Add duplicate search when adding files

Duplicates are searched for in the database as well as in the given
files (on the command line).

1 files changed, 89 insertions, 10 deletions

diff --git a/src/main.c b/src/main.c
index b83072f..09540b6 100644
--- a/src/main.c
+++ b/src/main.c
@@ -386,17 +386,34 @@ static void ftag_file_add_help(void)
 	printf("    $ find DIR -type f -print | xargs ftag file add -i\n");
 }
 
+/* Structure only used by ftad_file_add and get_sums_callback, for building a
+ * list of file sums. */
+struct known_sums_s {
+	uint32_t *array;
+	int len;
+	int capacity;
+};
+
+static int get_sums_callback(void *_known_sums, int, char **cols, char **)
+{
+	struct known_sums_s *known_sums = _known_sums;
+	assert(cols[0]);
+	known_sums->array[known_sums->len] = (uint32_t)atol(cols[0]);
+	known_sums->len++;
+	assert(known_sums->len <= known_sums->capacity);
+	return 0;
+}
+
 /* Add new files to the database. If directories are given, every file in the
  * directory will be added, does _not_ recurse into subdirectories. */
 static void ftag_file_add(int argc, char **argv)
 {
+	/* step 0: parse options */
 	if (argc == 0) {
 		ftag_file_add_usage();
 		exit(EXIT_FAILURE);
 	}
-	sqlite3 *db = NULL;
 	int interactive = 0;
-	int rc;
 	while ((argc > 0) && (argv[0][0] == '-')) {
 		switch (argv[0][1]) {
 		case 'i':
@@ -412,24 +429,86 @@ static void ftag_file_add(int argc, char **argv)
 		argv++;
 		argc--;
 	}
+
+	/* step 1: compute file sums */
+	uint32_t *sums = malloc(argc*sizeof(*sums));
+	struct stat st;
+	int rc;
+	for (int i = 0; i < argc; i++) {
+		char *file = argv[i];
+		rc = stat(file, &st);
+		if (rc == -1) {
+			fprintf(stderr, "stat: \"%s\": ", file);
+			perror("");
+			exit(EXIT_FAILURE);
+		}
+		if (st.st_mode & S_IFREG) {
+			sums[i] = sum(file);
+		}
+		else {
+			/* skip non-file arguments */
+			argv[i] = NULL;
+		}
+	}
+
+	/* step 2: retrieve sums of files already in the database */
+	sqlite3 *db;
 	rc = sqlite3_open(DATABASE_PATH, &db);
 	sqlite3_check(rc, db);
+
 	int next_id = table_next_id(db, "files");
+	int file_count = next_id; /* number of files already in the database */
+	uint32_t known_sums_array[file_count];
+	struct known_sums_s known_sums = {
+		.array = known_sums_array,
+		.len = 0,
+		.capacity = file_count
+	};
+
+	rc = sqlite3_exec(db, "SELECT sum FROM files;", get_sums_callback, &known_sums, NULL);
+	sqlite3_check(rc, db);
+
+	/* step 3: eliminate duplicates with respect to file sum */
+	for (int i = 0; i < argc; i++) {
+		const uint32_t original = sums[i];
+		/* step 3.1: eliminate duplicates found in database */
+		for (int j = 0; j < file_count; j++) {
+			const uint32_t duplicate = known_sums_array[j];
+			if (duplicate == original) {
+				/* sum already known in database */
+				printf("%s debug: file \"%s\" found in database\n",
+				       __func__, argv[i]);
+				argv[i] = NULL;
+				break;
+			}
+		}
+		/* step 3.2: eliminate duplicates within given files */
+		if (argv[i] == NULL)
+			continue;
+		for (int j = i+1; j < argc; j++) {
+			const uint32_t duplicate = sums[j];
+			if (duplicate == original) {
+				printf("%s debug: file \"%s\" found in given files\n",
+				       __func__, argv[j]);
+				argv[j] = NULL;
+			}
+		}
+	}
+
+	/* step 4: perform addition to database */
 	for (int i = 0; i < argc; i++) {
 		char *file = argv[i];
+		if (file == NULL)
+			continue;
 		if (interactive) {
 			printf("Add \"%s\" to database ?", file);
 			if (!prompt_yes_no())
-				continue; /* do not add this file */
-		}
-		struct stat st;
-		rc = stat(file, &st);
-		assert(rc == 0);
-		if (st.st_mode & S_IFREG) {
-			ftag_add_one_file(db, &next_id, file);
-			next_id++;
+				continue;
 		}
+		ftag_add_one_file(db, &next_id, file, sums[i]);
+		next_id++;
 	}
+	free(sums);
 	sqlite3_close(db);
 }