Detect utf8 bom from meson build files

Some text editors on Windows may use utf8bom encoding by default.
Prevent crash and properly report misencoded files.

Fixes #12766.

10 files changed, 42 insertions, 4 deletions

diff --git a/mesonbuild/interpreterbase/interpreterbase.py b/mesonbuild/interpreterbase/interpreterbase.py
index 365e2fd4b..47dd46f32 100644
--- a/mesonbuild/interpreterbase/interpreterbase.py
+++ b/mesonbuild/interpreterbase/interpreterbase.py
@@ -107,10 +107,11 @@ class InterpreterBase:
             self.handle_meson_version_from_ast()
         except mparser.ParseException as me:
             me.file = mesonfile
-            # try to detect parser errors from new syntax added by future
-            # meson versions, and just tell the user to update meson
-            self.ast = me.ast
-            self.handle_meson_version_from_ast()
+            if me.ast:
+                # try to detect parser errors from new syntax added by future
+                # meson versions, and just tell the user to update meson
+                self.ast = me.ast
+                self.handle_meson_version_from_ast()
             raise me
 
     def parse_project(self) -> None:
diff --git a/mesonbuild/mparser.py b/mesonbuild/mparser.py
index 28235b613..baccd2753 100644
--- a/mesonbuild/mparser.py
+++ b/mesonbuild/mparser.py
@@ -96,6 +96,10 @@ class Token(T.Generic[TV_TokenTypes]):
 
 class Lexer:
     def __init__(self, code: str):
+        if code.startswith(codecs.BOM_UTF8.decode('utf-8')):
+            line, *_ = code.split('\n', maxsplit=1)
+            raise ParseException('Builder file must be encoded in UTF-8 (with no BOM)', line, lineno=0, colno=0)
+
         self.code = code
         self.keywords = {'true', 'false', 'if', 'else', 'elif',
                          'endif', 'and', 'or', 'not', 'foreach', 'endforeach',
diff --git a/test cases/failing/130 utf8 with bom/meson.build b/test cases/failing/130 utf8 with bom/meson.build
new file mode 100644
index 000000000..492a0c691
--- /dev/null
+++ b/test cases/failing/130 utf8 with bom/meson.build
@@ -0,0 +1,3 @@
+project('utf8 with bom')
+
+subdir('subdir')
diff --git a/test cases/failing/130 utf8 with bom/test.json b/test cases/failing/130 utf8 with bom/test.json
new file mode 100644
index 000000000..2292bebe3
--- /dev/null
+++ b/test cases/failing/130 utf8 with bom/test.json
@@ -0,0 +1,8 @@
+{
+    "stdout": [
+      {
+        "line": "test cases/failing/130 utf8 with bom/meson.build:0:0: ERROR: Builder file must be encoded in UTF-8 (with no BOM)"
+      }
+    ]
+  }
+  
\ No newline at end of file
diff --git a/test cases/failing/131 utf8 with bom subdir/meson.build b/test cases/failing/131 utf8 with bom subdir/meson.build
new file mode 100644
index 000000000..8d3bbd7e8
--- /dev/null
+++ b/test cases/failing/131 utf8 with bom subdir/meson.build
@@ -0,0 +1,3 @@
+project('utf8 with bom subdir')
+
+subdir('subdir')
diff --git a/test cases/failing/131 utf8 with bom subdir/subdir/meson.build b/test cases/failing/131 utf8 with bom subdir/subdir/meson.build
new file mode 100644
index 000000000..dbf2b9cc7
--- /dev/null
+++ b/test cases/failing/131 utf8 with bom subdir/subdir/meson.build
@@ -0,0 +1 @@
+a = 'Hello, World!'
diff --git a/test cases/failing/131 utf8 with bom subdir/test.json b/test cases/failing/131 utf8 with bom subdir/test.json
new file mode 100644
index 000000000..72cc5152d
--- /dev/null
+++ b/test cases/failing/131 utf8 with bom subdir/test.json
@@ -0,0 +1,8 @@
+{
+    "stdout": [
+      {
+        "line": "test cases/failing/131 utf8 with bom subdir/subdir/meson.build:0:0: ERROR: Builder file must be encoded in UTF-8 (with no BOM)"
+      }
+    ]
+  }
+  
\ No newline at end of file
diff --git a/test cases/failing/132 utf8 with bom options/meson.build b/test cases/failing/132 utf8 with bom options/meson.build
new file mode 100644
index 000000000..50413e031
--- /dev/null
+++ b/test cases/failing/132 utf8 with bom options/meson.build
@@ -0,0 +1 @@
+project('utf8 with bom options')
diff --git a/test cases/failing/132 utf8 with bom options/meson.options b/test cases/failing/132 utf8 with bom options/meson.options
new file mode 100644
index 000000000..250c03279
--- /dev/null
+++ b/test cases/failing/132 utf8 with bom options/meson.options
@@ -0,0 +1 @@
+option('someoption', type : 'string', value : 'optval', description : 'An option')
diff --git a/test cases/failing/132 utf8 with bom options/test.json b/test cases/failing/132 utf8 with bom options/test.json
new file mode 100644
index 000000000..c09f48e6a
--- /dev/null
+++ b/test cases/failing/132 utf8 with bom options/test.json
@@ -0,0 +1,8 @@
+{
+    "stdout": [
+      {
+        "line": "test cases/failing/132 utf8 with bom options/meson.options:0:0: ERROR: Builder file must be encoded in UTF-8 (with no BOM)"
+      }
+    ]
+  }
+  
\ No newline at end of file