Implement string_parser

4 files changed, 148 insertions, 1 deletions

diff --git a/tests/string-parser/buildfile b/tests/string-parser/buildfile
new file mode 100644
index 0000000..9ccf480
--- /dev/null
+++ b/tests/string-parser/buildfile
@@ -0,0 +1,7 @@
+# file      : tests/string-parser/buildfile
+# copyright : Copyright (c) 2014-2017 Code Synthesis Ltd
+# license   : MIT; see accompanying LICENSE file
+
+exe{driver}: cxx{driver} ../../butl/lib{butl} test{testscript}
+
+include ../../butl/
diff --git a/tests/string-parser/driver.cxx b/tests/string-parser/driver.cxx
new file mode 100644
index 0000000..2aad3a7
--- /dev/null
+++ b/tests/string-parser/driver.cxx
@@ -0,0 +1,93 @@
+// file      : tests/string-parser/driver.cxx -*- C++ -*-
+// copyright : Copyright (c) 2014-2017 Code Synthesis Ltd
+// license   : MIT; see accompanying LICENSE file
+
+#include <ios>      // ios::failbit, ios::badbit
+#include <vector>
+#include <string>
+#include <cassert>
+#include <iostream>
+
+#include <butl/utility>       // operator<<(ostream,exception)
+#include <butl/string-parser>
+
+using namespace std;
+using namespace butl;
+
+// Usage: argv[0] [-l] [-u] [-p]
+//
+// Read and parse lines into strings from STDIN and print them to STDOUT.
+//
+// -l  output each string on a separate line
+// -u  unquote strings
+// -p  output positions
+//
+int
+main (int argc, char* argv[])
+try
+{
+  bool spl (false);     // Print string per line.
+  bool unquote (false);
+  bool pos (false);
+
+  for (int i (1); i != argc; ++i)
+  {
+    string o (argv[i]);
+
+    if (o == "-l")
+      spl = true;
+    else if (o == "-u")
+      unquote = true;
+    else if (o == "-p")
+      pos = true;
+    else
+      assert (false);
+  }
+
+  // Do not throw when eofbit is set (end of stream reached), and when failbit
+  // is set (getline() failed to extract any character).
+  //
+  cin.exceptions  (ios::badbit);
+
+  cout.exceptions (ios::failbit | ios::badbit);
+
+  string l;
+  while (getline (cin, l))
+  {
+    vector<pair<string, size_t>> v (
+      string_parser::parse_quoted_position (l, unquote));
+
+    if (!spl)
+    {
+      for (auto b (v.cbegin ()), i (b), e (v.cend ()); i != e; ++i)
+      {
+        if (i != b)
+          cout << ' ';
+
+        if (pos)
+          cout << i->second << ":";
+
+        cout << i->first;
+      }
+
+      cout << endl;
+    }
+    else
+    {
+      for (const auto& s: v)
+      {
+        if (pos)
+          cout << s.second << ":";
+
+        cout << s.first << endl;
+      }
+    }
+  }
+
+  return 0;
+}
+catch (const invalid_string& e)
+{
+  cerr << e.position << ": " << e << endl;
+  return 1;
+}
diff --git a/tests/string-parser/testscript b/tests/string-parser/testscript
new file mode 100644
index 0000000..83c8346
--- /dev/null
+++ b/tests/string-parser/testscript
@@ -0,0 +1,42 @@
+# file      : tests/string-parser/testscript
+# copyright : Copyright (c) 2014-2017 Code Synthesis Ltd
+# license   : MIT; see accompanying LICENSE file
+
+: valid
+:
+{
+  test.options += -l
+  sp=' ' # For line-terminating spaces.
+
+  : quoted
+  :
+  {
+    $* -p <<"EOI" >>EOO
+    abc  "d 'ef " 'x  "y z'$sp
+    EOI
+    0:abc
+    5:"d 'ef "
+    14:'x  "y z'
+    EOO
+  }
+
+  : unquoted
+  :
+  {
+    $* -u <<"EOI" >>"EOO"
+    abc  "d 'ef " 'x  "y z'$sp
+    EOI
+    abc
+    d 'ef$sp
+    x  "y z
+    EOO
+  }
+}
+
+: invalid
+:
+{
+  : unterm-quoting
+  :
+  $* <'ab"c' 2>'4: unterminated quoted string' == 1
+}
diff --git a/tests/tab-parser/testscript b/tests/tab-parser/testscript
index 1b0a816..de3b167 100644
--- a/tests/tab-parser/testscript
+++ b/tests/tab-parser/testscript
@@ -45,5 +45,10 @@
 {
   : unterm-quoting
   :
-  $* <'ab"c' 2>'cin:1:5: error: unterminated quoted string' == 1
+  $* <<EOI >'123' 2>'cin:3:5: error: unterminated quoted string' == 1
+
+  123
+  ab"c
+  xyz
+  EOI
 }