Put the read buffers on the heap instead of the stack.

[finddup.git] / finddup.c

diff --git a/finddup.c b/finddup.c
index a8847f5..ab345d3 100644 (file)
--- a/finddup.c
+++ b/finddup.c
@@ -41,8 +41,8 @@
 #include <getopt.h>
 #include <fcntl.h>
 
-#define BUFFER_SIZE 4096
-#define LARGE_BUFFER_SIZE 65536
+/* 1M really helps compared to 64k */
+#define READ_BUFFER_SIZE (1024 * 1024)
 
 typedef int64_t size_sum_t;
 
@@ -68,6 +68,9 @@ int show_groups = 1; /* 1 means to show the group IDs when printing
 int ignore_same_inode = 0; /* 1 means that comparison between two file
                               with same inode will always be false */
 
+int tty_width = -1; /* Positive value means what width to use to show
+                       the progress bar */
+
 /********************************************************************/
 
 /* malloc with error checking.  */
@@ -112,7 +115,7 @@ struct file_with_size {
   size_t size;
   ino_t inode;
   struct file_with_size *next;
-  int group_id;
+  int group_id, dir_id;
 };
 
 void file_list_delete(struct file_with_size *head) {
@@ -136,17 +139,17 @@ int file_list_length(struct file_with_size *head) {
 
 /**********************************************************************/
 
-int same_content(struct file_with_size *f1, struct file_with_size *f2) {
+int same_content(struct file_with_size *f1, struct file_with_size *f2,
+                 char *buffer1, char *buffer2) {
   int fd1, fd2, s1, s2;
-  char buffer1[LARGE_BUFFER_SIZE], buffer2[LARGE_BUFFER_SIZE];
 
   fd1 = open(f1->filename, O_RDONLY);
   fd2 = open(f2->filename, O_RDONLY);
 
   if(fd1 >= 0 && fd2 >= 0) {
     while(1) {
-      s1 = read(fd1, buffer1, LARGE_BUFFER_SIZE);
-      s2 = read(fd2, buffer2, LARGE_BUFFER_SIZE);
+      s1 = read(fd1, buffer1, READ_BUFFER_SIZE);
+      s2 = read(fd2, buffer2, READ_BUFFER_SIZE);
 
       if(s1 < 0 || s2 < 0) {
         close(fd1);
@@ -191,11 +194,12 @@ int same_content(struct file_with_size *f1, struct file_with_size *f2) {
   }
 }
 
-int same_files(struct file_with_size *f1, struct file_with_size *f2) {
+int same_files(struct file_with_size *f1, struct file_with_size *f2,
+               char *buffer1, char *buffer2) {
   if(ignore_same_inode && f1->inode == f2->inode) {
     return 0;
   }
-  return f1->size == f2->size && same_content(f1, f2);
+  return f1->size == f2->size && same_content(f1, f2, buffer1, buffer2);
 }
 
 /**********************************************************************/
@@ -237,6 +241,7 @@ struct file_with_size *scan_directory(struct file_with_size *tail,
           tmp->size = sb.st_size;
           tmp->inode = sb.st_ino;
           tmp->group_id = -1;
+          tmp->dir_id = -1;
           tail = tmp;
         }
       }
@@ -276,7 +281,13 @@ int compare_nodes(const void *x1, const void *x2) {
   } else if((*f1)->group_id > (*f2)->group_id) {
     return 1;
   } else {
-    return 0;
+    if((*f1)->dir_id < (*f2)->dir_id) {
+      return -1;
+    } else if((*f1)->dir_id > (*f2)->dir_id) {
+      return 1;
+    } else {
+      return 0;
+    }
   }
 }
 
@@ -326,11 +337,34 @@ void print_result(struct file_with_size *list1, struct file_with_size *list2) {
   free(nodes);
 }
 
+void print_progress(int max, int n, int *pp) {
+  int p, k;
+  int width;
+  if(show_progress && tty_width > 0) {
+    width = tty_width - 7;
+    p = (width * n) / max;
+    if(p > *pp) {
+      for(k = 0; k < p; k++) {
+        fprintf(stderr, "+");
+      }
+      for(; k < width; k++) {
+        fprintf(stderr, "-");
+      }
+      *pp = p;
+      p = (100 * n) / max;
+      fprintf(stderr, " [% 3d%%]\r", p);
+    }
+  }
+}
+
 void start(const char *dirname1, const char *dirname2) {
   struct file_with_size *list1, *list2;
   struct file_with_size *node1, *node2;
   int not_in, found;
-  int k, p, pp, l1, n;
+  int k, pp, l1, n;
+
+  char *buffer1 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
+  char *buffer2 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE);
 
   not_in = 0;
 
@@ -343,6 +377,9 @@ void start(const char *dirname1, const char *dirname2) {
   if(dirname2) {
     if(strncmp(dirname2, "not:", 4) == 0) {
       not_in = 1;
+      /* We should show groups even in that mode. However they are not
+         properly calculated for now, so we force it off. */
+      show_groups = 0;
       dirname2 += 4;
     }
     if(show_progress) {
@@ -364,19 +401,13 @@ void start(const char *dirname1, const char *dirname2) {
 
   if(not_in) {
     for(node1 = list1; node1; node1 = node1->next) {
-      if(show_progress) {
-        p = (100 * n)/l1;
-        if(p > pp) {
-          fprintf(stderr, "%d%%\n", p);
-          pp = p;
-        }
-        n++;
-      }
+      print_progress(l1, n, &pp);
+      n++;
 
       found = 0;
 
       for(node2 = list2; !found && node2; node2 = node2->next) {
-        if(same_files(node1, node2)) {
+        if(same_files(node1, node2, buffer1, buffer2)) {
           found = 1;
         }
       }
@@ -392,28 +423,24 @@ void start(const char *dirname1, const char *dirname2) {
 
   } else {
     for(node1 = list1; node1; node1 = node1->next) {
-      if(show_progress) {
-        p = (100 * n)/l1;
-        if(p > pp) {
-          fprintf(stderr, "%d%%\n", p);
-          pp = p;
-        }
-        n++;
-      }
+      print_progress(l1, n, &pp);
+      n++;
 
       for(node2 = list2; node2; node2 = node2->next) {
         if(node1->group_id < 0 || node2->group_id < 0) {
-          if(same_files(node1, node2)) {
+          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 = k;
+                node1->dir_id = 1;
                 k++;
               }
             }
             if(node2->group_id < 0) {
               node2->group_id = node1->group_id;
+              node2->dir_id = 2;
             }
           }
         }
@@ -421,6 +448,10 @@ void start(const char *dirname1, const char *dirname2) {
     }
   }
 
+  if(show_progress) {
+    fprintf(stderr, "\n");
+  }
+
   if(dirname2) {
     print_result(list1, list2);
     file_list_delete(list1);
@@ -429,6 +460,9 @@ void start(const char *dirname1, const char *dirname2) {
     print_result(list1, 0);
     file_list_delete(list1);
   }
+
+  free(buffer1);
+  free(buffer2);
 }
 
 void print_help(FILE *out) {
@@ -453,13 +487,14 @@ void print_help(FILE *out) {
 int main(int argc, char **argv) {
   int c;
   struct file_with_size *root;
+  struct winsize win;
 
   root = 0;
 
   setlocale (LC_ALL, "");
 
   while (1) {
-    c = getopt(argc, argv, "hrcgd0p");
+    c = getopt(argc, argv, "hircgd0p");
     if (c == -1)
       break;
 
@@ -504,6 +539,12 @@ int main(int argc, char **argv) {
     }
   }
 
+  if(show_progress &&
+     isatty(STDOUT_FILENO) &&
+     !ioctl (STDOUT_FILENO, TIOCGWINSZ, (char *) &win)) {
+    tty_width = win.ws_col;
+  }
+
   if(optind + 2 == argc) {
     start(argv[optind], argv[optind + 1]);
   } else if(optind + 1 == argc) {