From de1eb76336836641808fe6d4374d1dc4d28d2b3c Mon Sep 17 00:00:00 2001
From: Michael Uleysky <uleysky@poi.dvo.ru>
Date: Tue, 30 Aug 2016 15:17:19 +1000
Subject: [PATCH] Code for compare strings with templates.

---
 modules/gmt/cmpstr.c++     | 130 +++++++++++++++++++
 modules/gmt/cmpstr_alt.c++ | 258 +++++++++++++++++++++++++++++++++++++
 2 files changed, 388 insertions(+)
 create mode 100644 modules/gmt/cmpstr.c++
 create mode 100644 modules/gmt/cmpstr_alt.c++

diff --git a/modules/gmt/cmpstr.c++ b/modules/gmt/cmpstr.c++
new file mode 100644
index 0000000..1d180b1
--- /dev/null
+++ b/modules/gmt/cmpstr.c++
@@ -0,0 +1,130 @@
+#include <iostream>
+#include <set>
+#include <string>
+#include <vector>
+
+bool CmpStrEx(const std::string& expr, const std::string& str)
+{
+ if(expr.empty() || str.empty()) return false;
+
+ struct State
+ {
+  struct Block
+  {
+   const size_t b,len;
+   const bool optional;
+   Block(const size_t bb,const size_t ee,const bool o):b(bb),len(ee-bb),optional(o) {}
+  };
+
+  struct Cursor
+  {
+   size_t block,offset;
+   bool operator <(const struct Cursor& c) const
+   {
+    if(block<c.block) return true;
+    if(block>c.block) return false;
+    if(offset<c.offset) return true;
+    return false;
+   }
+  };
+
+  typedef std::set<struct Cursor> Cursors;
+  typedef std::set<struct Cursor>::iterator pCursor;
+
+  std::vector<struct Block> blockchain;
+  Cursors cursors;
+  const std::string& s;
+
+  Cursors InitCursors(size_t block) const
+  {
+   Cursors cs;
+   for(size_t i=block; i<blockchain.size(); ++i)
+   {
+    cs.insert({i,0});
+    if(!blockchain[i].optional) break;
+   }
+   return cs;
+  }
+
+  State(const std::string& str):s(str)
+  {
+   size_t cur=0;
+   size_t bpos=0;
+
+   // Parse blocks
+   while(cur<s.length())
+   {
+    if('['==s[cur] || ']'==s[cur])
+    {
+     // Add current block to blockchain
+     if(cur>bpos) blockchain.push_back(Block(bpos,cur,(']'==s[cur])));
+     cur++;
+     bpos=cur;
+     continue;
+    }
+    cur++;
+   }
+   // Add last block
+   if(bpos<s.length()) blockchain.push_back(Block(bpos,s.length(),false));
+
+   // Creating cursors for the first symbol
+   cursors=InitCursors(0);
+  }
+
+  bool CmpSmb(const char c)
+  {
+   pCursor p=cursors.begin();
+   bool res=false;
+
+   // Compare symbol with all cursors
+   while(p!=cursors.end())
+   {
+    const Block& bl=blockchain[p->block];
+    if(c==s[bl.b+p->offset]) {res=true; ++p;}
+    else p=cursors.erase(p);
+   }
+
+   // Increment cursors
+   Cursors upd;  // New cursors
+   p=cursors.begin();
+   while(p!=cursors.end())
+   {
+    // Increment cursor on one position
+    if(p->offset+1>=blockchain[p->block].len) // Go to next block
+    {
+     Cursors cs=InitCursors(p->block+1); // Get cursors for next block
+     p=cursors.erase(p); // Erase current cursor
+     for(const auto& cur: cs) upd.insert(cur); // Copy cursors to new set
+     continue;
+    }
+    else upd.insert({p->block,p->offset+1});
+    if(blockchain[p->block].optional) // If current block is optional next symbol may be from next block
+    {
+     Cursors cs=InitCursors(p->block+1); // Get cursors for next blockchain
+     for(const auto& cur: cs) upd.insert(cur); // Copy cursors to new set
+    }
+    ++p;
+   }
+
+   cursors=upd;
+   return res;
+  }
+
+ };
+
+ struct State st(expr);
+ for(size_t pos=0; pos<str.length(); ++pos)
+ {
+  if(!st.CmpSmb(str[pos])) return false;
+ }
+
+ return true;
+}
+
+
+int main(int argc, char** argv)
+{
+ if(argc!=3) return 1;
+ std::cout<<"Compare "<<argv[1]<<" with template "<<argv[2]<<": "<<(CmpStrEx(argv[2],argv[1])?"match":"not match")<<std::endl;
+ return 0;
+}
diff --git a/modules/gmt/cmpstr_alt.c++ b/modules/gmt/cmpstr_alt.c++
new file mode 100644
index 0000000..bc9cdda
--- /dev/null
+++ b/modules/gmt/cmpstr_alt.c++
@@ -0,0 +1,258 @@
+#include <iostream>
+#include <memory>
+#include <set>
+#include <string>
+#include <vector>
+
+bool CmpStrEx(const std::string& expr, const std::string& str)
+{
+ if(expr.empty() || str.empty()) return false;
+
+ struct State
+ {
+  struct Block
+  {
+   using pBlock=std::unique_ptr<struct Block>;
+   enum Type {NOTDEF,TEXT,OPTIONAL,VARIANTS,DELIM};
+   size_t b,e;
+   Type type;
+   const struct Block* parent;
+   pBlock next,child;
+
+   Block() = delete;
+   Block(const struct Block&) = delete;
+   Block(struct Block&&) = delete;
+
+   bool isText() const {return type==TEXT;}
+
+   Block(Type t, const struct Block* p=nullptr):b(0),e(0),type(t),parent(p),next(nullptr),child(nullptr) {}
+   Block(size_t bb, size_t ee, const struct Block* p=nullptr):b(bb),e(ee),type(TEXT),parent(p),next(nullptr),child(nullptr) {}
+  };
+
+  Block::pBlock Parse(const std::string& str)
+  {
+   size_t e=0;
+   struct Block* blk;
+   struct Block* root;
+   auto c=str[e];
+   bool init,next;
+
+   // First symbol
+   if('['==c || '('==c) blk=new Block(('['==c)?Block::OPTIONAL:Block::VARIANTS);
+   else blk=new Block(e,e+1);
+   root=blk;
+   init=!root->isText();
+   next=!init;
+
+   e++;
+   while(e<str.length())
+   {
+    c=str[e];
+
+    switch(c)
+    {
+     case '[':
+     case '(':
+     {
+      (next?blk->next:blk->child).reset(new Block(('['==c)?Block::OPTIONAL:Block::VARIANTS,(next?blk->parent:blk)));
+      blk=(next?blk->next:blk->child).get();
+      init=true;
+      next=false;
+      break;
+     }
+     case ']':
+     case ')':
+     {
+      blk=const_cast<struct Block*>(blk->parent);
+      init=true;
+      next=true;
+      break;
+     }
+     case '|': {blk->next.reset(new Block(Block::DELIM,blk->parent)); blk=blk->next.get(); init=next=true; break;}
+     default:
+     {
+      if(init)
+      {
+       (next?blk->next:blk->child).reset(new Block(e,e+1,(next?blk->parent:blk)));
+       blk=(next?blk->next:blk->child).get();
+      }
+      else blk->e=e+1;
+      init=false;
+      next=true;
+     }
+    }
+    e++;
+   }
+   return Block::pBlock(root);
+  }
+
+  struct Cursor
+  {
+   const struct Block* block;
+   size_t offset;
+   bool operator <(const struct Cursor& c) const {return block<c.block;}
+   bool compare(const std::string& str, const char c) const {return c==str[block->b+offset];}
+   Cursor(const struct Block* b, size_t o):block(b),offset(o) {}
+  };
+
+  using Cursors=std::set<struct Cursor>;
+  using pCursor=std::set<struct Cursor>::const_iterator;
+
+  Block::pBlock root;
+  Cursors cursors;
+  const std::string& s;
+
+  void InitCursors(const struct Block* blk)
+  {
+   switch(blk->type)
+   {
+    case Block::TEXT: {cursors.insert({blk,0}); break;}
+    case Block::OPTIONAL:
+    {
+     auto b=blk;
+     InitCursors(b->child.get());
+     while(true)
+     {
+      if(!b->next) break;
+      b=b->next.get();
+      InitCursors(b);
+      if(b->type!=Block::OPTIONAL) break;
+     }
+     break;
+    }
+    case Block::VARIANTS:
+    {
+     auto b=blk->child.get();
+     while(true)
+     {
+      InitCursors(b);
+      while(b->type!=Block::DELIM)
+      {
+       if(!b->next) break;
+       b=b->next.get();
+      }
+      if(!b->next) break;
+      b=b->next.get();
+     }
+    }
+    default: {}
+   }
+  }
+
+
+  State(const std::string& str):root(Parse(str)),s(str) {InitCursors(root.get());}
+
+
+  bool CmpSmb(const char c)
+  {
+   pCursor p=cursors.cbegin();
+   bool res=false;
+
+   // Compare symbol with all cursors
+
+   std::cout<<"--------------\n";
+   std::cout<<"Symbol: "<<c<<"\n";
+
+   while(p!=cursors.end())
+   {
+    std::cout<<p->block->b+p->offset<<" "<<s[p->block->b+p->offset]<<"\n";
+    if(p->compare(s,c)) {res=true; p++;}
+    else p=cursors.erase(p);
+   }
+
+   Cursors old(std::move(cursors));
+   for(const auto& p:old)
+   {
+    auto blk=p.block;
+    // Increment cursors
+    if(blk->b+p.offset+1<blk->e) cursors.insert({blk,p.offset+1}); // Advance in current block
+    else // Move to next block
+    {
+     if(!blk->next || blk->next->type==Block::DELIM) // End of chain, must go up
+     {
+      while(blk->parent!=nullptr)
+      {
+       blk=blk->parent;
+       if(!blk->next || blk->next->type==Block::DELIM) continue;
+       InitCursors(blk->next.get());
+      }
+     }
+     else InitCursors(blk->next.get()); // Next block in chain
+    }
+
+    // Try to leave optional blocks
+    blk=p.block;
+    while(blk->parent!=nullptr)
+    {
+     blk=blk->parent;
+     if(blk->type!=Block::OPTIONAL || !blk->next || blk->next->type==Block::DELIM) continue;
+     InitCursors(blk->next.get());
+    }
+   }
+
+   return res;
+  }
+ };
+
+
+// Main function
+ struct State st(expr);
+
+ // Dump code
+ {
+  std::cerr<<"digraph main\n{\nrankdir=TB pack=\"true\" packmode=\"node\";\n";
+  std::cerr<<"node [fontsize=30, shape=rectangle, style=\"filled,rounded\", fillcolor=azure2, fixedsize=\"false\", margin=0.2, width=1, penwidth=3, fontname=\"Times New Roman\"];\n";
+  std::cerr<<"edge [arrowsize=1.5, penwidth=3];\n";
+
+  const struct State::Block* r=st.root.get();
+  std::set<const struct State::Block*> visited;
+  visited.insert(nullptr);
+  while(true)
+  {
+   if(visited.count(r)==0)
+   {
+    std::string label;
+    if(r->type==State::Block::OPTIONAL) label="[]";
+    if(r->type==State::Block::VARIANTS) label="()";
+    if(r->type==State::Block::DELIM) label="|";
+    if(r->type==State::Block::TEXT) label=expr.substr(r->b,r->e-r->b);
+    std::cerr<<"\""<<r<<"\" [label=\""<<label<<"\""<<((r->type==State::Block::TEXT)?",fontcolor=\"red\"":"")<<"];\n";
+    if(r->child) std::cerr<<"\""<<r<<"\" -> \""<<r->child.get()<<"\";\n";
+    if(r->next)
+    {
+     std::cerr<<"\""<<r<<"\" -> \""<<r->next.get()<<"\" [color=\"blue\"];\n";
+     std::cerr<<"{rank=same; "<<"\""<<r<<"\"; \""<<r->next.get()<<"\";}\n";
+    }
+    if(r->parent!=nullptr) std::cerr<<"\""<<r<<"\" -> \""<<r->parent<<"\" [color=\"green\", penwidth=1];\n";
+    visited.insert(r);
+   }
+   if(r->child && visited.count(r->child.get())==0) r=r->child.get();
+   else if(r->next) r=r->next.get();
+   else
+   {
+    while(r->parent!=nullptr)
+    {
+     r=r->parent;
+     if(r->next) {r=r->next.get(); break;}
+    }
+   }
+   if(visited.count(r)!=0 && r->parent==nullptr && visited.count(r->next.get())!=0 && visited.count(r->child.get())!=0) break;
+  }
+  std::cerr<<"}\n";
+ }
+
+ for(size_t pos=0; pos<str.length(); ++pos)
+ {
+  if(!st.CmpSmb(str[pos])) return false;
+ }
+
+ return true;
+}
+
+
+int main(int argc, char** argv)
+{
+ if(argc!=3) return 1;
+ std::cout<<"Compare "<<argv[1]<<" with template "<<argv[2]<<": "<<(CmpStrEx(argv[2],argv[1])?"match":"not match")<<std::endl;
+ return 0;
+}