k-rus commented on code in PR #4038:
URL: https://github.com/apache/cassandra/pull/4038#discussion_r2050534395
##########
src/java/org/apache/cassandra/schema/SchemaConstants.java:
##########
@@ -75,14 +77,54 @@ public final class SchemaConstants
*/
public static final int NAME_LENGTH = 48;
+ /**
+ * Longest acceptable file name. Longer names lead to too long file name
error.
+ */
+ public static final int FILENAME_LENGTH = 255;
+
+ /**
+ * Longest acceptable table name, so it can be used in a directory
+ * name constructed with a suffix of a table id and a separator.
+ */
+ public static final int TABLE_NAME_LENGTH = FILENAME_LENGTH - 32 - 1;
+
// 59adb24e-f3cd-3e02-97f0-5b395827453f
public static final UUID emptyVersion;
public static final List<String> LEGACY_AUTH_TABLES =
Arrays.asList("credentials", "users", "permissions");
+ /**
+ * Validates that a name is valid to be used in files. Assumes that the
name length
+ * should fit the default, {@link #NAME_LENGTH}.
+ * See {@link #isValidName(String, int)} for more details.
+ *
+ * @param name the name to check
+ * @return whether the name is safe for use in file paths and file names
+ */
public static boolean isValidName(String name)
{
- return name != null && !name.isEmpty() && name.length() <= NAME_LENGTH
&& PATTERN_WORD_CHARS.matcher(name).matches();
+ return isValidName(name, NAME_LENGTH);
+ }
+
+ /**
+ * Names such as keyspace, table, index names are used in file paths and
file names,
+ * so, they need to be safe for the use there, i.e., short enough and
+ * containing only alphanumeric characters and underscores.
+ * Allows to provide the length of names, since it varies for different
database objects,
+ * especially, they were not historically controlled for {@link
#NAME_LENGTH}, see,
+ * e.g., CASSANDRA-20389.
+ * There are cases when the length cannot be controlled by a single value.
In such case
+ * the length validation is skipped if the given length is smaller than 1.
+ *
+ * @param name the name to check
+ * @param maxLength max acceptable length for the given name. For the
cases when it cannot
+ * be limited, 0 or negative number should be supplied.
Review Comment:
> > Also, I think that !name.isEmpty() is not enough. What if name is
- meaning e.g. 5 spaces? Then name.isEmpty() is false but is hardly a
valid file name so you would execute pattern matching unnecessarily. You should
use !name.isBlank() instead which is there from Java 11.
>
> @smiklosovic Thank you for sharing about `isBlank()`. Spaces and other
non-alphanumeric and non-underscore are already covered by
`PATTERN_WORD_CHARS.matcher(name).matches()`. From your comment it sounds that
the current condition does not fail ` `, while it does. What do I miss? (note I
am not arguing about `isBlank` vs `isEmpty`, but trying to understand what I am
missing)
Actually looking to a bigger picture, I don't think it's good idea to
replace `is Empty` with `isBlank`. Some identifier names are used in file names
(e.g., table and index names) and thus are restricted by `name.isEmpty` and
`PATTERN_WORD_CHARS.matcher(name).matches()`, while other identifier names are
not used in file names (e.g., column names), thus only restricted by
`name.isEmpty`. The latter allows name like `\" \"`, thus `name.isBlank`
cannot be used. Thus to me it will be more logical to keep `isEmpty`:
- Identifier names used in file names are limited by `name.isEmpty` and
`PATTERN_WORD_CHARS.matcher(name).matches()`
- Identifier names not used in file names are limited by `name.isEmpty` only
(I left out length limit from this comment)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: pr-unsubscr...@cassandra.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: pr-unsubscr...@cassandra.apache.org
For additional commands, e-mail: pr-h...@cassandra.apache.org
