From 10cfcbed676f9005053a656a0aa543425c151089 Mon Sep 17 00:00:00 2001
From: barredterra <14891507+barredterra@users.noreply.github.com>
Date: Fri, 19 Jan 2024 16:22:54 +0100
Subject: [PATCH] feat: turn jinja extractor into a generic HTML template
 extractor

... supporting both jinja and JS microtemplates
---
 babel_extractors.csv                       |  2 +-
 frappe/gettext/extractors/html_template.py | 26 ++++++++++++++++++++++
 frappe/gettext/extractors/jinja2.py        | 11 ---------
 3 files changed, 27 insertions(+), 12 deletions(-)
 create mode 100644 frappe/gettext/extractors/html_template.py
 delete mode 100644 frappe/gettext/extractors/jinja2.py

diff --git a/babel_extractors.csv b/babel_extractors.csv
index 133a13e736..37858338ba 100644
--- a/babel_extractors.csv
+++ b/babel_extractors.csv
@@ -6,4 +6,4 @@ hooks.py,frappe.gettext.extractors.navbar.extract
 **/report/*/*.json,frappe.gettext.extractors.report.extract
 **.py,frappe.gettext.extractors.python.extract
 **.js,frappe.gettext.extractors.javascript.extract
-**.html,frappe.gettext.extractors.jinja2.extract
\ No newline at end of file
+**.html,frappe.gettext.extractors.html_template.extract
\ No newline at end of file
diff --git a/frappe/gettext/extractors/html_template.py b/frappe/gettext/extractors/html_template.py
new file mode 100644
index 0000000000..34f51e4032
--- /dev/null
+++ b/frappe/gettext/extractors/html_template.py
@@ -0,0 +1,26 @@
+from jinja2.ext import babel_extract
+
+from .utils import extract_messages_from_code
+
+
+def extract(*args, **kwargs):
+	"""Extract messages from Jinja and JS microtemplates.
+
+	Reuse the babel_extract function from jinja2.ext, but handle our own implementation of `_()`.
+	To handle JS microtemplates, parse all code again using regex."""
+	fileobj = args[0] or kwargs["fileobj"]
+	print(fileobj.name)
+	code = fileobj.read().decode("utf-8")
+
+	for lineno, funcname, messages, comments in babel_extract(*args, **kwargs):
+		if funcname == "_" and isinstance(messages, tuple) and len(messages) > 1:
+			funcname = "pgettext"
+			messages = (messages[-1], messages[0])  # (context, message)
+
+		yield lineno, funcname, messages, comments
+
+	for lineno, message, context in extract_messages_from_code(code):
+		if context:
+			yield lineno, "pgettext", (context, message), []
+		else:
+			yield lineno, "_", message, []
diff --git a/frappe/gettext/extractors/jinja2.py b/frappe/gettext/extractors/jinja2.py
deleted file mode 100644
index ee07ac6cee..0000000000
--- a/frappe/gettext/extractors/jinja2.py
+++ /dev/null
@@ -1,11 +0,0 @@
-from jinja2.ext import babel_extract
-
-
-def extract(*args, **kwargs):
-	"""Reuse the babel_extract function from jinja2.ext, but handle our own implementation of `_()`"""
-	for lineno, funcname, messages, comments in babel_extract(*args, **kwargs):
-		if funcname == "_" and isinstance(messages, tuple) and len(messages) > 1:
-			funcname = "pgettext"
-			messages = (messages[-1], messages[0])  # (context, message)
-
-		yield lineno, funcname, messages, comments