Add support for portable path modifer and dot character escaping inversion

4 files changed, 198 insertions, 63 deletions

diff --git a/unit-tests/test/script/parser/driver.cxx b/unit-tests/test/script/parser/driver.cxx
index 271fd2c..f641dc4 100644
--- a/unit-tests/test/script/parser/driver.cxx
+++ b/unit-tests/test/script/parser/driver.cxx
@@ -189,6 +189,13 @@ namespace build2
                                   string (),
                                   trace));
 
+          value& v (
+            tt.assign (
+              var_pool.insert<target_triplet> (
+                "test.target", variable_visibility::project)));
+
+          v = cast<target_triplet> ((*global_scope)["build.host"]);
+
           testscript& st (
             targets.insert<testscript> (work,
                                         dir_path (),
diff --git a/unit-tests/test/script/parser/redirect.test b/unit-tests/test/script/parser/redirect.test
index 5a91354..3db684a 100644
--- a/unit-tests/test/script/parser/redirect.test
+++ b/unit-tests/test/script/parser/redirect.test
@@ -6,6 +6,82 @@
 # @@ Does it make sense to split into separate files - one per redirect type?
 #
 
+: str
+:
+{
+  : literal
+  :
+  {
+    : portable-path
+    :
+    $* <<EOI >>EOO
+    cmd </foo >/bar 2>/baz
+    EOI
+    cmd </foo >/bar 2>/baz
+    EOO
+  }
+
+  : regex
+  :
+  {
+    : portable-path
+    :
+    $* <<EOI >>EOO
+    cmd >/~%foo% 2>/~%bar%
+    EOI
+    cmd >/~%foo% 2>/~%bar%
+    EOO
+  }
+}
+
+: doc
+:
+{
+  : literal
+  :
+  {
+    : portable-path
+    :
+    $* <<EOI >>EOO
+    cmd <</EOI_ >/EOO_ 2>/EOE_
+    foo
+    EOI_
+    bar
+    EOO_
+    baz
+    EOE_
+    EOI
+    cmd <</EOI_ >/EOO_ 2>/EOE_
+    foo
+    EOI_
+    bar
+    EOO_
+    baz
+    EOE_
+    EOO
+  }
+
+  : regex
+  :
+  {
+    : portable-path
+    :
+    $* <<EOI >>EOO
+    cmd >/~%EOF% 2>/~%EOE%
+    foo
+    EOF
+    bar
+    EOE
+    EOI
+    cmd >/~%EOF% 2>/~%EOE%
+    foo
+    EOF
+    bar
+    EOE
+    EOO
+  }
+}
+
 : file
 :
 {
diff --git a/unit-tests/test/script/parser/regex.test b/unit-tests/test/script/parser/regex.test
index f9101c9..058a5bc 100644
--- a/unit-tests/test/script/parser/regex.test
+++ b/unit-tests/test/script/parser/regex.test
@@ -26,12 +26,30 @@
     testscript:1:7: error: no closing introducer character in stdout regex redirect
     EOE
 
+    : portable-path-introducer
+    :
+    $* <'cmd >/~/foo/' 2>>EOE != 0
+    testscript:1:8: error: portable path modifier and '/' introducer in stdout regex redirect
+    EOE
+
     : empty
     :
     $* <'cmd >~//' 2>>EOE != 0
     testscript:1:7: error: stdout regex redirect is empty
     EOE
 
+    : no-flags
+    :
+    $* <'cmd >~/fo*/' >'cmd >~/fo*/'
+
+    : idot
+    :
+    $* <'cmd >~/fo*/d' >'cmd >~/fo*/d'
+
+    : icase
+    :
+    $* <'cmd >~/fo*/i' >'cmd >~/fo*/i'
+
     : invalid-flags1
     :
     $* <'cmd >~/foo/z' 2>>EOE != 0
@@ -44,24 +62,6 @@
     testscript:1:7: error: junk at the end of stdout regex redirect
     EOE
 
-    : malformed
-    :
-    : Note that old versions of libc++ (for example 1.1) do not detect some
-    : regex errors. For example '*' is parsed successfully.
-    :
-    $* <'cmd >~/foo[/' 2>>~/EOE/ != 0
-    /testscript:1:7: error: invalid stdout regex redirect.*/
-      info: regex: /foo[/
-    EOE
-
-    : without-flags
-    :
-    $* <'cmd >~/fo*/' >'cmd >~/fo*/'
-
-    : with-flags
-    :
-    $* <'cmd >~/fo*/i' >'cmd >~/fo*/i'
-
     : no-newline
     :
     $* <'cmd >:~/fo*/' >'cmd >:~/fo*/'
@@ -106,46 +106,24 @@
     testscript:1:8: error: expected here-document regex end marker
     EOE
 
-    : unterminated-line-char
+    : portable-path-introducer
     :
     $* <<EOI 2>>EOE != 0
-    cmd >>~/EOO/
-    /
+    cmd >>/~/EOO/
+    foo
     EOO
     EOI
-    testscript:2:1: error: regex introducer without regex
-      info: consider changing regex introducer '/' in here-document end marker
+    testscript:1:5: error: portable path modifier and '/' introducer in here-document regex end marker
     EOE
 
-    : invalid-syntax-char
+    : unterminated-line-char
     :
     $* <<EOI 2>>EOE != 0
     cmd >>~/EOO/
-    /x
-    EOO
-    EOI
-    testscript:2:1: error: invalid line-regex syntax character 'x'
-    EOE
-
-    : invalid-char-regex
-    :
-    $* <<EOI 2>>~/EOE/ != 0
-    cmd >>~/EOO/
-    /foo[/
-    EOO
-    EOI
-    /testscript:2:1: error: invalid regex.*/
-    EOE
-
-    : invalid-line-regex
-    :
-    $* <<EOI 2>>~/EOE/ != 0
-    cmd >>~/EOO/
-    a
-    /{
+    /
     EOO
     EOI
-    /testscript:4:1: error: invalid here-document regex.*/
+    testscript:2:1: error: no syntax line characters
     EOE
 
     : empty
@@ -157,7 +135,7 @@
     testscript:2:1: error: empty here-document regex
     EOE
 
-    : valid
+    : no-flags
     :
     $* <<EOI >>EOO
     cmd 2>>~/EOE/
@@ -184,6 +162,47 @@
     //*
     EOE
     EOO
+
+    : no-newline
+    :
+    $* <'cmd >:~/fo*/' >'cmd >:~/fo*/'
+    $* <<EOI >>EOO
+    cmd 2>>:~/EOE/
+    foo
+    EOE
+    EOI
+    cmd 2>>:~/EOE/
+    foo
+    EOE
+    EOO
+
+    : end-marker-restore
+    :
+    {
+      : idot
+      :
+      $* <<EOI >>EOO
+      cmd 2>>~/EOE/d
+      foo
+      EOE
+      EOI
+      cmd 2>>~/EOE/d
+      foo
+      EOE
+      EOO
+
+      : icase
+      :
+      $* <<EOI >>EOO
+      cmd 2>>~/EOE/i
+      foo
+      EOE
+      EOI
+      cmd 2>>~/EOE/i
+      foo
+      EOE
+      EOO
+    }
   }
 
   : stderr
diff --git a/unit-tests/test/script/regex/driver.cxx b/unit-tests/test/script/regex/driver.cxx
index 6ff4d75..2680672 100644
--- a/unit-tests/test/script/regex/driver.cxx
+++ b/unit-tests/test/script/regex/driver.cxx
@@ -16,6 +16,8 @@ main ()
   using lc = line_char;
   using ls = line_string;
   using lr = line_regex;
+  using cf = char_flags;
+  using cr = char_regex;
 
   // Test line_char.
   //
@@ -54,7 +56,7 @@ main ()
     assert (lc ('0') != '1');
     assert (lc ('n') != mp);
     assert (lc ('0') != lc ("0", p));
-    assert (lc ('0') != lc (regex ("0"), p));
+    assert (lc ('0') != lc (cr ("0"), p));
 
     assert (lc ('0') < lc ('1'));
     assert (lc ('0') < '1');
@@ -77,24 +79,54 @@ main ()
     assert (char (lc ("a", p))     == '\a');
 
     assert (lc ("a", p)   != lc ("b", p));
-    assert (!(lc ("a", p) != lc (regex ("a"), p))); // Matches.
-    assert (lc ("a", p)   != lc (regex ("b"), p));
+    assert (!(lc ("a", p) != lc (cr ("a"), p)));
+    assert (lc ("a", p)   != lc (cr ("b"), p));
 
     assert (lc ("a", p)   < lc ("b", p));
-    assert (!(lc ("a", p) < lc (regex ("a"), p))); // Matches.
+    assert (!(lc ("a", p) < lc (cr ("a"), p)));
 
     assert (lc ("a", p) <= lc ("b", p));
-    assert (lc ("a", p) <= lc (regex ("a"), p));
-    assert (lc ("a", p) <  lc (regex ("c"), p));
+    assert (lc ("a", p) <= lc (cr ("a"), p));
+    assert (lc ("a", p) <  lc (cr ("c"), p));
 
     // Regex roundtrip.
     //
-    assert (regex_match ("abc", *lc (regex ("abc"), p).regex ()));
+    assert (regex_match ("abc", *lc (cr ("abc"), p).regex ()));
+
+    // Regex flags.
+    //
+    // icase
+    //
+    assert (regex_match ("ABC", cr ("abc", cf::icase)));
+
+    // idot
+    //
+    assert (!regex_match ("a", cr ("[.]",  cf::idot)));
+    assert (!regex_match ("a", cr ("[\\.]", cf::idot)));
+
+    assert (regex_match  ("a", cr (".")));
+    assert (!regex_match ("a", cr (".", cf::idot)));
+    assert (regex_match  ("a", cr ("\\.", cf::idot)));
+    assert (!regex_match ("a", cr ("\\.")));
+
+    // regex::transform()
+    //
+    // The function is static and we can't test it directly. So we will test
+    // it indirectly via regex matches.
+    //
+    // @@ Would be nice to somehow address the inability to test internals (not
+    //    exposed via headers). As a part of utility library support?
+    //
+    assert (regex_match  (".a[.", cr (".\\.\\[[.]",   cf::idot)));
+    assert (regex_match  (".a[.", cr (".\\.\\[[\\.]", cf::idot)));
+    assert (!regex_match ("ba[.", cr (".\\.\\[[.]",   cf::idot)));
+    assert (!regex_match (".a[b", cr (".\\.\\[[.]",   cf::idot)));
+    assert (!regex_match (".a[b", cr (".\\.\\[[\\.]", cf::idot)));
 
     // Regex comparison.
     //
-    assert (lc ("a", p)           == lc (regex ("a|b"), p));
-    assert (lc (regex ("a|b"), p) == lc ("a", p));
+    assert (lc ("a", p)        == lc (cr ("a|b"), p));
+    assert (lc (cr ("a|b"), p) == lc ("a", p));
   }
 
   // Test char_traits<line_char>.
@@ -159,7 +191,7 @@ main ()
     const ct& t (use_facet<ct> (l));
     line_pool p;
 
-    assert (t.is (ct::digit, '0'));
+    assert (t.is  (ct::digit, '0'));
     assert (!t.is (ct::digit, '?'));
     assert (!t.is (ct::digit, lc ("0", p)));
 
@@ -216,7 +248,7 @@ main ()
   //
   {
     line_pool p;
-    lr r1 ({lc ("foo", p), lc (regex ("ba(r|z)"), p)}, move (p));
+    lr r1 ({lc ("foo", p), lc (cr ("ba(r|z)"), p)}, move (p));
 
     lr r2 (move (r1));
     assert (regex_match (ls ({lc ("foo", r2.pool), lc ("bar", r2.pool)}), r2));
@@ -239,11 +271,12 @@ main ()
     assert (regex_match (ls ({bar, foo}),
                          lr ({'(', foo, '|', bar, ')', '+'})));
 
-    assert (regex_match (ls ({foo, foo}), lr ({'(', foo, ')', '\\', '1'})));
+    assert (regex_match (ls ({foo, foo, bar}),
+                         lr ({'(', foo, ')', '\\', '1', bar})));
 
-    assert (regex_match (ls ({foo}),   lr ({lc (regex ("fo+"), p)})));
-    assert (regex_match (ls ({foo}),   lr ({lc (regex (".*"), p)})));
-    assert (regex_match (ls ({blank}), lr ({lc (regex (".*"), p)})));
+    assert (regex_match (ls ({foo}),   lr ({lc (cr ("fo+"), p)})));
+    assert (regex_match (ls ({foo}),   lr ({lc (cr (".*"), p)})));
+    assert (regex_match (ls ({blank}), lr ({lc (cr (".*"), p)})));
 
     assert (regex_match (ls ({blank, blank, foo}),
                          lr ({blank, '*', foo, blank, '*'})));