From d061c88ae81eb5e1354526f07f9f8d90d0732656 Mon Sep 17 00:00:00 2001
From: Karen Arutyunov <karen@codesynthesis.com>
Date: Fri, 1 Nov 2024 11:58:42 +0200
Subject: Add comments parameter to string_parser functions

---
 libbutl/string-parser.cxx      | 27 +++++++++++++++++++++++----
 libbutl/string-parser.hxx      | 10 +++++++---
 tests/string-parser/driver.cxx | 24 +++++++++++++++++++-----
 tests/string-parser/testscript | 30 ++++++++++++++++++++++++++++++
 4 files changed, 79 insertions(+), 12 deletions(-)

diff --git a/libbutl/string-parser.cxx b/libbutl/string-parser.cxx
index af5c1b3..aa7d52c 100644
--- a/libbutl/string-parser.cxx
+++ b/libbutl/string-parser.cxx
@@ -18,16 +18,34 @@ namespace butl
     }
 
     vector<pair<string, size_t>>
-    parse_quoted_position (const string& s, bool unquote)
+    parse_quoted_position (const string& s, bool unquote, bool comments)
     {
       vector<pair<string, size_t>> r;
+
+      bool newline (true);
       for (auto b (s.begin ()), i (b), e (s.end ()); i != e; )
       {
-        for (; i != e && space (*i); ++i) ; // Skip spaces.
+        // Skip spaces.
+        //
+        for (; i != e && space (*i); ++i)
+        {
+          if (*i == '\n')
+            newline = true;
+        }
+
+        // Skip comment line.
+        //
+        if (comments && newline && i != e && *i == '#')
+        {
+          for (++i; i != e && *i != '\n'; ++i) ;
+          continue;
+        }
 
         if (i == e) // No more strings.
           break;
 
+        newline = false;
+
         string s;
         char quoting ('\0'); // Current quoting mode, can be used as bool.
         size_t pos (i - b);  // String position.
@@ -74,9 +92,10 @@ namespace butl
     }
 
     vector<string>
-    parse_quoted (const string& s, bool unquote)
+    parse_quoted (const string& s, bool unquote, bool comments)
     {
-      vector<pair<string, size_t>> sp (parse_quoted_position (s, unquote));
+      vector<pair<string, size_t>> sp (
+        parse_quoted_position (s, unquote, comments));
 
       vector<string> r;
       r.reserve (sp.size ());
diff --git a/libbutl/string-parser.hxx b/libbutl/string-parser.hxx
index 9fc20c0..74280a2 100644
--- a/libbutl/string-parser.hxx
+++ b/libbutl/string-parser.hxx
@@ -26,18 +26,22 @@ namespace butl
 
     // Parse a whitespace-separated list of strings. Can contain single or
     // double quoted substrings. No escaping is supported. If unquote is true,
-    // return one-level unquoted values. Throw invalid_string in case of
+    // return one-level unquoted values. Optionally, assume that the passed
+    // string may contain comment lines (lines with `#` as the first non-
+    // whitespace character) and ignore them. Throw invalid_string in case of
     // invalid quoting.
     //
     LIBBUTL_SYMEXPORT std::vector<std::string>
-    parse_quoted (const std::string&, bool unquote);
+    parse_quoted (const std::string&, bool unquote, bool comments = false);
 
     // As above but return a list of string and zero-based position pairs.
     // Position is useful for issuing diagnostics about an invalid string
     // during second-level parsing.
     //
     LIBBUTL_SYMEXPORT std::vector<std::pair<std::string, std::size_t>>
-    parse_quoted_position (const std::string&, bool unquote);
+    parse_quoted_position (const std::string&,
+                           bool unquote,
+                           bool comments = false);
 
     // Remove a single level of quotes. Note that the format or the
     // correctness of the quotation is not validated.
diff --git a/tests/string-parser/driver.cxx b/tests/string-parser/driver.cxx
index 8cba912..93d2088 100644
--- a/tests/string-parser/driver.cxx
+++ b/tests/string-parser/driver.cxx
@@ -14,13 +14,14 @@
 using namespace std;
 using namespace butl::string_parser;
 
-// Usage: argv[0] [-l] [-u] [-p]
+// Usage: argv[0] [-l] [-u] [-p] [-c]
 //
 // 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
+// -c  comments
 //
 int
 main (int argc, char* argv[])
@@ -29,6 +30,7 @@ try
   bool spl (false);     // Print string per line.
   bool unquote (false);
   bool pos (false);
+  bool comments (false);
 
   for (int i (1); i != argc; ++i)
   {
@@ -40,6 +42,8 @@ try
       unquote = true;
     else if (o == "-p")
       pos = true;
+    else if (o == "-c")
+      comments = true;
     else
       assert (false);
   }
@@ -51,11 +55,8 @@ try
 
   cout.exceptions (ios::failbit | ios::badbit);
 
-  string l;
-  while (getline (cin, l))
+  auto print = [spl, pos] (const vector<pair<string, size_t>>& v)
   {
-    vector<pair<string, size_t>> v (parse_quoted_position (l, unquote));
-
     if (!spl)
     {
       for (auto b (v.cbegin ()), i (b), e (v.cend ()); i != e; ++i)
@@ -81,6 +82,19 @@ try
         cout << s.first << endl;
       }
     }
+  };
+
+  if (!comments)
+  {
+    string l;
+    while (getline (cin, l))
+      print (parse_quoted_position (l, unquote));
+  }
+  else
+  {
+    string s;
+    getline (cin, s, '\0');
+    print (parse_quoted_position (s, unquote, true /* comments */));
   }
 
   return 0;
diff --git a/tests/string-parser/testscript b/tests/string-parser/testscript
index d484e01..05c2807 100644
--- a/tests/string-parser/testscript
+++ b/tests/string-parser/testscript
@@ -30,6 +30,36 @@
     x  "y z
     EOO
   }
+
+  : comments
+  :
+  {
+    $* -c <<EOI >>EOO
+    # Comment 1
+    #
+    abc #xyz
+
+      # Comment 2
+      #
+      abc#
+
+    "# not a comment 3" #not-a-comment4
+
+    "abc
+    # not a comment 5
+    "
+    # Comment 6
+    EOI
+    abc
+    #xyz
+    abc#
+    "# not a comment 3"
+    #not-a-comment4
+    "abc
+    # not a comment 5
+    "
+    EOO
+  }
 }
 
 : invalid
-- 
cgit v1.1