From 00ffde6896006fe15051ef2f3036571c4d4b6404 Mon Sep 17 00:00:00 2001
From: Francois Fleuret <francois@fleuret.org>
Date: Sun, 7 Mar 2010 16:12:37 +0100
Subject: [PATCH] First working version. When sizes match, runs full
 comparison.

---
 finddup.c | 145 ++++++++++++++++++------------------------------------
 1 file changed, 49 insertions(+), 96 deletions(-)

diff --git a/finddup.c b/finddup.c
index d3e5a12..09c50d1 100644
--- a/finddup.c
+++ b/finddup.c
@@ -97,72 +97,33 @@ void print_size_sum(size_sum_t s) {
   printf(a);
 }
 
-size_sum_t file_or_dir_size(const char *name) {
-  DIR *dir;
-  struct dirent *dir_e;
-  struct stat dummy;
-  size_sum_t result;
-  char subname[BUFFER_SIZE];
-
-  result = 0;
-
-  if(lstat(name, &dummy) != 0) {
-    printf("Can not stat %s: %s\n", name, strerror(errno));
-    exit(EXIT_FAILURE);
-  }
-
-  if(S_ISLNK(dummy.st_mode)) {
-    return 0;
-  }
-
-  dir = opendir(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);
-      }
-    }
-    closedir(dir);
-  } else {
-    if(S_ISREG(dummy.st_mode)) {
-      result += dummy.st_size;
-    }
-  }
-
-  return result;
-}
-
 /**********************************************************************/
 
 struct file_with_size {
   char *filename;
   size_t size;
+  ino_t inode;
   struct file_with_size *next;
   int error;
 };
 
-struct file_list {
-  struct file_with_size *head, *tail;
-};
-
-void create(struct file_list *list, const char *name) {
-  list->head = safe_malloc(sizeof(struct file_with_size));
-  list->tail = 0;
-  list->head->filename = strdup(name);
-  list->head->size = file_or_dir_size(name);
-  list->head->next = 0;
+void file_list_delete(struct file_with_size *head) {
+  struct file_with_size *next;
+  while(head) {
+    next = head->next;
+    free(head->filename);
+    free(head);
+    head = next;
+  }
 }
 
-void destroy(struct file_with_size *node) {
-  struct file_with_size *next;
-  while(node) {
-    next = node->next;
-    free(node->filename);
-    free(node);
-    node = next;
+int file_list_length(struct file_with_size *head) {
+  int l = 0;
+  while(head) {
+    l++;
+    head = head->next;
   }
+  return l;
 }
 
 /**********************************************************************/
@@ -172,52 +133,42 @@ int same_size(struct file_with_size *f1, struct file_with_size *f2) {
 }
 
 int same_content(struct file_with_size *f1, struct file_with_size *f2) {
-  int fd1, fd2, eq, s1, s2;
+  int fd1, fd2, s1, s2;
   char buffer1[BUFFER_SIZE], buffer2[BUFFER_SIZE];
 
   fd1 = open(f1->filename, O_RDONLY);
   fd2 = open(f2->filename, O_RDONLY);
-  eq = 1;
 
   if(fd1 >= 0 && fd2 >= 0) {
-    while(eq) {
+    while(1) {
       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) {
+        close(fd1);
+        close(fd2);
+        return 0;
       }
-      if(s1 > 0 && s2 > 0) {
-        if(s1 == s2) {
-          eq = (strncmp(buffer1, buffer2, s1) == 0);
+
+      if(s1 == s2) {
+        if(s1 == 0) {
+          return 1;
         } else {
-          fprintf(stderr,
-                  "Different read size without error on files of same size.\n");
-          exit(EXIT_FAILURE);
+          if(strncmp(buffer1, buffer2, s1)) {
+            close(fd1);
+            close(fd2);
+            return 0;
+          }
         }
+      } else {
+        fprintf(stderr,
+                "Different read size without error on files of same size.\n");
+        exit(EXIT_FAILURE);
       }
     }
-  }
-
-  if(fd1 >= 0) {
-    close(fd1);
-  } else {
-    f1->error = 1;
-    eq = 0;
-  }
-
-  if(fd2 >= 0) {
-    close(fd2);
   } else {
-    f2->error = 1;
-    eq = 0;
+    return 0;
   }
-
-  return eq;
 }
 
 int same_files(struct file_with_size *f1, struct file_with_size *f2) {
@@ -226,7 +177,8 @@ int same_files(struct file_with_size *f1, struct file_with_size *f2) {
 
 /**********************************************************************/
 
-struct file_with_size *scan_directory(struct file_with_size *tail, const char *name) {
+struct file_with_size *scan_directory(struct file_with_size *tail,
+                                      const char *name) {
   DIR *dir;
   struct dirent *dir_e;
   struct stat dummy;
@@ -257,7 +209,8 @@ struct file_with_size *scan_directory(struct file_with_size *tail, const char *n
       tmp = safe_malloc(sizeof(struct file_with_size));
       tmp->next = tail;
       tmp->filename = strdup(name);
-      tmp->size = file_or_dir_size(name);
+      tmp->size = dummy.st_size;
+      tmp->inode = dummy.st_ino;
       tail = tmp;
     }
   }
@@ -267,20 +220,20 @@ struct file_with_size *scan_directory(struct file_with_size *tail, const char *n
 
 void start(const char *dirname1, const char *dirname2) {
   struct file_with_size *list1, *list2;
-  struct file_with_size *node;
+  struct file_with_size *node1, *node2;
   list1 = scan_directory(0, dirname1);
   list2 = scan_directory(0, dirname2);
 
-  for(node = list1; node; node = node->next) {
-    printf("list1: %s\n", node->filename);
-  }
-
-  for(node = list2; node; node = node->next) {
-    printf("list2: %s\n", node->filename);
+  for(node1 = list1; node1; node1 = node1->next) {
+    for(node2 = list2; node2; node2 = node2->next) {
+      if(node1->inode != node2->inode && same_files(node1, node2)) {
+        printf("%s %s \n", node1->filename, node2->filename);
+      }
+    }
   }
 
-  destroy(list1);
-  destroy(list2);
+  file_list_delete(list1);
+  file_list_delete(list2);
 }
 
 /**********************************************************************/
-- 
2.20.1