X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=finddup.c;h=8f2b6c9e07396f279d8fd343b52c216da7acf911;hb=ab7b6e26f35ac1dfc88d9bf1e09dd289a30ea782;hp=0d5d5afd039d472c3c11010142d491371c218689;hpb=003e2e7bd441526160ec1d1ab68c685f1aeb39db;p=finddup.git

diff --git a/finddup.c b/finddup.c
index 0d5d5af..8f2b6c9 100644
--- a/finddup.c
+++ b/finddup.c
@@ -1,7 +1,8 @@
 
 /*
- *  finddup is a simple utility to display the files and directories
- *  according to their total disk occupancy.
+ *  finddup is a simple utility to find duplicated files, files common
+ *  to several directories, or files present in one directory and not
+ *  in another one.
  *
  *  Copyright (c) 2010 Francois Fleuret
  *  Written by Francois Fleuret <francois@fleuret.org>
@@ -22,10 +23,13 @@
  *
  */
 
+#define VERSION_NUMBER "0.9"
+
 #define _BSD_SOURCE
 
 #include <sys/types.h>
 #include <sys/stat.h>
+#include <sys/param.h>
 #include <dirent.h>
 #include <stdlib.h>
 #include <stdio.h>
@@ -37,7 +41,10 @@
 #include <getopt.h>
 #include <fcntl.h>
 
-#define BUFFER_SIZE 4096
+/* 1M really helps compared to 64k */
+#define READ_BUFFER_SIZE (1024 * 1024)
+
+#define PROGRESS_BUFFER_SIZE 1024
 
 typedef int64_t size_sum_t;
 
@@ -46,19 +53,23 @@ typedef int64_t size_sum_t;
 int ignore_dotfiles = 0; /* 1 means ignore files and directories
                             starting with a dot */
 
-int forced_width = 0; /* -1 means no width limit, strictly positive
-                         means limit, 0 means not active */
+int ignore_empty_files = 0; /* 1 means ignore empty files */
 
-int forced_height = 0; /* -1 means no height limit, strictly positive
-                           means limit, 0 means not active */
+int show_realpaths = 0; /* 1 means ignore files and directories
+                            starting with a dot */
 
-int fancy_size_display = 0; /* 1 means to use floating values with K, M and G
-                               as units */
+int show_progress = 0; /* 1 means show a progress bar when we are in a
+                          tty */
 
-int reverse_sorting = 0; /* 1 means to show the large ones first */
+int show_hits = 1; /* 1 means to show the files from dir2
+                      corresponding to the ones from dir1 */
 
-int show_top = 0; /* 1 means to show the top of the sorted list
-                     instead of the bottom */
+int show_groups = 1; /* 1 means to show the group IDs when printing
+                        file names */
+
+int same_inodes_are_different = 0; /* 1 means that comparison between
+                                      two files with same inode will
+                                      always be false */
 
 /********************************************************************/
 
@@ -79,40 +90,123 @@ int ignore_entry(const char *name) {
   return
     strcmp(name, ".") == 0 ||
     strcmp(name, "..") == 0 ||
-    (ignore_dotfiles && name[0] == '.');
+    (ignore_dotfiles && name[0] == '.' && name[1] != '/');
+}
+
+/**********************************************************************/
+
+struct file_node {
+  struct file_node *next;
+  char *name;
+  size_t size;
+  ino_t inode;
+  int group_id; /* one per identical file content */
+  int dir_id; /* 1 for DIR1, and 2 for DIR2 */
+};
+
+void file_list_delete(struct file_node *head) {
+  struct file_node *next;
+  while(head) {
+    next = head->next;
+    free(head->name);
+    free(head);
+    head = next;
+  }
 }
 
-void print_size_sum(size_sum_t s) {
-  char tmp[100];
-  char *a = tmp + sizeof(tmp)/sizeof(char);
-  *(--a) = '\0';
-  if(s) {
-    while(s) {
-      *(--a) = s%10 + '0';
-      s /= 10;
+int file_list_length(struct file_node *head) {
+  int l = 0;
+  while(head) {
+    l++;
+    head = head->next;
+  }
+  return l;
+}
+
+/**********************************************************************/
+
+int same_content(struct file_node *f1, struct file_node *f2,
+                 char *buffer1, char *buffer2) {
+  int fd1, fd2, s1, s2;
+
+  fd1 = open(f1->name, O_RDONLY);
+  fd2 = open(f2->name, O_RDONLY);
+
+  if(fd1 >= 0 && fd2 >= 0) {
+    while(1) {
+      s1 = read(fd1, buffer1, READ_BUFFER_SIZE);
+      s2 = read(fd2, buffer2, READ_BUFFER_SIZE);
+
+      if(s1 < 0 || s2 < 0) {
+        close(fd1);
+        close(fd2);
+        return 0;
+      }
+
+      if(s1 == s2) {
+        if(s1 == 0) {
+          close(fd1);
+          close(fd2);
+          return 1;
+        } else {
+          if(memcmp(buffer1, buffer2, s1)) {
+            /* printf("size_to_read = %d\n", size_to_read); */
+            close(fd1);
+            close(fd2);
+            return 0;
+          }
+        }
+      } else {
+        fprintf(stderr,
+                "Different read size without error on files of same size.\n");
+        exit(EXIT_FAILURE);
+      }
     }
   } else {
-    *(--a) = '0';
+
+    if(fd1 < 0) {
+      fprintf(stderr,
+              "Can not open \"%s\" error: %s\n",
+              f1->name,
+              strerror(errno));
+    }
+
+    if(fd2 < 0) {
+      fprintf(stderr,
+              "Can not open \"%s\" error: %s\n",
+              f2->name,
+              strerror(errno));
+    }
+
+    exit(EXIT_FAILURE);
   }
-  printf(a);
 }
 
-size_sum_t file_or_dir_size(const char *name) {
+int same_files(struct file_node *f1, struct file_node *f2,
+               char *buffer1, char *buffer2) {
+  if(same_inodes_are_different && f1->inode == f2->inode) {
+    return 0;
+  }
+
+  return f1->size == f2->size && same_content(f1, f2, buffer1, buffer2);
+}
+
+/**********************************************************************/
+
+struct file_node *scan_directory(struct file_node *tail, const char *name) {
   DIR *dir;
   struct dirent *dir_e;
-  struct stat dummy;
-  size_sum_t result;
-  char subname[BUFFER_SIZE];
-
-  result = 0;
+  struct stat sb;
+  struct file_node *tmp;
+  char subname[PATH_MAX + 1];
 
-  if(lstat(name, &dummy) != 0) {
-    printf("Can not stat %s: %s\n", name, strerror(errno));
+  if(lstat(name, &sb) != 0) {
+    fprintf(stderr, "Can not stat \"%s\": %s\n", name, strerror(errno));
     exit(EXIT_FAILURE);
   }
 
-  if(S_ISLNK(dummy.st_mode)) {
-    return 0;
+  if(S_ISLNK(sb.st_mode)) {
+    return tail;
   }
 
   dir = opendir(name);
@@ -120,206 +214,388 @@ size_sum_t file_or_dir_size(const char *name) {
   if(dir) {
     while((dir_e = readdir(dir))) {
       if(!ignore_entry(dir_e->d_name)) {
-        snprintf(subname, BUFFER_SIZE, "%s/%s", name, dir_e->d_name);
-        result += file_or_dir_size(subname);
+        snprintf(subname, PATH_MAX, "%s/%s", name, dir_e->d_name);
+        tail = scan_directory(tail, subname);
       }
     }
     closedir(dir);
   } else {
-    if(S_ISREG(dummy.st_mode)) {
-      result += dummy.st_size;
+    if(S_ISREG(sb.st_mode)) {
+      if(!ignore_entry(name)) {
+        if(!ignore_empty_files || sb.st_size > 0) {
+          tmp = safe_malloc(sizeof(struct file_node));
+          tmp->next = tail;
+          tmp->name = strdup(name);
+          tmp->size = sb.st_size;
+          tmp->inode = sb.st_ino;
+          tmp->group_id = -1;
+          tmp->dir_id = -1;
+          tail = tmp;
+        }
+      }
     }
   }
 
-  return result;
+  return tail;
 }
 
-/**********************************************************************/
+void print_file(struct file_node *node) {
+  char tmp[PATH_MAX + 1];
+  if(show_realpaths) {
+    if(realpath(node->name, tmp)) {
+      if(show_groups) {
+        printf("%d %s\n", node->group_id, tmp);
+      } else {
+        printf("%s\n", tmp);
+      }
+    } else {
+      printf("Can not get the realpath of \"%s\": %s\n",
+             node->name,
+             strerror(errno));
+      exit(EXIT_FAILURE);
+    }
+  } else {
+    if(show_groups) {
+      printf("%d %s\n", node->group_id, node->name);
+    } else {
+      printf("%s\n", node->name);
+    }
+  }
+}
 
-struct file_with_size {
-  char *filename;
-  size_t size;
-  struct file_with_size *next;
-  int error;
-};
+int compare_nodes(const void *x1, const void *x2) {
+  const struct file_node **f1, **f2;
 
-struct file_list {
-  struct file_with_size *head, *tail;
-};
+  f1 = (const struct file_node **) x1;
+  f2 = (const struct file_node **) x2;
 
-void merge(struct *file_list list1, struct *file_list list2) {
-  if(list1->head && list2->head) {
-    struct file_list result;
-    result.head = list1.head;
+  if((*f1)->group_id < (*f2)->group_id) {
+    return -1;
+  } else if((*f1)->group_id > (*f2)->group_id) {
+    return 1;
   } else {
-    if(list1.head) {
-      return list1;
+    if((*f1)->dir_id < (*f2)->dir_id) {
+      return -1;
+    } else if((*f1)->dir_id > (*f2)->dir_id) {
+      return 1;
     } else {
-      return list2;
+      return 0;
     }
   }
 }
 
-struct file_with_size *create(char *name, struct file_with_size *current) {
-  struct file_with_size *result;
-  result = safe_malloc(sizeof(struct file_with_size));
-  result->filename = strdup(name);
-  result->size = file_or_dir_size(name);
-  result->next = current;
-  return result;
-}
 
-void destroy(struct file_with_size *node) {
-  struct file_with_size *next;
-  while(node) {
-    next = node->next;
-    free(node->filename);
-    free(node);
-    node = next;
-  }
-}
+void print_result(struct file_node *list1, struct file_node *list2) {
+  struct file_node *node1, *node2;
+  struct file_node **nodes;
+  int nb, n;
 
-/**********************************************************************/
+  nb = 0;
+  for(node1 = list1; node1; node1 = node1->next) {
+    if(node1->group_id >= 0) { nb++; }
+  }
 
-int same_size(struct file_with_size *f1, struct file_with_size *f2) {
-  return f1->size == f2->size;
-}
+  if(list2) {
+    for(node2 = list2; node2; node2 = node2->next) {
+      if(node2->group_id >= 0) { nb++; }
+    }
+  }
 
-int same_content(struct file_with_size *f1, struct file_with_size *f2) {
-  int fd1, fd2, eq, s1, s2;
-  char buffer1[BUFFER_SIZE], buffer2[BUFFER_SIZE];
+  nodes = safe_malloc(nb * sizeof(struct file_node *));
 
-  fd1 = open(f1->filename, O_RDONLY);
-  fd2 = open(f2->filename, O_RDONLY);
-  eq = 1;
+  n = 0;
+  for(node1 = list1; node1; node1 = node1->next) {
+    if(node1->group_id >= 0) {
+      nodes[n++] = node1;
+    }
+  }
 
-  if(fd1 >= 0 && fd2 >= 0) {
-    while(eq) {
-      s1 = read(fd1, buffer1, BUFFER_SIZE);
-      if(s1 < 0) {
-        f1->error = 1;
-        eq = 0;
-      }
-      s2 = read(fd2, buffer2, BUFFER_SIZE);
-      if(s2 < 0) {
-        f2->error = 1;
-        eq = 0;
-      }
-      if(s1 > 0 && s2 > 0) {
-        if(s1 == s2) {
-          eq = (strncmp(buffer1, buffer2, s1) == 0);
-        } else {
-          fprintf(stderr,
-                  "Different read size without error on files of same size.\n");
-          exit(EXIT_FAILURE);
-        }
+  if(list2) {
+    for(node2 = list2; node2; node2 = node2->next) {
+      if(node2->group_id >= 0) {
+        nodes[n++] = node2;
       }
     }
   }
 
-  if(fd1 >= 0) {
-    close(fd1);
-  } else {
-    f1->error = 1;
-    eq = 0;
-  }
+  qsort(nodes, nb, sizeof(struct file_node *), compare_nodes);
 
-  if(fd2 >= 0) {
-    close(fd2);
-  } else {
-    f2->error = 1;
-    eq = 0;
+  for(n = 0; n < nb; n++) {
+    if(!show_groups && n > 0 && nodes[n]->group_id != nodes[n-1]->group_id) {
+      printf("\n");
+    }
+    print_file(nodes[n]);
   }
 
-  return eq;
+  free(nodes);
 }
 
-int same_files(struct file_with_size *f1, struct file_with_size *f2) {
-  return same_size(f1, f2) && same_content(f1, f2);
+struct progress_state {
+  int bar_width;
+  int nb_values, value;
+  int last_position;
+};
+
+void print_progress(struct progress_state *state) {
+  int position, k;
+  struct winsize win;
+  char buffer[PROGRESS_BUFFER_SIZE];
+  char *s;
+
+  if(show_progress) {
+    /* We use the previous bar_width to compute the position, so that
+       we avoid doing too many ioctls */
+    position = (state->bar_width * state->value) / (state->nb_values - 1);
+    if(state->bar_width <= 0 || position != state->last_position) {
+      if(!ioctl (STDERR_FILENO, TIOCGWINSZ, (char *) &win)) {
+        /* Something weird is going on if the previous test is wrong */
+        if(win.ws_col >= PROGRESS_BUFFER_SIZE - 3) {
+          state->bar_width = PROGRESS_BUFFER_SIZE - 10;
+        } else {
+          state->bar_width = win.ws_col - 7;
+        }
+        position = (state->bar_width * state->value) / (state->nb_values - 1);
+        state->last_position = position;
+        s = buffer;
+        for(k = 0; k < position; k++) {
+          *(s++) = '+';
+        }
+        for(; k < state->bar_width; k++) {
+          *(s++) = '-';
+        }
+
+        /* We need four % because of the fprintf that follows */
+        sprintf(s, " [%3d%%%%]\r",
+                (100 * state->value) / (state->nb_values - 1));
+
+        fprintf(stderr, buffer);
+      }
+    }
+  }
 }
 
-/**********************************************************************/
+void start(const char *dirname1, const char *dirname2) {
+  struct file_node *list1, *list2;
+  struct file_node *node1, *node2;
+  struct progress_state progress_state;
+  int not_in, found;
+  int nb_groups, nb_nodes;
+  int list1_length, list2_length, previous_progress;
 
-struct file_list scan_directory(const char *name) {
-  DIR *dir;
-  struct dirent *dir_e;
-  struct file_list result, sublist;
-  struct stat dummy;
-  char subname[BUFFER_SIZE];
+  char *buffer1 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
+  char *buffer2 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
 
-  result.head = 0;
-  result.tail = 0;
+  not_in = 0;
 
-  if(lstat(name, &dummy) != 0) {
-    fprintf(stderr, "Can not stat %s: %s\n", name, strerror(errno));
-    exit(EXIT_FAILURE);
+  if(show_progress) {
+    fprintf(stderr, "Scanning '%s' ... ", dirname1);
   }
 
-  if(S_ISLNK(dummy.st_mode)) {
-    return result;
+  list1 = scan_directory(0, dirname1);
+
+  list1_length = file_list_length(list1);
+
+  if(show_progress) {
+    fprintf(stderr, "done (%d file%s).\n",
+            list1_length, (list1_length > 1 ? "s" : ""));
   }
 
-  dir = opendir(name);
+  if(dirname2) {
+    if(strncmp(dirname2, "not:", 4) == 0) {
+      not_in = 1;
+      /* groups are not computed in the not: mode */
+      show_groups = 0;
+      dirname2 += 4;
+    } else if(strncmp(dirname2, "and:", 4) == 0) {
+      dirname2 += 4;
+    }
+    if(show_progress) {
+      fprintf(stderr, "Scanning '%s' ... ", dirname2);
+    }
+    list2 = scan_directory(0, dirname2);
+    if(show_progress) {
+      list2_length = file_list_length(list2);
+      fprintf(stderr, "done (%d file%s).\n",
+              list2_length, (list2_length > 1 ? "s" : ""));
+    }
+  } else {
+    list2 = list1;
+  }
 
-  if(dir) {
-    while((dir_e = readdir(dir))) {
-      if(!ignore_entry(dir_e->d_name)) {
-        snprintf(subname, BUFFER_SIZE, "%s/%s", name, dir_e->d_name);
-        sublist = scan_directory(subname);
-        result = concat(result, sublist);
+  if(show_progress) {
+    fprintf(stderr, "Now looking for identical files (this may take a while).\n");
+  }
+
+  nb_groups = 0;
+  previous_progress = -1;
+  nb_nodes = 0;
+
+  progress_state.bar_width = -1;
+  progress_state.last_position = -1;
+  progress_state.nb_values = list1_length;
+
+  if(not_in) {
+    for(node1 = list1; node1; node1 = node1->next) {
+      progress_state.value = nb_nodes;
+      print_progress(&progress_state);
+      nb_nodes++;
+
+      found = 0;
+
+      for(node2 = list2; !found && node2; node2 = node2->next) {
+        if(same_files(node1, node2, buffer1, buffer2)) {
+          found = 1;
+        }
+      }
+
+      if(!found) {
+        if(show_realpaths) {
+          printf("%s\n", realpath(node1->name, 0));
+        } else {
+          printf("%s\n", node1->name);
+        }
       }
     }
-    closedir(dir);
+
   } else {
-    if(S_ISREG(dummy.st_mode)) {
-      result += dummy.st_size;
+    for(node1 = list1; node1; node1 = node1->next) {
+      progress_state.value = nb_nodes;
+      print_progress(&progress_state);
+      nb_nodes++;
+
+      for(node2 = list2; node2; node2 = node2->next) {
+        if(node1->group_id < 0 || node2->group_id < 0) {
+          if(same_files(node1, node2, buffer1, buffer2)) {
+            if(node1->group_id < 0) {
+              if(node2->group_id >= 0) {
+                node1->group_id = node2->group_id;
+              } else {
+                node1->group_id = nb_groups;
+                node1->dir_id = 1;
+                nb_groups++;
+              }
+            }
+            if(node2->group_id < 0) {
+              node2->group_id = node1->group_id;
+              node2->dir_id = 2;
+            }
+          }
+        }
+      }
     }
   }
 
-  return result;
+  if(show_progress) {
+    fprintf(stderr, "\n");
+  }
+
+  if(dirname2) {
+    print_result(list1, list2);
+    file_list_delete(list1);
+    file_list_delete(list2);
+  } else {
+    print_result(list1, 0);
+    file_list_delete(list1);
+  }
+
+  free(buffer1);
+  free(buffer2);
 }
 
-void start(const char *dirname1, const char *dirname2) {
-  struct file_with_size *list1, *list2;
-  list1 = scan_directory(dirname1);
-  list2 = scan_directory(dirname2);
-  destroy(list1);
-  destroy(list2);
+void usage(FILE *out) {
+  fprintf(out, "Usage: finddup [OPTION]... DIR1 [[and:|not:]DIR2]\n");
+  fprintf(out, "Version %s (%s)\n", VERSION_NUMBER, UNAME);
+  fprintf(out, "Without DIR2, lists duplicated files found in DIR1. With DIR2, lists files common to both directories. With the not: prefix, lists files found in DIR1 which do not exist in DIR2. The and: prefix is the default and should be used only if you have a directory starting with 'not:'\n");
+  fprintf(out, "\n");
+  /*            01234567890123456789012345678901234567890123456789012345678901234567890123456789*/
+  fprintf(out, "   -h, --help                 show this help\n");
+  fprintf(out, "   -d, --ignore-dots          ignore dot files and directories\n");
+  fprintf(out, "   -0, --ignore-empty         ignore empty files\n");
+  fprintf(out, "   -c, --hide-matchings       do not show which files in DIR2 corresponds to\n");
+  fprintf(out, "                              those in DIR1\n");
+  fprintf(out, "   -g, --no-group-ids         do not show the file groups\n");
+  fprintf(out, "   -p, --show-progress        show progress\n");
+  fprintf(out, "   -r, --real-paths           show the real file paths\n");
+  fprintf(out, "   -i, --same-inodes-are-different\n");
+  fprintf(out, "                              consider files with same inode as different\n");
+  fprintf(out, "\n");
+  fprintf(out, "Report bugs and comments to <francois@fleuret.org>.\n");
 }
 
 /**********************************************************************/
 
+static struct option long_options[] = {
+  { "help", no_argument, 0, 'h' },
+  { "same-inodes-are-different", no_argument, 0, 'i' },
+  { "real-paths", no_argument, 0, 'r' },
+  { "hide-matchings", no_argument, 0, 'c' },
+  { "no-group-ids", no_argument, 0, 'g' },
+  { "ignore-dots", no_argument, 0, 'd' },
+  { "ignore-empty", no_argument, 0, '0' },
+  { "show-progress", no_argument, 0, 'p' },
+  { 0, 0, 0, 0 }
+};
+
 int main(int argc, char **argv) {
   int c;
-  struct file_with_size *root;
-
-  root = 0;
 
   setlocale (LC_ALL, "");
 
-  while (1) {
-    c = getopt(argc, argv, "h");
-    if (c == -1)
-      break;
-
+  while ((c = getopt_long(argc, argv, "hircgd0pm",
+                          long_options, NULL)) != -1) {
     switch (c) {
 
     case 'h':
-      printf("Usage: finddup [OPTION]... [FILE]...\n");
-      printf("Report bugs and comments to <francois@fleuret.org>\n");
+      usage(stdout);
       exit(EXIT_SUCCESS);
 
       break;
 
+    case 'd':
+      ignore_dotfiles = 1;
+      break;
+
+    case '0':
+      ignore_empty_files = 1;
+      break;
+
+    case 'r':
+      show_realpaths = 1;
+      break;
+
+    case 'i':
+      same_inodes_are_different = 1;
+      break;
+
+    case 'g':
+      show_groups = 0;
+      break;
+
+    case 'p':
+      show_progress = 1;
+      break;
+
+    case 'c':
+      show_hits = 0;
+      break;
+
     default:
+      usage(stderr);
       exit(EXIT_FAILURE);
     }
   }
 
-  if (optind + 2 < argc) {
-    start(argv[optind + 1], argv[optind + 2]);
+  if(!isatty(STDERR_FILENO)) {
+    show_progress = 0;
+  }
+
+  if(optind + 2 == argc) {
+    start(argv[optind], argv[optind + 1]);
+  } else if(optind + 1 == argc) {
+    same_inodes_are_different = 1;
+    start(argv[optind], 0);
   } else {
-    fprintf(stderr, "%s [OPTIONS] <dir1> <dir2>\n", argv[0]);
+    usage(stderr);
     exit(EXIT_FAILURE);
   }