JP2005215951A - Encoding or decoding method for document data, and program therefor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a code processing method for document data and a program, allowing improvement of an encoding compressibility by use of a grammatical characteristic of the XML document data. <P>SOLUTION: For the purpose of encoding the XML original document data by use of a conversion sheet of XML designated with a code corresponding to an element name and an attribute name, the element name, the attribute name, and a value corresponding to the attribute name are detected in tag units of the original document data, first. Next, a first code obtained by combining the code derived from the element name by use of the conversion sheet and a code wherein presence/absence of the attribute name is expressed by a bit is sequentially recorded in a structure table. Next, a second code obtained by encoding the value corresponding to the present attribute name according to appearance frequency is recorded in an entry of a value table linked to an entry of the structure table recording the first code. The structure table and the value table generated by repeating from the first step to the third step are set as code data. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a document data encoding or decoding method and program for encoding or decoding original document data described in an extensible text format structured description language.

  Conventionally, there is a method of encoding and decoding document data in order to reduce the amount of data to be transmitted. In order to realize this method, each of the transmission device and the reception device needs to have a conversion table. The conversion table is a one-to-one correspondence between a structured description language and code data. The transmission device encodes the document data into code data based on the conversion table. On the other hand, the receiving device decodes the code data into document data based on the conversion table. Such a method is also effective from the viewpoint of security on the Internet. This is because a client that does not have a conversion table cannot decode the encoded data.

  Specifically, there is a method of encoding XML / SGML document data using an encoding table compliant with XML (eXtensible Markup Language) or SGML (Standard Generalized Markup Language) (see, for example, Patent Documents 1 and 2). . This method defines a code length and a code for the element name, element value, attribute name, and attribute value items in the conversion table described in the XML / SGML structured type description language, and sets the first element name to the first element name. The element name of 2 defines the code length and code indicating the parent-child relationship. By encoding the document data using this conversion table, the data transmission amount can be reduced. Further, since the decoding side apparatus does not restore the original document data from the code data, it does not require a parser.

  FIG. 1 is a system configuration diagram including an encoding server.

  The server 4 transmits document data in XML format to the encoding server 6. The encoding server 6 encodes the document data using the conversion table received from the conversion table server 7. The code data is transmitted to the client 5. The client 5 performs document processing using the conversion table received from the conversion table server 7. According to FIG. 1, the encoding server can be used as a proxy server without changing an existing server that transmits XML-format document data.

JP 2002-259194 A Kobayashi, Takagi, Muramatsu, Baba, Matsumoto, Inoue, “XML document general-purpose encoding method“ xeus ”, IEICE Tech. 65-72 Kobayashi, Matsumoto, Inoue “Performance evaluation of general-purpose XML document encoding method“ XEUS ””, FIT (Information Science and Technology Forum), 2003, LE008, pp. 99-100

  Compared to such a conventional technique, it has been a problem to further improve the code compression rate. In the prior art, XML document data is encoded in a permutation for each tag, and does not use the nature of the XML grammar. For XML, the conditions that the attribute with the same name per element can be described only once, the correlation of the logical structure of the encoding target document, and the characteristics of each value of the encoding target document are fully utilized. It was not an encoding method.

  In ordinary XML document data, although the logical structure portion is often the same or similar tags in many cases, in order to encode a series of tags, after all, in the encoded bit string itself. In many cases, the same bit string continued.

  SUMMARY OF THE INVENTION Accordingly, it is an object of the present invention to provide a document data encoding method and program in which the encoding compression rate is further improved by utilizing the grammatical characteristics of XML document data.

The encoding method in the present invention is:
A conversion sheet that is written in a structural description language in an extensible text format, and that has an element name and a code corresponding to an attribute name that can be specified in the element name;
A structure table that records the structure code for each element in the entry;
Corresponding to the entries in the structure table, using the value table that records the sign of the attribute value or element value in the entry, in the tag unit order of the original document data,
A first step of detecting an element name, presence or absence of a child element or element value, presence or absence of an attribute name, and existing attribute value or element value;
A structure code combining a code derived from an element name using a conversion sheet, a code indicating the presence / absence of a child element or element value of the element name, and a code indicating the presence / absence of an attribute name is stored in the structure table entry. A second step of recording in sequence;
A third step of recording a code for an existing attribute value or element value in a value table, and a structure table and a value table generated by repeating the first step to the third step as code data It is characterized by doing.

  According to another embodiment of the encoding method of the present invention, the method further includes a value assignment table in which short codes are assigned in order from the attribute value or element value having the highest appearance frequency, and the attribute value or element value in the third step. The code for is derived using a value assignment table, and the value assignment table is also preferably part of the code data.

  According to another embodiment of the encoding method of the present invention, the conversion sheet defines data type information corresponding to the attribute value or element value, and a plurality of value assignment tables are provided according to the data type information. It is also preferable that the code for the attribute value or element value in the third step is derived using a value assignment table corresponding to the data type information in the attribute value or element value.

  Furthermore, according to another embodiment of the encoding method of the present invention, the structure table assigns short codes as tree codes in order from the structure code having the highest appearance frequency, and the tree code is assigned to the element of the original document data. A tree table arranged in the order of appearance is further provided, and the tree table is also preferably a part of the code data.

  Furthermore, according to another embodiment of the encoding method of the present invention, the code indicating the presence or absence of an attribute name is represented by a bit string in which each attribute is assigned to 1 bit in the order of attribute names that can be specified for the element name in the conversion sheet. It is also preferable to specify the presence or absence of the attribute.

  Furthermore, according to another embodiment of the encoding method of the present invention, the encoding method of the present invention further includes an attribute table in which short codes are assigned in order from a code indicating the presence / absence of an attribute name having a high appearance frequency. It is also preferable that a code representing the presence or absence of an attribute name is derived using a table, and the attribute table is also part of the code data.

  Furthermore, according to another embodiment of the encoding method of the present invention, selection type information corresponding to an attribute value or element value can be defined in the conversion sheet, and the attribute value or element value in the third step is defined. It is also preferable that the code for is represented by the minimum number of bits corresponding to the number of selections.

  Furthermore, according to another embodiment of the encoding method of the present invention, group structure type information in which attribute values or element values consist of a plurality of groups can be defined in the conversion sheet, and the attribute in the third step It is also preferable that the code for the value or the element value is encoded in the group unit for the previous group and then only the difference value for the previous group is encoded for the subsequent group.

  Furthermore, according to another embodiment of the encoding method of the present invention, a separator can be defined in the conversion sheet to represent attribute values or element values as a plurality of groups, and the attribute value in the third step. Or it is also preferable that the code | symbol with respect to an element value is divided into a group unit with a separator.

  Furthermore, according to another embodiment of the encoding method of the present invention, the conversion sheet can define a fixed value portion for the attribute value or element value, and the attribute value or element value in the third step. It is also preferable that the fixed value portion of the sign for is omitted.

The decoding method in the present invention is:
The code data is composed of a structure table in which the structure code for each element is recorded in the entry, and a value table in which the attribute value or element value code is recorded in the entry corresponding to the entry in the structure table.
Using a conversion sheet that is described in a structural description language in an extensible text format, and that has an element name and a code corresponding to an attribute name that can be specified for the element name,
A first step of extracting a corresponding structure code from the structure table;
A second step of detecting an entry in the value table corresponding to the structure code, and using the conversion sheet to identify an element name, presence or absence of a child element or element value, and an existing attribute name from the structure code;
A third step of specifying an existing attribute value or element value from the value code of the corresponding entry in the value table, and decoding the original document data by repeating the third step from the first step It is characterized by.

  According to another embodiment of the decoding method of the present invention, the code data further includes a value assignment table in which short codes are assigned in order from an attribute value or an element value having a high appearance frequency, in the third step. The attribute value or element value is preferably derived using a value assignment table.

  According to another embodiment of the decoding method of the present invention, the conversion sheet defines data type information corresponding to attribute values or element values, and a plurality of value assignment tables are provided according to the data type information. It is also preferable that the attribute value or element value in the third step is derived using a value assignment table corresponding to the data type information in the attribute value or element value.

  Furthermore, according to another embodiment of the decoding method of the present invention, the code data is such that the structure table is assigned with a short code as a tree code in order from the structure code having the highest appearance frequency. It is also preferable to further include a tree table arranged in the order in which the document data elements appear.

  Furthermore, according to another embodiment of the decoding method of the present invention, the code indicating the presence or absence of an attribute name is represented by a bit string in which each attribute is assigned to one bit in the order of attribute names that can be specified for the element name in the conversion sheet. It is also preferable that the presence or absence of the attribute is specified.

  Furthermore, according to another embodiment of the decoding method of the present invention, the code table further includes an attribute table to which a short code is assigned in order from a code indicating the presence / absence of an attribute name having a high appearance frequency. It is also preferable to derive the presence / absence of an attribute name using the attribute table.

  Furthermore, according to another embodiment of the decoding method of the present invention, the conversion sheet can specify selection type information corresponding to the attribute value or element value, and the attribute value or element value in the third step. It is also preferable that the code for is represented by the minimum number of bits corresponding to the number of selections.

  Furthermore, according to another embodiment of the decoding method of the present invention, group structure type information in which attribute values or element values are composed of a plurality of groups can be defined in the conversion sheet, and the attribute in the third step The value code of the value or element value is obtained by encoding only the difference value with respect to the previous group for the subsequent group after encoding the previous group and decoding for the previous group. It is also preferable that the difference value is added to the value obtained by decoding the later group later.

  Furthermore, according to another embodiment of the decoding method of the present invention, a separator can be defined in the conversion sheet to represent attribute values or element values as a plurality of groups, and the attribute value in the third step. Alternatively, the value code of the element value is preferably divided into group units by a separator.

  Furthermore, according to another embodiment of the decoding method of the present invention, the conversion sheet can define a fixed value portion for the attribute value or element value, and the attribute value or element value in the third step. It is also preferable to derive the value code by restoring the omitted fixed value portion using a conversion sheet.

The encoding program in the present invention is:
A conversion sheet that is written in a structural description language in an extensible text format, and that has an element name and a code corresponding to an attribute name that can be specified in the element name;
A structure table that records the structure code for each element in the entry;
Corresponding to the entries in the structure table, using the value table that records the sign of the attribute value or element value in the entry, in the tag unit order of the original document data,
A first step of detecting an element name, presence or absence of a child element or element value, presence or absence of an attribute name, and existing attribute value or element value;
A structure code combining a code derived from an element name using a conversion sheet, a code indicating the presence / absence of a child element or element value of the element name, and a code indicating the presence / absence of an attribute name is stored in the structure table entry. A second step of recording in sequence;
A code for an existing attribute value or element value is executed as a third step of recording in the value table, and a structure table and a value table generated by repeating the first step to the third step are used as code data. It is characterized by that.

The decryption program in the present invention is:
The code data is composed of a structure table in which the structure code for each element is recorded in the entry, and a value table in which the attribute value or element value code is recorded in the entry corresponding to the entry in the structure table.
Using a conversion sheet that is described in a structural description language in an extensible text format, and that has an element name and a code corresponding to an attribute name that can be specified for the element name,
A first step of extracting a corresponding structure code from the structure table;
A second step of detecting an entry in the value table corresponding to the structure code, and using the conversion sheet to identify an element name, presence or absence of a child element or element value, and an existing attribute name from the structure code;
It is executed as a third step for specifying an existing attribute value or element value from the value code of the corresponding entry in the value table, and decoding into the original document data by repeating the third step from the first step. Features.

  According to the document data encoding method and program of the present invention, the encoding compression rate can be further improved. In particular, since the logical structure portion and the value portion of the document data that are encoded together are encoded separately, it is possible to realize the optimal encoding for the characteristics of each portion.

  In particular, since encoding can be performed using a bias between the appearance frequency of the logical structure and the filing frequency of values, the code amount can be reduced by approximately 20% compared to the conventional method. Such an encoding method can be applied to encoding of all document data conforming to XML and SGML, and in particular, a change amount between coordinate values is small and continuous or identical, such as a list of coordinate values of map content. The code amount reduction effect is extremely large for content in which approximate tags are continuous.

  In addition, such a reduction in the code amount is extremely effective for a terminal connected to a narrow-band wireless link such as a mobile phone. Since such an encoding method does not require parsing processing, it is extremely effective for a terminal having a low CPU capability such as a mobile phone.

  Hereinafter, the best mode for carrying out the present invention will be described in detail with reference to the drawings.

  FIG. 2 shows a document data code processing method according to the present invention.

  According to FIG. 2, the document data 12 is expanded by a plurality of document data 120 and 121. On the other hand, the conversion table 11 also defines link information of a plurality of conversion tables 110 and 111 corresponding to the expanded document data. Thereby, the document data 12 in the XML format is encoded 10 using the conversion table 11.

  In addition, according to FIG. 2, the code data is directly processed 30 using the conversion table 21 and displayed on the browser 24. According to the present invention, the code data includes the logical structure of the elements. Accordingly, there is no need to decrypt the document data, and there is no need to analyze the logical structure by the parser 23.

  FIG. 3 is a sample of document data. Hereinafter, this document data will be described as an example.

  FIG. 4 shows various encoding tables corresponding to the document data of FIG.

  According to FIG. 4, it has a tree table, a structure table, and an attribute table to show the logical structure of document data, and has a value table to show values. As described above, the present invention is characterized in that the part indicating the logical structure and the value part are encoded separately.

  Here, the code for linking the tree table and the structure table and the code for presence / absence of the attribute are represented by a Huffman code. The Huffman code is a method in which a short code is assigned to an item with a high appearance frequency, and a long code is assigned to an item that does not. By efficiently performing such variable length coding, efficient data compression can be realized as compared with a method of assigning a code of a certain size to each piece of information.

  FIG. 5 is a Huffman table (value assignment table) corresponding to the document data of FIG.

  According to FIG. 5, it has an integer Huffman table, a character string Huffman table, and an occupied bit length table. Similar to FIG. 4, these tables are also assigned by Huffman codes. Therefore, each table is sorted according to the appearance frequency, and short codes are assigned in order from the one with the highest appearance frequency. This index code is used for the sign of the value. By using a variable length coding method with appearance frequency such as Huffman coding as the index code allocation method, the code amount can be reduced. According to the prior art, the encoding of values was performed using the codes defined in the encoding rules. However, according to the present invention, a value appearance frequency table is created for each data type, and each value is The code length is dynamically calculated based on the appearance frequency of and the encoding is performed.

  6A and 6B are conversion sheets.

  The conversion sheet is preliminarily held by an apparatus for encoding or decoding without depending on document data, and defines a correspondence relationship between element names or attribute names and codes.

  FIG. 7 is an explanatory diagram showing a process of encoding the document data in FIG. Below, based on FIG. 7, it demonstrates, referring each table.

[T1] <svg>
->"101"
First, for “svg”, a code “00” is derived from [x1] of the conversion sheet (FIG. 6A). At this time, since there is a child element of svg, it is set to “1”, and since there is no attribute, it is set to “0”.
For “svg”, “0010” is recorded in the structure table (F), and since there is no value, it is not recorded in the corresponding value table. Further, the code “101” is assigned to the entry F of the structure table. This code “101” is recorded in the tree table in association with “svg”.

[T2] <g id = "300" stroke = "# 1e90ff" stroke-width = "1" fill = "none">
->"011","100,1110,111,00,0"
For “g”, the code “01” is derived as a child element of svg by the conversion sheet [x3]. At this time, since there is a child element of g, it is set to “1”. Further, it can be understood from the conversion sheets [x7] to [x10] that there are four attributes, and since all these four attributes are included in [T2], the code is “1111”. Each bit represents id, stroke, stroke-width, and fill. According to the XML specification, the same attribute cannot be described twice in one element, and since it is either the presence or absence of each attribute, such description is possible. Here, referring to the attribute table, “1111” corresponds to the code “01”. Therefore, “g” can be encoded as “01101” as a whole. For “g”, “01101” is recorded in the structure table (D). Here, “011” is assigned to the entry D of this structure table. This code “011” is recorded in the tree table in association with “g”.

According to the present invention, instead of assigning a code to each element name and attribute name like “g” element = “010101”, “id” element = “101010” as in the prior art, “g” Is "
By assigning 4 bits to the presence / absence of the “id, stroke, stroke-width, fill” attribute, only that value needs to be encoded.

For “id”, the attribute value “300” is set to “300”-> “1110” by the integer Huffman table, and the size of 4 bits is set to the code “100” by the occupied bit length table.
For "stroke", the attribute value "# 1e90ff" derives the code "111" from the character string Huffman table.
For "stroke-width", the attribute value "1" derives the code "00" from the integer Huffman table.
As for “fill”, the attribute value “none” is a selection type according to the conversion sheet [x11], and therefore the code “0” is derived.
Finally, the value code “100,1110,111,00,0” of “g” is recorded in entry D of the corresponding value table.

Here, according to the conversion sheets [x11] [x12], a data type called a selection type is designated for “fill”. According to the prior art, even if the candidates (options) that the value can take are limited, it is necessary to encode the value with a data type such as a numerical value or a character. According to the present invention, since the attribute value of the fill attribute of the g element has only a character string of none or 7 characters, if the value is none, only the code “0” is sufficient. If it is ASCII, 4 bytes can be encoded with 1 bit.
Thereby, it can encode with the index code allocated to the choice, and can implement | achieve code amount reduction.

[T3] <polyline points = "88,99 96,102 108,109"/>
->"000","11111,11110,0100,0101,0110,0111"
For “polyline”, the code “10” is derived as a child element of “g” by the conversion sheet [x14]. At this time, since there is no child element of polyline, it is set to "0". Further, it can be understood from the conversion sheets [x17] to [x21] that there are five attributes, and since only the attributes of the points are included in [T2], the code is “10000”. Each bit represents "points, stroke, stroke-width, fill, stroke-dasharray". From this code “10000”, the code “100” is derived using the attribute table. These bit strings “100100” are recorded in the structure table (A). A code “000” is assigned to this entry A in the structure table, and the code “000” is recorded in the tree table in association with “polyline”.

  It can be understood that the value of “points” is a coordinate “coordinates” type. Therefore, it can be understood that “= 88,99 96,102 108,109” is a coordinate composed of the group of (88,99) (96,102) (108,109). Here, "88,99 96,102 108,109" is not directly encoded as six numerical value groups, but by using a coordinate value type as a data type, a difference value from the previous coordinate value group, that is, "88,99" , +8, +3, +12, +7 ", and encode. Using the integer Huffman table, the first values "88"-> "11111", "99"-> "11110" can be encoded. Next, “96,102” is expressed as a difference value “+ 8, + 3” of the previous coordinate value group, and can be encoded as “0100,0101” using the integer Huffman table. Next, “108,109” is expressed as a difference value “+ 12, + 7” of the previous coordinate value group, and can be encoded as “0110,0111” using the integer Huffman table. Such encoding is extremely effective for data (for example, map data) with a small amount of change between coordinates. Finally, the value code “11111,11110,0100,0101,0110,0111” of “polyline” is recorded in entry A of the corresponding value table.

  As for a numerical value and a numerical value and a separator (separator), a space “” is usually described, but according to the conversion sheet [x17], a comma “,” is defined. Depending on the value, the separator is also different. In such a case, instead of encoding as a character string as in the prior art, it is also possible to define a separator in the attribute as an encoding rule and thereby divide and encode the value. Here, for the value of “points”, the X and Y coordinates are separated by “,” and the pair is further separated by “”. Thus, high encoding efficiency can be realized by combining a plurality of separators according to value attributes.

[T4] </ g>
->"011"
The logical structure code “011” of “/” is recorded in the logical structure table.

[T5] <g id = "500" fill = "# ff4500">
->"100","01,111,1,111"
For “g”, as a child element of svg, a code “01” is derived from the conversion sheet [x3]. At this time, since there is a child element of g, it is set to “1”. Further, it can be understood from the conversion sheets [x7] to [x10] that there are four attributes, and since only the attributes of id and fill are included in [T5], the code is “1001”. Each bit represents "id, stroke, stroke-width, fill". Here, referring to the attribute table, “1001” corresponds to the code “11”. Therefore, “g” can be encoded into “01111” as a whole, and “01111” is recorded in the structure table (E). Here, “100” is assigned to the entry E of this structure table. This code “100” is recorded in correspondence with “g” in the tree table.

For "id", the attribute value "500" is "500"->"111" from the integer Huffman table, and its size is 3 bits as code "01".
For "fill", for attribute value "# ff4500", "fill" is set as the selection type. For 7 characters "# ff4500", the ASCII code "# ff4500" immediately after option code "1""Is encoded. At this time, “# ff4500” is assigned the code “111” by the character string Huffman table.
Finally, the value code “01,111,1,111” of “g” is recorded in entry E of the corresponding value table.

[T6] <text x = "64" y = "22"> ABC </ text>
->"001","110,01,110,110"
->"010","1000,1001,1010"
For “text”, the code “11” is derived from the conversion sheet [x15] as a child element of g. At this time, since there is no child element of text, it is set to "0". Further, it can be understood from the conversion sheets [x23] to [x26] that there are four attributes, and since only the attributes of x and y are included in [T6], the code is “1001”. Each bit represents id, stroke, stroke-width, and fill. Here, referring to the attribute table, “1001” corresponds to the code “11”. Therefore, “text” can be encoded as “11011” as a whole, and “11011” is recorded in the structure table (B). Here, “001” is assigned to the entry B of this structure table. This code “001” is recorded in the tree table in association with “text”.

For “x”, for the attribute value “64”, a code “01” is derived from the integer Huffman table, and its size of 2 bits is set to a code “110”.
For “y”, the attribute value “22” is represented by the code “110” according to the integer Huffman table, and the size 3 bits is represented by the code “110”.
Finally, the value code “110,01,110,110” of “text” is recorded in entry B of the corresponding value table.

  For the element value “ABC” of the text element, since this node is an element value, there is no element name, its child element or its element value, or attribute. Therefore, the element value “ABC” can be encoded as “NULL, 0, 0”, and is recorded in the structure table (C). Here, “010” is assigned to the entry (C) of this structure table. This code “010” is recorded in the tree table in association with “[value]”.

  Furthermore, “ABC” is represented as “1000, 1001, 1010” from the character string Huffman table. Record the tree table “010” corresponding to the structure table. Codes “1000, 1001, 1010” are recorded in entry C of the value table corresponding to the structure table (C).

</ text>
->"011"
The code “011” which is the logical structure of “/” is recorded in the tree table.

[T7] </ g>
->"011"
The logical structure code “011” of “/” is recorded in the tree table.

[T8] </ svg>
->"011"
The logical structure code “011” of “/” is recorded in the tree table.

  The other functions of the present invention will be described.

First, an implicit separator will be described. For example, in the case of a value of “100 yen”, when the data type changes such as “numerical value-> character”, a separator existing between “100” and “yen” is defined as an implicit separator. Although the selection type has been described with the conversion sheets [x11] and [x12], for example, when the value “100 yen” is encoded, the following conversion sheet can be prepared.
<attr name = "price" explicit_separator = "true">
<value>
<number data = "UI" qt = "1"/>
<char data = "Shift_JIS" length = "implied"/>
<choice>
<list code = "0"> Yen </ list>
<list code = "1"> $ </ list>
</ choice>
</ value>
</ attr>

Secondly, a case where only a character string is always defined will be described. For example, assume that a sample document has the following description.
<value>
http://www.kddi.com/
</ value>

In this case, since http://www.kddi.com/ can only be a character string, it can be converted into a very small number of codes by defining it in the character string Huffman table. In particular, since http: // and the last “/” are fixed, encoding of these parts can be omitted. In this case, the following rules can be written on the conversion sheet.
<value>
http: //
<char data = "ASCII" length = "implied"/>
<value>

  FIG. 8 is a flowchart of the encoding process in the present invention.

Assume that XML document data is input.
(S801) All elements and element values are encoded for each element in order from the first tag of the document data to S812.
(S802) First, it is determined whether or not it is an element value.
(S803) If it is not an element value, the code | symbol of the element of the said tag is derived | led-out using a conversion sheet.
(S804) Next, whether or not the element has a child element or element value is encoded with 1 bit.
(S805) According to the characteristics of XML, according to the conversion sheet, the element has several attributes, and the order of the elements is defined in advance. Therefore, a bit string corresponding to the number of attributes is arranged, and the presence / absence of each attribute is expressed by bits.
(S806) A series of code strings obtained in S803 to S805 is recorded in the structure table as a structure code. At this time, the structure table is Huffman-coded according to the appearance frequency. Therefore, a structural code having a high appearance frequency is represented by a short code, and a structural code having a low appearance frequency is represented by a long code.
(S807) The code specified by the structure table is recorded together with the element in the tree table.
(S808) Next, it is determined whether or not the element has an attribute. If there is no attribute, the element is processed next (S801).
(S809) If there is an attribute, all attributes are encoded for each attribute between the first attribute of the document data and S811.
(S810) Since the existence of the attribute has already been grasped by the attribute bit, it is sufficient to encode only the value of the attribute. Here, an occupied bit length table, an integer Huffman table, and a character string Huffman table are used. Since the attribute type is already known from the conversion sheet, the table can be selected according to the data type. Also here, since Huffman coding is used, a structural code having a high appearance frequency is represented by a short code, and a structural code having a low appearance frequency is represented by a long code.
(S811) It repeats until it encodes about all the attributes between S808.
(S812) If it is determined in S802 that the value is an element value, a code string “no element name, no child element, no attribute” is recorded in the structure table as a structure code. As described above, the structure table is Huffman-coded according to the appearance frequency.
(S813) The code specified by the structure table is recorded together with the element in the tree table.
(S814) The element value is encoded using the occupied bit length table, the integer Huffman table, and the character string Huffman table.
(S815) The code indicating the end of element “/” is recorded in the tree table.
(S816) It repeats until it encodes about all the elements between S801 and element values.
(S817) It repeats until it encodes about all the elements between S801.

  FIG. 9 is a flowchart of the decoding process in the present invention.

Assume that code data is input. The code data basically includes a tree table and a value table. Of course, a structure table, an attribute table, an occupied bit length table, an integer Huffman table, and a character string Huffman table are also required, but these tables may be stored in advance on the encoding side and the decoding side. Good.
(S901) All elements are encoded for each element in order from the beginning of the code data to S908.
(S902) First, the code of the element structure is cut out with reference to the tree table.
(S903) With reference to the structure table, a bit string of element name, child element presence / absence and attribute presence / absence in the entry corresponding to the code is searched. The element name bit string can be specifically grasped using the conversion sheet.
(S904) As the bit string indicating the presence / absence of an attribute, a bit string specifically indicating the presence / absence of the attribute is derived from the attribute table. From the conversion sheet, the attribute name corresponding to the element name can also be grasped, and which attribute is present can be grasped by the attribute presence / absence bit.
(S905) Next, the value table corresponding to the entry in the structure table is referenced. Thereby, the bit string of the value about the existing attribute can be cut out using the conversion sheet.
(S906) The value and character string of the extracted bit string can be derived from the occupied bit length table, integer Huffman table, and character string Huffman table.
(S907) Since the structure of one element can be grasped by the processing from S902, document processing is performed for the tag.
(S908) Iterate until it decodes about all the elements between S901.

  According to various embodiments of the document data encoding method and system of the present invention described above, those skilled in the art can easily make various changes, corrections and omissions in the technical idea and scope of the present invention. . The above description is merely an example, and is not intended to be restrictive. The invention is limited only as defined in the following claims and the equivalents thereto.

It is a system configuration diagram including an encoding server. 3 is a document data code processing method according to the present invention. This is a sample of document data. 4 is various encoding tables corresponding to the document data of FIG. 4 is a Huffman table corresponding to the document data in FIG. 3. It is the front | former part of a conversion sheet. It is the latter part of a conversion sheet. It is explanatory drawing showing the process of the encoding in the document data of FIG. It is a flowchart of the encoding process in this invention. It is a flowchart of the decoding process in this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Encoding 11, 21, 110, 111, 210, 211 Conversion table 12, 22, 120, 121 Text format document data 23 Parser 24 Browser display screen 30 Document processing 4 Existing server 5 Client 6 Code data 7 Conversion table Server 8 Internet

Claims (22)

In a document data encoding method for encoding original document data described in an extensible textual structured description language,
A conversion sheet that is written in a structural description language in an extensible text format, and that has an element name and a code corresponding to an attribute name that can be specified in the element name;
A structure table that records the structure code for each element in the entry;
Corresponding to the entries in the structure table, using a value table that records the sign of the attribute value or element value in the entry, in the tag unit order of the original document data,
A first step of detecting an element name, presence or absence of a child element or element value, presence or absence of an attribute name, and existing attribute value or element value;
The structure code obtained by combining a code derived from the element name using the conversion sheet, a code representing the presence or absence of a child element or element value of the element name, and a code representing the presence or absence of the attribute name, A second step of sequentially recording entries in the structure table;
A third step of recording a code for the existing attribute value or element value in the value table, and the structure table generated by repeating the third step from the first step, and A document data encoding method, wherein a value table is encoded data. An allocation table in which short codes are allocated in order from the attribute value or the element value having a high appearance frequency;
The sign for the attribute value or element value in the third step is derived using the value assignment table,
The method according to claim 1, wherein the value assignment table is also a part of the code data. The conversion sheet defines data type information corresponding to the attribute value or element value,
A plurality of the value allocation tables exist according to the data type information,
3. The code for the attribute value or element value in the third step is derived using a value assignment table corresponding to the data type information in the attribute value or element value. The method described in 1. In the structure table, a short code is assigned as a tree code in order from the structure code having the highest appearance frequency,
A tree table in which the tree codes are arranged in the order of appearance of the elements of the original document data;
4. The method according to claim 1, wherein the tree table is also a part of the code data.   2. The code representing the presence / absence of the attribute name specifies the presence / absence of the attribute by a bit string in which each attribute is assigned to 1 bit in the order of attribute names that can be specified for the element name in the conversion sheet. 5. The method according to any one of items 1 to 4. In order from the code representing the presence or absence of the attribute name having a high appearance frequency, further having an attribute table assigned a short code,
The third step derives a code representing the presence or absence of the attribute name using the attribute table,
6. The method according to claim 5, wherein the attribute table is also a part of the code data. In the conversion sheet, selection type information corresponding to the attribute value or element value can be defined,
7. The method according to claim 1, wherein the sign for the attribute value or the element value in the third step is expressed by a minimum number of bits corresponding to the number of selections. . The conversion sheet can specify group structure type information in which the attribute value or element value is composed of a plurality of groups,
The code for the attribute value or element value in the third step is characterized in that, after encoding for the previous group, only the difference value for the previous group is encoded for the subsequent group. The method according to any one of claims 1 to 7. In the conversion sheet, a separator can be defined to represent the attribute value or element value as a plurality of groups,
9. The method according to claim 8, wherein the sign for the attribute value or element value in the third step is divided into groups by the separator. In the conversion sheet, a fixed value part can be defined for the attribute value or the element value,
The method according to claim 1, wherein the fixed value portion is omitted from the sign for the attribute value or the element value in the third step. In a document data decoding method for decoding encoded data of original document data described in a structural description language in an extensible text format,
The code data includes a structure table in which a structure code for each element is recorded in an entry, and a value table in which an attribute value or a code of an element value is recorded in the entry corresponding to the entry in the structure table.
Using a conversion sheet that is described in a structural description language in an extensible text format, and that has an element name and a code corresponding to an attribute name that can be specified for the element name,
A first step of extracting a corresponding structure code from the structure table;
A second entry that detects an entry in the value table corresponding to the structure code and identifies an element name, the presence or absence of a child element or an element value, and an existing attribute name from the structure code using the conversion sheet Steps,
A third step of specifying an existing attribute value or element value from the value code of the corresponding entry in the value table, and by repeating the third step from the first step, the original document data is converted into the original document data. A method for decrypting document data, comprising decrypting the document data. The code data further includes a value assignment table in which short codes are assigned in order from the attribute value or element value having a high appearance frequency,
The method according to claim 11, wherein the attribute value or element value in the third step is derived using the value assignment table. The conversion sheet defines data type information corresponding to the attribute value or element value,
A plurality of the value allocation tables exist according to the data type information,
The attribute value or element value in the third step is derived using a value assignment table corresponding to the data type information in the attribute value or element value. the method of.   In the code data, the structure table is assigned with a short code as a tree code in order from the structure code having the highest appearance frequency, and a tree table in which the tree code is arranged in the order of appearance of the elements of the original document data is further provided. 14. A method according to any one of claims 11 to 13, characterized in that it comprises.   The code indicating the presence / absence of the attribute name is characterized in that the presence / absence of the attribute is specified by a bit string in which each attribute is assigned to 1 bit in order of attribute names that can be specified for the element name in the conversion sheet. Item 15. The method according to any one of Items 11 to 14. The code table further includes an attribute table to which a short code is assigned in order from a code representing the presence or absence of the attribute name having a high appearance frequency,
The method according to claim 15, wherein the third step derives the presence or absence of the attribute name using the attribute table. In the conversion sheet, selection type information corresponding to the attribute value or element value can be defined,
The method according to any one of claims 11 to 16, wherein the sign for the attribute value or the element value in the third step is represented by a minimum number of bits corresponding to the number of selections. . The conversion sheet can specify group structure type information in which the attribute value or element value is composed of a plurality of groups,
The value code of the attribute value or element value in the third step is obtained by encoding only the difference value with respect to the previous group for the subsequent group after encoding the previous group in the group unit. 18. The method according to claim 11, wherein after decoding the previous group, the difference value is added to a value obtained by decoding the subsequent group. Method. In the conversion sheet, a separator can be defined to represent the attribute value or element value as a plurality of groups,
The method according to claim 18, wherein the value code of the attribute value or element value in the third step is divided into groups by the separator. In the conversion sheet, a fixed value part can be defined for the attribute value or the element value,
21. The value code of the attribute value or element value in the third step is derived by restoring the omitted fixed value portion using the conversion sheet. 2. The method according to item 1. In a document data encoding program that encodes original document data described in a structural description language in an extensible text format,
A conversion sheet that is written in a structural description language in an extensible text format, and that has an element name and a code corresponding to an attribute name that can be specified in the element name;
A structure table that records the structure code for each element in the entry;
Corresponding to the entries in the structure table, using a value table that records the sign of the attribute value or element value in the entry, in the tag unit order of the original document data,
A first step of detecting an element name, presence or absence of a child element or element value, presence or absence of an attribute name, and existing attribute value or element value;
The structure code obtained by combining a code derived from the element name using the conversion sheet, a code representing the presence or absence of a child element or element value of the element name, and a code representing the presence or absence of the attribute name, A second step of sequentially recording entries in the structure table;
The structure table and the value generated by executing the third step of recording a sign for the existing attribute value or element value in the value table and repeating the third step from the first step An encoding program for document data, characterized in that the table is encoded data. In a document data decoding program for decoding code data of original document data described in an extensible textual structured description language,
The code data includes a structure table in which a structure code for each element is recorded in an entry, and a value table in which an attribute value or a code of an element value is recorded in the entry corresponding to the entry in the structure table.
Using a conversion sheet that is described in a structural description language in an extensible text format, and that has an element name and a code corresponding to an attribute name that can be specified for the element name,
A first step of extracting a corresponding structure code from the structure table;
A second entry that detects an entry in the value table corresponding to the structure code and identifies an element name, the presence or absence of a child element or an element value, and an existing attribute name from the structure code using the conversion sheet Steps,
This is executed as a third step of specifying an existing attribute value or element value from the value code of the corresponding entry in the value table, and is decoded into original document data by repeating the third step from the first step. A program for decrypting document data.

Images