[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");
}
Commit	Line	Data
38b6fc57 JW	1	#include <stdlib.h>
	2	#include <sys/stat.h>
	3	#include <sys/types.h>
	4	#include <fcntl.h>
	5	#include <dirent.h>
	6	#include <stdio.h>
	7	#include <time.h>
	8	#include <string.h>
	9	#include <errno.h>
	10
	11	#define ARCHIVE_PREFIX "Archive-"
	12
	13	char *months[] = { "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec" };
	14	char fromfolder[512];
	15	char fromdirectory[512];
	16	int skipped = 0;
	17
	18	struct folder_list_t {
	19	char *name;
	20	int emails;
	21	struct folder_list_t *next;
	22	};
	23
	24	struct folder_list_t *folderhead;
	25
	26	void move_email(const char name, const char folder)
	27	{
	28	struct folder_list_t *lp;
	29	int foundfolder = 0;
	30	char oldfilename[512];
	31	char filename[512];
	32
	33	for (lp = folderhead; lp->next; lp = lp->next)
	34	if (!strcmp(folder, lp->name))
	35	break;
	36
	37	if (!lp->next)
	38	{
	39	int fd;
	40
	41	lp = malloc(sizeof(*folderhead));
	42	lp->next = folderhead;
	43	lp->emails = 0;
	44	lp->name = strdup(folder);
	45	folderhead = lp;
	46
	47	/* create the folder */
	48	sprintf(filename, ".INBOX.%s", folder);
	49	mkdir(filename, 0700);
	50
	51	sprintf(filename, ".INBOX.%s/cur", folder);
	52	if (mkdir(filename, 0700) && (errno != EEXIST))
	53	{
	54	perror("mkdir(cur)");
	55	exit(1);
	56	}
	57
	58	sprintf(filename, ".INBOX.%s/new", folder);
	59	if (mkdir(filename, 0700) && (errno != EEXIST))
	60	{
	61	perror("mkdir(new)");
	62	exit(1);
	63	}
	64
65	sprintf(filename, ".INBOX.%s/tmp", folder);
66	if (mkdir(filename, 0700) && (errno != EEXIST))
67	{
68	perror("mkdir(tmp)");
69	exit(1);
70	}
71
72	sprintf(filename, ".INBOX.%s/maildirfolder", folder);
73	if ((fd = open(filename, O_CREAT \| O_WRONLY, 0700)) < 0)
74	{
75	perror("open(maildirfolder)");
76	exit(1);
77	}
78	close(fd);
79	}
80
81	lp->emails++;
82
83	sprintf(oldfilename, "%s/%s", fromdirectory, name);
84	sprintf(filename, ".INBOX.%s/cur/%s", folder, name);
85
86	/* skipped if it's the same thing, since the link won't catch it, and the unlink will kill it */
87	if (!strcmp(oldfilename, filename))
88	{
89	skipped++;
90	return;
91	}
92
93	if (link(oldfilename, filename) && (errno != EEXIST))
94	{
95	char theerror[1024];
96	sprintf(theerror, "link(%s, %s)", oldfilename, filename);
97	perror(theerror);
98	exit(1);
99	}
100
101	if (unlink(oldfilename))
102	{
103	char theerror[1024];
104	sprintf(theerror, "unlink(%s)", oldfilename);
105	perror(theerror);
106	exit(1);
107	}
108	}
109
110	int spinnerstate = 0;
111	char spinner[] = "/-\\\|";
112	#define SPINNERSTATES 4
113	#define MAXAGE 72460*60
114
115	void main(int argc, char **argv)
116	{
117	DIR *cur;
118	struct dirent *de;
119	off_t dirstart;
120	int emails = 0, notemails = 0, toonew = 0, starttime = time(NULL);
121
122	if (argc == 2)
123	{
124	strcpy(fromfolder, argv[1]);
125	sprintf(fromdirectory, ".INBOX.%s/cur", fromfolder);
126	} else if (argc == 1) {
127	strcpy(fromdirectory, "cur");
128	strcpy(fromfolder, "");
129	} else {
130	printf("maildir-archiver: Usage: %s [<folder name>]\n", argv[0]);
131	exit(1);
132	}
133
134	folderhead = malloc(sizeof(*folderhead));
135	folderhead->next = NULL;
136
137	printf("maildir-archiver: Opening %s/... ", fromdirectory);
138	fflush(stdout);
139	cur = opendir(fromdirectory);
140	if (!cur)
141	{
142	perror("opendir(cur)");
143	exit(1);
144	}
145	dirstart = telldir(cur);
146	printf("done\n");
147
148	printf("maildir-archiver: Doing preliminary scan of e-mails... .");
149	fflush(stdout);
150	while ((de = readdir(cur)))
151	{
152	time_t tt;
153	struct tm *tm;
154	char archivefolder[128];
155
156	tt = atoi(de->d_name);
157	if (tt == 0)
158	{
159	notemails++;
160	continue;
161	}
162	if ((starttime - tt) < MAXAGE)
163	{
164	toonew++;
165	continue;
166	}
167	emails++;
168	if ((emails % 10000) == 0)
169	{
170	printf("\b%c", spinner[(spinnerstate++) % SPINNERSTATES]);
171	fflush(stdout);
172	}
173	}
174	printf("\bdone (%d mails to archive, %d too new, %d invalid)\n", emails, toonew, notemails);
175	seekdir(cur, dirstart);
176
177	printf("maildir-archiver: Archiving e-mail... [> ]");
178	fflush(stdout);
179	while ((de = readdir(cur)))
180	{
181	static int progresscounter = 0;
182	static int archived = 0;
183	time_t tt;
184	struct tm *tm;
185	char archivefolder[128];
186	int islist = 0;
187	char fn[256];
188	FILE *fp;
189
190	tt = atoi(de->d_name);
191	if (tt == 0)
192	continue;
193	if ((starttime - tt) < MAXAGE)
194	continue;
195
196	tm = localtime(&tt);
197
198	/* Open the file to determine if it's a mailing list or what */
199	sprintf(fn, "%s/%s", fromdirectory, de->d_name);
200	fp = fopen(fn, "r");
201	if (!fp)
202	{
203	perror("fopen(fn, r)");
204	exit(1);
205	}
206	{
207	char line[512];
208	while (fgets(line, 512, fp))
209	{
210	if (!strncmp(line, "X-BeenThere: ", 13))
211	islist++;
212	if (!strncmp(line, "List-Unsubscribe: ", 18))
213	islist++;
214	if (!strncmp(line, "X-Mailing-List", 14))
215	islist++;
216	if (!strncmp(line, "Precedence: bulk", 16))
217	islist++;
218
219	if (!line[0])
220	break;
221	}
222	}
223	fclose(fp);
224
225	/* sprintf(archivefolder, ARCHIVE_PREFIX "%d-%02d", 1900 + tm->tm_year, tm->tm_mon + 1); */
226	sprintf(archivefolder, ARCHIVE_PREFIX "%d-Q%d-%s", 1900 + tm->tm_year, (tm->tm_mon) / 3 + 1, islist ? "List" : "Personal");
227	move_email(de->d_name, archivefolder);
228	archived++;
229
230	if (((archived % 10) == 0) \|\| (archived == emails))
231	{
232	int i;
233	progresscounter = archived * 32 / emails;
234	printf("\rmaildir-archiver: Archiving e-mail... [");
235	for (i=0; i < progresscounter; i++)
236	printf("=");
237	if (progresscounter != 32)
238	printf(">");
239	for (i=0; i < (31 - progresscounter); i++)
240	printf(" ");
241	printf("] ");
242	printf("%c ", spinner[(spinnerstate++) % SPINNERSTATES]);
243	printf("%d/%d", archived, emails);
244	fflush(stdout);
245	}
246	}
247	printf("\n");
248	printf("maildir-archiver: Statistics:\n");
249	printf("maildir-archiver: All operations performed in %d seconds\n", time(NULL)-starttime);
250	printf("maildir-archiver:\n");
251	printf("maildir-archiver: -------------------------------+----------\n");
252	printf("maildir-archiver: %-30s \| Emails\n", "Status");
253	printf("maildir-archiver: -------------------------------+----------\n");
254	printf("maildir-archived: %-30s \| %6d\n", "Processed", emails);
255	printf("maildir-archived: %-30s \| %6d\n", "Too new", toonew);
256	printf("maildir-archived: %-30s \| %6d\n", "Not an e-mail", notemails);
257	printf("maildir-archived: %-30s \| %6d\n", "Already in correct folder", skipped);
258	printf("maildir-archiver: -------------------------------+----------\n");
259	printf("maildir-archiver:\n");
260	printf("maildir-archiver: -------------------------------+----------\n");
261	printf("maildir-archiver: %-30s \| Emails\n", "Folder");
262	printf("maildir-archiver: -------------------------------+----------\n");
263	{
264	struct folder_list_t *lp;
265	int total = 0;
266
267	for (lp = folderhead; lp->next; lp = lp->next)
268	{
269	printf("maildir-archiver: %-30s \| %6d\n", lp->name, lp->emails);
270	total += lp->emails;
271	}
272	printf("maildir-archiver: %-30s \| %6d\n", "Total", total);
273	}
274	printf("maildir-archiver: -------------------------------+----------\n");
275	}