JPH0926902A - Method for compressing and restoring file - Google Patents

Abstract

PROBLEM TO BE SOLVED: To inhibit a part of the free area of a RAM from being used for a restored data file and to effectively utilize a RAM area by adding the position information of respective records after compression to files after the compression. SOLUTION: One character is taken out from a compressed file and the bit of '1' or '1' attached to the branch of a Huffman tree corresponding to the taken-out character is counted in a record position counter. A bit string (code word) outputted to an operation area is outputted to the compression file. Compression to one character is performed, and every time compression for one record is ended, the value (a total bit number inside one record) of the record position counter is added to the compression file. As a result, for the compression file, the positions (terminating positions) of the respective records 1-(n) in a compression data part 91 are added as a record position information part 92 in the order of record numbers as a bit number from the head of the compression file.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to data compression and decompression of a data file of a computer or a personal computer, and more particularly to a handy size information device for collecting, processing and storing data, and its auxiliary storage device,
That is, the file compression method called the Huffman method is used for the purpose of effectively using the RAM area.

[0002]

2. Description of the Related Art A small portable terminal as a handy-sized information device has a structure shown in FIG. In FIG. 18, the small mobile terminal 1 includes a CPU 1a that controls each unit.
A ROM 1b that stores an operating system program that controls the basic operation of the CPU 1a; a RAM 1c that is loaded with an application program for the corresponding business and that temporarily stores various register values as a work area of the program; A display circuit 1d composed of a liquid crystal display element or the like for displaying various data supplied from the CPU 1a via an internal bus, and a manipulator 1 composed of a numeric keypad or a function key for inputting numbers.
e and a memory card 1f that is detachably attached. Note that 1g is a communication control circuit configured of, for example, a modem and controlling serial data communication.

A data file used by the small portable terminal 1 (a file such as screen information used for special reading from the system, which is a random access file having a fixed-length record) is created by a personal computer, and is downsized by downloading. It is used by being copied to the RAM 1c or the memory card 1f of the mobile terminal 1.

However, the RAM 1c and the memory card 1f
Has a limited capacity, and when a small portable terminal 1 tries to operate a large-scale system that handles a large amount of data files, the files including the data files cannot be stored in the auxiliary storage device, and the system operation is unsuccessful. It is expected that this will be possible (see FIG. 17).

[0005]

In order to avoid the above problems, a method of compressing a data file, downloading it to a small terminal, and restoring it for use in a small portable terminal system is conceivable. When a compressed data file is restored to the original data file, the restored data file is stored, so a free area is required in the RAM.

There is a problem that the restoration processing takes a long time (see FIG. 16).

[0007]

According to the present invention, basically, when a file is compressed on a personal computer and a compressed data file is downloaded and copied to a small portable terminal, the system uses it. Only necessary records are restored on the main memory (see FIG. 15).

For the above-mentioned purpose, the present invention is realized by effectively utilizing a compression algorithm called Huffman method.

The Huffman method, which will be described in detail later, uses a code conversion table called a Huffman tree to convert a character having a high frequency of occurrence into a code word (compressed character) having a short bit length into a code word having a short bit length. Encoding is performed by converting low-character characters into code words with long bit length. Therefore, the generated codeword does not have a constant length in bit units. Further, the Huffman tree is uniquely determined for one character string and does not change. Also, since uniquely decodable encoding can be performed, it is possible to restore from the beginning of each codeword.

Therefore, basically, the following (a), (b),
By performing step (c), any record (codeword string) in the compressed file can be restored.

(A) At the time of compression, the data length after compression is counted, the position information of each record in the compressed file is obtained, and added to the compressed file.

(B) The position information of the record of (a) is read from the compressed file, and only the desired record is restored based on this information.

(C) At the time of restoration, information such as a desired record position and data length is obtained from the position information of adjacent records, and the desired record is restored by restoring the data based on the information.

First, when the file to be compressed is composed of records of a predetermined length (the number of character strings), the following (d) is added. It should be noted that when the record is composed of fixed-length records, each record has a fixed predetermined length, and thus is included in the above case.

(D) At the time of compression, the number of characters in the record is counted, and when this reaches the record length, it is determined that the end of each record, and the position information of each record in (a) is fixed. To do.

Further, instead of (c), the following (e),
By adding (f), the restoration process can be simplified.

(E) Predetermined record length information of the compressed file is added to the compressed file.

(F) At the time of decompression, the record length information of (e) is read from the compressed file and only the character string of that record length is decompressed, so that only the desired record can be decompressed.

Next, when the compressed file is composed of records of variable length, the end of each record is usually separated by a delimiter (delimiter), so (a),
The following (g) is added to (b) and (c).

(G) At the time of compression, when the character of the record matches the above delimiter, it is determined to be the end of the record, and the position information of each record of (a) is fixed.

Further, the restoration process can be simplified by adding the following (h) instead of (c).

(H) Upon restoration, when the character of the record matches the above delimiter, it is judged to be the end of the record and the restoration is terminated, so that only the desired record is restored.

According to the present invention, the data file compressed by the Huffman method compression algorithm is downloaded onto the personal computer and copied to the small portable terminal, and only the records necessary for the system to use this are stored in the main memory. Restore. As a result, some of the free space in RAM is not used for the restored data file. Moreover, since the record length is sufficiently shorter than the file length, the restoration processing time is short.

[0024]

BEST MODE FOR CARRYING OUT THE INVENTION The present invention will be described below with reference to the drawings in the use of a compressed file in a small portable terminal. FIG. 18 is a block diagram showing the configuration of the small mobile terminal according to the present invention. The present invention is applied to a small mobile terminal having such a configuration. Conventionally, as shown in FIG. 17, the small mobile terminal 10 reads a record of the data file 1 copied to the small mobile terminal auxiliary storage device 12 (A1). Alternatively, FIG.
, The compressed data file 1 is read into the small portable terminal (B1) and restored, and then the auxiliary storage device 1
The data is written in the empty area 2 (B2), and the record of the restored data file 1 is read by the small mobile terminal 10 (B3).

In the present invention, as shown in FIG. 15, the small portable terminal 10 reads (C1) and restores only necessary records of the data file 1 compressed by the compression method using the Huffman method. The restored record is placed in main memory.

Here, the compression algorithm called the Huffman method will be briefly described. The Huffman method uses a code conversion table called a Huffman tree to convert a character having a high appearance frequency in a character string into a code word having a short bit length (compressed). (Encoded characters), the characters having a low appearance frequency are converted into code words having a long bit length to perform encoding. Therefore, the generated codeword does not have a constant length in bit units.

In the Huffman tree, a character (called a leaf) that appears in a character string or a node is bundled by a node from a node having a small number of appearances, and the total number of occurrences is used as the number of appearances of a node. It is obtained by repeating until the end (the last section is called the root). Now, "ADCABAEBA
When the character string "ADC" is encoded by the Huffman method, the constructed Huffman tree is as shown in FIG.

In FIG. 1, the number in () of the leaf is the number of appearances of each character in the data file 1. The numbers in parentheses () for nodes and roots are the sum of the numbers in parentheses () for leaves or nodes.

A Huffman tree that connects roots, nodes, and leaves is called a branch. Then, at each branch point of the root and node, 1 is given to the branch extending to the lower right, 0 is given to the branch extending to the lower left,
A Huffman code (codeword by the Huffman method) is a bit string in which the sequence of 1s and 0s on the branches traced from the root to the leaves is used as a bit string. The character string "ADCABAEB"
The Huffman code for "AADC" is as shown in FIG.
The code word string (compressed character string) is as shown in FIG.
Since the Huffman tree is used for the compressed file at the time of decompression, the Huffman tree information is added.

Since the Huffman tree is created by checking the appearance frequencies of all the characters in the compressed file, it is uniquely determined for one character string and does not change. Further, since uniquely decodable coding can be performed, it is possible to restore from the beginning of each codeword.

The procedure for compressing one character and the procedure for compressing the entire file, which are the basis of the present invention, are shown in a flow chart in FIG. 4 and FIG. 5, respectively.

That is, while recording 1 or 0 attached to the branch of the Huffman tree in FIG. 1, trace back from the leaf (letter) to the root, and reverse the row of 1,0 recorded when reaching the root, A method of outputting to a compressed file while converting into a bit string is adopted. However, the output of the code word of the compressed file is performed in byte units. In addition, compression of a file may be performed by compressing one character from the beginning to the end of the file.

The present invention is carried out based on the compression (encoding) / decompression (decoding) algorithm by the Huffman method.

Further, in the present invention, since the Huffman tree is uniquely determined and does not change for one character string, it is possible to restore from any code word of the code word string from the beginning of each code word. In order to restore only an arbitrary record based on the above, it is necessary to add the following (a) and (b) to the flowcharts of FIGS. 4 and 5.

(A) Count the total number of bits of a character string compressed based on the Huffman tree.

(B) Each time one record of compression is completed, the total number of bits of the code word sequence is recorded in a work area such as an array and added to the compressed file at the end of compression.

That is, as shown in FIG. 9, the conventional Huffman method compressed file has the Huffman tree information 82 added to the encoded compressed data portion 81. Since each record 1, 2, ..., N in the compressed data portion 81 has a variable bit length, the random access to the record becomes impossible.

In the present invention, since only an arbitrary record is restored at the time of restoration, each record (1, 2 ,,) in the compressed data portion 91 is added to the conventional compressed file as shown in FIG.
.., n,) positions are added as the record position information section 92 in the order of the record numbers.

Embodiments will be described below in accordance with the claims.

(Embodiment 1) One embodiment of a method of compressing a file when the file to be compressed is composed of a plurality of records of a predetermined length (the number of character strings) according to claim 1. An example will be described.

First, compression for one character shown in the flowchart of FIG. 6 starts (ST60). In addition, it is necessary to form a Huffman tree for one record of the compressed file in advance.

Extract one character from the compressed file (ST
61). The Huffman tree leaf corresponding to the character extracted in step ST61 is found (ST62). The 0 or 1 attached to the branch traced back to the node immediately above this leaf is output to the work area (ST63). The "0" or "1" bit output in step ST63 is counted by the record position counter (ST64). Steps ST63 and ST64 are continued until the root is reached (ST6
5). Thus, the bit string (codeword) output to the work area is output to the compressed file (ST66). This is 1
The compression on the character ends.

Next, the compression of the file is performed by the flow shown in FIG. 7, and the compression of the file starts (ST70). The compression for one character shown in the flow of FIG. 6 is performed, and the number of compression processes is counted in the counter B (ST72, ST73). Prior to this, the initial value of the counter B is set to 0 (ST71). Steps ST72 and ST73 are continued until the value of the counter B reaches the record length (ST74). At this time, each time the compression of one record is completed, the value of the record position counter (the total number of bits in one record) is output to the work area (ST7).
5). Then, the counter B is initialized (ST76).
The above steps ST72 to ST76 are continued until one file is completed (ST77). The value of each record position counter output to the work area in step ST75 is added to the compressed file (ST78), and the compression of the file ends. As a result, the compressed file is stored in the compressed data section 91 as shown in FIG.
.., n,) positions (end positions) are added as a record position information section 92 in the order of record numbers as the number of bits from the head of the compressed file.

It is appropriate to input a numerical value for the record length of the file to be compressed each time it is compressed. However, if the record length is always constant, it may be input to the compression program.

Further, the present method can be applied even when the length of each record of the compressed file is different, for example, by switching the record length of ST74 of FIG. 7 in the record order.

(Embodiment 2) An embodiment of a file compression method in the case where the file to be compressed is composed of a plurality of records of variable length according to claim 2 will be described.

In a file composed of such variable length records, the end of each record is usually delimited by a delimiter (delimiter), and compression of the file is performed by the flow shown in FIG. 1 shown in the flow of FIG.
The characters are compressed (ST81), and the delimiter inserted at the end of the record is compressed (ST82). The value of the record position counter at this time is output to the work area (ST83). Steps ST81-ST above
83 is continued until one file is completed (ST8
4). The value of each record position counter output to the work area in step ST83 is added to the compressed file (S
(T85) The compression for the file ends. As a result,
The compressed file has a compressed data section 91 as shown in FIG.
The position (end position) of each record (1, 2, ..., N,) in the inside is added as the record position information section 92 in the order of the record number as the number of bits from the head of the compressed file.

(Embodiment 3) An embodiment of the method for decompressing a compressed file according to the embodiment 1 or 2 will be described.

First, a method of restoring only an arbitrary record from the compressed file will be described. Since the "position of each record" data of the compressed file shown in FIG. 10 is the end position data of each record, assuming that the record to be restored is the i-th, "i-1 position of record" data is read from the compressed file. .

This data is i-1 from the top of the file.
It is represented by the total number of bits up to the end of the second record. When the compressed data section 91 of FIG. 10 is enlarged, it becomes as shown in FIG. 11, so that the total number of bits of the data is (j-1).
It can be converted into a byte unit of [byte] + k [bit].

Next, a method for extracting the i-th record from the compressed data section using the file pointer will be described. The file pointer is a logical address in C language and can be moved only in byte units, so the above conversion is performed. As shown in FIG. 11, the file pointer is set to (j-1) [byte] from the start end of the compressed data section 91.
After moving, the (k + 1) th bit of the jth [byte]
If the restoration process is started from the eye and continued for the record length, only the i-th record can be restored.

Since the "i-th record position" of the compressed file is the data up to the end of the i-th record, the record length of the i-th record is the "i-th record position" data. Then, it is obtained by subtracting the “i−1th record position” data.

The Huffman tree information is also used for restoration.

Next, the flowchart will be described in more detail.

First, the restoration for one character will be described. To restore one character, as shown in the flow of FIG. 12A, one bit is extracted from the compressed file (STa
121). The Huffman tree shown in FIG. 1 is lowered to the node one level lower than the fetched bit, that is, when the fetched bit is "0", the lower left node is lowered, and when it is "1", the lower right node is lowered (STa122). . At this time, the number of processing times for one bit is counted in the counter A '(STa12
3). In step STa122, the process is repeated until the leaves of the Huffman tree are reached (STa124). The character corresponding to the leaf is output to the memory (STa125), and the restoration for one character is completed.

Next, the record restoration will be described.
The restoration flow is shown in FIG. In the compressed file in this case, the record position information is, for example, the end of the (i-1) th record is the total number of bits from the beginning of the compressed file, and the end of the i-th record is from the beginning of the compressed file. The total number of bits is m and stored in the "position of record i" and the "position of record i-1" in the compressed file. The record length of the “record i” after compression is obtained by mn.

The position information of the record to be restored and the record length information (converted from the position information of the record adjacent to the record to be restored as described above) are extracted from the compressed file (ST141). In order to read the position of the extracted i-th record, the file pointer is moved by (j-1) bytes as described in FIG. 11 (ST143). j
Extract the (k + 1) th bit from the beginning of the byte (S
(T144), restoration of one character is started from the (k + 1) bit (ST145). When the counter value A ′ of the number of processing times for one bit becomes the record length after compression (ST146), the decompression of the i-th record ends.

When the compressed file is composed of records of variable length and the end of each record is delimited by a delimiter (delimiter), when the characters of the record match the above delimiter. Alternatively, only the desired record may be restored by determining the end of the record and ending the restoration.

(Embodiment 4) A compression and decompression method corresponding to claim 4 and claim 5 when the file to be compressed is composed of a plurality of records of a predetermined length (the number of character strings). An example will be described. This method is characterized in that the predetermined record length information of the compressed file is added to the compressed file, and the program software of the decompression method is simplified as compared with the method of the above embodiment. As a result, there is an advantage that the processing speed can be increased. The record length information is preferably added to, for example, the header of the compressed file, but may be added to another location, such as the record position information section.

First, the method of compressing a file according to claim 4 is the same as the method of the first embodiment. Next, a method of restoring the compressed file according to claim 5 will be described.

FIG. 12 shows a flow of restoring one character and a flow of restoring a record in this case, respectively.
(B) and FIG. In FIG. 12B, one character is extracted from the compressed file to restore one character (STb121). The extracted bit lowers the Huffman tree shown in FIG. 1 to the next lower node. That is, when the taken out bit is "0", it is lowered to the lower left node, and when it is "1", it is lowered to the lower right node (STb122). Step STb1
22 until the Huffman tree leaves are reached (ST
b123). Output the characters corresponding to the leaves to the memory (S
(Tb124) The restoration for one character ends. FIG.
At, the position information and record length of the record to be restored are extracted from the compressed file (ST131). In order to read the position of the taken out i-th record, the file pointer is moved by (j-1) bytes as described in FIG. 11 (ST133). J in the compressed data part of the compressed file
Extract the (k + 1) th bit from the beginning of the byte (S
(T134), the restoration for one character is started from the (k + 1) bit (ST135). At this time, counting of the number of restoration processes is started (ST136), and when the value of the counter A reaches the record length before compression (ST137), i
Restoration of the th record ends.

It is desirable to input a numerical value for the record length of the file to be compressed each time it is compressed or restored.
If the record length is always constant, it may be input to the compression or decompression program.

In the above embodiment, the position information of each record after compression is represented by the number of bits from the beginning of the data of the first record to the end of the data of each record, but other information, for example, The number of bits up to the beginning of each record, the data length of each record, etc. may be used.

[0064]

As described above, the Huffman tree is uniquely determined for one character string and does not change. Further, since uniquely decodable encoding can be performed, compressed data can be restored from any record (codeword string) from the beginning of each codeword.

Therefore, by recording the position information of each record at the time of compression and adding it to the compressed file, any record can be restored.

For example, a data file compressed by the Huffman method on a personal computer can be downloaded and copied to a small portable terminal, and only the records required when using this can be restored on the main memory. As a result, the free area of the RAM, which is the auxiliary storage device of the small mobile terminal, is not used for the restored data file.

Further, since only the records that are sufficiently shorter than the file length are restored, the processing time is shorter than the restoration for the entire compressed file.

[Brief description of drawings]

FIG. 1 shows a Huffman tree of a compression algorithm.

FIG. 2 shows a Huffman code for the character string shown in FIG.

FIG. 3 shows a compressed character string.

FIG. 4 is a flowchart showing a conventional compression operation for one character.

FIG. 5 is a flowchart showing a conventional compression operation for files.

FIG. 6 is a flowchart showing a compression operation for one character according to the embodiment of the present invention.

FIG. 7 is a flowchart showing a compression operation for a file according to the embodiment of the present invention.

FIG. 8 is a flowchart showing a compressing operation for a file according to another embodiment of the present invention.

FIG. 9 is an explanatory diagram of a conventional compressed file structure.

FIG. 10 is an explanatory diagram of a compressed file structure according to the embodiment of the present invention.

FIG. 11 is an explanatory diagram of a method of restoring only an arbitrary record at the time of restoration according to the embodiment of the present invention.

FIG. 12A and FIG. 12B are flowcharts showing a restoring operation for one character according to the embodiment of the present invention in different examples.

FIG. 13 is a flowchart showing a record restoring operation according to the embodiment of the present invention.

FIG. 14 is a flow chart diagram showing a restoring operation for a record according to another embodiment of the present invention.

FIG. 15 is an explanatory diagram of a method of using the data file of the present invention.

FIG. 16 is an explanatory diagram of a conventional method of using a data file.

FIG. 17 is an explanatory diagram of a conventional method of using a data file.

FIG. 18 is a block diagram showing a configuration example of a small mobile terminal 1.

[Explanation of symbols]

 1 small portable terminal 10 small portable terminal system 12 small portable terminal auxiliary storage device (RAM) 80, 90 compressed files

Claims (5)

[Claims] 1. A method of compressing a file composed of a plurality of records of a predetermined length (the number of character strings) by the Huffman method, wherein the record length (the number of character strings) before the compression is compressed. ) And the data length after compression, the position information of each record after compression is obtained, and the position information of each record is added to the file after compression. 2. A method of compressing a file composed of a plurality of variable-length records by the Huffman method, wherein the file has a delimiter inserted at the end of each record, and the characters of the record are compressed at the time of compression. By determining the end of each record when it matches the delimiter and counting the data length of the compressed record,
A file compression method, wherein the position information of each record after compression is obtained and the position information of each record is added to the file after compression. 3. A file restoration method for restoring only a desired record from the compressed file of claim 1 or the compressed file of claim 2 based on the position information of each record after compression. 4. The file compression method according to claim 1, wherein information on each record length of the compressed file is added to the compressed file. 5. A file restoration, wherein only a desired record is restored from the compressed file of claim 4 based on the position information of each record after compression and the information of each record length of the compressed file. Method.