/* $Header$
 *
 *  (c) Copyright 1988 by the Massachusetts Institute of Technology.
 *  For copying and distribution information, please see the file
 *  <mit-copyright.h>.
 */

#include <mit-copyright.h>
#include <moira.h>
#include <stdio.h>
#include <stdlib.h>
#include "dbck.h"
EXEC SQL INCLUDE sqlca;

static char phase1_qc_rcsid[] = "$Header$";

EXEC SQL WHENEVER SQLERROR DO dbmserr();

int show_user_id(struct user *u)
{
  printf("User %s (%s, status %d) has duplicate ID\n",
	 u->login, u->fullname, u->status);
  return 0;
}

handle_duplicate_logins(struct save_queue *sq)
{
  struct user *u, *uu, *tmp;

  uu = NULL;
  if (sq_get_data(sq, &uu))
    {
      while (sq_get_data(sq, &u))
	{
	  if (!strcmp(u->login, uu->login))
	    {
	      if (uu->status == 1 || u->status == 0)
		{
		  tmp = u;
		  u = uu;
		  uu = tmp;
		}
	      printf("User %s (%s, status %d) and\n",
		     u->login, u->fullname, u->status);
	      printf("User %s (%s, status %d) have duplicate logins\n",
		     uu->login, uu->fullname, uu->status);
	      if (!strcmp(u->fullname, uu->fullname) &&
		  single_fix("Delete the second one"))
		single_delete("users", "users_id", uu->users_id);
	      else if (single_fix("Unregister the second one"))
		{
		  EXEC SQL BEGIN DECLARE SECTION;
		  int id = uu->users_id, rowcount;
		  EXEC SQL END DECLARE SECTION;

		  EXEC SQL UPDATE users
		    SET login = '#' || CHAR(users.unix_uid), status = 0
		    WHERE users_id = :id;
		  rowcount = sqlca.sqlerrd[2];
		  if (rowcount > 0)
		    {
		      printf("%d entr%s fixed\n", rowcount,
			     rowcount == 1 ? "y" : "ies");
		    }
		  else
		    printf("Not fixed\n");
		  modified("users");
		}
	    }
	  else
	    uu = u;
	}
    }
}

fix_user_id(struct user *u)
{
  u->users_id = generic_fix_id("users", "users_id", "login",
			       u->users_id, u->login);
}


cant_fix(int id)
{
  printf("Sorry, don't know how to fix that\n");
}

int show_mach_id(struct machine *m)
{
  printf("Machine %s has duplicate ID %d\n", m->name, m->mach_id);
  return 0;
}

int show_mach_name(struct machine *m)
{
  printf("Machine %s (%d) has duplicate name\n", m->name, m->mach_id);
  return 0;
}

fix_mach_id(struct machine *m)
{
  m->mach_id = generic_fix_id("machine", "mach_id", "name",
			      m->mach_id, m->name);
}

int show_snet_name(struct subnet *s)
{
  printf("Subnet %s (%d) has duplicate name\n", s->name, s->snet_id);
  return 0;
}

int show_clu_id(struct cluster *c)
{
  printf("Cluster %s has duplicate ID %d\n", c->name, c->clu_id);
  return 0;
}

int show_clu_name(struct cluster *c)
{
  printf("Cluster %s (%d) has duplicate name\n", c->name, c->clu_id);
  return 0;
}

fix_clu_id(struct cluster *c)
{
  c->clu_id = generic_fix_id("cluster", "clu_id", "name", c->clu_id, c->name);
}

int show_list_id(struct list *l)
{
  printf("List %s has duplicate ID %d\n", l->name, l->list_id);
  return 0;
}

int show_list_name(struct list *l)
{
  printf("List %s (%d) has duplicate name\n", l->name, l->list_id);
  return 0;
}

fix_list_id(struct list *l)
{
  l->list_id = generic_fix_id("list", "list_id", "name", l->list_id, l->name);
}

int show_fs_id(struct filesys *f)
{
  printf("Filesys %s has duplicate ID %d\n", f->name, f->filsys_id);
  return 0;
}

fix_fs_id(struct filesys *f)
{
  f->filsys_id = generic_fix_id("filesys", "filsys_id", "label",
				f->filsys_id, f->name);
}

int show_fs_name(struct filesys *fs)
{
  printf("Filesys %s (%d) has duplicate name\n", fs->name, fs->filsys_id);
  return 0;
}

int show_np_id(struct nfsphys *n)
{
  printf("NfsPhys %s:%s has duplicate ID %d\n",
	 ((struct machine *)hash_lookup(machines, n->mach_id))->name,
	 n->dir, n->nfsphys_id);
  return 0;
}

fix_np_id(struct nfsphys *n)
{
  n->nfsphys_id = generic_fix_id("nfsphys", "nfsphys_id", "dir",
				 n->nfsphys_id, n->dir);
}

int show_str_id(struct string *s)
{
  printf("String %s has duplicate ID %d\n", s->name, s->string_id);
  return 0;
}

int print_str_id(int id)
{
  printf("String %d is a duplicate\n", id);
  return 0;
}

print_dup_map(int key, int data, char *hint)
{
  printf("String %d is a duplicate of string %d\n", key, data);
}

phase1(void)
{
  EXEC SQL BEGIN DECLARE SECTION;
  char name[81], name1[81], last[17], first[17], buf[257];
  int id, id1, id2, id3, aid, aid2, status;
  int sid, sid2, sid3, sid4, sid5;
  EXEC SQL END DECLARE SECTION;
  int i, q, retval, tmp;
  struct save_queue *sq;
  struct user *u;
  struct machine *m;
  struct subnet *sn;
  struct list *l;
  struct cluster *c;
  struct string *s;
  struct filesys *f;
  struct nfsphys *n;

  printf("Phase 1 - Looking for duplicates\n");

  /* self-join strings table on "string" to get duplicate strings, then
     build a duplicates table to merge them. */

  dprintf("Looking for duplicate strings...\n");
  string_dups = create_hash(100);
  if (!string_dups)
    out_of_mem("storing duplicate strings");

  EXEC SQL DECLARE csr116 CURSOR FOR
    SELECT s1.string_id, s2.string_id FROM strings s1, strings s2
    WHERE s1.string = s2.string and s1.string_id < s2.string_id;
  EXEC SQL OPEN csr116;
  /*  The SELECT gives us two columns, both with non-negative integers.
   *  The number in the left column is always the smaller of the two,
   *  and each row includes string IDs for identical strings.  We use
   *  them to make a mapping from id-to-delete to id-to-keep for all
   *  superflous IDs.
   */
  q = 0;
  while (1)
    {
      EXEC SQL FETCH csr116 INTO :id1, :id2;
      if (sqlca.sqlcode)
	break;
      q++;
      /*  If id2 is already stored, skip this row. */
      i = int_hash_lookup(string_dups, id2);
      if (i > 0)
	continue;
      /*  Follow the chain of id1 equivalent IDs back to the lowest one. */
      id = id1;
      while ((tmp = int_hash_lookup(string_dups, id)) > 0)
	id = tmp;
      int_hash_store(string_dups, id2, id);
    }
  EXEC SQL CLOSE csr116;
  dprintf("found %d duplicates\n", q);
  int_hash_step(string_dups, print_dup_map, NULL);
  /* We don't want to delete the duplicates now because if the dbck
     is cancelled, a LOT of state will be lost. So, we'll just let
     them not get marked as used and then phase3 will clean them up */

  dprintf("Loading strings...\n");
  sq = sq_create();
  strings = create_hash(75000);
  if (!sq || !strings)
    out_of_mem("loading strings");

  EXEC SQL DECLARE csr101 CURSOR FOR
    SELECT string_id, string FROM strings ORDER BY string_id;
  EXEC SQL OPEN csr101;
  q = 0;
  while (1)
    {
      EXEC SQL FETCH csr101 INTO :id, :buf;
      if (sqlca.sqlcode)
	break;
      q++;
      s = malloc(sizeof(struct string));
      if (!s)
	out_of_mem("storing strings");
      s->name = strsave(strtrim(buf));
      s->string_id = id;
      s->refc = 0;
      retval = hash_store(strings, id, s);
      if (retval == -1)
	out_of_mem("storing strings in hash table");
      else if (retval == 1) /* duplicate string_id */
	{
	  sq_save_data(sq, hash_lookup(strings, id));
	  sq_save_data(sq, s);
	}
    }
  EXEC SQL CLOSE csr101;
  /* keep string id 0 (the empty string) even if unreferenced */
  string_check(0);

  printf("Loaded %d strings\n", q);

  dprintf("Loading users...\n");
  sq = sq_create();
  users = create_hash(30000);
  if (!sq || !users)
    out_of_mem("loading users");

  EXEC SQL DECLARE csr102 CURSOR FOR
    SELECT users_id, login, last, first, status, potype, pop_id, box_id,
    modby, fmodby, pmodby, comments, sigwho FROM users
    ORDER BY users_id;
  EXEC SQL OPEN csr102;
  while (1)
    {
      EXEC SQL FETCH csr102 INTO :id, :name, :last, :first, :status,
	:buf, :id2, :id3, :sid, :sid2, :sid3, :sid4, :sid5;
      if (sqlca.sqlcode)
	break;

      u = malloc(sizeof(struct user));
      if (!u)
	out_of_mem("storing users");
      strcpy(u->login, strtrim(name));
      u->potype = buf[0];
      sprintf(buf, "%s, %s", strtrim(last), strtrim(first));
      u->fullname = strsave(buf);
      u->status = status;
      u->users_id = id;
      u->modby = sid;
      u->fmodby = sid2;
      u->pmodby = sid3;
      u->comment = sid4;
      u->sigwho = sid5;
      switch (u->potype)
	{
	case 'P':
	  u->pobox_id = id2;
	  break;
	case 'S':
	  /*  If potype is SMTP, box_id is a string_id for the strings tbl */
	  u->pobox_id = id3;
	  break;
	default:
	  u->pobox_id = 0;
	}
      retval = hash_store(users, id, u);
      if (retval == -1)
	out_of_mem("storing users in hash table");
      else if (retval == 1)
	{
	  sq_save_data(sq, hash_lookup(users, id));
	  sq_save_data(sq, u);
	}
    }
  EXEC SQL CLOSE csr102;

  generic_fix(sq, show_user_id, "Change ID", fix_user_id, 0);

  if (!fast)
    {
      sq = sq_create();
      if (!sq)
	out_of_mem("finding duplicate logins");

      EXEC SQL DECLARE csr103 CURSOR FOR
	SELECT u1.users_id FROM users u1, users u2
	WHERE u1.login = u2.login and u1.rowid != u2.rowid;
      EXEC SQL OPEN csr103;
      while (1)
	{
	  EXEC SQL FETCH csr103 INTO :id;
	  if (sqlca.sqlcode)
	    break;
	  sq_save_data(sq, hash_lookup(users, id));
	}
      EXEC SQL CLOSE csr103;
      handle_duplicate_logins(sq);
    }

  if (!fast)
    {
      dprintf("Scanning krbmap...\n");

      EXEC SQL DECLARE csr113 CURSOR FOR
	SELECT k1.users_id FROM krbmap k1, krbmap k2
	WHERE k1.users_id = k2.users_id AND k1.rowid != k2.rowid;
      EXEC SQL OPEN csr113;
      while (1)
	{
	  EXEC SQL FETCH csr113 INTO :id;
	  if (sqlca.sqlcode)
	    break;

	  printf("User %d is in the krbmap more than once!\n", id);
	  printf("Not fixing this error\n");
	}
      EXEC SQL CLOSE csr113;

      EXEC SQL DECLARE csr114 CURSOR FOR
	SELECT k1.string_id FROM krbmap k1, krbmap k2
	WHERE k1.string_id = k2.string_id AND k1.rowid != k2.rowid;
      EXEC SQL OPEN csr114;
      while (1)
	{
	  EXEC SQL FETCH csr114 INTO :id;
	  if (sqlca.sqlcode)
	    break;

	  printf("Principal %d is in the krbmap more than once!\n", id);
	  printf("Not fixing this error\n");
	}
      EXEC SQL CLOSE csr114;
    }

  dprintf("Loading machines...\n");
  sq = sq_create();
  machines = create_hash(20000);
  if (!sq || !machines)
    out_of_mem("loading machines");

  EXEC SQL DECLARE csr104 CURSOR FOR
    SELECT mach_id, name, snet_id, owner_type, owner_id,
    acomment, ocomment, creator, modby
    FROM machine ORDER BY mach_id;
  EXEC SQL OPEN csr104;
  while (1)
    {
      EXEC SQL FETCH csr104 INTO :id, :name, :id2, :buf, :id3, :sid2,
	:sid3, :sid4, :sid;
      if (sqlca.sqlcode)
	break;

      m = malloc(sizeof(struct machine));
      if (!m)
	out_of_mem("storing machines");
      strcpy(m->name, strtrim(name));
      m->owner_type = buf[0];
      m->owner_id = id3;
      m->snet_id = id2;
      m->mach_id = id;
      m->clucount = 0;
      m->acomment = sid2;
      m->ocomment = sid3;
      m->creator = sid4;
      m->modby = sid;
      retval = hash_store(machines, id, m);
      if (retval == -1)
	out_of_mem("storing machines in hash table");
      else if (retval == 1)
	{
	  sq_save_data(sq, hash_lookup(machines, id));
	  sq_save_data(sq, m);
	}
    }
  EXEC SQL CLOSE csr104;
  generic_fix(sq, show_mach_id, "Change ID", fix_mach_id, 0);

  if (!fast)
    {
      sq = sq_create();
      if (!sq)
	out_of_mem("looking for duplicate machine names");

      EXEC SQL DECLARE csr105 CURSOR FOR
	SELECT m1.mach_id FROM machine m1, machine m2
	WHERE m1.name = m2.name AND m1.rowid != m2.rowid;
      EXEC SQL OPEN csr105;
      while (1)
	{
	  EXEC SQL FETCH csr105 INTO :id;
	  if (sqlca.sqlcode)
	    break;

	  sq_save_data(sq, hash_lookup(machines, id));
	}
      EXEC SQL CLOSE csr105;
      generic_fix(sq, show_mach_name, "Change name", cant_fix, 0);

      EXEC SQL DECLARE csr_hal1 CURSOR FOR
	SELECT h1.name, m1.mach_id, m2.mach_id
	FROM hostalias h1, machine m1, hostalias h2, machine m2
	WHERE h1.name = h2.name AND h1.mach_id != h2.mach_id
	AND m1.mach_id = h1.mach_id AND m2.mach_id = h2.mach_id;
      EXEC SQL OPEN csr_hal1;
      while (1)
	{
	  EXEC SQL FETCH csr_hal1 INTO :name, :id1, :id2;
	  if (sqlca.sqlcode)
	    break;
	  printf("Aliases for machines %d and %d have duplicate name %s\n",
		 id1, id2, strtrim(name));
	  cant_fix(0);
	}
      EXEC SQL CLOSE csr_hal1;

      EXEC SQL DECLARE csr_hal2 CURSOR FOR
	SELECT h1.name, m1.mach_id, m2.mach_id
	FROM hostalias h1, machine m1, machine m2
	WHERE h1.name = m1.name AND h1.mach_id = m2.mach_id;
      EXEC SQL OPEN csr_hal2;
	while (1)
	  {
	    EXEC SQL FETCH csr_hal2 INTO :name, :id1, :id2;
	    if (sqlca.sqlcode)
	      break;
	    printf("Machine %d has alias %s that conflicts with machine %d\n",
		   id2, strtrim(name), id1);
	    cant_fix(0);
	  }
	EXEC SQL CLOSE csr_hal2;
    }

  dprintf("Loading subnets...\n");
  subnets = create_hash(254);
  if (!subnets)
    out_of_mem("loading subnets");

  EXEC SQL DECLARE csr115 CURSOR FOR
    SELECT snet_id, name, owner_type, owner_id, modby from subnet;
  EXEC SQL OPEN csr115;
  while (1)
    {
      EXEC SQL FETCH csr115 INTO :id, :name, :buf, :id2, :sid;
      if (sqlca.sqlcode)
	break;

      sn = malloc(sizeof(struct machine));
      if (!sn)
	out_of_mem("storing subnets");
      strcpy(sn->name, strtrim(name));
      sn->owner_type = buf[0];
      sn->owner_id = id2;
      sn->snet_id = id;
      sn->modby = sid;
      retval = hash_store(subnets, id, sn);
      if (retval == -1)
	out_of_mem("storing subnets in hash table");
      else if (retval == 1)
	{
	  printf("Duplicate subnet ID: %d (%s)\n", id, name);
	  /* should add code to delete */
	  cant_fix(0);
	}
    }
  EXEC SQL CLOSE csr115;

  if (!fast)
    {
      sq = sq_create();
      if (!sq)
	out_of_mem("looking for duplicate subnet names");

      EXEC SQL DECLARE csr117 CURSOR FOR
	SELECT s1.snet_id FROM subnet s1, subnet s2
	WHERE s1.name = s2.name AND s1.rowid != s2.rowid;
      EXEC SQL OPEN csr117;
      while (1)
	{
	  EXEC SQL FETCH csr117 INTO :id;
	  if (sqlca.sqlcode)
	    break;

	  sq_save_data(sq, hash_lookup(subnets, id));
	}
      EXEC SQL CLOSE csr117;
      generic_fix(sq, show_snet_name, "Change name", cant_fix, 0);
    }

  dprintf("Loading clusters...\n");
  sq = sq_create();
  clusters = create_hash(100);
  if (!sq || !clusters)
    out_of_mem("loading clusters");

  EXEC SQL DECLARE csr106 CURSOR FOR
    SELECT clu_id, name, modby FROM clusters;
  EXEC SQL OPEN csr106;
  while (1)
    {
      EXEC SQL FETCH csr106 INTO :id, :name, :sid;
      if (sqlca.sqlcode)
	break;

      c = malloc(sizeof(struct cluster));
      if (!c)
	out_of_mem("storing clusters");
      strcpy(c->name, strtrim(name));
      c->clu_id = id;
      c->modby = sid;
      retval = hash_store(clusters, id, c);
      if (retval == -1)
	out_of_mem("storing clusters in hash table");
      else if (retval == 1)
	{
	  sq_save_data(sq, hash_lookup(clusters, id));
	  sq_save_data(sq, c);
	}
    }
  EXEC SQL CLOSE csr106;
  generic_fix(sq, show_clu_id, "Change ID", fix_clu_id, 0);

  if (!fast)
    {
      sq = sq_create();
      if (!sq)
	out_of_mem("looking for duplicate cluster names");

      EXEC SQL DECLARE csr107 CURSOR FOR
	SELECT c1.clu_id FROM clusters c1, clusters c2
	WHERE c1.name = c2.name AND c1.rowid != c2.rowid;
      EXEC SQL OPEN csr107;
      while (1)
	{
	  EXEC SQL FETCH csr107 INTO :id;
	  if (sqlca.sqlcode)
	    break;

	  sq_save_data(sq, hash_lookup(clusters, id));
	}
      EXEC SQL CLOSE csr107;
      generic_fix(sq, show_clu_name, "Change name", cant_fix, 0);
    }

  dprintf("Loading lists...\n");
  sq = sq_create();
  lists = create_hash(50000);
  if (!sq || !lists)
    out_of_mem("loading lists");

  EXEC SQL DECLARE csr108 CURSOR FOR
    SELECT list_id, name, acl_id, acl_type, modby  FROM list
    ORDER BY list_id;
  EXEC SQL OPEN csr108;
  while (1)
    {
      EXEC SQL FETCH csr108 INTO :id, :name, :aid, :buf, :sid;
      if (sqlca.sqlcode)
	break;
      l = malloc(sizeof(struct list));
      if (!l)
	out_of_mem("storing lists");
      strcpy(l->name, strtrim(name));
      l->acl_type = buf[0];
      l->acl_id = aid;
      l->list_id = id;
      l->members = 0;
      retval = hash_store(lists, id, l);
      if (retval == -1)
	out_of_mem("storing lists in hash table");
      else if (retval == 1)
	{
	  sq_save_data(sq, hash_lookup(lists, id));
	  sq_save_data(sq, l);
	}
    }
  EXEC SQL CLOSE csr108;
  generic_fix(sq, show_list_id, "Change ID", fix_list_id, 0);

  if (!fast)
    {
      sq = sq_create();
      if (!sq)
	out_of_mem("looking for duplicate list names");

      EXEC SQL DECLARE csr109 CURSOR FOR
	SELECT l1.list_id FROM list l1, list l2
	WHERE l1.name = l2.name AND l1.rowid != l2.rowid;
      EXEC SQL OPEN csr109;
      while (1)
	{
	  EXEC SQL FETCH csr109 INTO :id;
	  if (sqlca.sqlcode)
	    break;

	  sq_save_data(sq, hash_lookup(lists, id));
	}
      EXEC SQL CLOSE csr109;
      generic_fix(sq, show_list_name, "Change name", cant_fix, 0);
    }

  dprintf("Loading filesys...\n");
  sq = sq_create();
  filesys = create_hash(30000);
  if (!sq || !filesys)
    out_of_mem("loading filesys");

  EXEC SQL DECLARE csr110 CURSOR FOR
    SELECT filsys_id, label, owner, owners, phys_id, mach_id,
    type, name, modby FROM filesys ORDER BY filsys_id;
  EXEC SQL OPEN csr110;
  while (1)
    {
      EXEC SQL FETCH csr110 INTO :id, :name, :aid, :aid2, :id2, :id3,
	:buf, :name1, :sid;
      if (sqlca.sqlcode)
	break;

      f = malloc(sizeof(struct filesys));
      if (!f)
	out_of_mem("storing filesystems");
      strcpy(f->name, strtrim(name));
      strcpy(f->dir, strtrim(name1));
      f->filsys_id = id;
      f->owner = aid;
      f->owners = aid2;
      f->phys_id = id2;
      f->mach_id = id3;
      f->type = buf[0];
      retval = hash_store(filesys, id, f);
      if (retval == -1)
	out_of_mem("storing filesys in hash table");
      else if (retval == 1)
	{
	  sq_save_data(sq, hash_lookup(filesys, id));
	  sq_save_data(sq, f);
	}
    }
  EXEC SQL CLOSE csr110;

  generic_fix(sq, show_fs_id, "Change ID", fix_fs_id, 0);

  if (!fast)
    {
      sq = sq_create();
      if (!sq)
	out_of_mem("looking for duplicate filesys names");

      EXEC SQL DECLARE csr118 CURSOR FOR
	SELECT fs1.filsys_id FROM filesys fs1, filesys fs2
	WHERE fs1.label = fs2.label AND fs1.rowid != fs2.rowid;
      EXEC SQL OPEN csr118;
      while (1)
	{
	  EXEC SQL FETCH csr118 INTO :id;
	  if (sqlca.sqlcode)
	    break;

	  sq_save_data(sq, hash_lookup(filesys, id));
	}
      EXEC SQL CLOSE csr118;
      generic_fix(sq, show_fs_name, "Change name", cant_fix, 0);
    }

  dprintf("Loading nfsphys...\n");
  sq = sq_create();
  nfsphys = create_hash(500);
  if (!sq || !nfsphys)
    out_of_mem("loading nfsphs");

  EXEC SQL DECLARE csr111 CURSOR FOR
    SELECT nfsphys_id, dir, mach_id, allocated, modby FROM nfsphys;
  EXEC SQL OPEN csr111;
  while (1)
    {
      EXEC SQL FETCH csr111 INTO :id, :name, :id2, :id3, :sid;
      if (sqlca.sqlcode)
	break;

      n = malloc(sizeof(struct nfsphys));
      if (!n)
	out_of_mem("storing nfsphys");
      strcpy(n->dir, strtrim(name));
      n->mach_id = id2;
      n->nfsphys_id = id;
      n->allocated = id3;
      n->count = 0;
      retval = hash_store(nfsphys, id, n);
      if (retval == -1)
	out_of_mem("storing nfsphys in hash table");
      else if (retval == 1)
	{
	  sq_save_data(sq, hash_lookup(nfsphys, id));
	  sq_save_data(sq, n);
	}
    }
  EXEC SQL CLOSE csr111;

  generic_fix(sq, show_np_id, "Change ID", fix_np_id, 0);

  dprintf("Checking printcap...\n");

  EXEC SQL DECLARE csr119 CURSOR FOR
    SELECT p1.name FROM printcap p1, printcap p2
    WHERE p1.name = p2.name AND p1.rowid < p2.rowid;
  EXEC SQL OPEN csr119;
  while (1)
    {
      EXEC SQL FETCH csr119 INTO :name;
      if (sqlca.sqlcode)
	break;

      printf("Printer %s has duplicate name\n", name);
      cant_fix(0);
    }
  EXEC SQL CLOSE csr119;
}