From 3ec43c7de5edfc2fa489720d7c105b84313781f2 Mon Sep 17 00:00:00 2001
From: Michael Uleysky <uleysky@poi.dvo.ru>
Date: Tue, 30 Aug 2016 16:35:11 +1000
Subject: [PATCH] Gmt module: Add class for comparision strings with templates.

---
 modules/gmt/modgmt_strcomp.cpp | 139 +++++++++++++++++++++++++++++++++
 modules/gmt/modgmt_strcomp.h   |  67 ++++++++++++++++
 2 files changed, 206 insertions(+)
 create mode 100644 modules/gmt/modgmt_strcomp.cpp
 create mode 100644 modules/gmt/modgmt_strcomp.h

diff --git a/modules/gmt/modgmt_strcomp.cpp b/modules/gmt/modgmt_strcomp.cpp
new file mode 100644
index 0000000..db5803b
--- /dev/null
+++ b/modules/gmt/modgmt_strcomp.cpp
@@ -0,0 +1,139 @@
+#include "modgmt_strcomp.h"
+
+TemplateComparator::Block::pBlock TemplateComparator::Parse()
+{
+ size_t e=0;
+ struct Block* blk;
+ struct Block* root;
+ auto c=s[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->type!=Block::TEXT;
+ next=!init;
+
+ e++;
+ while(e<s.length())
+ {
+  c=s[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);
+}
+
+void TemplateComparator::InitCursors(const struct TemplateComparator::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: {}
+ }
+}
+
+bool TemplateComparator::CmpSmb(const char c)
+{
+ pCursor p=cursors.cbegin();
+ bool res=false;
+
+ // Compare symbol with all cursors
+ while(p!=cursors.end())
+ {
+  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;
+}
diff --git a/modules/gmt/modgmt_strcomp.h b/modules/gmt/modgmt_strcomp.h
new file mode 100644
index 0000000..143822c
--- /dev/null
+++ b/modules/gmt/modgmt_strcomp.h
@@ -0,0 +1,67 @@
+#include <iostream>
+#include <memory>
+#include <set>
+#include <string>
+#include <vector>
+
+class TemplateComparator
+{
+ TemplateComparator() = delete;
+ TemplateComparator(const TemplateComparator&) = delete;
+ TemplateComparator(TemplateComparator&&) = delete;
+
+ 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;
+
+  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) {}
+ };
+
+ 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 Parse();
+ void InitCursors(const struct Block* blk);
+ bool CmpSmb(const char c);
+
+ Block::pBlock root;
+ Cursors cursors;
+ const std::string s;
+
+ public:
+ TemplateComparator(const char* str):s(str) {root=Parse();}
+ TemplateComparator(const std::string& str):s(str) {root=Parse();}
+ TemplateComparator(std::string&& str):s(std::move(str)) {root=Parse();}
+
+ bool Compare(const std::string& str)
+ {
+  Reset();
+  for(size_t pos=0; pos<str.length(); ++pos) if(!CmpSmb(str[pos])) return false;
+  return true;
+ }
+
+ void Reset()
+ {
+  cursors.clear();
+  InitCursors(root.get());
+ }
+};