/* inventory.c:
 *
 ****************************************************************
 * Copyright (C) 2003 Tom Lord
 *
 * See the file "COPYING" for further information about
 * the copyright and warranty status of this work.
 */


#include "hackerlab/char/str.h"
#include "hackerlab/char/char-class.h"
#include "hackerlab/mem/mem.h"
#include "hackerlab/sort/qsort.h"
#include "hackerlab/arrays/ar.h"
#include "hackerlab/fs/file-names.h"
#include "hackerlab/vu/safe.h"
#include "tla/libarch/inventory.h"


/* __STDC__ prototypes for static functions */
static int cmp_traversal_files (void * va, void * vb, void * ign);
static t_uchar * explicit_tag (t_uchar * id_file, t_uchar * prefix, t_uchar * postfix);
static t_uchar * implicit_tag (struct arch_id_program * prog, t_uchar * file);
static long smash_non_graphical (t_uchar * buf, long amt);




void
arch_tree_traversal (t_uchar * root,
                     arch_traversal_callback callback,
                     void * closure)
{
  DIR * dir;
  struct arch_traversal_file * files = 0;
  int x;


  safe_opendir (&dir, root);

  while (1)
    {
      char * name = 0;
      struct arch_traversal_file * this;

      safe_readdir (&name, dir);
      if (!name)
        break;

      if ((name[0] == '.') && ((name[1] == '.') || !name[1]))
        {
          lim_free (0, name);
          continue;
        }

      this = (struct arch_traversal_file *)ar_push ((void **)&files, 0, sizeof (*files));

      this->name = name;
      this->path = file_name_in_vicinity (0, root, name);
      safe_stat (this->path, &this->stat);
      /* name freed from files->name */
    }

  safe_closedir (dir);

  quicksort ((void *)files, ar_size ((void *)files, 0, sizeof (*files)), sizeof (*files), cmp_traversal_files, 0);


  for (x = 0; x < ar_size ((void *)files, 0, sizeof (*files)); ++x)
    {
      if (callback (&files[x], closure) && S_ISDIR (files[x].stat.st_mode))
        {
          arch_tree_traversal (files[x].path, callback, closure);
        }
    }


  for (x = 0; x < ar_size ((void *)files, 0, sizeof (*files)); ++x)
    {
      lim_free (0, files[x].name);
    }
  ar_free ((void **)&files, 0);
}


static int
cmp_traversal_files (void * va, void * vb, void * ign)
{
  struct arch_traversal_file * a = (struct arch_traversal_file *)va;
  struct arch_traversal_file * b = (struct arch_traversal_file *)vb;

  return str_cmp (a->path, b->path);
}



void
arch_id_program_compile_external (arch_id_program * prog)
{
  prog->check_external_tags = 1;
}

void
arch_id_program_compile_untagged_disposition (arch_id_program * prog,
                                              enum arch_inventory_category category)
{
  prog->untagged_source_category = category;
}


void
arch_id_program_compile_basename_embedded (arch_id_program * prog)
{
  prog->check_embedded_tags = 1;
  prog->check_basename_embedded_tags = 1;
}


void
arch_id_program_compile_embedded (arch_id_program * prog, t_uchar * regexp_source)
{
  regex_t * new_re;
  int reg_error;

  new_re = (regex_t *)ar_push ((void **)&prog->regexp_list, 0, sizeof (*new_re));
  mem_set0 ((void *)new_re, sizeof (*new_re));
  reg_error = regcomp (new_re, regexp_source, REG_EXTENDED | REG_NOSUB | REG_NEWLINE);

  if (reg_error)
    {
      char err_msg[256];

      regerror (reg_error, new_re, err_msg, sizeof (err_msg));
      safe_printfmt (2, "arch: illegal regexp in =tagging-method (%s)\n  regexp: %s\n", err_msg, regexp_source);
      exit (2);
    }
}

void
arch_free_id_program_data (arch_id_program * prog)
{
  int x;

  for (x = 0; x < ar_size ((void *)prog->regexp_list, 0, sizeof (regex_t)); ++x)
    {
      regfree (&prog->regexp_list[x]);
    }
}


t_uchar *
arch_inventory_id (enum arch_inventory_category * category_ret,
                   arch_id_program * prog, t_uchar * path, struct stat * opt_lstat)
{
  t_uchar * as_file = 0;
  t_uchar * dir = 0;
  t_uchar * dir_basename = 0;
  t_uchar * basename = 0;
  t_uchar * answer = 0;
  struct stat lstat;

  if (!path)
    return 0;

  as_file = file_name_from_directory (0, path);
  if (!as_file)
    return 0;

  basename = file_name_tail (0, as_file);
  dir = file_name_directory_file (0, as_file);
  if (!dir)
    dir = str_save (0, ".");

  dir_basename = file_name_tail (0, dir);

  if (opt_lstat)
    lstat = *opt_lstat;
  else
    safe_lstat (as_file, &lstat);


  /* Special control file tags
   *
   * If we're using external or embedded tags, arch control
   * files are treated specially.
   */

  if (prog->check_external_tags || prog->check_embedded_tags)
    {
      if (!str_cmp (dir_basename, ".arch-ids"))
        {
          /* Explicit tag files use their contents as tag, with the
           * prefix 'E'.
           */
          answer = explicit_tag (as_file, "E_", 0);
        }
      else if (!str_cmp (basename, ".arch-ids"))
        {
          /* Explicit tag file directories:
           */
          long amt;

          answer = str_alloc_cat (0, "D_", as_file);
          amt = smash_non_graphical (answer, str_length (answer));
          answer[amt] = 0;
        }
      else if (   !str_cmp_prefix ("./{arch}/", as_file)
               || !str_cmp ("./{arch}", as_file))
        {
          /* Paths beginning with "./{arch}" are tagged with their own
           * path name, with the prefix "A_".  The presumptions are that these
           * files never move, and that if a file is present, its contents are
           * invariant.
           */
          long amt;

          answer = str_alloc_cat (0, "A_", as_file);
          amt = smash_non_graphical (answer, str_length (answer));
          answer[amt] = 0;
        }
    }

  if (answer)
    goto done;


  /* Next, check for explicit tags (but not =default explicit tags, yet.
   */

  if (prog->check_external_tags)
    {
      t_uchar * id_file = 0;

      if (S_ISDIR (lstat.st_mode))
        {
          id_file = file_name_in_vicinity (0, as_file, ".arch-ids/=id");
        }
      else
        {
          id_file = file_name_in_vicinity (0, dir, ".arch-ids/");
          id_file = str_realloc_cat_many (0, id_file, basename, ".id", str_end);
        }

      answer = explicit_tag (id_file, "x_", 0);

      if (!answer)
        {
          /* Is there a .arch-ids/=all file? */

          lim_free (0, id_file);

          if (S_ISDIR (lstat.st_mode))
            {
              id_file = file_name_in_vicinity (0, as_file, ".arch-ids/=all");
            }
          else
            {
              id_file = file_name_in_vicinity (0, dir, ".arch-ids/=all");
            }

          answer = explicit_tag (id_file, "a_", (S_ISDIR (lstat.st_mode) ? (t_uchar *)"./." : basename));
        }

      lim_free (0, id_file);
    }

  if (answer)
    goto done;

  /* Next, check for embedded tags.
   */

  if (prog->check_embedded_tags)
    {
      answer = implicit_tag (prog, as_file);
    }

  if (answer)
    goto done;

  /* check for an =default tag
   */
  if (prog->check_external_tags)
    {
      t_uchar * id_file = 0;

      if (S_ISDIR (lstat.st_mode))
        {
          id_file = file_name_in_vicinity (0, as_file, ".arch-ids/=default");
        }
      else
        {
          id_file = file_name_in_vicinity (0, dir, ".arch-ids/=default");
        }

      answer = explicit_tag (id_file, "w_", (S_ISDIR (lstat.st_mode) ? (t_uchar *)"./." : basename));

      lim_free (0, id_file);
    }

  if (answer)
    goto done;

  /* check for a =dont-care tag
   */
  if (prog->check_embedded_tags)
    {
      t_uchar * id_file = 0;
      long amt;

      id_file = file_name_in_vicinity (0, dir, ".arch-ids/=dont-care");
      if (!safe_access (id_file, F_OK))
        {
          answer = str_alloc_cat (0, "k_", as_file);
        }
      else
        {
          answer = str_alloc_cat (0, "?_", as_file);
        }

      amt = smash_non_graphical (answer, str_length (answer));
      answer[amt] = 0;

      lim_free (0, id_file);
    }

  if (answer)
    goto done;

  /* apply an optional `names' tag
   */

  if (prog->untagged_source_category == arch_inventory_source)
    {
      answer = str_alloc_cat (0, "?", as_file);
    }


 done:

  lim_free (0, as_file);
  lim_free (0, dir);
  lim_free (0, basename);

  if (answer)
    *category_ret = arch_inventory_source;
  else
    *category_ret = prog->untagged_source_category;


  return answer;
}



static t_uchar *
explicit_tag (t_uchar * id_file, t_uchar * prefix, t_uchar * postfix)
{
  int errn;
  int id_fd;
  t_uchar * answer;
  char buf[1024];
  long amt;

  id_fd = vu_open (&errn, id_file, O_RDONLY, 0);

  if (id_fd < 0)
    return 0;

  answer = str_save (0, prefix);

  while (1)
    {
      t_uchar * eol;

      amt = vu_read_retry (&errn, id_fd, buf, sizeof (buf));

      if (amt < 0)
        {
          lim_free (0, answer);
          vu_close (&errn, id_fd);
          return 0;
        }

      if (!amt)
        break;

      eol = str_chr_index_n (buf, amt, '\n');
      if (!eol)
        {
          t_uchar * old_answer;
          amt = smash_non_graphical (buf, amt);
          old_answer = answer;
          answer = str_realloc_cat_n (0, answer, buf, amt);
        }
      else
        {
          t_uchar * old_answer;

          amt = eol - (t_uchar *)buf;
          amt = smash_non_graphical (buf, amt);
          old_answer = answer;
          answer = str_realloc_cat_n (0, answer, buf, amt);
        }
    }

  answer = str_realloc_cat (0, answer, (postfix ? postfix : (t_uchar *)""));

  vu_close (&errn, id_fd);
  return answer;
}

static t_uchar *
implicit_tag (struct arch_id_program * prog, t_uchar * file)
{
  t_uchar * answer = 0;
  int errn;
  t_uchar * basename = 0;
  int file_fd;
  char buf[1025];
  int amt;
  int line;
  int bottom;

  file_fd = vu_open (&errn, file, O_RDONLY, 0);
  if (file_fd < 0)
    return 0;

  if (prog->check_basename_embedded_tags)
    basename = file_name_tail (0, file);

  for (bottom = 1; bottom >= 0; --bottom)
    {
      if (!bottom)
        {
          if (0 > vu_lseek (&errn, file_fd, 0, SEEK_SET))
            {
              int ign;
            error_return:
              vu_close (&ign, file_fd);
              return 0;
            }
          amt = vu_read_retry (&errn, file_fd, buf, sizeof (buf) - 1);
          if (amt < 0)
            goto error_return;
        }
      else
        {
          struct stat file_stat_buf;
          char * x;

          if (0 > vu_fstat (&errn, file_fd, &file_stat_buf))
            goto error_return;
          if (file_stat_buf.st_size > sizeof (buf))
            amt = sizeof (buf);
          else
            continue;
          if (0 > vu_lseek (&errn, file_fd, -1026, SEEK_END))
            goto error_return;
          amt = vu_read_retry (&errn, file_fd, buf, sizeof (buf));
          if (amt < 0)
            goto error_return;
          x = str_chr_index_n (buf, amt, '\n');
          if (!x)
            continue;
          amt = amt - (1 + x - buf);
          mem_move (buf, x + 1, amt);
        }

      buf[amt] = 0;
      line = 0;

      while (1)
        {
          char * eol_pos;
          int eol;
          int x;

          /* preconditions:
           *
           * line <= amt
           * buf + line is the start of a line in the file
           * buf[amt] == 0
           */

          if (line >= amt)
            break;

          eol_pos = str_chr_index (buf + line, '\n');
          if (eol_pos)
            eol = eol_pos - buf;
          else
            eol = amt;


          /* Does this line match any of the tag regexps?
           */
          for (x = 0; x < ar_size ((void *)prog->regexp_list, 0, sizeof (regex_t)); ++x)
            {
              regmatch_t match_pos;
              regmatch_t * match_pos_addr = &match_pos;

              if (regnexec (&prog->regexp_list[x], buf + line, eol - line, (size_t)1, &match_pos_addr, 0))
                {
                  int start_of_tag;

                  /* found the tag.  It begins after any whitespace that follows
                   * the end of the match.
                   */

                  for (start_of_tag = line + match_pos.rm_eo;
                       (start_of_tag < eol) && char_is_blank (buf[start_of_tag]);
                       ++start_of_tag)
                    ;

                  {
                    long size;

                    size = smash_non_graphical (buf + start_of_tag, eol - start_of_tag);
                    answer = str_alloc_cat_n (0, "i_", buf + start_of_tag, size);
                    break;
                  }
                }
            }


          /* Does this line use an old-style basename tag?
           */
          if (prog->check_basename_embedded_tags)
            {
              int start_of_tag;

              start_of_tag = line;

              /* skip punctuation and blanks at the start of the line
               */
              while ((start_of_tag < eol) && (char_is_punct (buf[start_of_tag]) || char_is_blank (buf[start_of_tag])))
                ++start_of_tag;

              /* Looks dangerous but remember: buf[amt] == 0
               */
              if (str_cmp_prefix (basename, buf + start_of_tag))
                {
                  start_of_tag += str_length (basename);

                  while ((start_of_tag < eol) && char_is_blank (buf[line]))
                    ++line;

                  if ((start_of_tag < eol) && (buf[start_of_tag] == '-'))
                    {
                      long size;

                      ++start_of_tag;
                      size = smash_non_graphical (buf + start_of_tag, eol - start_of_tag);
                      answer = str_alloc_cat_n (0, "i_", buf + start_of_tag, size);
                      break;
                    }
                }
            }

          line = eol + 1;
        }
    }

  if (0 > vu_close (&errn, file_fd))
    goto error_return;

  lim_free (0, basename);

  return answer;
}

static long
smash_non_graphical (t_uchar * buf, long amt)
{
  long x;

  while (amt > 0)
    {
      if (!char_is_graph (buf[amt - 1]))
        --amt;
      else
        break;
    }

  for (x = 0; x < amt; ++x)
    {
      if (!char_is_graph (buf[x]))
        buf[x] = '_';
    }

  return amt;
}






/* tag: Tom Lord Tue Jul 22 13:17:12 2003 (inventory.c)
 */