impl: fix --multiline anchored match bug

This fixes a bug where using \A or (?-m)^ in combination with
-U/--multiline would permit matches that aren't anchored to the
beginning of the file. The underlying cause was an optimization that
occurred when mmaps couldn't be used. Namely, ripgrep tries to still
read the input incrementally if it knows the pattern can't match through
a new line. But the detection logic was flawed, since it didn't account
for line anchors. This commit fixes that.

Fixes #1878, Fixes #1879

CHANGELOG.md

@@ -23,6 +23,8 @@ Bug fixes:
   Fix stdin detection when using PowerShell in UNIX environments.
 * [BUG #1866](https://github.com/BurntSushi/ripgrep/issues/1866#issuecomment-841635553):
   Fix bug when computing column numbers in `--vimgrep` mode.
+* [BUG #1878](https://github.com/BurntSushi/ripgrep/issues/1878):
+  Fix bug where `\A` could produce unanchored matches in multiline search.
 
 
 12.1.1 (2020-05-29)

crates/regex/src/non_matching.rs

@@ -13,7 +13,10 @@ pub fn non_matching_bytes(expr: &Hir) -> ByteSet {
 /// the given expression.
 fn remove_matching_bytes(expr: &Hir, set: &mut ByteSet) {
     match *expr.kind() {
-        HirKind::Empty | HirKind::Anchor(_) | HirKind::WordBoundary(_) => {}
+        HirKind::Empty | HirKind::WordBoundary(_) => {}
+        HirKind::Anchor(_) => {
+            set.remove(b'\n');
+        }
         HirKind::Literal(hir::Literal::Unicode(c)) => {
             for &b in c.encode_utf8(&mut [0; 4]).as_bytes() {
                 set.remove(b);

tests/regression.rs

@@ -882,3 +882,26 @@ test:3:5:foo quux
 ";
     eqnice!(expected, cmd.stdout());
 });
+
+rgtest!(r1878, |dir: Dir, _: TestCommand| {
+    dir.create("test", "a\nbaz\nabc\n");
+
+    // Since ripgrep enables (?m) by default, '^' will match at the beginning
+    // of a line, even when -U/--multiline is used.
+    let args = &["-U", "--no-mmap", r"^baz", "test"];
+    eqnice!("baz\n", dir.command().args(args).stdout());
+    let args = &["-U", "--mmap", r"^baz", "test"];
+    eqnice!("baz\n", dir.command().args(args).stdout());
+
+    // But when (?-m) is disabled, or when \A is used, then there should be no
+    // matches that aren't anchored to the beginning of the file.
+    let args = &["-U", "--no-mmap", r"(?-m)^baz", "test"];
+    dir.command().args(args).assert_err();
+    let args = &["-U", "--mmap", r"(?-m)^baz", "test"];
+    dir.command().args(args).assert_err();
+
+    let args = &["-U", "--no-mmap", r"\Abaz", "test"];
+    dir.command().args(args).assert_err();
+    let args = &["-U", "--mmap", r"\Abaz", "test"];
+    dir.command().args(args).assert_err();
+});