file_processor: Skip files already in the DB with same size and mtime

1 files changed, 130 insertions, 1 deletions

diff --git a/src/database_interaction.c b/src/database_interaction.c
index 5188a35..9e29c17 100644
--- a/src/database_interaction.c
+++ b/src/database_interaction.c
@@ -923,6 +923,136 @@ int dbi_insert_fileinfo(struct df_fileinfo *fi) {
     return rc;
 };
 
+/**
+ * Fill the given struct (path and filename required) with database content if available.
+ *
+ * This can be used to speed up a rescan if the stat matches and last mtime was before last_seen
+ */
+int dbi_fill_fileinfo(struct df_fileinfo *fi) {
+    int rc = 0;
+    int64_t fname_id, pname_id, existing_entry = 0;
+
+
+    DBCONN_CHECK(-1);
+
+    if (fi == NULL) {
+        LOGERR("ERROR: No fileinfo given.\n");
+        return -1;
+    }
+
+    fname_id = dbi_select_filename_by_name(fi->name);
+    pname_id = dbi_select_path_by_pathname(fi->path);
+
+    /* Any problems with the selects? */
+    if (fname_id <0 || pname_id <0) {
+        DBGTRC("ERROR: abort due to previous error.\n");
+        return -1;
+    }
+
+    /* Abort if no matching entry in the table */
+    existing_entry = dbi_select_fileinfo_by_path_filename_ids(pname_id, fname_id);
+    if (existing_entry < 0) {
+        fi->id = 0;
+        return 0;
+    }
+
+    /* TODO: The following has be moved to a dbi_select_fileinfo_by_id_resolved */
+    rc = dbi_select_fileinfo_by_id_resolved(existing_entry, fi);
+
+    return rc;
+}
+
+
+
+int dbi_select_fileinfo_by_id_resolved(int64_t id, struct df_fileinfo *fi) {
+    int strc = 0;
+    size_t len;
+    char *tmp = NULL;
+    const unsigned char *txt = NULL;
+    sqlite3_stmt *st = select_fileinfo_by_id_resolved;
+
+
+    DBCONN_CHECK(-1);
+
+    if (id <=0) {
+        LOGERR("ERROR: No id given.\n");
+        return -1;
+    }
+
+    if (fi == NULL) {
+        LOGERR("ERROR: No fileinfo struct given.\n");
+        return -1;
+    }
+
+
+    sqlite3_clear_bindings(st);
+    sqlite3_reset(st);
+
+    if (sqlite3_bind_int64(st, 1,id) != SQLITE_OK) {
+        LOGERR("ERROR: Failed to bind id to prepared statement: %s\n", sqlite3_errmsg(dbconn));
+        return -1;
+    }
+
+    strc = sqlite3_step(st);
+
+    if (strc == SQLITE_DONE) {
+        return -1;
+    }
+
+    if (strc != SQLITE_ROW) {
+        LOGERR("ERROR: Failed step: %s\n", sqlite3_errmsg(dbconn));
+        return -1;
+    }
+
+    if (fi->path == NULL) {
+        txt = sqlite3_column_text(st, 0); /* paths.pathname */
+        len = (size_t)sqlite3_column_bytes(st, 0);
+        if ((tmp = calloc(len+1, sizeof(char))) == NULL) {
+            LOGERR("ERROR: Failed to allocate a few bytes memory\n");
+            return -1;
+        }
+        memcpy(tmp, txt, len);
+        fi->path = tmp;
+    }
+
+    if (fi->name == NULL) {
+        txt = sqlite3_column_text(st, 1); /* filenames.name */
+        len = (size_t)sqlite3_column_bytes(st, 1);
+        if ((tmp = calloc(len+1, sizeof(char))) == NULL) {
+            LOGERR("ERROR: Failed to allocate a few bytes memory\n");
+            return -1;
+        }
+        memcpy(tmp, txt, len);
+        fi->path = tmp;
+    }
+
+    txt = sqlite3_column_text(st, 2); /* hashes.blake2 */
+    memcpy(fi->hashes.blake2, txt, DF_STR_SIZE_512);
+    fi->hashes.blake2[DF_STR_SIZE_512] = '\0'; /* array has the additional nul byte space */
+
+    txt = sqlite3_column_text(st, 3); /* hashes.sha256 */
+    memcpy(fi->hashes.sha256, txt, DF_STR_SIZE_256);
+    fi->hashes.sha256[DF_STR_SIZE_256] = '\0'; /* array has the additional nul byte space */
+
+    txt = sqlite3_column_text(st, 4); /* hashes.sha512 */
+    memcpy(fi->hashes.sha512, txt, DF_STR_SIZE_512);
+    fi->hashes.sha512[DF_STR_SIZE_512] = '\0'; /* array has the additional nul byte space */
+
+    /* Ignore the item 5 from the query, it's the filesize we have in the stat struct */
+    
+    fi->last_seen = (time_t) sqlite3_column_int64(st, 6); /* last_seen */
+
+    /* struct stat statbuf */
+    memcpy(&(fi->statbuf), sqlite3_column_blob(st, 7), (size_t)sqlite3_column_bytes(st, 7));
+
+    fi->id = id;
+
+    sqlite3_clear_bindings(st);
+    sqlite3_reset(st);
+
+
+    return 0;
+}
 
 /**
  * Print the database content to the given filedescriptor in a CSV form which can be processed with grep awk etc.
@@ -1363,7 +1493,6 @@ int dbi_print_filenames(FILE *out) {
 
 #if 0
              *select_fileinfo_by_id,
-             *select_fileinfo_by_id_resolved,
              *select_fileinfo_by_path_id,
              *select_fileinfo_by_filename_id,
              *select_fileinfo_by_path_filename_ids,