/*joinu from C. Hammond, from puddle.mit.edu
31 October 1996
r groman

"It needs to be linked with outer.o (some variation - whatever you
are calling yours) and jgofs.a which is usually in ../lib." */

/*	Scheme/"interesting" details:					*/
/*	  1) Read 2nd object into memory.				*/
/*	  2) The variable list of the constructed object is the union	*/
/*	     of the variable lists of the 2 input objects.  The		*/
/*	     "match set" of variables is the intersection of the 2 lists */
/*	  3) Records of the 2 objects match when the values for all	*/
/*	     variables in the match set have the same value.		*/
/*	     Test is numeric if both vals are numeric; string else.	*/
/*	     Matching has nothing to do with levels - records that	*/
/*	     match are output w/the appropriate levelizing		*/
/*	  4) Number of records in constructed object is number of	*/
/*	     records in the 1st object that have a match with some	*/
/*	     record in 2nd object.					*/
/*	  5) It is possible that many records from each object have	*/
/*	     the same set of values for the match set of variables.	*/
/*	     In such cases, every record from the 1st object is		*/
/*	     matched with the first record from the 2nd object.  Other	*/
/*	     records from the 2nd object do not appear in the con-	*/
/*	     structed object.						*/
/*	  6) Except for the string test mentioned in 3, above, the	*/
/*	     functionality described in 1-5 was in the joinu code as	*/
/*	     of 30 Sep 98.						*/
/*				WJS  10 Nov 98				*/

char join_version[]="joinu version 2.1a  27 Sep 1999";
/*   27 Sep 99.  WJS							*/
/*	Bug fix: counting error.  Again, amazed it even seemed to work!	*/
/*		[Begin v 2.1a]						*/
/*    4 May 99.  WJS							*/
/*	Fix massive bug caused by assuming jdbread_ gives a "line"	*/
/*	  instead of a level						*/
/*	Add version info						*/
/*	Add iovaldouble_ entry						*/
/*      Define required functions					*/
/*	Change #includes to OPTIONS & INNEROPTIONS per other methods	*/
/*	  (this mod in official OO version but don't know how it got	*/
/*	  there)							*/
/*	Incorporate CLH fix of 13 Nov 98 made to official OO version	*/
/*		[Begin v 2.1]						*/
/*   17 Oct 97.  WJS							*/
/*	Dynamically allocate space for saved object			*/
/*	Correct bad parametrization					*/
/*    2 Oct 97.  WJS							*/
/*	Dynamically shape values2					*/
/*	Add alpha join (assumes alpha vals are illegal numerically)	*/
/*	Provide non-zero widths						*/
/*	Allow tokens of size DATUMSIZE (instead of DATUMSIZE-1)		*/
/*	JGOFS 1.5 mods:							*/
/*	  Call error_ instead of doing printf				*/
/*	  Change iovalreal_ alpha value test				*/
/*	  Add in_ to internal function names				*/
/*	  #include *default.h files & reparametrize accordingly		*/
/*   30 Sep 97.  WJS							*/
/*	Process any input width attributes, and add iowidths_ entry	*/
/*  (per JGOFS 1.5) to return this information.				*/
/*	Type the void io*_ entries					*/
/*	Notes:								*/
/*		1) requires PATH_INFO control if used on local objects	*/
/*		   Otherwise, if the environment variable is set to	*/
/*		   .html or .flat, program breaks			*/
/*		2) does not propagate widths (or any other attributes)	*/

#include <stdio.h>
#include <string.h>
#include <math.h>

#ifdef VMS
#include <ctype.h>
#include <stdlib.h>
#else
#include <malloc.h>
#endif

  /*  Next statement should really be "if exactly one of HP or IBM is	*/
  /*  defined and also TRUE"  but I'm tired...				*/
#if defined(HP) || defined(IBM)
#define jdbopen_ jdbopen
#define jdbreada_ jdbreada
#define jdbclose_ jdbclose
#define jdblevel_ jdblevel
#define jdbattributes_ jdbattributes
#define jdbcomments_ jdbcomments
#endif
int jdbopen_();
int jdbreada_();
int jdbclose_();
int jdblevel_();
int jdbattributes_();
int jdbcomments_();

#include OPTIONS
#include INNEROPTIONS

void error_();			/*  ... entry in outer			*/

/* #define DEBUG */
#ifndef MEMORY_ALLOCATION_CHUNK
#define MEMORY_ALLOCATION_CHUNK 41960
#endif
#define POINTERS_CHUNK 100
 
struct {
  int pntr,index,level;
} v[NVAR],tempv;

double strtod();

int nlevels[2];
char names[2][NVAR][VARNAMESIZE+1];
int fldwidths[NVAR];

int namesize=VARNAMESIZE+1;
int valuesize=DATUMSIZE+1;

char *values;	/* values[NVAR][DATUMSIZE+1]; ptr for convenience	*/

  /*  Dynamic memory scheme:						*/
  /*    Logically, we want   values2 [nlines2+1] [nv2] [valuesize]	*/
  /*	We could allocate, say, NLINES worth, and, after filling it,	*/
  /*	  allocate 2*NLINES, copy the original space into the new, free	*/
  /*	  the original, etc.  This is a lot of bytes to move, however.	*/
  /*    Instead, we make an array of pointers to chunks of size NLINES	*/
  /*	  The pointers start out NULL.  We allocate NLINES worth to the	*/
  /*	  first pointer.  When we exceed LINES, we allocate NLINES	*/
  /*	  worth to the second pointer, and start using that.  Logically	*/
  /*	  this breaks up the [nlines2+1] dimension into 2 dimensions.	*/
  /*	  One is "which chunk_size/NLINES chunk?".  The other runs	*/
  /*	  from 0 to NLINES-1 within each chunk.				*/
  /*	If we run out of pointers, we dynamically allocate more space,	*/
  /*	  copy over, free, etc, but we do this to the pointers array,	*/
  /*	  which is lots easier to move.					*/
  /*	There is a question of how much memory to allocate at once.	*/
  /*	  I chose 41960 bytes, but it can be specified as a		*/
  /*	  compilation constant (MEMORY_ALLOCATION_CHUNK).  It		*/
  /*	  should be tuned to the efficiency of the system's memory	*/
  /*	  allocation scheme (actually, the system's malloc routine's	*/
  /*	  allocation scheme).  We allocate that much, then fit an	*/
  /*	  integral number of lines into each chunk, and do not use the	*/
  /*	  end of the chunk.  Clearly it is possible to have some	*/
  /*	  values that interact horribly with a particular object and/	*/
  /*	  or system, but we don't do any work on this (beyond being	*/
  /*	  sure that the chunk will hold at least one line).		*/
char **values2;
char *val2lineptr;
int nlines2,nv2,linesize2;
int nlines2perchunk;

int handle[2]={ -1, -1};
int nt[2];
int match[NVAR];
int minlevelread;
char *comments;

/************************************************************************/

void in_get_width(width,object_selector,variable_number)
int *width;
int object_selector,variable_number;
/*  If width attribute for variable variable_number in object		*/
/*  object_selector specifies a width > &width, set &width to that	*/
/*  attribute value							*/
{
  int j;
  char str[ATTRSIZE];

  while (  jdbattributes_(handle+object_selector,
			  &variable_number,
			  str)			 )   {
    if (strncmp(str,"width=",6) == 0) {
      if (  (j = atoi(str+6)) > *width  ) *width = j;
      break;
    }
  }

  return;
}

int in_getrec(object_selector,buf)
int object_selector;
char *buf;
{
  return jdbreada_(handle+object_selector,buf,&valuesize);
}

int in_get_chunk (object_selector,chunk,linesize,nlines,level_splits)
int object_selector,linesize,nlines;
int *level_splits;
char *chunk;
/*  Reads, into chunk, up to nlines of size linesize from object	*/
/*  associated with object_selector.					*/
/*  Returns number of lines read  (!= nlines equivalent to EOF)		*/
{
  int linecnt,j,lev;
  char *start_this_line;
  static char *start_previous_line;

  start_this_line = chunk - linesize;
  for (linecnt=0; linecnt<nlines; linecnt++) {
    start_this_line += linesize;
    lev = in_getrec(object_selector,start_this_line);
    if (lev < 0) break;
      /*  Fill in lower levels from previous line.  Note that since	*/
      /*  the very first read comes back w/ lev = 0, and since level_	*/
      /*  splits[lev] is always 0, start_previous_line isn't used until	*/
      /*  it's defined							*/
    for (j = 0;  j < level_splits[lev]*valuesize; j += valuesize)
      strcpy(start_this_line + j, start_previous_line + j);
    start_previous_line = start_this_line;
  }

  return linecnt;
}

int in_get_entire_object
	(object_selector,data_array,linesize,nlinesperchunk,splits)
int object_selector;
char ***data_array;
int linesize;
int *nlinesperchunk,*splits;
{
  int linecnt,i;
  int chunk_pointer_base;	/*  Goes from 0 -> infinity in steps of	*/
				/*    POINTERS_CHUNK			*/
  int ichunk;			/*  Goes from 0 -> POINTERS_CHUNK.	*/
				/*  Actual chunk pointer is		*/
				/*   chunk_pointer_base	+ ichunk	*/
  char *ptr;
  char **temp;
  char **object_chunk_ptrs;
  int allocate;

  object_chunk_ptrs = *data_array;

  allocate = (MEMORY_ALLOCATION_CHUNK > linesize) ? 
				MEMORY_ALLOCATION_CHUNK : linesize;

  *nlinesperchunk = allocate/linesize;
  linecnt = 0;
  chunk_pointer_base = 0;
  ichunk = 0;

  while (1) {
    if (ichunk == 0) {
	/*  Out of pointers to chunks.  Allocate one POINTERS_CHUNK	*/
	/*  more than we have, move what we have, & free old space	*/
      temp = object_chunk_ptrs;
      object_chunk_ptrs = (char **) malloc(
		(chunk_pointer_base + POINTERS_CHUNK) * sizeof(char *)
									);
      if (object_chunk_ptrs == NULL)
        error_("Could not get memory for 'object_chunk_ptrs'","");
      if (chunk_pointer_base != 0) {
	memcpy(object_chunk_ptrs,temp,chunk_pointer_base*sizeof(char *));
	free (temp);
      }
    }

      /*  Allocate a chunk of memory and fill it with a chunk's worth	*/
      /*  of lines from the object.					*/
      /*  Save pointer to chunk in object_chunk_ptrs			*/
    ptr = (char *) malloc(allocate);
    if (ptr == NULL) 
      error_("Could not get memory for 'object_chunk_ptrs[i]'","");
    i = in_get_chunk(object_selector,ptr,linesize,*nlinesperchunk,splits);
    linecnt += i;
    object_chunk_ptrs[chunk_pointer_base+ichunk] = ptr;
    if (i != *nlinesperchunk) break;
    if (++ichunk == POINTERS_CHUNK) {
      chunk_pointer_base += POINTERS_CHUNK;
      ichunk = 0;
    }
  }

  *data_array = object_chunk_ptrs;
  jdbclose_(handle+object_selector);

  /*  Original code made an extra line of "nd"s for some reason.  If it	*/
  /*  supposed to "not match anything", "nd"s aren't really good - I	*/
  /*  suggest "**NoTrEaLdAtUm**" or some such.  In any case, though,	*/
  /*  the returned line count was never incremented, so that line 	*/
  /*  wasn't used.  Just in case, commented out code follows...		*/
  /*  If needed, it must be modified for dynamic memory scheme		*/
  /*	for (i=0; i<linesize; i += valuesize)				*/
  /*	  strcpy(object_array+i+linecnt*linesize,"**NoTrEaLdAtUm**");	*/

  return linecnt;
}

int in_iotest()
{
/*  sets val2lineptr, returns true if no match */
int i;
double v0,v1;
char *val2ptr,*endptr;
int ichunk,ilineinchunk;
int testline;

ichunk = -1;
ilineinchunk = nlines2perchunk-1;

for (testline = 0;  testline < nlines2;  testline++){
  if (++ilineinchunk == nlines2perchunk) {
    ichunk++;
    ilineinchunk = 0;
  }
    /*  Logically, this selects the start of line testline		*/
  val2lineptr = values2[ichunk] + ilineinchunk*linesize2;
  for (i = 0;  i < nt[0];  i++)
    if (match[i] >= 0) {
      val2ptr = val2lineptr + match[i]*valuesize;
      v0 = strtod(values + i*valuesize, &endptr);
      if (*endptr == '\0') v1 = strtod(val2ptr,&endptr);
      if (*endptr == '\0') {
	if (v0 != v1) break;
      } else {
	if (strcmp(values + i*valuesize, val2ptr) != 0) break;
      }
    }
  if (i == nt[0]) break;
}

return (testline==nlines2);
}

int ioopen_(params,nparams,ntotal)
char *params[];
int *nparams,*ntotal;
/*
    s[0..nparams-1]: parameter strings. Inner sets s[j][0]=0 for any 
    strings which it processes; others will be processed by outer. Thus
    selection/projections would normally be ignored by inner.
    nparams: number of parameter strings
    ntotal (returned): total number of variable names */
{
  char *tmp;
  int i,j,j1,k,maxlev;
  int ih;
    /*  Temp cumulative variable/level array for "second" object	*/
  int *nvar2;
  
    /*  Don't know why QUERY_STRING is squashed...			*/
  putenv("QUERY_STRING=");

    /*  Size data buffer big enough to hold a "line" of data		*/
  values = (char *) malloc (NVAR*valuesize);
  if (values == NULL) error_("Could not get memory for 'values' buffer","");

    /*  Size a couple of buffers that hold object spec(s)		*/
    /*  We haven't parametrized that in defaults.h or innerdefaults.h	*/
    /*  (could use PATHSIZE) but our idea is to go dynamic anyway (&	*/
    /*  dynamic is better!)						*/
    /*    comments never used, but leave it in for now ... length is	*/
    /*    length of "join ", a blank, a terminating null, and params	*/
  i = strlen(params[0]);
  j = strlen(params[1]);
  comments = (char *) malloc (5 + i + 1 + j + 1);
  if (comments == NULL) error_("Could not get memory for 'comments' buffer","");
  strcpy(comments,"join ");
  strcat(comments,params[0]);
  strcat(comments," ");
  strcat(comments,params[1]);

    /*    tmp needs to hold each spec consecutively, so size to max	*/
  tmp = (i > j) ? (char *)malloc(i+1) : (char *)malloc(j+1);
  if (tmp == NULL) error_("Could not get memory for 'tmp' buffer","");

  for (ih = 0;  ih <= 1;  ih++)  {
    strcpy(tmp,params[ih]);

    nt[ih] = -NVAR;
    nlevels[ih] = jdbopen_(handle+ih,tmp,names[ih],&namesize,nt+ih);    
    if (nlevels[ih]<0) error_("Bad Object",params[ih]);

    params[ih][0] = 0;
  }
  free (tmp);
  
  for (i = 0;  i < nt[0];  i++)  {
    for (j = 0;  j < nt[1];  j++) if (!strcmp(names[0][i],names[1][j])) break;
    if (j == nt[1]) match[i]= -1;
    else {
      match[i] = j;
      names[1][j][0] = '\0';
    }
  }
  k = 0;
  maxlev = 0;
  for (i = 0;  i < nt[0];  i++)
    if (match[i] == -1)  {
      v[k].level = jdblevel_(handle,&i);
      if (v[k].level > maxlev) maxlev=v[k].level;
      v[k].index = 0;
      v[k++].pntr = i;
    } else {
      j = jdblevel_(handle,&i);
      j1 = jdblevel_(handle+1,match+i);
      v[k].level = (j >= j1) ? j : j1;
      if (v[k].level > maxlev) maxlev=v[k].level;
      v[k].index = 0;
      v[k++].pntr = i;
    }
  for (i = 0;  i < nt[1]; i++)
    if (names[1][i][0]) {
      v[k].level = jdblevel_(handle+1,&i);
      if (v[k].level > maxlev) maxlev=v[k].level;
      v[k].index = 1;
      v[k++].pntr = i;
    }
  for (i = 0;  i < k-1;  i++)
    for (j = i+1;  j < k;  j++)
      if (v[i].level > v[j].level) {
	tempv = v[i];
	v[i] = v[j];
	v[j] = tempv;
      }

    /*    See if there are any width attributes in either input object	*/
    /*  for each variable.  Use the max of the length of the variable	*/
    /*  name, the attribute in object 1 (if any) and the attribute in	*/
    /*  object 2 (if any).						*/

  for (i = 0;  i < k;  i++) {
    fldwidths[i] = strlen  (  names [v[i].index] [v[i].pntr]  );
    in_get_width ( &fldwidths[i], v[i].index, v[i].pntr );
    if (match[i] >= 0)  in_get_width ( &fldwidths[i], 1, match[i] );
  }

  ih = 0;
  minlevelread = maxlev+1;

  nv2 = nt[1];
  linesize2 = nv2*valuesize;

    /*  For the in-memory object, want cumulative variable/level array 	*/
    /*  0th value is 0; = # vars before the 0th level; etc		*/
  nvar2 =  (int *)  malloc ( (nlevels[1]+1)*sizeof(int) );
  if (nvar2 == NULL) error_("Could not get memory for 'nvar2' buffer","");
    /*  Generate "level splits" for this object				*/
  for (i = 0;  i <= nlevels[1] + 1;  i++) nvar2[i] = 0;
  for (i = 0;  i < nv2;  i++)   ++ nvar2 [ jdblevel_(handle+1,&i)+1 ];

  nlines2 = in_get_entire_object(1,&values2,linesize2,&nlines2perchunk,nvar2);

  free (nvar2);

  *ntotal=k;
  return maxlev;
}

int ioreadrec_(level)
int *level;
/*
    Read record at appropriate level. Return 0 if end at that level. Return
    1 if ok. */
{
  static int eofflag = 0;
  int level0,itest;

  if (eofflag) return 0;

  if (*level == minlevelread) {
    minlevelread++;
    return 1;
  } else if (*level > minlevelread) return 0;
  
  itest=1;
  while (itest) {
    level0 = in_getrec(0,values);
    if (level0 < 0) {
      eofflag=1;
      return 0;
    }
    if (level0 <= *level) {
      minlevelread = level0;
      itest = in_iotest();
    }
  }

  if (minlevelread < *level) return 0;
  minlevelread++;
  return 1;
}

void ioclose_()
/*
    Close files */
{
  jdbclose_(handle);
}

void iovalstr_(vn,tmp)
int *vn;
char *tmp;
/*
    Return string value (tmp) for variable indexed by vn. */
/*  val2lineptr is implicit input from ioreadrec_.  It points to the	*/
/*  line from the entirely-read-in object that iovalstr_ is getting	*/
/*  values from.  A pointer to the line of the "normal" object is	*/
/*  implicit in that its values array is filled by ioreadrec_ from its	*/
/*  current line							*/
{
  char *s;

  s = (v[*vn].index == 0) ? values : val2lineptr;
  s += v[*vn].pntr*valuesize;
  strcpy ( tmp, s+strspn(s," ") );

  return;
}

void iovaldouble_(vn,f)
int *vn;
double *f;
{
  char tmp[DATUMSIZE+1];
  char *endptr;

  iovalstr_(vn,tmp);
  *f = strtod(tmp,&endptr);
  if (*endptr != '\0') *f = -9999.0;

  return;
}

void iovalreal_(vn,f)
int *vn;
float *f;
/*
    Return real value (f) for variable indexed by vn. -9999 for strings */
{
  double df;

  iovaldouble_(vn,&df);
  *f = df;

  return;
}

int iovarlevel_(vn)
int *vn;
/*
    Return level corresponding to variable indexed by vn.    */
{
return v[*vn].level;
}

int ioattrout_(vn,str)
int *vn;
char *str;
/*
    Output attributes for variable indexed by vn. */
{
  return 0;
}

void ioname_(vn,s)
int *vn;
char *s;
/*
    Return name (s) corresponding to variable number vn. */
{
  strcpy   (   s,  names [ v[*vn].index] [v[*vn].pntr]   );
  return;
}

int iocommout_(s)
char *s;
{
  return 0;
}

int iowidth_(vn)
int *vn;
/*  "Return length of variable field indexed by vn" [from outer, JGOFS 1.5]  */
{
  return fldwidths[*vn];
}