added unique and first modifiers to lookup subcommand

filetags.py

@@ -1,6 +1,6 @@
 #!/usr/bin/python3
 
-import sys, hashlib, os, sqlite3, shutil
+import sys, hashlib, os, sqlite3, shutil, json
 
 ################################################################################
 # program wide constants
@@ -126,7 +126,7 @@ def get_hashes_by_tag(storage_directory, tag):
 
   return file_hashes
 
-def file_is_in_storage(storage_directory, file_hash_or_path, compress):
+def file_hash_or_path_is_known_hash(storage_directory, file_hash_or_path, compress):
   suffix=""
   if compress:
     suffix=".xz"
@@ -141,6 +141,7 @@ def load_container_settings(storage_directory):
   cursor = db.cursor()
   cursor.execute("SELECT option, value FROM settings")
 
+  #TODO: check container settings properly instead of just assuming default values for things that aren't found
   # set default values and then read the db, just in case...
   parity             = False
   parity_bytes       = 512
@@ -301,19 +302,19 @@ if __name__ == "__main__":
     hash_allowed=True
     if 'file' in command:
       hash_allowed=False
-    if not any([hash_allowed and file_is_in_storage(storage_directory, file_hash_or_path, compress), os.path.isfile(file_hash_or_path)]):
+    if not any([hash_allowed and file_hash_or_path_is_known_hash(storage_directory, file_hash_or_path, compress), os.path.isfile(file_hash_or_path)]):
       print("Unknown file!", file=sys.stderr)
       print(USAGE, file=sys.stderr)
       sys.exit(PATH_ERROR)
 
     tags = sys.argv[4:]
 
-    if hash_allowed and file_is_in_storage(storage_directory, file_hash_or_path, compress):
+    if hash_allowed and file_hash_or_path_is_known_hash(storage_directory, file_hash_or_path, compress):
       file_hash = file_hash_or_path
-      print("File already in storage.")
+      print("Hash for file in storage: "+file_hash)
     else:
       file_hash = calculate_file_hash(checksum_algorithm, file_hash_or_path)
-      if file_is_in_storage(storage_directory, file_hash_or_path, compress):
+      if file_hash_or_path_is_known_hash(storage_directory, file_hash, compress):
         print("File already in storage.")
         #this assumes that the storage directory has not been tampered with or corrupted, FIXME!
         if 'move' in command:
@@ -344,9 +345,11 @@ if __name__ == "__main__":
   # lookup subcommand: return hash and tags of found files
   # arguments: <storage directory> <hash|exact tag|set of exact tags>
   # modifiers:
-  #  first - only return one file
+  #  first   - only return one file
   #  unique - return error if not found or multiple found
-  #TODO: modifiers
+  #  hash   - perform lookup by hash
+  #  tags   - perform lookup by tag or set of tags
+  #TODO: modifiers: first unique
   if command[0] == "lookup":
     if len(sys.argv)<4:
       print("Too few arguments!", file=sys.stderr)
@@ -364,27 +367,39 @@ if __name__ == "__main__":
       sys.exit(status)
 
     file_tags_or_hash = sys.argv[3:]
-    if file_is_in_storage(storage_directory, file_tags_or_hash[0], compress):
+    lookup_results = {}
-      tags = get_tags_by_hash(storage_directory, file_tags_or_hash[0])
+    if not 'tags' in command:
-      print("Tags for file:")
+      if file_hash_or_path_is_known_hash(storage_directory, file_tags_or_hash[0], compress):
-      print(tags)
+        lookup_results[file_tags_or_hash[0]] = get_tags_by_hash(storage_directory, file_tags_or_hash[0])
+
+    if not 'hash' in command:
+      # create a two dimensional array of all the files associated with each individual tag
+      file_hash_lists = []
+      for tag in file_tags_or_hash:
+        file_hash_lists = file_hash_lists + [get_hashes_by_tag(storage_directory, tag)]
+      # take the first of the arrays in the two dimensional array
+      common_file_hashes = file_hash_lists[0]
+      # iterate over the two dimensional array
+      for file_hash_list in file_hash_lists:
+        # check each element in common_file_hashes to ensure it is also in all other arrays in the two dimensional array, remove if it isn’t
+        for file_hash in common_file_hashes:
+          if not file_hash in file_hash_list:
+            common_file_hashes.remove(file_hash)
 
-    # create a two dimensional array of all the files associated with each individual tag
-    file_hash_lists = []
-    for tag in file_tags_or_hash:
-      file_hash_lists = file_hash_lists + [get_hashes_by_tag(storage_directory, tag)]
-    # take the first of the arrays in the two dimensional array
-    common_file_hashes = file_hash_lists[0]
-    # iterate over the two dimensional array
-    for file_hash_list in file_hash_lists:
-      # check each element in common_file_hashes to ensure it is also in all other arrays in the two dimensional array, remove if it isn’t
       for file_hash in common_file_hashes:
-        if not file_hash in file_hash_list:
+        lookup_results[file_hash] = get_tags_by_hash(storage_directory, file_hash)
-          common_file_hashes.remove(file_hash)
 
-    if not common_file_hashes == []:
+    if 'unique' in command:
-      print("Files for tag(s):")
+      if(len(lookup_results)==1):
-      print(common_file_hashes)
+        print(json.dumps(lookup_results))
+      else:
+        print("More than one matching file found.", file=sys.stderr)
+        sys.exit(GENERAL_ERROR)
+    elif 'first' in command:
+      file_hash, tags = list(lookup_results.items())[0]
+      print(json.dumps({file_hash : tags}))
+    else:
+      print(json.dumps(lookup_results))