fstrcmp.c

此程序是关于字符串 模糊相似度的计算程序（C语言）

/* Functions to make fuzzy comparisons between strings
   Copyright (C) 1988, 1989, 1992, 1993, 1995 Free Software Foundation, Inc.

   This program is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
   the Free Software Foundation; either version 2 of the License, or (at
   your option) any later version.

   This program 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
   General Public License for more details.

   You should have received a copy of the GNU General Public License
   along with this program; if not, write to the Free Software
   Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.


   Derived from GNU diff 2.7, analyze.c et al.

   The basic algorithm is described in:
   "An O(ND) Difference Algorithm and its Variations", Eugene Myers,
   Algorithmica Vol. 1 No. 2, 1986, pp. 251-266;
   see especially section 4.2, which describes the variation used below.

   The basic algorithm was independently discovered as described in:
   "Algorithms for Approximate String Matching", E. Ukkonen,
   Information and Control Vol. 64, 1985, pp. 100-118.

   Modified to work on strings rather than files
   by Peter Miller <[e-mail]>, October 1995

   Modified to accept a "minimum similarity limit" to stop analyzing the
   string when the similarity drops below the given limit by Marc Lehmann
   <[e-mail]>.

*/

#include <string.h>
#include <stdio.h>
#include <stdlib.h>
#include <limits.h>

#include "fstrcmp.h"


/*
 * Data on one input string being compared.
 */
struct string_data
{
  /* The string to be compared. */
  const char *data;

  /* The length of the string to be compared. */
  int data_length;

  /* The number of characters inserted or deleted. */
  int edit_count;
};

static struct string_data string[2];

static int max_edits; /* compareseq stops when edits > max_edits */

#ifdef MINUS_H_FLAG

/* This corresponds to the diff -H flag.  With this heuristic, for
   strings with a constant small density of changes, the algorithm is
   linear in the strings size.  This is unlikely in typical uses of
   fstrcmp, and so is usually compiled out.  Besides, there is no
   interface to set it true.  */
static int heuristic;

#endif


/* Vector, indexed by diagonal, containing 1 + the X coordinate of the
   point furthest along the given diagonal in the forward search of the
   edit matrix.  */
static int *fdiag;

/* Vector, indexed by diagonal, containing the X coordinate of the point
   furthest along the given diagonal in the backward search of the edit
   matrix.  */
static int *bdiag;

/* Edit scripts longer than this are too expensive to compute.  */
static int too_expensive;

/* Snakes bigger than this are considered `big'.  */
#define SNAKE_LIMIT	20

struct partition
{
  /* Midpoints of this partition.  */
  int xmid, ymid;

  /* Nonzero if low half will be analyzed minimally.  */
  int lo_minimal;

  /* Likewise for high half.  */
  int hi_minimal;
};


/* NAME
	diag - find diagonal path

   SYNOPSIS
	int diag(int xoff, int xlim, int yoff, int ylim, int minimal,
		struct partition *part);

   DESCRIPTION
	Find the midpoint of the shortest edit script for a specified
	portion of the two strings.

	Scan from the beginnings of the strings, and simultaneously from
	the ends, doing a breadth-first search through the space of
	edit-sequence.  When the two searches meet, we have found the
	midpoint of the shortest edit sequence.

	If MINIMAL is nonzero, find the minimal edit script regardless
	of expense.  Otherwise, if the search is too expensive, use
	heuristics to stop the search and report a suboptimal answer.

   RETURNS
	Set PART->(XMID,YMID) to the midpoint (XMID,YMID).  The diagonal
	number XMID - YMID equals the number of inserted characters
	minus the number of deleted characters (counting only characters
	before the midpoint).  Return the approximate edit cost; this is
	the total number of characters inserted or deleted (counting
	only characters before the midpoint), unless a heuristic is used
	to terminate the search prematurely.

	Set PART->LEFT_MINIMAL to nonzero iff the minimal edit script
	for the left half of the partition is known; similarly for
	PART->RIGHT_MINIMAL.

   CAVEAT
	This function assumes that the first characters of the specified
	portions of the two strings do not match, and likewise that the
	last characters do not match.  The caller must trim matching
	characters from the beginning and end of the portions it is
	going to specify.

	If we return the "wrong" partitions, the worst this can do is
	cause suboptimal diff output.  It cannot cause incorrect diff
	output.  */

static int diag PARAMS ((int, int, int, int, int, struct partition *));

static int
diag (xoff, xlim, yoff, ylim, minimal, part)
     int xoff;
     int xlim;
     int yoff;
     int ylim;
     int minimal;
     struct partition *part;
{
  int *const fd = fdiag;	/* Give the compiler a chance. */
  int *const bd = bdiag;	/* Additional help for the compiler. */
  const char *const xv = string[0].data;	/* Still more help for the compiler. */
  const char *const yv = string[1].data;	/* And more and more . . . */
  const int dmin = xoff - ylim;	/* Minimum valid diagonal. */
  const int dmax = xlim - yoff;	/* Maximum valid diagonal. */
  const int fmid = xoff - yoff;	/* Center diagonal of top-down search. */
  const int bmid = xlim - ylim;	/* Center diagonal of bottom-up search. */
  int fmin = fmid;
  int fmax = fmid;		/* Limits of top-down search. */
  int bmin = bmid;
  int bmax = bmid;		/* Limits of bottom-up search. */
  int c;			/* Cost. */
  int odd = (fmid - bmid) & 1;

  /*
	 * True if southeast corner is on an odd diagonal with respect
	 * to the northwest.
	 */
  fd[fmid] = xoff;
  bd[bmid] = xlim;
  for (c = 1;; ++c)
    {
      int d;			/* Active diagonal. */
      int big_snake;

      big_snake = 0;
      /* Extend the top-down search by an edit step in each diagonal. */
      if (fmin > dmin)
	fd[--fmin - 1] = -1;
      else
	++fmin;
      if (fmax < dmax)
	fd[++fmax + 1] = -1;
      else
	--fmax;
      for (d = fmax; d >= fmin; d -= 2)
	{
	  int x;
	  int y;
	  int oldx;
	  int tlo;
	  int thi;

	  tlo = fd[d - 1],
	    thi = fd[d + 1];

	  if (tlo >= thi)
	    x = tlo + 1;
	  else
	    x = thi;
	  oldx = x;
	  y = x - d;
	  while (x < xlim && y < ylim && xv[x] == yv[y])
	    {
	      ++x;
	      ++y;
	    }
	  if (x - oldx > SNAKE_LIMIT)
	    big_snake = 1;
	  fd[d] = x;
	  if (odd && bmin <= d && d <= bmax && bd[d] <= x)
	    {
	      part->xmid = x;
	      part->ymid = y;
	      part->lo_minimal = part->hi_minimal = 1;
	      return 2 * c - 1;
	    }
	}
      /* Similarly extend the bottom-up search.  */
      if (bmin > dmin)
	bd[--bmin - 1] = INT_MAX;
      else
	++bmin;
      if (bmax < dmax)
	bd[++bmax + 1] = INT_MAX;
      else
	--bmax;
      for (d = bmax; d >= bmin; d -= 2)
	{
	  int x;
	  int y;
	  int oldx;
	  int tlo;
	  int thi;

	  tlo = bd[d - 1],
	    thi = bd[d + 1];
	  if (tlo < thi)
	    x = tlo;
	  else
	    x = thi - 1;
	  oldx = x;
	  y = x - d;
	  while (x > xoff && y > yoff && xv[x - 1] == yv[y - 1])
	    {
	      --x;
	      --y;
	    }
	  if (oldx - x > SNAKE_LIMIT)
	    big_snake = 1;
	  bd[d] = x;
	  if (!odd && fmin <= d && d <= fmax && x <= fd[d])
	    {
	      part->xmid = x;
	      part->ymid = y;
	      part->lo_minimal = part->hi_minimal = 1;
	      return 2 * c;
	    }
	}

      if (minimal)
	continue;

#ifdef MINUS_H_FLAG
      /* Heuristic: check occasionally for a diagonal that has made lots
         of progress compared with the edit distance.  If we have any
         such, find the one that has made the most progress and return
         it as if it had succeeded.

         With this heuristic, for strings with a constant small density
         of changes, the algorithm is linear in the strings size.  */
      if (c > 200 && big_snake && heuristic)
	{
	  int best;

	  best = 0;
	  for (d = fmax; d >= fmin; d -= 2)
	    {
	      int dd;
	      int x;
	      int y;
	      int v;

	      dd = d - fmid;
	      x = fd[d];
	      y = x - d;
	      v = (x - xoff) * 2 - dd;

	      if (v > 12 * (c + (dd < 0 ? -dd : dd)))
		{
		  if
		    (
		      v > best
		      &&
		      xoff + SNAKE_LIMIT <= x
		      &&
		      x < xlim
		      &&
		      yoff + SNAKE_LIMIT <= y
		      &&
		      y < ylim
		    )
		    {
		      /* We have a good enough best diagonal; now insist
			 that it end with a significant snake.  */