Merge pull request #96 from marcusklaas/find-uncommented

Deal with problematic characters in comments

src/items.rs

@@ -11,7 +11,7 @@
 // Formatting top-level items - functions, structs, enums, traits, impls.
 
 use {ReturnIndent, BraceStyle};
-use utils::{format_visibility, make_indent};
+use utils::{format_visibility, make_indent, FindUncommented};
 use lists::{write_list, ListFormatting, SeparatorTactic, ListTactic};
 use visitor::FmtVisitor;
 use syntax::{ast, abi};
@@ -336,12 +336,11 @@ impl<'a> FmtVisitor<'a> {
         // FIXME If you thought the crap with the commas was ugly, just wait.
         // This is awful. We're going to look from the last item span to the
         // start of the return type span, then we drop everything after the
-        // first closing paren. Obviously, this will break if there is a
+        // first closing paren.
-        // closing paren in the comment.
         // The fix is comments in the AST or a span for the closing paren.
         let snippet = self.snippet(codemap::mk_sp(prev_end, next_span_start));
         let snippet = snippet.trim();
-        let snippet = &snippet[..snippet.find(terminator).unwrap_or(snippet.len())];
+        let snippet = &snippet[..snippet.find_uncommented(terminator).unwrap_or(snippet.len())];
         let snippet = snippet.trim();
         result.push(snippet.to_owned());
 
@@ -417,8 +416,7 @@ impl<'a> FmtVisitor<'a> {
         self.changes.push_str_span(span, &header_str);
 
         let enum_snippet = self.snippet(span);
-        // FIXME this will give incorrect results if there is a { in a comment.
+        let body_start = span.lo + BytePos(enum_snippet.find_uncommented("{").unwrap() as u32 + 1);
-        let body_start = span.lo + BytePos(enum_snippet.find('{').unwrap() as u32 + 1);
         let generics_str = self.format_generics(generics, body_start);
         self.changes.push_str_span(span, &generics_str);
 
@@ -542,8 +540,8 @@ impl<'a> FmtVisitor<'a> {
         self.changes.push_str_span(span, &generics_str);
 
         let struct_snippet = self.snippet(span);
-        // FIXME this will give incorrect results if there is a { in a comment.
+        // This will drop the comment in between the header and body.
-        self.last_pos = span.lo + BytePos(struct_snippet.find('{').unwrap() as u32 + 1);
+        self.last_pos = span.lo + BytePos(struct_snippet.find_uncommented("{").unwrap() as u32 + 1);
 
         self.block_indent += config!(tab_spaces);
         for (i, f) in struct_def.fields.iter().enumerate() {
@@ -632,8 +630,7 @@ impl<'a> FmtVisitor<'a> {
         // This hack makes sure we only add comments etc. after the comma, and
         // makes sure we don't repeat any commas.
         let hi = field.span.hi;
-        // FIXME a comma in a comment will break this hack.
+        let comma_pos = match struct_snippet[(hi.0 - struct_start.0) as usize..].find_uncommented(",") {
-        let comma_pos = match struct_snippet[(hi.0 - struct_start.0) as usize..].find(',') {
             Some(i) => i,
             None => 0,
         };

src/utils.rs

@@ -10,6 +10,72 @@
 
 use syntax::ast::Visibility;
 
+pub trait FindUncommented {
+    fn find_uncommented(&self, pat: &str) -> Option<usize>;
+}
+
+impl FindUncommented for str {
+    fn find_uncommented(&self, pat: &str) -> Option<usize> {
+        let mut needle_iter = pat.chars();
+        let mut possible_comment = false;
+
+        for (i, b) in self.char_indices() {
+            match needle_iter.next() {
+                Some(c) => {
+                    if b != c {
+                        needle_iter = pat.chars();
+                    }
+                },
+                None => return Some(i - pat.len())
+            }
+
+            if possible_comment {
+                if b == '/' {
+                    return self[(i+1)..].find('\n')
+                                        .and_then(|end| {
+                                            self[(end + i + 2)..].find_uncommented(pat)
+                                                                 .map(|idx| idx + end + i + 2)
+                                        });
+                } else if b == '*' {
+                    return self[(i+1)..].find("*/")
+                                        .and_then(|end| {
+                                            self[(end + i + 3)..].find_uncommented(pat)
+                                                                 .map(|idx| idx + end + i + 3)
+                                        });
+                } else {
+                    possible_comment = false;
+                }
+            } else {
+                possible_comment = b == '/';
+            }
+        }
+
+        // Handle case where the pattern is a suffix of the search string
+        match needle_iter.next() {
+            Some(_) => None,
+            None => Some(self.len() - pat.len())
+        }
+    }
+}
+
+#[test]
+fn test_find_uncommented() {
+    fn check(haystack: &str, needle: &str, expected: Option<usize>) {
+        assert_eq!(expected, haystack.find_uncommented(needle));
+    }
+
+    check("/*//*/test", "test", Some(6));
+    check("//test\ntest", "test", Some(7));
+    check("/* comment only */", "whatever", None);
+    check("/* comment */ some text /* more commentary */ result", "result", Some(46));
+    check("sup // sup", "p", Some(2));
+    check("sup", "x", None);
+    check("π? /**/ π is nice!", "π is nice", Some(9));
+    check("/*sup yo? \n sup*/ sup", "p", Some(20));
+    check("hel/*lohello*/lo", "hello", None);
+    check("acb", "ab", None);
+}
+
 #[inline]
 pub fn prev_char(s: &str, mut i: usize) -> usize {
     if i == 0 { return 0; }
@@ -63,6 +129,7 @@ pub fn round_up_to_power_of_two(mut x: usize) -> usize {
     x + 1
 }
 
+#[inline]
 #[cfg(target_pointer_width="32")]
 pub fn round_up_to_power_of_two(mut x: usize) -> usize {
     x -= 1;

tests/target/fn.rs

@@ -4,13 +4,13 @@ fn foo(a: AAAA, b: BBB, c: CCC) -> RetType {
 
 }
 
-fn foo(a: AAAA, b: BBB, c: CCC) -> RetType
+fn foo(a: AAAA, b: BBB /* some, weird, inline comment */, c: CCC) -> RetType
     where T: Blah
 {
 
 }
 
-fn foo(a: AAA)
+fn foo(a: AAA /* (comment) */)
     where T: Blah
 {