path: root/riemann.fmi.uni-sofia.bg/programs/mkhtmlindex.c

/*
   mkhtmlindex.c 
   gcc -O2 -s -o mkhtmlindex mkhtmlindex.c -Wall -W
   diet -Os gcc -nostdinc -O2 -s -o mkhtmlindex mkhtmlindex.c -Wall -W

   large file support:
   gcc -O2 -DSUPPORT_64BIT_FILESIZE -s -o mkhtmlindex mkhtmlindex.c -Wall -W
   diet -Os gcc -nostdinc -O2 -DSUPPORT_64BIT_FILESIZE -s -o mkhtmlindex mkhtmlindex.c -Wall -W

   source: http://riemann.fmi.uni-sofia.bg/programs/
   binary: http://riemann.fmi.uni-sofia.bg/utils/
*/

#ifdef SUPPORT_64BIT_FILESIZE
#define _FILE_OFFSET_BITS 64
typedef unsigned long long humank_t;
#else 
typedef unsigned long humank_t;
#endif

#include <unistd.h>
#include <fcntl.h>
#include <time.h>
#include <string.h>
#include <errno.h>
#include <sys/stat.h>
#include <limits.h>
#include <dirent.h>
#include <alloca.h>
#include <stdlib.h>
#include <stdio.h>  /* gcc rename */
#define INC_HEADER	".#header"
#define INC_TAILER	".#tailer"
#define EXCLUDE		".#exclude"
#define IDX_TMP		".#index.html"
#define IDX_DENY	".#deny"

#ifndef __dietlibc__
#ifndef O_DIRECTORY
#define O_DIRECTORY	0200000	/* must be a directory */
#endif
#endif

typedef struct stralloc {
  char* s;
  unsigned long int len;
  unsigned long int a;
} stralloc;

unsigned int fmt_ulong(register char *s, register unsigned long u) {
  register unsigned int len; register unsigned long q;
  len = 1; q = u;
  while (q > 9) { ++len; q /= 10; }
  if (s) {
    s += len;
    do { *--s = '0' + (u % 10); u /= 10; } while(u); /* handles u == 0 */
  }
  return len;
}

unsigned int fmt_humank(char* dest, humank_t l) {
  char unit;
  int i;
  if (l<1000) return fmt_ulong(dest,l);
#ifdef SUPPORT_64BIT_FILESIZE
  if (l>1024*1024*1024*1024ull) {
    l=(l+(1024*1024*1024*1024ull/20))/(1024*1024*1024*1024ull/10);
    unit='T';
  } else 
#endif
  if (l>1024*1024*1024) {
    l=(l+(1024*1024*1024/20))/(1024*1024*1024/10);
    unit='G';
  } else if (l>1024*1024) {
    l=(l+(1024*1024/20))/(1024*1024/10);
    unit='M';
  } else {
    l=(l+(1024/20))/(1024/10);
    unit='k';
  }
  i=fmt_ulong(dest,l/10);
  if (i<2) {
    if (dest) dest[i]='.'; i++;
    if (dest) dest[i]=(l%10)+'0'; i++;
  }
  if (dest) dest[i]=unit; i++;
  return i;
}

struct tm *gmtime_r(const time_t *t, struct tm *tm) {
  unsigned long day, mon, year, yday=0, tod;

  tod = (unsigned long)(*t) % 86400;
  day = (unsigned long)(*t) / 86400;
  tm->tm_wday = ((day+4) % 7);
  tm->tm_sec  = tod%60; tod /= 60;
  tm->tm_min  = tod%60; 
  tm->tm_hour = tod/60;

  year = 4*day + 2;
  year /= 1461;

  day += 671;
  day %= 1461;           /* day 0 is march 1, 1972 */
  if (day < 306) yday = 1;
  if (day == 1460) { day = 365; yday = 59; }
  else { day %= 365;   yday += (day + 59) % 365; }

  day *= 10;
  mon = (day + 5) / 306;
  day = day + 5 - 306 * mon;
  day /= 10;
  if (mon >= 10) mon -= 10;
  else mon += 2;      /* day 0,1,30, mon 0..11,  year 1970=0,1,2 */

  tm->tm_year = year+70;
  tm->tm_mon  = mon;
  tm->tm_mday = day+1;
  tm->tm_yday = yday;
  return tm;
}

static int fmt_hour(char *dest, unsigned int h, unsigned int m) {
  dest[0] = '0' + (h/10);
  dest[1] = '0' + (h%10);
  dest[2] = ':';
  dest[3] = '0' + (m/10);
  dest[4] = '0' + (m%10);
  return 5;
}

typedef struct buffer {
  char *x;        /* actual buffer space */
  unsigned int p; /* current position */
  unsigned int n; /* string postition */
  unsigned int a; /* allocated buffer size */
  int fd;
  int (*op)(int,char *,unsigned int);
} buffer;

extern int safe_write(int fd,char *s,unsigned int len);
char out_space[4096];
buffer out_buf = { out_space,0,0,sizeof(out_space),1,safe_write };
buffer dir_buf = { out_space,0,0,sizeof(out_space),-1,safe_write };

static int buffer_flush(buffer* b) {
  register int p;
  if (!(p=b->p)) return 0; /* buffer already empty */
  b->p=0;
  if (p != b->op(b->fd,b->x,p)) return -1;
  return 0;
}

int buffer_put(buffer* b,char* buf,unsigned int len) {
  if (len>b->a-b->p) {	/* doesn't fit */
    if (buffer_flush(b)==-1) return -1;
    if (len>b->a) {
      if (len != (unsigned int)b->op(b->fd,buf,len)) return -1;
      return 0;
    }
  }
  memcpy(b->x+b->p, buf, len);
  b->p+=len;
  return 0;
}

int out(char *s) {return buffer_put(&out_buf,s, strlen(s));}
int outb(char* s,unsigned int len) {return buffer_put(&out_buf,s,len);}

int fd=-1,fd_err=1;
static char *months =  "JanFebMarAprMayJunJulAugSepOctNovDec";
static char *hex = "0123456789abcdef";

void err(char *s) 
  {write(fd_err,"error: ",7); write(fd_err,s,strlen(s)); _exit(1);}
int safe_write(int fd,char *s, unsigned int len) {
  int r;
  unsigned int u=len;
  while (u) {
    if ((r=write(fd,s,u))==0) break;
    if (r==-1) {if (errno==EINTR) continue; break;}
    s += r; u -= r;
  }
  if (u) err("write output\n");
  return len-u;
}
void outh(char *s) {
  for (; *s; s++) 
    switch (*s) {
    case '\"': out("&quot;"); break;
    case '&': out("&amp;"); break;
    case '>': out("&gt;"); break;
    case '<': out("&lt;"); break;
    default: outb(s,1);
    }
}
void outu(char *s) {
  char c, x[3];
  x[0] = '%';
  for (; *s; s++) {
    if ((c=*s)!='"' && c!='%' && c>=' ' && c!='+') {
      outb(s,1);
    } else {
      x[1] = hex[c>>4];
      x[2] = hex[c&15];
      outb(x,3);
    }
  }
}

int include_fd(int fd) {
  char buf[1024];
  int r;
  for (;;) {
    if ((r=read(fd,buf,sizeof buf))==0) break;
    if (r==-1) {if (errno==EINTR) continue; break;}
    outb(buf,r);
  }
  return r;
}

struct d {
  struct d *next;
  humank_t size;
  time_t time;
  char mode[4];
};

int byname(struct d** a,struct d** b) {
  return strcmp((char*)(*a)+sizeof(struct d),
		(char*)(*b)+sizeof(struct d));
}

int bysize(struct d** a,struct d** b) {
  if ((*a)->size>(*b)->size) return 1;
  else if ((*a)->size<(*b)->size) return -1;
  else return byname(a,b);
}

int bytime(struct d** a,struct d** b) {
  int res=(*b)->time-(*a)->time;
  if (res) return res;
  return byname(a,b);
}

int (*compare)(const void *a,const void *b)=(int (*)(const void *,const void *))byname;

void fmt_mode(char *x, mode_t mode) {
  memset(x,'-',4);
  if (S_ISLNK(mode)) x[0]='l';
  else if (S_ISDIR(mode)) x[0]='d';
  if ((mode&0444) == 0444) x[1]='r';
  if (mode&2) x[2]='w';
  if (mode&1) x[3]='x';
}

mode_t do_chmod(char *s, mode_t mode) {
  mode_t m=0644;
  if (S_ISDIR(mode)) m = 0755;
  else {
    char *x = strrchr(s, '.');
    if (x && !strcmp(x,".cgi")) m = 0711;
  }
  if ((mode & 07777) != m) {
    chmod(s, m);
    mode =  (mode & ~07777) | m;
  }
  return mode;
}

void print_link(char *s) {
  char link[512];
  int r=readlink(s,link,sizeof(link)-1);
  outh(" -> "); 
  if (r > 0) { link[r] = 0; outh(link); }
}

struct dirent* read_dir(buffer *d) {
  if (d->p >= d->n || 
      (d->p += ((struct dirent*)(d->x+d->p))->d_reclen) >= d->n) {
#ifdef __dietlibc__
    int res = getdents(d->fd, (struct dirent *)d->x, d->a-1);
#else 
    off_t x;
    int res = getdirentries(d->fd, d->x, d->a-1, &x);
#endif
    if (res<=0) return 0;
    d->n=res; d->p=0;
  }
  return (struct dirent*)(d->x+d->p);
}

struct d *do_qsort(struct d *root) {
  unsigned int k;
  struct d **x, **map, *tmp;
  if (!root) return root;

  for (k=0,tmp=root; tmp; tmp=tmp->next) k++;
  if (!(map = alloca((k+1)*sizeof(struct d*)))) err("alloca");
  for (x=map,tmp=root ;; tmp=tmp->next) { *x++ = tmp; if (!tmp) break; }

  qsort(map,k,sizeof(struct d*),compare);

  for (x=map; *x;) { tmp = *x++; tmp->next = *x; }
  return map[0];
}

static stralloc mem_block;
static void *alloc_a(stralloc *sa, unsigned long len) {
  len = (len + 3) & (~3);
  if (!sa->s || sa->len < len) return 0;
  sa->len -= len;
  return (void *)sa->s + sa->len;
}

int main(int argc, char **argv) {
  struct stat st;
  int ifd,i=0;
  struct tm *tm, tm_;
  time_t now;
  int flagdir;
  int flaglink;
  int flagmode=0;
  int flagdirent=0;
  int flagsort = 1;
  char x[16],*p;
  char *title=0;
  char **ee, **exclude=argv;
  int excl_len=0;
  struct dirent *de;
  struct d *tmp, *root_last=0, *root=0;

  for (optind=1; argv[optind] && argv[optind][0]=='-'; optind++) {
    switch (argv[optind][1]) {
    case 'm': flagmode=1; break;
    case 'o': fd=1; fd_err=2; break; 
    case 't': title=argv[optind]+2; break;
    case 'e': exclude[excl_len++]=argv[optind]+2; break;
    case 'd': flagdirent=1;
    case 's':
      switch (argv[optind][2]) {
      case 'n': compare=(int (*)(const void *,const void *))byname; break;
      case 't': compare=(int (*)(const void *,const void *))bytime; break;
      case 's': compare=(int (*)(const void *,const void *))bysize; break;
      case 'r': flagsort=0; break;
      }
      break;
    case 'h':
    default: goto usage;
    }
  }

  argc -= (optind);
  if (!argc && !flagdirent) {
  usage:
    err("usage: mkhtmlindex [-m] [-o] [-d] [-tTitle] [-sntsr] [-eName] [file(s)]\n"
	"\t-m chmod 644/711;  -o write to stdout;  "
	"-d read files from disk\n"
	"\t-sx sort by Name  Time  Size  Raw\n"
	"\t-ex.gz exclude x.gz;  "
	"similary if file '" EXCLUDE  "' exists\n"
	"\tif file '" INC_HEADER  "' exists it is included for header\n"
);
  }

  exclude[excl_len] = 0;
  if (!stat(EXCLUDE, &st)) {
    int fd,r;
    char *map,*p;
    if ((fd=open(EXCLUDE, O_RDONLY)) >= 0) {
      if (!(map = alloca(st.st_size + 1))) err("alloca");
      r=read(fd,map,st.st_size);
      close(fd);
      if (r>0) {
	map[r]=0;
	for (r=excl_len + 4, p=map; *p; ++p) if (*p == '\n') r++;
	if (!(exclude = alloca(i*sizeof(char *)))) err("alloca");

	memcpy(exclude , argv, excl_len*sizeof(char*));
	for (ee=exclude+excl_len,*ee=map,p=map; *p; p++)
	  if (*p == '\n') { *p=0; ++ee, *ee = p+1; }
	ee[1]=0;
      }
    }
  }

  argv += (optind); 

  if (flagdirent &&
      (dir_buf.fd=open(".", O_RDONLY | O_DIRECTORY))==-1) err("open '.'");
  for (i=0;;) {
    char *name=0;
    unsigned int len;
    if (flagdirent) {if ((de = read_dir(&dir_buf))) name = de->d_name;}
    else name = *argv++;
    if (name==0) break;
    if (name[0]=='.') continue;
    if (fd != 1 && !strcmp(name, "index.html")) continue;
    if (lstat(name,&st)) continue;
    for (ee=exclude; *ee; ee++)
      if (!strcmp(*ee,name)) break;
    if (*ee) continue;

    len = strlen(name) +1+ sizeof(struct d);
    if (!(tmp = alloc_a(&mem_block, len))) {
      mem_block.a = len + 2000;
      if (!(mem_block.s = alloca(mem_block.a))) err("alloca");
      mem_block.len = mem_block.a;
      tmp = alloc_a(&mem_block,len);  /* already allocated */
    }

    tmp->next = 0;
    if (!root) root = tmp;
    else root_last->next = tmp;
    root_last = tmp;

    strcpy((char*)tmp + sizeof(struct d), name);
    tmp->time = st.st_mtime;
    tmp->size = st.st_size;
    if (flagmode && (S_ISLNK(st.st_mode) == 0))
      st.st_mode = do_chmod(name, st.st_mode);
    fmt_mode(tmp->mode, st.st_mode);
    if (tmp->mode[0] == 'd') tmp->size=0;
  }
  if (flagdirent) close(dir_buf.fd);
  if (flagsort) root = do_qsort(root);

  if (fd == -1) {
    fd = open(IDX_TMP, O_WRONLY | O_CREAT | O_TRUNC, 0600);
    if (fd == -1) err(IDX_TMP "\n");
    out_buf.fd=fd;
  }
  ifd=open(INC_HEADER,O_RDONLY);

  if (title || ifd==-1) {
    if (title && !title[0]) title = 0;
    out("<html>\n<head><title>Index");
    if (title) {out(" of "); out(title);}
    out("</title></head>\n");
    out("<body bgcolor=\"#99cc99\" text=\"#000000\" link=\"#2020ff\" vlink=\"red\">");
    out("\n<h2>Index");
    if (title) {out(" of "); out(title);}
    out("</h2>\n<font>\n");
    out("Parent <a href=../>directory</a>\n");
  } else {
    if (include_fd(ifd)==-1) err("read " INC_HEADER);
    close(ifd);
  }
  
  out("<pre>mode   size  last-change   name<hr>\n");

  for (now=time(0),tmp=root; tmp; tmp=tmp->next) {
    flaglink= (tmp->mode[0] == 'l');
    flagdir = (tmp->mode[0] == 'd');

    outb(tmp->mode,4);

    memset(x,' ',sizeof x);
    if (!flaglink && !flagdir) {
      i = fmt_humank(0, tmp->size);
      fmt_humank(x+7-i, tmp->size);
    }
    outb(x,9);
    
    tm = gmtime_r((time_t*)&tmp->time, &tm_);
    memset(x,' ',sizeof x);  
    p = x+1;

    outb(months +3*tm->tm_mon, 3);
    if (tm->tm_mday<10) p++;
    p += fmt_ulong(p, tm->tm_mday) + 1;
      
    if (now - tmp->time >= 330*86400)
      p += 1 + fmt_ulong(p+1, 1900+tm->tm_year);
    else
      p += fmt_hour(p, tm->tm_hour, tm->tm_min);
    outb(x,11);

    p = (char*)tmp + sizeof(struct d);
    out("<a href=\""); 
    outu(p);
    if (flagdir) out("/index.html");
    out("\">");

    outh(p);
    if (flagdir) out("/");
    out("</a>");

    if (flaglink) print_link(p);
    out("\n");
  }

  ifd = open(INC_TAILER,O_RDONLY);
  if (ifd != -1) {
    if (include_fd(ifd)==-1) err("read " INC_TAILER);
    close(ifd);
  } else {
    char date[32];
    out("\nindex genereted on <b>");
    tm = gmtime_r(&now, &tm_);
    outb(months +3*tm->tm_mon,3);
    p=date+1;
    memset(date,' ',32);
    if (tm->tm_mday <10) ++p;
    p += 1+fmt_ulong(p, tm->tm_mday);
    p += 1+fmt_ulong(p, 1900+tm->tm_year);
    p += fmt_hour(p, tm->tm_hour, tm->tm_min);
    outb(date,p-date+1);
    out("GMT</b> by <a href=http://riemann.fmi.uni-sofia.bg/programs/mkhtmlindex.c>mkhtmlindex</a>");
    out("</pre>\n</font></body></html>\n");
  }
  
  buffer_flush(&out_buf);
  if (close(fd)) err("close\n");
  if (fd!=1) {
    if (chmod(IDX_TMP,0644)) err("chmod\n");
    if (open(IDX_DENY, O_RDONLY)==-1)
      if (rename(IDX_TMP, "index.html")) err("rename\n");
  }
  return(0);
}