flds[n]
-            if not f:
-                continue
-            if f.type == "many2one" and isinstance(v, int):
-                ovr[n] = self.env[f.comodel_name].browse(v)
-            else:
-                ovr[n] = v
-        return _EvalRecordProxy(base, ovr)
+        """Return (draft_record, record_id, persisted_record)."""
+        Model = self.env[model]
+        vals = vals or {}
+        if res_id:
+            persisted = Model.browse(int(res_id))
+            base_vals = persisted.read(list(vals.keys()))[0] if vals else {}
+            draft = Model.new({**base_vals, **vals})
+            return draft, persisted, persisted.id
+        # new record (no res_id yet): persisted is an empty recordset, not None
+        return Model.new(vals), Model, False
 
     @api.model
     def _run_rule_code(self, rule, eval_ctx):
@@ -259,16 +226,19 @@ def compute_message(self, rule_id, model, res_id, form_vals=None):
         rule = self.browse(int(rule_id)).sudo()
         if not rule.exists() or not rule.active:
             return {"visible": False}
-        vals = self._sanitize_draft(model, form_vals)
-        record = self._build_eval_record(model, res_id, vals)
+        values = self._sanitize_values(model, form_vals)
+        # record, current_id = self._build_eval_record(model, res_id, vals)
+        draft, record, record_id = self._build_eval_record(model, res_id, values)
         eval_ctx = self._get_eval_context(record)
-        # expose changes for rule code that wants direct access to raw values
         eval_ctx.update(
-            {"changes": vals, "current_id": int(res_id) if res_id else False}
+            {
+                "draft": draft,  # DB base + simple field overrides
+                "record_id": record_id,
+            }
         )
         out = self._run_rule_code(rule, eval_ctx) or {}
         severity = out.get("severity", rule.severity or "danger")
-        visible = out.get("visible", True)  # default True like before
+        visible = out.get("visible", True)
         if not visible:
             return {"visible": False}
         values = out.get("values") or {

web_form_banner/readme/ROADMAP.rst

@@ -1,3 +1,17 @@
+Banner presentation inside `<group>`
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
 Placing a full-width inline banner inside `<group>` is currently not supported. The
-banner will be limited to 50% of the group's width, and its label/value ratio will be
-forced to 1:1.
+presentation of the banner and the child fields will be distorted.
+
+Limitations of `draft` eval context variable
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+* `draft` is always available in the eval context, but for new records (`record_id` =
+  `False`) it only contains the trigger fields from the banner rules.
+* For existing records, `draft` overlays the trigger field values on top of the
+  persisted record; all other fields come from `Model.new` defaults rather than the
+  database.
+* Only simple field types are included: `char`, `text`, `html`, `selection`, `boolean`,
+  `integer`, `float`, `monetary`, `date`, `datetime`, and `many2one` (normalized to an
+  integer ID). **x2many/reference/other types are omitted.**