From e47659149591da19ec844f691260ae4b0e75647a Mon Sep 17 00:00:00 2001
From: Robin Jadoul <robin.jadoul@gmail.com>
Date: Mon, 25 Apr 2016 20:39:38 +0200
Subject: [PATCH] Finish regex parsing

---
 src/re.cpp | 121 ++++++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 115 insertions(+), 6 deletions(-)

diff --git a/src/re.cpp b/src/re.cpp
index a51b875..84318b4 100644
--- a/src/re.cpp
+++ b/src/re.cpp
@@ -119,7 +119,7 @@ namespace lxs {
             stk.push(tp);
         }
 
-        std::shared_ptr<RE> parseEscapeChar(char c) {
+        char parseEscapeChar(char c) {
             switch (c)
             {
                 case '\\':
@@ -154,7 +154,113 @@ namespace lxs {
                 default:
                     throw SyntaxError(("Invalid escape sequence: \\" + std::string(1, c)).c_str());
             }
-            return std::make_shared<SingleRE>(c);
+            return c;
+        }
+
+        void sumREs(std::vector<std::shared_ptr<RE> >& res)
+        {
+            for (std::size_t step = 2; step < res.size(); step <<= 1)
+            {
+                for (std::size_t i = 0; i < res.size(); i += step)
+                {
+                    if (i + step / 2 < res.size())
+                        res[i] = std::make_shared<PlusRE>(res[i], res[i + step / 2]);
+                }
+            }
+        }
+
+        std::shared_ptr<RE> parseCharacterClass(string& input, size_t& idx) {
+            if (idx >= input.size())
+                throw SyntaxError("Unclosed character class");
+            std::vector<std::shared_ptr<RE> > chars;
+            std::set<char> used_chars;
+
+            bool invert = false;
+
+            if (input[idx] == '^')
+            {
+                invert = true;
+                idx++;
+            }
+
+            if (idx >= input.size())
+                throw SyntaxError("Unclosed character class");
+
+
+            if (input[idx] == ']')
+            {
+                used_chars.insert(']');
+                idx++;
+            }
+            
+            if (idx >= input.size())
+                throw SyntaxError("Unclosed character class");
+
+            if (input[idx] == '-')
+            {
+                used_chars.insert('-');
+                idx++;
+            }
+
+            if (idx >= input.size())
+                throw SyntaxError("Unclosed character class");
+
+            for (; idx < input.size() && input[idx] != ']'; idx++)
+            {
+                if (input[idx] == '-')
+                {
+                    idx++;
+                    
+                    if (idx >= input.size())
+                        throw SyntaxError("Unclosed character class");
+
+                    if (input[idx] == ']')
+                    {
+                        used_chars.insert('-');
+                    }
+                    else
+                    {
+                        for (int i = ((SingleRE*)(chars[chars.size() - 1].get()))->c + 1; i <= input[idx]; i++)
+                        {
+                            used_chars.insert((char) i);
+                        }
+                    }
+                }
+                else if (input[idx] == '\\')
+                {
+                    idx++;
+                    if (idx >= input.size())
+                        throw SyntaxError("Unclosed character classe");
+                    used_chars.insert(parseEscapeChar(input[idx]));
+                }
+                else
+                {
+                    used_chars.insert(input[idx]);
+                }
+            }
+            
+            if (idx >= input.size())
+                throw SyntaxError("Unclosed character class");
+            idx++; //Eat the ]
+
+            for (int i = 0; i < 256; i++)
+            {
+                if (invert ^ (used_chars.count((char) i) > 0))
+                    chars.push_back(std::make_shared<SingleRE>((char) i));
+            }
+
+            sumREs(chars);
+
+            return chars[0];
+        }
+
+        std::shared_ptr<RE> dotChar() {
+            std::vector<std::shared_ptr<RE> > any;
+            for (int i = 0; i < 256; i++)
+                if ((char) i != '\n') //Dot matches anything except newlines
+                    any.push_back(std::make_shared<SingleRE>((char) i));
+            sumREs(any);
+            return any[0];
         }
 
         std::shared_ptr<RE> parseRE(string& input, size_t& idx)
@@ -170,16 +276,19 @@ namespace lxs {
                         if (idx >= input.length())
                             throw SyntaxError("Escape sequence at the end of the string");
                         else
-                            throw SyntaxError(("invalid escape sequence: \\" + string(1, input[idx])).c_str());
-                        stk.push(parseEscapeChar(input[idx++]));
+                            stk.push(std::make_shared<SingleRE>(parseEscapeChar(input[idx])));
                         break;
 
                     case '[':
-                        //TODO: parse character classes
+                        stk.push(parseCharacterClass(input, ++idx));
                         break;
 
                     case '.':
-                        //TODO: any character
+                        for (int c = 0; c <= 256; c++)
+                        {
+                            stk.push(dotChar());
+                        }
+                        break;
 
                     case ']':
                         throw SyntaxError("Unopened ']'");