Initial import

[maildir-archive.git] / maildir-archive.c

#include <stdlib.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <fcntl.h>
#include <dirent.h>
#include <stdio.h>
#include <time.h>
#include <string.h>
#include <errno.h>

#define ARCHIVE_PREFIX "Archive-"

char *months[] = { "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec" };
char fromfolder[512];
char fromdirectory[512];
int skipped = 0;

struct folder_list_t {
  char *name;
  int emails;
  struct folder_list_t *next;
};

struct folder_list_t *folderhead;

void move_email(const char *name, const char *folder)
{
  struct folder_list_t *lp;
  int foundfolder = 0;
  char oldfilename[512];
  char filename[512];
  
  for (lp = folderhead; lp->next; lp = lp->next)
    if (!strcmp(folder, lp->name))
      break;
  
  if (!lp->next)
  {
    int fd;
    
    lp = malloc(sizeof(*folderhead));
    lp->next = folderhead;
    lp->emails = 0;
    lp->name = strdup(folder);
    folderhead = lp;
    
    /* create the folder */
    sprintf(filename, ".INBOX.%s", folder);
    mkdir(filename, 0700);
    
    sprintf(filename, ".INBOX.%s/cur", folder);
    if (mkdir(filename, 0700) && (errno != EEXIST))
    {
      perror("mkdir(cur)");
      exit(1);
    } 

    sprintf(filename, ".INBOX.%s/new", folder);
    if (mkdir(filename, 0700) && (errno != EEXIST))
    {
      perror("mkdir(new)");
      exit(1);
    } 

    sprintf(filename, ".INBOX.%s/tmp", folder);
    if (mkdir(filename, 0700) && (errno != EEXIST))
    {
      perror("mkdir(tmp)");
      exit(1);
    } 
    
    sprintf(filename, ".INBOX.%s/maildirfolder", folder);
    if ((fd = open(filename, O_CREAT | O_WRONLY, 0700)) < 0)
    {
      perror("open(maildirfolder)");
      exit(1);
    }
    close(fd);
  }
  
  lp->emails++;
  
  sprintf(oldfilename, "%s/%s", fromdirectory, name);
  sprintf(filename, ".INBOX.%s/cur/%s", folder, name);
  
  /* skipped if it's the same thing, since the link won't catch it, and the unlink will kill it */
  if (!strcmp(oldfilename, filename))
  {
    skipped++;
    return;
  }

  if (link(oldfilename, filename) && (errno != EEXIST))
  {
    char theerror[1024];
    sprintf(theerror, "link(%s, %s)", oldfilename, filename);
    perror(theerror);
    exit(1);
  }
  
  if (unlink(oldfilename))
  {
    char theerror[1024];
    sprintf(theerror, "unlink(%s)", oldfilename);
    perror(theerror);
    exit(1);
  }
}

int spinnerstate = 0;
char spinner[] = "/-\\|";
#define SPINNERSTATES 4
#define MAXAGE 7*24*60*60

void main(int argc, char **argv)
{
  DIR *cur;
  struct dirent *de;
  off_t dirstart;
  int emails = 0, notemails = 0, toonew = 0, starttime = time(NULL);
  
  if (argc == 2)
  {
    strcpy(fromfolder, argv[1]);
    sprintf(fromdirectory, ".INBOX.%s/cur", fromfolder);
  } else if (argc == 1) {
    strcpy(fromdirectory, "cur");
    strcpy(fromfolder, "");
  } else {
    printf("maildir-archiver: Usage: %s [<folder name>]\n", argv[0]);
    exit(1);
  }
  
  folderhead = malloc(sizeof(*folderhead));
  folderhead->next = NULL;
  
  printf("maildir-archiver: Opening %s/... ", fromdirectory);
  fflush(stdout);
  cur = opendir(fromdirectory);
  if (!cur)
  {
    perror("opendir(cur)");
    exit(1);
  }
  dirstart = telldir(cur);
  printf("done\n");
  
  printf("maildir-archiver: Doing preliminary scan of e-mails... .");
  fflush(stdout);
  while ((de = readdir(cur)))
  {
    time_t tt;
    struct tm *tm;
    char archivefolder[128];
    
    tt = atoi(de->d_name);
    if (tt == 0)
    {
      notemails++;
      continue;
    }
    if ((starttime - tt) < MAXAGE)
    {
      toonew++;
      continue;
    }
    emails++;
    if ((emails % 10000) == 0)
    {
      printf("\b%c", spinner[(spinnerstate++) % SPINNERSTATES]);
      fflush(stdout);
    }
  }
  printf("\bdone (%d mails to archive, %d too new, %d invalid)\n", emails, toonew, notemails);
  seekdir(cur, dirstart);
  
  printf("maildir-archiver: Archiving e-mail... [>                               ]");
  fflush(stdout);
  while ((de = readdir(cur)))
  {
    static int progresscounter = 0;
    static int archived = 0;
    time_t tt;
    struct tm *tm;
    char archivefolder[128];
    int islist = 0;
    char fn[256];
    FILE *fp;
    
    tt = atoi(de->d_name);
    if (tt == 0)
      continue;
    if ((starttime - tt) < MAXAGE)
      continue;
    
    tm = localtime(&tt);
    
    /* Open the file to determine if it's a mailing list or what */
    sprintf(fn, "%s/%s", fromdirectory, de->d_name);
    fp = fopen(fn, "r");
    if (!fp)
    {
      perror("fopen(fn, r)");
      exit(1);
    }
    {
      char line[512];
      while (fgets(line, 512, fp))
      {
        if (!strncmp(line, "X-BeenThere: ", 13))
          islist++;
        if (!strncmp(line, "List-Unsubscribe: ", 18))
          islist++;
        if (!strncmp(line, "X-Mailing-List", 14))
          islist++;
        if (!strncmp(line, "Precedence: bulk", 16))
          islist++;
        
        if (!line[0])
          break;
      }
    }
    fclose(fp);
    
    /* sprintf(archivefolder, ARCHIVE_PREFIX "%d-%02d", 1900 + tm->tm_year, tm->tm_mon + 1); */
    sprintf(archivefolder, ARCHIVE_PREFIX "%d-Q%d-%s", 1900 + tm->tm_year, (tm->tm_mon) / 3 + 1, islist ? "List" : "Personal");
    move_email(de->d_name, archivefolder);
    archived++;
    
    if (((archived % 10) == 0) || (archived == emails))
    {
      int i;
      progresscounter = archived * 32 / emails;
      printf("\rmaildir-archiver: Archiving e-mail... [");
      for (i=0; i < progresscounter; i++)
        printf("=");
      if (progresscounter != 32)
        printf(">");
      for (i=0; i < (31 - progresscounter); i++)
        printf(" ");
      printf("] ");
      printf("%c ", spinner[(spinnerstate++) % SPINNERSTATES]);
      printf("%d/%d", archived, emails);
      fflush(stdout);
    }
  }
  printf("\n");
  printf("maildir-archiver: Statistics:\n");
  printf("maildir-archiver: All operations performed in %d seconds\n", time(NULL)-starttime);
  printf("maildir-archiver:\n");
  printf("maildir-archiver: -------------------------------+----------\n");
  printf("maildir-archiver: %-30s | Emails\n", "Status");
  printf("maildir-archiver: -------------------------------+----------\n");
  printf("maildir-archived: %-30s | %6d\n", "Processed", emails);
  printf("maildir-archived: %-30s | %6d\n", "Too new", toonew);
  printf("maildir-archived: %-30s | %6d\n", "Not an e-mail", notemails);
  printf("maildir-archived: %-30s | %6d\n", "Already in correct folder", skipped);
  printf("maildir-archiver: -------------------------------+----------\n");
  printf("maildir-archiver:\n");
  printf("maildir-archiver: -------------------------------+----------\n");
  printf("maildir-archiver: %-30s | Emails\n", "Folder");
  printf("maildir-archiver: -------------------------------+----------\n");
  {
    struct folder_list_t *lp;
    int total = 0;
    
    for (lp = folderhead; lp->next; lp = lp->next)
    {
      printf("maildir-archiver: %-30s | %6d\n", lp->name, lp->emails);
      total += lp->emails;
    }
    printf("maildir-archiver: %-30s | %6d\n", "Total", total);
  }
  printf("maildir-archiver: -------------------------------+----------\n");
}