fix: detect go package clause after leading comments

mdsf/src/filetype/generated_file_types.rs

@@ -336,6 +336,7 @@ pub fn language_to_ext(language: &str) -> Option<&'static str> {
         "fortran free form" => Some(".f90"),
         "fortran" => Some(".f"),
         "foxpro" => Some(".prg"),
+        "fpp" => Some(".fpp"),
         "freebasic" => Some(".bi"),
         "freemarker" => Some(".ftl"),
         "frege" => Some(".fr"),
@@ -851,6 +852,7 @@ pub fn language_to_ext(language: &str) -> Option<&'static str> {
         "qsharp" => Some(".qs"),
         "qt script" => Some(".qs"),
         "quakec" => Some(".qc"),
+        "quartus simulation ip" => Some(".sip"),
         "quickbasic" => Some(".bas"),
         "quint" => Some(".qnt"),
         "r" => Some(".r"),
@@ -888,6 +890,7 @@ pub fn language_to_ext(language: &str) -> Option<&'static str> {
         "restructuredtext" => Some(".rst"),
         "rexx" => Some(".rexx"),
         "rez" => Some(".r"),
+        "rhai" => Some(".rhai"),
         "rhtml" => Some(".erb"),
         "rich text format" => Some(".rtf"),
         "ring" => Some(".ring"),
@@ -946,6 +949,7 @@ pub fn language_to_ext(language: &str) -> Option<&'static str> {
         "shen" => Some(".shen"),
         "sieve" => Some(".sieve"),
         "simple file verification" => Some(".sfv"),
+        "sip" => Some(".sip"),
         "slang" => Some(".slang"),
         "slash" => Some(".sl"),
         "slice" => Some(".ice"),
@@ -1085,6 +1089,8 @@ pub fn language_to_ext(language: &str) -> Option<&'static str> {
         "vento" => Some(".vto"),
         "verilog" => Some(".v"),
         "verse" => Some(".verse"),
+        "vespa schema definition" => Some(".sd"),
+        "vespa" => Some(".sd"),
         "vhdl" => Some(".vhdl"),
         "vhs" => Some(".tape"),
         "vim help file" => Some(".txt"),
@@ -1520,6 +1526,7 @@ mod test_language_to_ext {
             ("fortran free form", ".f90"),
             ("fortran", ".f"),
             ("foxpro", ".prg"),
+            ("fpp", ".fpp"),
             ("freebasic", ".bi"),
             ("freemarker", ".ftl"),
             ("frege", ".fr"),
@@ -2035,6 +2042,7 @@ mod test_language_to_ext {
             ("qsharp", ".qs"),
             ("qt script", ".qs"),
             ("quakec", ".qc"),
+            ("quartus simulation ip", ".sip"),
             ("quickbasic", ".bas"),
             ("quint", ".qnt"),
             ("r", ".r"),
@@ -2072,6 +2080,7 @@ mod test_language_to_ext {
             ("restructuredtext", ".rst"),
             ("rexx", ".rexx"),
             ("rez", ".r"),
+            ("rhai", ".rhai"),
             ("rhtml", ".erb"),
             ("rich text format", ".rtf"),
             ("ring", ".ring"),
@@ -2130,6 +2139,7 @@ mod test_language_to_ext {
             ("shen", ".shen"),
             ("sieve", ".sieve"),
             ("simple file verification", ".sfv"),
+            ("sip", ".sip"),
             ("slang", ".slang"),
             ("slash", ".sl"),
             ("slice", ".ice"),
@@ -2269,6 +2279,8 @@ mod test_language_to_ext {
             ("vento", ".vto"),
             ("verilog", ".v"),
             ("verse", ".verse"),
+            ("vespa schema definition", ".sd"),
+            ("vespa", ".sd"),
             ("vhdl", ".vhdl"),
             ("vhs", ".tape"),
             ("vim help file", ".txt"),

mdsf/src/parser/mod.rs

@@ -32,17 +32,39 @@ pub fn parse_generic_codeblock(lines: &mut Enumerate<Lines>) -> (bool, String, u
 pub fn parse_go_codeblock(lines: &mut Enumerate<Lines>) -> (bool, String, usize) {
     let (is_snippet, mut code_snippet, snippet_lines) = parse_generic_codeblock(lines);
 
-    if is_snippet && !GO_PACKAGE_RE.is_match(&code_snippet) {
+    if is_snippet && !has_go_package(&code_snippet) {
         code_snippet.insert_str(0, GO_TEMPORARY_PACKAGE_NAME);
     }
 
     (is_snippet, code_snippet, snippet_lines)
 }
 
-// TODO: check for multiline comments
 pub static GO_PACKAGE_RE: std::sync::LazyLock<Regex> =
     std::sync::LazyLock::new(|| Regex::new(r"^\s*package\s+\w").unwrap());
 
+/// A Go package clause is the first token of a file that is neither a comment
+/// nor whitespace, so leading comments are skipped before looking for it.
+#[inline]
+pub fn has_go_package(snippet: &str) -> bool {
+    let mut rest = snippet.trim_start();
+
+    loop {
+        if let Some(after) = rest.strip_prefix("//") {
+            rest = after
+                .split_once('\n')
+                .map_or("", |(_, tail)| tail)
+                .trim_start();
+        } else if let Some(after) = rest.strip_prefix("/*") {
+            rest = after
+                .split_once("*/")
+                .map_or("", |(_, tail)| tail)
+                .trim_start();
+        } else {
+            return GO_PACKAGE_RE.is_match(rest);
+        }
+    }
+}
+
 #[inline]
 pub fn remove_go_package(snippet: String) -> String {
     if snippet.contains(GO_TEMPORARY_PACKAGE_NAME) {
@@ -99,3 +121,33 @@ mod test_go_package_re {
         }
     }
 }
+
+#[cfg(test)]
+mod test_has_go_package {
+    use crate::parser::has_go_package;
+
+    #[test]
+    fn it_should_match() {
+        for s in [
+            "package mdsf",
+            "// mdsf.go\npackage mdsf",
+            "//go:build integration\n\npackage mdsf",
+            "// mdsf.go\n// second line\npackage mdsf",
+            "/*\nI like my\npackage manager\n*/\npackage mdsf",
+        ] {
+            assert!(has_go_package(s), "'{s}' did not match");
+        }
+    }
+
+    #[test]
+    fn it_should_not_match() {
+        for s in [
+            "const foo = 1",
+            "// package mdsf",
+            "// mdsf.go\nconst foo = 1",
+            "/*\nI like my\npackage manager\n*/\n\nconst foo = 1",
+        ] {
+            assert!(!has_go_package(s), "'{s}' matched");
+        }
+    }
+}