apparmor/parser/libapparmor_re/expr-tree.h

/*
 * (C) 2006, 2007 Andreas Gruenbacher <agruen@suse.de>
 * Copyright (c) 2003-2008 Novell, Inc. (All rights reserved)
 * Copyright 2009-2010 Canonical Ltd.
 *
 * The libapparmor library is licensed under the terms of the GNU
 * Lesser General Public License, version 2.1. Please see the file
 * COPYING.LGPL.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public License
 * along with this program.  If not, see <http://www.gnu.org/licenses/>.
 *
 *
 * Functions to create/manipulate an expression tree for regular expressions
 * that have been parsed.
 *
 * The expression tree can be used directly after the parse creates it, or
 * it can be factored so that the set of important nodes is smaller.
 * Having a reduced set of important nodes generally results in a dfa that
 * is closer to minimum (fewer redundant states are created).  It also
 * results in fewer important nodes in a the state set during subset
 * construction resulting in less memory used to create a dfa.
 *
 * Generally it is worth doing expression tree simplification before dfa
 * construction, if the regular expression tree contains any alternations.
 * Even if the regular expression doesn't simplification should be fast
 * enough that it can be used with minimal overhead.
 */
#ifndef __LIBAA_RE_EXPR_H
#define __LIBAA_RE_EXPR_H

#include <map>
#include <set>
#include <stack>
#include <ostream>

#include "apparmor_re.h"

using namespace std;

typedef unsigned char uchar;
typedef set<uchar> Chars;

ostream& operator<<(ostream& os, uchar c);

/* Compute the union of two sets. */
template<class T>
set<T> operator+(const set<T>& a, const set<T>& b)
{
	set<T> c(a);
	c.insert(b.begin(), b.end());
	return c;
}

/**
 * When creating DFAs from regex trees, a DFA state is constructed from
 * a set of important nodes in the syntax tree. This includes AcceptNodes,
 * which indicate that when a match ends in a particular state, the
 * regular expressions that the AcceptNode belongs to match.
 */
class Node;
class ImportantNode;
typedef set <ImportantNode *> NodeSet;

/**
 * Text-dump a state (for debugging).
 */
ostream& operator<<(ostream& os, const NodeSet& state);

/**
 * Out-edges from a state to another: we store the follow-set of Nodes
 * for each input character that is not a default match in
 * cases (i.e., following a CharNode or CharSetNode), and default
 * matches in otherwise as well as in all matching explicit cases
 * (i.e., following an AnyCharNode or NotCharSetNode). This avoids
 * enumerating all the explicit tranitions for default matches.
 */
typedef struct NodeCases {
	typedef map<uchar, NodeSet *>::iterator iterator;
	iterator begin() { return cases.begin(); }
	iterator end() { return cases.end(); }

	NodeCases() : otherwise(0) { }
	map<uchar, NodeSet *> cases;
	NodeSet *otherwise;
} NodeCases;


ostream& operator<<(ostream& os, Node& node);

/* An abstract node in the syntax tree. */
class Node {
public:
	Node() :
	    nullable(false) { child[0] = child[1] = 0; }
	Node(Node *left) :
	    nullable(false) { child[0] = left; child[1] = 0; }
	Node(Node *left, Node *right) :
	    nullable(false) { child[0] = left; child[1] = right; }
	virtual ~Node()
	{
		if (child[0])
			child[0]->release();
		if (child[1])
			child[1]->release();
	}

	/**
	 * See the "Dragon Book" for an explanation of nullable, firstpos,
	 * lastpos, and followpos.
	 */
	virtual void compute_nullable() { }
	virtual void compute_firstpos() = 0;
	virtual void compute_lastpos() = 0;
	virtual void compute_followpos() { }
	virtual int eq(Node *other) = 0;
	virtual ostream& dump(ostream& os) = 0;
	void dump_syntax_tree(ostream& os);

	bool nullable;
	NodeSet firstpos, lastpos, followpos;
	/* child 0 is left, child 1 is right */
	Node *child[2];

	unsigned int label;	/* unique number for debug etc */
	/**
	 * We indirectly release Nodes through a virtual function because
	 * accept and Eps Nodes are shared, and must be treated specially.
	 * We could use full reference counting here but the indirect release
	 * is sufficient and has less overhead
	 */
	virtual void release(void) { delete this; }
};


class InnerNode : public Node {
public:
	InnerNode() : Node() { };
	InnerNode(Node *left) : Node(left) {};
	InnerNode(Node *left, Node *right) : Node(left, right) { };
};

class OneChildNode : public InnerNode {
public:
	OneChildNode(Node *left) : InnerNode(left) { };
};

class TwoChildNode : public InnerNode {
public:
	TwoChildNode(Node *left, Node *right) :  InnerNode(left, right) { };
};

class LeafNode : public Node {
public:
	LeafNode() : Node() { };
};

/* Match nothing (//). */
class EpsNode : public LeafNode {
public:
	EpsNode() : LeafNode()
	{
		nullable = true;
		label = 0;
	}
	void release(void)
	{
		/* don't delete Eps nodes because there is a single static
		 * instance shared by all trees.  Look for epsnode in the code
		 */
	}

	void compute_firstpos() { }
	void compute_lastpos() { }
	int eq(Node *other)
	{
		if (dynamic_cast<EpsNode *>(other))
			return 1;
		return 0;
	}
	ostream& dump(ostream& os)
	{
		return os << "[]";
	}
};

/**
 * Leaf nodes in the syntax tree are important to us: they describe the
 * characters that the regular expression matches. We also consider
 * AcceptNodes import: they indicate when a regular expression matches.
 */
class ImportantNode : public LeafNode {
public:
	ImportantNode() : LeafNode() { }
	void compute_firstpos()
	{
		firstpos.insert(this);
	}
	void compute_lastpos() {
		lastpos.insert(this);
	}
	virtual void follow(NodeCases& cases) = 0;
};

/* common base class for all the different classes that contain
 * character information.
 */
class CNode : public ImportantNode {
public:
	CNode() : ImportantNode() { }
};

/* Match one specific character (/c/). */
class CharNode : public CNode {
public:
	CharNode(uchar c) : c(c) { }
	void follow(NodeCases& cases)
	{
		NodeSet **x = &cases.cases[c];
		if (!*x) {
			if (cases.otherwise)
				*x = new NodeSet(*cases.otherwise);
			else
				*x = new NodeSet;
		}
		(*x)->insert(followpos.begin(), followpos.end());
	}
	int eq(Node *other)
	{
		CharNode *o = dynamic_cast<CharNode *>(other);
		if (o) {
			return c == o->c;
		}
		return 0;
	}
	ostream& dump(ostream& os)
	{
		return os << c;
	}

	uchar c;
};

/* Match a set of characters (/[abc]/). */
class CharSetNode : public CNode {
public:
	CharSetNode(Chars& chars) : chars(chars) { }
	void follow(NodeCases& cases)
	{
		for (Chars::iterator i = chars.begin(); i != chars.end(); i++) {
			NodeSet **x = &cases.cases[*i];
			if (!*x) {
				if (cases.otherwise)
					*x = new NodeSet(*cases.otherwise);
				else
					*x = new NodeSet;
			}
			(*x)->insert(followpos.begin(), followpos.end());
		}
	}
	int eq(Node *other)
	{
		CharSetNode *o = dynamic_cast<CharSetNode *>(other);
		if (!o || chars.size() != o->chars.size())
			return 0;

		for (Chars::iterator i = chars.begin(), j = o->chars.begin();
		     i != chars.end() && j != o->chars.end();
		     i++, j++) {
			if (*i != *j)
				return 0;
		}
		return 1;
	}
	ostream& dump(ostream& os)
	{
		os << '[';
		for (Chars::iterator i = chars.begin(); i != chars.end(); i++)
			os << *i;
		return os << ']';
	}

	Chars chars;
};

/* Match all except one character (/[^abc]/). */
class NotCharSetNode : public CNode {
public:
	NotCharSetNode(Chars& chars) : chars(chars) { }
	void follow(NodeCases& cases)
	{
		if (!cases.otherwise)
			cases.otherwise = new NodeSet;
		for (Chars::iterator j = chars.begin(); j != chars.end(); j++) {
			NodeSet **x = &cases.cases[*j];
			if (!*x)
				*x = new NodeSet(*cases.otherwise);
		}
		/* Note: Add to the nonmatching characters after copying away
		 * the old otherwise state for the matching characters.
		 */
		cases.otherwise->insert(followpos.begin(), followpos.end());
		for (NodeCases::iterator i = cases.begin(); i != cases.end();
		     i++) {
			if (chars.find(i->first) == chars.end())
				i->second->insert(followpos.begin(),
						  followpos.end());
		}
	}
	int eq(Node *other)
	{
		NotCharSetNode *o = dynamic_cast<NotCharSetNode *>(other);
		if (!o || chars.size() != o->chars.size())
			return 0;

		for (Chars::iterator i = chars.begin(), j = o->chars.begin();
		     i != chars.end() && j != o->chars.end();
		     i++, j++) {
			if (*i != *j)
				return 0;
		}
		return 1;
	}
	ostream& dump(ostream& os)
	{
		os << "[^";
		for (Chars::iterator i = chars.begin(); i != chars.end(); i++)
			os << *i;
		return os << ']';
	}

	Chars chars;
};

/* Match any character (/./). */
class AnyCharNode : public CNode {
public:
	AnyCharNode() { }
	void follow(NodeCases& cases)
	{
		if (!cases.otherwise)
			cases.otherwise = new NodeSet;
		cases.otherwise->insert(followpos.begin(), followpos.end());
		for (NodeCases::iterator i = cases.begin(); i != cases.end();
		     i++)
			i->second->insert(followpos.begin(), followpos.end());
	}
	int eq(Node *other)
	{
		if (dynamic_cast<AnyCharNode *>(other))
			return 1;
		return 0;
	}
	ostream& dump(ostream& os) {
		return os << ".";
	}
};

/**
 * Indicate that a regular expression matches. An AcceptNode itself
 * doesn't match anything, so it will never generate any transitions.
 */
class AcceptNode : public ImportantNode {
public:
	AcceptNode() {}
	void release(void)
	{
		/* don't delete AcceptNode via release as they are shared, and
		 * will be deleted when the table the are stored in is deleted
		 */
	}

	void follow(NodeCases& cases __attribute__((unused)))
	{
		/* Nothing to follow. */
	}

	/* requires accept nodes to be common by pointer */
	int eq(Node *other)
	{
		if (dynamic_cast<AcceptNode *>(other))
			return (this == other);
		return 0;
	}
};

/* Match a node zero or more times. (This is a unary operator.) */
class StarNode : public OneChildNode {
public:
	StarNode(Node *left) : OneChildNode(left)
	{
		nullable = true;
	}
	void compute_firstpos()
	{
		firstpos = child[0]->firstpos;
	}
	void compute_lastpos()
	{
		lastpos = child[0]->lastpos;
	}
	void compute_followpos()
	{
		NodeSet from = child[0]->lastpos, to = child[0]->firstpos;
		for(NodeSet::iterator i = from.begin(); i != from.end(); i++) {
			(*i)->followpos.insert(to.begin(), to.end());
		}
	}
	int eq(Node *other) {
		if (dynamic_cast<StarNode *>(other))
			return child[0]->eq(other->child[0]);
		return 0;
	}
	ostream& dump(ostream& os)
	{
		os << '(';
		child[0]->dump(os);
		return os << ")*";
	}
};

/* Match a node one or more times. (This is a unary operator.) */
class PlusNode : public OneChildNode {
public:
	PlusNode(Node *left) : OneChildNode(left) { }
	void compute_nullable()
	{
		nullable = child[0]->nullable;
	}
	void compute_firstpos()
	{
		firstpos = child[0]->firstpos;
	}
	void compute_lastpos()
	{
		lastpos = child[0]->lastpos;
	}
	void compute_followpos()
	{
		NodeSet from = child[0]->lastpos, to = child[0]->firstpos;
		for(NodeSet::iterator i = from.begin(); i != from.end(); i++) {
			(*i)->followpos.insert(to.begin(), to.end());
		}
	}
	int eq(Node *other)
	{
		if (dynamic_cast<PlusNode *>(other))
			return child[0]->eq(other->child[0]);
		return 0;
	}
	ostream& dump(ostream& os)
	{
		os << '(';
		child[0]->dump(os);
		return os << ")+";
	}
};

/* Match a pair of consecutive nodes. */
class CatNode : public TwoChildNode {
public:
	CatNode(Node *left, Node *right) : TwoChildNode(left, right) { }
	void compute_nullable()
	{
		nullable = child[0]->nullable && child[1]->nullable;
	}
	void compute_firstpos()
	{
		if (child[0]->nullable)
			firstpos = child[0]->firstpos + child[1]->firstpos;
		else
			firstpos = child[0]->firstpos;
	}
	void compute_lastpos()
	{
		if (child[1]->nullable)
			lastpos = child[0]->lastpos + child[1]->lastpos;
		else
			lastpos = child[1]->lastpos;
	}
	void compute_followpos()
	{
		NodeSet from = child[0]->lastpos, to = child[1]->firstpos;
		for(NodeSet::iterator i = from.begin(); i != from.end(); i++) {
			(*i)->followpos.insert(to.begin(), to.end());
		}
	}
	int eq(Node *other) {
		if (dynamic_cast<CatNode *>(other)) {
			if (!child[0]->eq(other->child[0]))
				return 0;
			return child[1]->eq(other->child[1]);
		}
		return 0;
	}
	ostream& dump(ostream& os)
	{
		child[0]->dump(os);
		child[1]->dump(os);
		return os;
	}
};

/* Match one of two alternative nodes. */
class AltNode : public TwoChildNode {
public:
	AltNode(Node *left, Node *right) : TwoChildNode(left, right) { }
	void compute_nullable()
	{
		nullable = child[0]->nullable || child[1]->nullable;
	}
	void compute_lastpos()
	{
		lastpos = child[0]->lastpos + child[1]->lastpos;
	}
	void compute_firstpos()
	{
		firstpos = child[0]->firstpos + child[1]->firstpos;
	}
	int eq(Node *other) {
		if (dynamic_cast<AltNode *>(other)) {
			if (!child[0]->eq(other->child[0]))
				return 0;
			return child[1]->eq(other->child[1]);
		}
		return 0;
	}
	ostream& dump(ostream& os)
	{
		os << '(';
		child[0]->dump(os);
		os << '|';
		child[1]->dump(os);
		os << ')';
		return os;
	}
};


/* Traverse the syntax tree depth-first in an iterator-like manner. */
class depth_first_traversal {
	stack<Node *> pos;
	void push_left(Node *node)
	{
		pos.push(node);

		while (dynamic_cast<InnerNode *>(node)) {
			pos.push(node->child[0]);
			node = node->child[0];
		}
	}

public:
	depth_first_traversal(Node *node)
	{
		push_left(node);
	}
	Node *operator*()
	{
		return pos.top();
	}
	Node* operator->()
	{
		return pos.top();
	}
	operator bool()
	{
		return !pos.empty();
	}
	void operator++(int)
	{
		Node *last = pos.top();
		pos.pop();

		if (!pos.empty()) {
			/* no need to dynamic cast, as we just popped a node so
			 * the top node must be an inner node */
			InnerNode *node = (InnerNode *)(pos.top());
			if (node->child[1] && node->child[1] != last) {
				push_left(node->child[1]);
			}
		}
	}
};

struct node_counts {
	int charnode;
	int charset;
	int notcharset;
	int alt;
	int plus;
	int star;
	int any;
	int cat;
};

extern EpsNode epsnode;

int debug_tree(Node *t);
Node *simplify_tree(Node *t, dfaflags_t flags);
void label_nodes(Node *root);
unsigned long hash_NodeSet(NodeSet *ns);


/* Comparison operator for sets of <NodeSet *>.
 * Compare set hashes, and if the sets have the same hash
 * do compare pointer comparison on set of <Node *>, the pointer comparison
 * allows us to determine which Sets of <Node *> we have seen already from
 * new ones when constructing the DFA.
 */
struct deref_less_than {
	bool operator()(pair <unsigned long, NodeSet *> const & lhs,
			pair <unsigned long, NodeSet *> const & rhs) const
		{
			if (lhs.first == rhs.first)
				return *(lhs.second) < *(rhs.second);
			else
				return lhs.first < rhs.first;
		}
};

#endif /* __LIBAA_RE_EXPR */