diff --git a/src/oca_pre_commit_hooks/checks_odoo_module_xml.py b/src/oca_pre_commit_hooks/checks_odoo_module_xml.py
index dfe0cfa..def64f4 100644
--- a/src/oca_pre_commit_hooks/checks_odoo_module_xml.py
+++ b/src/oca_pre_commit_hooks/checks_odoo_module_xml.py
@@ -679,14 +679,16 @@ def check_xml_double_quotes_py(self):
                     and (new_py_code := self.is_compatible_single_quote(py_code))
                 ):
                     # Process text <field name="context">{}</field>
-                    node_content = node_xml.NodeContent(manifest_data["filename"], elem)
+                    node_content = node_xml.NodeContent(
+                        manifest_data["filename"], elem, locator=self._get_tag_locator(manifest_data)
+                    )
                     if b"&quot;" in node_content.content_node:
                         self.register_error(
                             code="xml-double-quotes-py",
                             message='Escaped double quotes " for python code detected',
                             info=f"Use single quote instead: `{new_py_code}`",
                             filepath=manifest_data["filename_short"],
-                            line=elem.sourceline,
+                            line=node_content.start_sourceline or elem.sourceline,
                         )
                         during2 = node_content.content_node.replace(b"&quot;", b"'")
                         if self.autofix and during2 != node_content.content_node:
@@ -702,15 +704,22 @@ def check_xml_double_quotes_py(self):
                         continue
                     if not (new_py_code := self.is_compatible_single_quote(attr_value)):
                         continue
-                    node_content = node_xml.NodeContent(manifest_data["filename"], elem)
+                    locator = self._get_tag_locator(manifest_data)
+                    node_content = node_xml.NodeContent(manifest_data["filename"], elem, locator=locator)
                     if b"&quot;" not in node_content.content_node:
                         continue
+                    # Use the attribute's exact line from the locator (most precise)
+                    attr_span = locator.get_attr(elem, attr_name)
+                    if attr_span is not None:
+                        line_no = locator.content[: attr_span.name_start].count(b"\n") + 1
+                    else:
+                        line_no = node_content.start_sourceline or elem.sourceline
                     self.register_error(
                         code="xml-double-quotes-py",
                         message='Escaped double quotes " for python code detected use',
                         info=f"Use single quote instead: `{new_py_code}`",
                         filepath=manifest_data["filename_short"],
-                        line=elem.sourceline,
+                        line=line_no,
                     )
                     during2 = node_content.content_node.replace(b"&quot;", b"'")
                     if self.autofix and during2 != node_content.content_node:
diff --git a/src/oca_pre_commit_hooks/node_xml.py b/src/oca_pre_commit_hooks/node_xml.py
index 004cd69..ec600da 100644
--- a/src/oca_pre_commit_hooks/node_xml.py
+++ b/src/oca_pre_commit_hooks/node_xml.py
@@ -1,4 +1,5 @@
 # Based on https://github.com/mitsuhiko/sloppy-xml-py
+import re
 from dataclasses import dataclass
 
 
@@ -217,15 +218,19 @@ def rewrite_start_tag(
 class NodeContent:
     """Represents the content and metadata of an XML node."""
 
-    def __init__(self, filename, node):
+    def __init__(self, filename, node, locator=None):
         """Initialize by reading and parsing the node from the file.
 
         Args:
             filename: Path to the XML file
             node: lxml node element to extract
+            locator: Optional XMLStartTagLocator already built for this file.
+                     When provided it is used as the authoritative source for
+                     tag start position and line, avoiding fragile heuristics.
         """
         self.filename = filename
         self.node = node
+        self._locator = locator
 
         # Initialize attributes
         self.content_before = b""
@@ -240,28 +245,43 @@ def __init__(self, filename, node):
 
     def _read_node(self):  # noqa:C901 pylint:disable=too-complex
         """Internal method to read the content of the file and extract node information."""
-        # TODO: Get the sourceline of a particular attribute
-        # Determine the search start line
-        if (node_previous := self.node.getprevious()) is not None:
-            search_start_line = node_previous.sourceline + 1
-        elif (node_parent := self.node.getparent()) is not None:
-            search_start_line = node_parent.sourceline + 1
-        else:
-            search_start_line = 2  # first element and it is the root
-
-        search_end_line = self.node.sourceline
         node_tag = self.node.tag.encode() if isinstance(self.node.tag, str) else self.node.tag
 
         # Read all lines from file
         with open(self.filename, "rb") as f_content:
             all_lines = list((i, line) for i, line in enumerate(f_content, start=1))
 
-        # Find the actual node start by looking for the tag
+        # --- Determine tag start using the locator when available ---
+        # XMLStartTagLocator already did a full byte-accurate parse of the file,
+        # so use it as the authoritative source instead of re-scanning with heuristics.
         node_start_idx = None
-        for idx, (no_line, line) in enumerate(all_lines):
-            if search_start_line <= no_line <= search_end_line:
-                stripped_line = line.lstrip()
-                if stripped_line.startswith(b"<" + node_tag):
+        if self._locator is not None:
+            tag_info = self._locator.get_tag(self.node)
+            if tag_info is not None:
+                self.start_sourceline = tag_info.line
+                # Find the matching line index in all_lines
+                for idx, (no_line, _line) in enumerate(all_lines):
+                    if no_line == tag_info.line:
+                        node_start_idx = idx
+                        break
+
+        # --- Fallback: scan backwards from sourceline using a regex ---
+        # This handles tags that appear mid-line (e.g. "<li>text<strong").
+        if node_start_idx is None:
+            if (node_previous := self.node.getprevious()) is not None:
+                search_start_line = node_previous.sourceline + 1
+            elif (node_parent := self.node.getparent()) is not None:
+                search_start_line = node_parent.sourceline + 1
+            else:
+                search_start_line = 2  # first element and it is the root
+
+            search_end_line = self.node.sourceline
+            node_start_re = re.compile(b"<" + node_tag + b"(?:[ /><\n\r\t]|$)")
+            end_idx = min(search_end_line - 1, len(all_lines) - 1)
+            start_idx = max(search_start_line - 1, 0)
+            for idx in range(end_idx, start_idx - 1, -1):
+                no_line, line = all_lines[idx]
+                if node_start_re.search(line):
                     node_start_idx = idx
                     self.start_sourceline = no_line
                     break
diff --git a/test_repo/test_module/model_view.xml b/test_repo/test_module/model_view.xml
index 1fef119..b5179df 100644
--- a/test_repo/test_module/model_view.xml
+++ b/test_repo/test_module/model_view.xml
@@ -22,5 +22,23 @@
             <filter name="in_filter" string="In Filter" domain="[(&quot;state&quot;, &quot;=&quot;, &quot;done&quot;)]"/>
         </record>
 
+    <template id="template_doc2">
+        <t t-call="base.external_layout">
+            <t t-set="o" t-value="o.with_context(lang=lang)" />
+            <div class="page o_base_execution_request_page" t-translation="off">
+                <ul class="bullet-list o_base_execution_request_bullet_tab" t-translation="off">
+                    <li>Name:  <strong t-field="o.name" />, date <strong
+                            t-field="o.date_start"
+                        /></li>
+                    <li>Percentage  <strong
+                            t-out="o.coverage*100"
+                            t-options="{&quot;widget&quot;: &quot;float&quot;, &quot;precision&quot;: 2}"
+                        />
+                            %</li>
+                </ul>
+            </div>
+        </t>
+    </template>
+
     </data>
 </openerp>
diff --git a/tests/test_checks.py b/tests/test_checks.py
index 9b863ca..b4c5535 100644
--- a/tests/test_checks.py
+++ b/tests/test_checks.py
@@ -38,7 +38,7 @@
     "xml-deprecated-qweb-directive-15": 4,
     "xml-deprecated-qweb-directive": 2,
     "xml-deprecated-tree-attribute": 3,
-    "xml-double-quotes-py": 3,
+    "xml-double-quotes-py": 5,
     "xml-duplicate-fields": 3,
     "xml-duplicate-record-id": 2,
     "xml-not-valid-char-link": 2,
diff --git a/tests/test_node_xml.py b/tests/test_node_xml.py
index 455e35d..268ce6a 100644
--- a/tests/test_node_xml.py
+++ b/tests/test_node_xml.py
@@ -93,3 +93,36 @@ def test_xml_record_id_autofixes_preserve_menuitem_layout():
         xml_content = (module_dst / "model_view_odoo2.xml").read_text()
         assert "<menuitem id='menu_root' name=\"Root\" />" in xml_content
         assert '<menuitem id=\'menu_root2\'\n        name="Root 2"\n        parent="menu_root"' in xml_content
+
+
+def test_xml_double_quotes_py_reports_correct_line_for_inline_and_multiline_tags():
+    """Regression test: xml-double-quotes-py must report the line where the
+    *attribute* is declared, not the closing line of the opening tag.
+
+    Covers two scenarios:
+    1. model_view.xml line 35: a <strong> tag with t-options spanning multiple
+       lines and starting AFTER other text on the same line as <li>.
+    2. website_templates.xml line 33: same pattern in a different file.
+    """
+    manifest_path = str(REPO_ROOT / "test_repo/test_module/__openerp__.py")
+    result = checks_odoo_module.run(
+        [manifest_path],
+        enable={"xml-double-quotes-py"},
+        no_exit=True,
+    )
+    errors = [e for e in result if getattr(e, "code", None) == "xml-double-quotes-py"]
+
+    # model_view.xml: <strong t-out="o.coverage*100" t-options="{&quot;...&quot;}">
+    # The tag <strong starts after <li>text on line 33, t-options is on line 35
+    model_view_lines = {e.position.line for e in errors if "model_view.xml" in e.position.filepath}
+    assert 35 in model_view_lines, (
+        f"Expected xml-double-quotes-py error on model_view.xml:35 (t-options with &quot;), "
+        f"got lines: {sorted(model_view_lines)}"
+    )
+
+    # website_templates.xml: <strong t-options on line 33
+    tmpl_lines = {e.position.line for e in errors if "website_templates.xml" in e.position.filepath}
+    assert 33 in tmpl_lines, (
+        f"Expected xml-double-quotes-py error on website_templates.xml:33 (t-options with &quot;), "
+        f"got lines: {sorted(tmpl_lines)}"
+    )