X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=finddup.c;h=acbbddc58e8a2f487d05a7c9ca3fe0253993cec8;hb=1edf3fa264c9455b7f6f8c2b55714bfa4558dbc1;hp=aa16dac99582b2c1365c06b97f9c9552044df26b;hpb=0c962f1c0fabf85606069ab40c7b943a28c61abd;p=finddup.git diff --git a/finddup.c b/finddup.c index aa16dac..acbbddc 100644 --- a/finddup.c +++ b/finddup.c @@ -23,7 +23,7 @@ * */ -#define VERSION_NUMBER "0.8" +#define VERSION_NUMBER "0.9" #define _BSD_SOURCE @@ -40,10 +40,15 @@ #include #include #include +#ifdef WITH_MD5 +#include +#endif /* 1M really helps compared to 64k */ #define READ_BUFFER_SIZE (1024 * 1024) +#define PROGRESS_BUFFER_SIZE 1024 + typedef int64_t size_sum_t; /* Yeah, global variables! */ @@ -65,11 +70,13 @@ int show_hits = 1; /* 1 means to show the files from dir2 int show_groups = 1; /* 1 means to show the group IDs when printing file names */ -int ignore_same_inode = 0; /* 1 means that comparison between two file - with same inode will always be false */ +int same_inodes_are_different = 0; /* 1 means that comparison between + two files with same inode will + always be false */ -int tty_width = -1; /* Positive value means what width to use to show - the progress bar */ +#ifdef WITH_MD5 +int use_md5 = 0; /* 1 means we keep an MD5 signature for each file */ +#endif /********************************************************************/ @@ -102,6 +109,10 @@ struct file_node { ino_t inode; int group_id; /* one per identical file content */ int dir_id; /* 1 for DIR1, and 2 for DIR2 */ +#ifdef WITH_MD5 + int md5_computed; + unsigned char md5[MD5_DIGEST_LENGTH]; +#endif }; void file_list_delete(struct file_node *head) { @@ -129,6 +140,25 @@ int same_content(struct file_node *f1, struct file_node *f2, char *buffer1, char *buffer2) { int fd1, fd2, s1, s2; +#ifdef WITH_MD5 + MD5_CTX c1, c2; + + if(use_md5) { + if(f1->md5_computed && f2->md5_computed) { + if(!memcmp(f1->md5, f2->md5, MD5_DIGEST_LENGTH)) { + return 0; + } + } else { + if(!f1->md5_computed) { + MD5_Init(&c1); + } + if(!f2->md5_computed) { + MD5_Init(&c2); + } + } + } +#endif + fd1 = open(f1->name, O_RDONLY); fd2 = open(f2->name, O_RDONLY); @@ -147,6 +177,18 @@ int same_content(struct file_node *f1, struct file_node *f2, if(s1 == 0) { close(fd1); close(fd2); +#ifdef WITH_MD5 + if(use_md5) { + if(!f1->md5_computed) { + MD5_Final(f1->md5, &c1); + f1->md5_computed = 1; + } + if(!f2->md5_computed) { + MD5_Final(f2->md5, &c2); + f2->md5_computed = 1; + } + } +#endif return 1; } else { if(memcmp(buffer1, buffer2, s1)) { @@ -154,6 +196,16 @@ int same_content(struct file_node *f1, struct file_node *f2, close(fd2); return 0; } +#ifdef WITH_MD5 + if(use_md5) { + if(!f1->md5_computed) { + MD5_Update(&c1, buffer1, s1); + } + if(!f2->md5_computed) { + MD5_Update(&c2, buffer2, s2); + } + } +#endif } } else { fprintf(stderr, @@ -183,9 +235,10 @@ int same_content(struct file_node *f1, struct file_node *f2, int same_files(struct file_node *f1, struct file_node *f2, char *buffer1, char *buffer2) { - if(ignore_same_inode && f1->inode == f2->inode) { + if(same_inodes_are_different && f1->inode == f2->inode) { return 0; } + return f1->size == f2->size && same_content(f1, f2, buffer1, buffer2); } @@ -228,6 +281,9 @@ struct file_node *scan_directory(struct file_node *tail, const char *name) { tmp->inode = sb.st_ino; tmp->group_id = -1; tmp->dir_id = -1; +#ifdef WITH_MD5 + tmp->md5_computed = 0; +#endif tail = tmp; } } @@ -328,22 +384,46 @@ void print_result(struct file_node *list1, struct file_node *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 - 1); - if(p > *pp) { - for(k = 0; k < p; k++) { - fprintf(stderr, "+"); - } - for(; k < width; k++) { - fprintf(stderr, "-"); +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); } - *pp = p; - p = (100 * n) / (max - 1); - fprintf(stderr, " [%3d%%]\r", p); } } } @@ -351,10 +431,10 @@ void print_progress(int max, int n, int *pp) { 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, previous_progress; - struct winsize win; + int list1_length, list2_length, previous_progress; char *buffer1 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE); char *buffer2 = safe_malloc(sizeof(char) * READ_BUFFER_SIZE); @@ -362,15 +442,13 @@ void start(const char *dirname1, const char *dirname2) { not_in = 0; if(show_progress) { - if(isatty(STDOUT_FILENO) && - !ioctl (STDOUT_FILENO, TIOCGWINSZ, (char *) &win)) { - tty_width = win.ws_col; - } fprintf(stderr, "Scanning %s ... ", dirname1); } list1 = scan_directory(0, dirname1); + list1_length = file_list_length(list1); + if(dirname2) { if(strncmp(dirname2, "not:", 4) == 0) { not_in = 1; @@ -390,16 +468,30 @@ void start(const char *dirname1, const char *dirname2) { if(show_progress) { fprintf(stderr, "done.\n"); + fprintf(stderr, + "%s: %d file%s.\n", + dirname1, list1_length, (list1_length > 1 ? "s" : "")); + if(dirname2) { + list2_length = file_list_length(list2); + fprintf(stderr, + "%s: %d file%s.\n", + dirname2, list2_length, (list2_length > 1 ? "s" : "")); + } + fprintf(stderr, "Now looking for identical files.\n"); } nb_groups = 0; previous_progress = -1; nb_nodes = 0; - list1_length = file_list_length(list1); + + 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) { - print_progress(list1_length, nb_nodes, &previous_progress); + progress_state.value = nb_nodes; + print_progress(&progress_state); nb_nodes++; found = 0; @@ -421,7 +513,8 @@ void start(const char *dirname1, const char *dirname2) { } else { for(node1 = list1; node1; node1 = node1->next) { - print_progress(list1_length, nb_nodes, &previous_progress); + progress_state.value = nb_nodes; + print_progress(&progress_state); nb_nodes++; for(node2 = list2; node2; node2 = node2->next) { @@ -463,7 +556,7 @@ void start(const char *dirname1, const char *dirname2) { free(buffer2); } -void print_help(FILE *out) { +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"); @@ -479,6 +572,9 @@ void print_help(FILE *out) { 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"); +#ifdef WITH_MD5 + fprintf(out, " -m, --md5 use MD5 hashing\n"); +#endif fprintf(out, "\n"); fprintf(out, "Report bugs and comments to .\n"); } @@ -494,6 +590,7 @@ static struct option long_options[] = { { "ignore-dots", no_argument, 0, 'd' }, { "ignore-empty", no_argument, 0, '0' }, { "show-progress", no_argument, 0, 'p' }, + { "md5", no_argument, 0, 'm' }, { 0, 0, 0, 0 } }; @@ -502,12 +599,12 @@ int main(int argc, char **argv) { setlocale (LC_ALL, ""); - while ((c = getopt_long(argc, argv, "hircgd0p", + while ((c = getopt_long(argc, argv, "hircgd0pm", long_options, NULL)) != -1) { switch (c) { case 'h': - print_help(stdout); + usage(stdout); exit(EXIT_SUCCESS); break; @@ -525,7 +622,7 @@ int main(int argc, char **argv) { break; case 'i': - ignore_same_inode = 1; + same_inodes_are_different = 1; break; case 'g': @@ -540,18 +637,34 @@ int main(int argc, char **argv) { show_hits = 0; break; + case 'm': +#ifdef WITH_MD5 + use_md5 = 1; +#else + fprintf(stderr, + "finddup has not been compiled with MD5 hashing.\n"); + usage(stderr); + exit(EXIT_FAILURE); +#endif + break; + default: + usage(stderr); exit(EXIT_FAILURE); } } + if(!isatty(STDERR_FILENO)) { + show_progress = 0; + } + if(optind + 2 == argc) { start(argv[optind], argv[optind + 1]); } else if(optind + 1 == argc) { - ignore_same_inode = 1; + same_inodes_are_different = 1; start(argv[optind], 0); } else { - print_help(stderr); + usage(stderr); exit(EXIT_FAILURE); }