Merge pull request #29 from cgag/lua-fix

fix lua counting

Author: cgag

Cargo.lock

@@ -5,7 +5,7 @@ extern crate memmap;
 extern crate memchr;
 
 use std::path::Path;
-use std::cmp;
+use std::cmp::{min, max};
 use std::fmt;
 
 use memmap::{Mmap, Protection};
@@ -40,9 +40,7 @@ pub enum LineConfig<'a> {
         multi_start: &'a str,
         multi_end: &'a str,
     },
-    SingleOnly {
-        single_start: &'a str,
-    },
+    SingleOnly { single_start: &'a str },
     MultiOnly {
         multi_start: &'a str,
         multi_end: &'a str,
@@ -584,6 +582,7 @@ pub fn count_single(filepath: &str, single_start: &str) -> Count {
     c
 }
 
+// TODO(cgag): don't forget to update this when fixing the lua bug
 pub fn count_multi(filepath: &str, multi_start: &str, multi_end: &str) -> Count {
     // this is a duplicate of count_single_multi without the check for single comment.
     // Basically removes one branch.  Probably pointless: benchmark.
@@ -630,7 +629,7 @@ pub fn count_multi(filepath: &str, multi_start: &str, multi_end: &str) -> Count
         let mut found_code = false;
         'outer: while pos < trimmed.len() {
             // TODO(cgag): must be a less stupid way to do this
-            for i in pos..(pos + cmp::max(start_len, end_len) + 1) {
+            for i in pos..pos + min(max(start_len, end_len) + 1, trimmed.len() - pos) {
                 if !trimmed.is_char_boundary(i) {
                     pos += 1;
                     continue 'outer;
@@ -642,7 +641,7 @@ pub fn count_multi(filepath: &str, multi_start: &str, multi_end: &str) -> Count
                 pos += start_len;
                 in_comment = true;
             } else if in_comment && pos + end_len <= trimmed.len() &&
-               &trimmed[pos..(pos + end_len)] == multi_end {
+                      &trimmed[pos..(pos + end_len)] == multi_end {
                 pos += end_len;
                 in_comment = false;
                 // TODO(cgag): should we bother handling whitespace here?
@@ -718,13 +717,17 @@ pub fn count_single_multi(filepath: &str,
         };
         c.lines += 1;
 
+
         let trimmed = line.trim_left();
         if trimmed.is_empty() {
             c.blank += 1;
             continue;
         };
 
-        if !in_comment && trimmed.starts_with(single_start) {
+        // TODO(cgag): Could be more efficient by only doing this third check when
+        // multi_start starts with the same chars as single_start, such as with lua (--, --[, ]]).
+        // Not sure it's necessary
+        if !in_comment && trimmed.starts_with(single_start) && !trimmed.starts_with(multi_start) {
             c.comment += 1;
             continue;
         }
@@ -748,7 +751,7 @@ pub fn count_single_multi(filepath: &str,
             // TODO(cgag): must be a less stupid way to do this.  At the
             // very least don't recalculate max over and over.  LLVM probably
             // optimizes this but it seems dumb to depend on it?
-            for i in pos..(pos + cmp::max(start_len, end_len) + 1) {
+            for i in pos..pos + min(max(start_len, end_len) + 1, trimmed.len() - pos) {
                 if !trimmed.is_char_boundary(i) {
                     pos += 1;
                     continue 'outer;
@@ -760,7 +763,7 @@ pub fn count_single_multi(filepath: &str,
                 pos += start_len;
                 in_comment = true;
             } else if in_comment && pos + end_len <= trimmed_len &&
-               &trimmed[pos..(pos + end_len)] == multi_end {
+                      &trimmed[pos..(pos + end_len)] == multi_end {
                 pos += end_len;
                 in_comment = false;
                 // TODO(cgag): should we bother handling whitespace here?

src/lib.rs

@@ -11,6 +11,7 @@ const PLASMA_EXPECTED: Count = Count {
 };
 
 
+
 #[test]
 fn test_plasma_count() {
     assert_eq!(PLASMA_EXPECTED, count(PLASMA));
@@ -172,3 +173,37 @@ fn test_ipl_blank() {
 fn test_ipl_lines() {
     assert_eq!(IPL_EXPECTED.lines, count(IPL).lines);
 }
+
+// TODO(cgag): find or make a better testing tool? Or add some simple macros?
+const LUA: &'static str = "tests/data/lua.lua";
+const LUA_EXPECTED: Count = Count {
+    code: 7,
+    blank: 1,
+    comment: 8,
+    lines: 7 + 8 + 1,
+};
+
+#[test]
+fn test_lua_count() {
+    assert_eq!(LUA_EXPECTED, count(LUA));
+}
+
+#[test]
+fn test_lua_code() {
+    assert_eq!(LUA_EXPECTED.code, count(LUA).code);
+}
+
+#[test]
+fn test_lua_comment() {
+    assert_eq!(LUA_EXPECTED.comment, count(LUA).comment);
+}
+
+#[test]
+fn test_lua_blank() {
+    assert_eq!(LUA_EXPECTED.blank, count(LUA).blank);
+}
+
+#[test]
+fn test_lua_lines() {
+    assert_eq!(LUA_EXPECTED.lines, count(LUA).lines);
+}

tests/count.rs

@@ -0,0 +1,16 @@
+--[[ This
+     is
+     a
+     multi-line
+     comment,
+     not
+     code. ]]
+
+-- build table
+statetab = {}
+local w1, w2 = NOWORD, NOWORD
+for w in allwords() do
+  insert(prefix(w1, w2), w)
+  w1 = w2; w2 = w;
+end
+insert(prefix(w1, w2), NOWORD)