A public remove method which takes a key parameter.

$1$ Find the path of the digest of the key equal to the given object $($calling steps K and L in order$).$

$2$ Check if a file exists at the path of the digest to determine a hit or a miss.

$3$ On a hit, readRow from the path of the old row, deleteRow at the same path, writeInt to the size and fingerprint

paths under the metadata directory to update them $($like in previous modules$),$ and return the old list of fields.

$4$ On a miss, return null.

R A public iterator method which returns an iterator of rows.

$1$ Walk the data directory by streaming as follows:

i Filter the stream to include only files which aren$$t directories.

ii Map each path in the stream to the row stored there using readRow.

$2$ Return the iterator from the stream.

S A public toString method.

$1$ Return an unsorted view of the table by calling toTabularView.

page $3/5$

$3$RD QUARTER: CUSTOM BYTE ENCODINGS

A$\text{'}$ This quarter improves the efficiency of the table in a measurable way and can be enabled and disabled by a flag.

$1$ The $1$

st and $2$

nd quarters should pass $100\%$ of the unit tests and be committed before starting this quarter.

$2$ The $1$

st and $2$

nd quarters write and read objects using the Serializable API, which is time and space inefficient.

$3$ The $3$

rd quarter adopts custom byte encodings, which are time and space efficient.

B$\text{'}$ Modify the following private fields:

$1$ A static constant boolean flag for whether the table should use custom byte encodings instead of serializable objects,

which should be initialized to true when this quarter is in use.

C$\text{'}$ Modify the writeInt method.

$1$ If the custom byte encoding flag is true, allocate a byte buffer with a capacity which fits a $32-$bit integer value, put the given

integer into it$,$ convert it to a byte array, and write the corresponding bytes to the given path.

$2$ Otherwise, execute the original code from step E for serializable integers.

D$\text{'}$ Modify the readInt method.

$1$ If the custom byte encoding flag is true, read all bytes from the given path, wrap the corresponding byte array within a

byte buffer, get the $32-$bit integer value from the buffer, and return the integer.

$2$ Otherwise, execute the original code from step F for serializable integers.

E$\text{'}$ In the Row record, modify the following features:

$1$ A public getBytes method which returns a byte array.

i Create a new list of objects which starts with the row$$s key and continues with each field of the row$$s list of fields.

ii Use a loop over the list to predict the total number of bytes needed to encode the list as a record.

iii Allocate a byte buffer with a capacity equal to the total number of bytes predicted.

iv For each object in the list, encode the corresponding record bytes and put them into the buffer.

v Return the byte array from the buffer.

$2$ A public static fromBytes method which takes a byte array parameter and returns a row.

i Create a new list of objects which is empty.

ii Wrap a byte buffer around the given byte array.

iii While the buffer has remaining record bytes, decode the corresponding objects and add them into the list.

iv Return a new row with the first object of the list as the key and the rest of the objects of the list as the row.

$3$ For steps $1$ and $2$ above, use the following custom byte encodings:

i Each field starts with a $1-$byte tag $($also called a descriptor$).$

ii For a string field, the tag stores the length of the string in character bytes $($either zero or a positive number$),$ and the

rest stores the corresponding character bytes $($using the default UTF$8$ either explicitly or implicitly$).$

iii For an integer field, the tag stores a unique negative constant, and the rest stores the $32-$bit value of the number.

iv For a floating point $($double$)$ field, the tag stores a unique negative constant, and the rest stores the $64-$bit value of

the number.

v For a boolean field which is true, the tag stores a unique negative constant, and there is nothing more.

vi For a boolean field which is false, the tag stores a unique negative constant, and there is nothing more.

vii For a null field, the tag stores a unique negative constant, and there is nothing more.

F$\text{'}$ Modify the writeRow method.

$1$ If the custom byte encoding flag is true, create a byte array from the given row by calling its getBytes method, and write

the corresponding bytes to the given path.

$2$ Otherwise, execute the original code from step H for serializable rows.

G$\text{'}$ Modify the readRow method.

$1$ If the custom byte encoding flag is true, read all bytes from the given path, create a new row by calling the static

fromBytes method with the corresponding byte array, and return the new row.

$2$ Otherwise, execute the original code from step I for serializable rows.