JP4565064B2 - Fast matching method - Google Patents

Description

  The present invention relates to a high-speed matching method, and more particularly to a high-speed matching method for searching a character string that matches a search character string from text.

  Conventional string search methods include a simple method, a KMP (Knuth-Morris-Pratt) method, a BM (Boyer-Moore) method, and a BMH (Boyer-Moore-Horspool) method. In the simple method, as shown in FIG. 10A, the text and the search character string are compared one by one, and if they do not match, the search character string is shifted to the right by one character. Even in the simple method, there is not so much problem in practical use, but the calculation amount in the worst case is proportional to (search character string length m × text character string length n). In order to increase the speed of character string matching, it is desirable to avoid as many unnecessary comparisons as possible when a match or mismatch is known.

  The KMP method is the first algorithm that shows that the string pattern matching problem can be solved in linear time. This is a method of obtaining a shift amount according to the number of matching characters between a text and a search character string, making it a skip table, and shifting while referring to the skip table. Before scanning the text, a skip table is constructed from the search character string as preprocessing. When a mismatch is found after matching only k characters in the search character string, it is checked in advance which character in the search character string should be compared with which character in the search character string. Using the fact that the left k characters are known from the mismatched position, settings are made so as to avoid unnecessary comparisons. As shown in FIG. 10 (b), this is recorded in the form of a skip table. In the calculation for obtaining the skip table, only the search character string is examined. The skip table can be represented by an array of size (m + 1) × 3, where m is the search character string length. The time required for the configuration of the skip table is O (m) time. When the text length is n, the number of character comparisons is at most 2n. Therefore, the text scanning time is O (n) time. However, it is inferior in practical performance compared to the string matching algorithm by the BM method.

  The BM method is a collation algorithm that is even faster than the KMP method. This is a method in which a skip table 1 and a skip table 2 are provided to select the one with the larger shift amount. If the search character string is longer than a certain length (usually about 5 characters or more), it is said to be the fastest character string matching algorithm. Similar to the KMP method, the computational complexity is O (n) at worst. As with the KMP method, when collating, the information obtained as a result of the comparison performed before that is used as much as possible to reduce the number of comparisons. The advantage of the BM method is that you may not have to examine most of the characters in the text. In simple and KMP methods, each character in the text must be examined once. There are at least n comparisons. On the other hand, in the BM method, it may be necessary to compare only with n / m characters (where n is the length of the text and m is the length of the search character string). This can be expected to greatly reduce the calculation time, especially when the search character string is long. Due to this feature, the BM method is regarded as an extremely important algorithm for practical use.

  In the BM method, before scanning the text, the character component and the character arrangement in the search character string are examined, and the skip table 1 and the skip table 2 are prepared. In creating the skip table 1, it is checked whether the partial character string from the last position of the search character string is included before the search character string, and the shift amount for the repeated partial character string is calculated. Keep it. The skip table 1 is calculated in the same manner as in the KMP method, and does not consider any text characters that do not match the search character string. The skip table 1 is calculated only from the character configuration in the search character string. The shift amount is determined by whether or not the text character string partially matched with the search character string is included in another position of the search character string. The time required for the configuration of the skip table 1 is O (q + m) time when the search character string length is m and the number of characters used is q.

  The skip table 2 considers whether the search character string includes text characters that do not match the search character string. In creating the skip table 2, the component characters of the search character string are obtained, and the shortest number of characters from the last position in the search character string of each component character is calculated as the shift amount. In the skip table 2, the shift amount is determined by the text-side character that does not match in the character comparison with the search character string. In order to create the skip table 2, it is necessary to calculate the shift amount for all the characters that may be included in the text. Therefore, the BM method complicates skip table creation. The time required for the configuration of the skip table 2 is O (m) time. An example of the skip table is shown in FIG.

  As with the simple method, the BM method examines text in order from the left. However, once the position on the text is determined, the search string is examined from right to left. Examining the search string in the reverse direction is the main point of the BM method. The idea of comparing characters from the right end of the search string can often skip some of the characters in the text and is known as the fastest algorithm. When evaluating the efficiency with the number of text character references, the average number of references is less than the text length. At worst, O (m × n) times of reference is performed.

  In the character string matching, the text character string facing the search character string is compared from the last part of the search character string. If there is a mismatch, the shift amount of the skip table 2 is obtained from the mismatched character on the text side. The shift amount of the skip table 1 is obtained from the length of the matching partial character string or the mismatched character position. Select the larger one and shift the text string. In the example shown in FIG. 11 (a), 0 characters are initially matched and the mismatched character is a. Since the shift amount of the skip table 1 is 1 and the shift amount of the skip table 2 is 2, the character is shifted by two characters. In the second collation, four characters match and the mismatch character is b. Since the shift amount of the skip table 1 is 5, and the shift amount of the skip table 2 is 1, the character is shifted by 5 characters. Thereafter, similarly, the character comparison is repeated from the last part of the search character string. In most cases, the number of character comparisons required for character string matching is less than O (n + m) times.

  The BMH method is simplified to use only the skip table 2. As shown in FIG. 11B, the shift amount is obtained by drawing the skip table with the rightmost character of the partial character string. In the example shown in FIG. 11B, since the rightmost character is initially h, the shift amount is 6. In the second collation, since the rightmost character is g, the shift amount is 6.

  Of the conventional character string search methods described above, the method for creating a skip table is complicated in the high-speed BM method. The simple method is slow to search. The present inventor proposed a simpler and faster method in Non-Patent Document 1. This method is an improvement of the BMH method. This is called the BK method and will be briefly explained. As shown in the flowchart of FIG. 12, if the character of the text is in the search character string, the shift amount of the first character of the search character is set to (number of search characters m). The shift amount of the next character is set to (number of search characters m-1). The shift amount of the last character of the search character is set to 1. For text characters that are not in the search character string, the shift amount is (number of search characters m + 1).

The character string search is executed as shown in the flowchart of FIG. Set the pointer to point to the next of the substring of the text to be compared with the search string. If the search character string and the partial character string do not match, the character of the text pointed to by the pointer is used as an argument, the skip table is drawn, the shift amount is obtained, and the position of the partial character string to be compared with the search character string is shifted. Repeat the search. An example of the skip table and search is shown in FIG.
Yuebin Bai, Hidetsune Kobayashi: "New String Matching Technology for Network Security", Proceedings of 17th International Conference on Advanced Information Networking and Applications (AINA'03), Mrch 27-29, 2003, Xi'an, China, pp.198- 201.

  However, the conventional high-speed matching method (BK method) has a problem that it takes time to create a skip table in proportion to the number of text characters and is not sufficiently fast. If the same character is found in the text, the same thing is repeated and there is a waste. An object of the present invention is to solve the above-mentioned conventional problems and to realize a high-speed matching method capable of creating a skip table by a simple method.

  In order to solve the above problems, in the present invention, the high-speed matching method is configured as follows. The shift amount corresponding to the character in the text and the search string in each character of the used character set is the reverse number of the search string, and the shift amount corresponding to the character not in the text and the character not in the search string is (search Create a skip table as the number of characters + 1), subtract the skip table using the rightmost character of the partial character string in the text to be compared with the search character string as an argument, obtain the shift amount, and then compare the next character in the text to be compared with the search character string. Is determined, and the partial character string is compared with the search character string.

  Moreover, the high-speed matching method was comprised as follows. The shift amount corresponding to the character to the right of the rightmost wild card in the search character string is set as the reverse number of the search character string, and the shift amount corresponding to the character to the left including the rightmost wild card is Create a skip table as the number of reverse search numbers for the rightmost wild card, subtract the skip table from the right-most character of the substring in the text to be compared with the search string, find the shift amount, and search Determine the next substring in the text to be compared with the string and compare the substring with the search string.

  In the present invention, the configuration as described above eliminates the need to repeatedly perform the same operation on the same character in the text, so that the skip table can be created in a short time and the time required for the search is shortened. Since the shift amount can be calculated and the skip table can be created as long as the search character string is determined, it is possible to search in real time for a network protocol or the like that requires real-time performance, and analysis for each search character string becomes possible. A specific virus pattern can be searched in real time from signals flowing on the network, or illegal packets given by a specific search character string can be searched in real time. In addition, when a specific IP is discovered, the packet can be deleted immediately and not passed. Similar images (moving images, still images) can be detected in consideration of ambiguity.

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

  In the first embodiment of the present invention, the shift amount corresponding to the character in the text and the search character string among the characters of the used character set is set as the reverse number of the search character string, and the character not in the text and the character not in the search character string Create a skip table with the shift amount corresponding to (number of search characters + 1), subtract the skip table using the rightmost character of the partial character string in the text to be compared with the search character string as the argument, and obtain the shift amount. This is a fast matching method in which the next partial character string in the text to be compared with the sequence is determined and the partial character string is compared with the search character string.

  A configuration of an apparatus that executes the high-speed matching method according to the first embodiment of the present invention will be described. FIG. 1 is a conceptual diagram of an apparatus that executes a high-speed matching method in Embodiment 1 of the present invention. In FIG. 1, an input device 1 is a means for inputting various data and commands, and is an input means such as a keyboard and a mouse and an input means capable of taking in a large amount of data such as a modem and a CDROM device. The used character set storage unit 2 is a memory for storing a used character set. The text storage unit 3 is a memory for storing search target text. The search character string storage unit 4 is a memory for storing a search character string. The skip table creation unit 5 is a calculation unit that creates a skip table while referring to the used character set, text, and search character string. The skip table storage unit 6 is a memory for storing a skip table. The character string search unit 7 is a calculation unit that searches for a search character string from text while referring to a skip table. The search result storage unit 8 is a memory that stores the position of the partial character string of the text that matches the search character string. The output device 9 is a display device that outputs search results.

  The operation procedure of the high-speed matching method in Embodiment 1 of the present invention will be described. FIG. 2 is a flowchart showing a skip table creation procedure of the fast matching method. FIG. 3 is a flowchart showing a search procedure of the fast matching method. FIG. 4 is a diagram illustrating a skip table and a search example of the fast matching method according to the first embodiment of the present invention. FIG. 5 is a table showing experimental results of the high-speed matching method.

  An outline of the operation procedure will be described. First determine the character set to use. For example, it is assumed that there are 26 ASCII single-byte lowercase letters. Check one character at a time from "a". If the character is in both the text and the search character string, the shift amount is registered in the skip table. The shift amount is the number of reverse search string numbers. That is, if the number of search characters is m, the shift amount corresponding to the first character of the search character is m, the shift amount corresponding to the next character is (m−1), and the shift amount corresponding to the last character. Is 1. If it is in the text but not in the search character string, it is registered as another character in the skip table. The shift amount is (number of search characters + 1). If the character is not in the text, it is ignored and nothing is registered in the skip table. If there is a search character in which no shift amount is registered in the skip table, it can be understood that there is no search character string in the text without performing a search. Search using this skip table in the same way as the conventional BK method.

  The skip table creation procedure of the high-speed matching method will be described with reference to FIG. In step 1, the length of the text is compared with the length of the search character string, and if the search character string is long, no match can be made, so the process ends. In step 2, one character is extracted from the character set to be used, and in step 3, it is checked whether it is in the text. Nothing if not in the text. In step 4, it is checked whether or not it is in the search character string. If there is, it is registered in the skip table in step 5. If not, it is registered in step 6 as other characters. In step 7, it is checked whether or not the text ends. If not, the process returns to step 2 to process the next used character.

  You may omit the check whether the character of a character set to be used exists in a text. If the character set used is in the search string, it is registered in the skip table. If it is not in the search string, it is registered as other characters in the skip table. Even if there is no character in the used character set in the text, it is registered in the skip table, but no matching occurs at the search stage, so no error occurs. The processing time may not vary greatly depending on conditions, but may be shortened. With this method, it is possible to create a skip table without text, so it is possible to search for a search character while receiving text.

  The search procedure of the fast matching method will be described with reference to FIG. In step 11, the search character string is set to the head of the text, and the pointer is set to the next character. In step 12, the search character string is compared with the partial character string of the text. If there is a match in step 13, a match result is output in step 14. If the text does not end in step 15, in step 16, the skip table is drawn using the character pointed to by the pointer as an argument to obtain the shift amount. In step 17, the search character string is shifted and the pointer is moved. Repeat this until the text ends.

  With reference to FIG. 4, a skip table and a search example of the fast matching method will be described. As shown in FIG. 4A, the character set to be used is 26 ASCII lower-case letters “az”. The search character string is “string”. The skip table is as shown in the figure. At the time of the first comparison, since the pointer points to “s”, the shift amount is 6. Next, since the pointer points to “f”, the shift amount is 7. As shown in FIG. 4B, if the search character string is “strong” and there is no “o” in the text, “o” is not registered in the skip table. If you compare the skip table with the search string before the search, you can see that there is no match. As shown in FIG. 4C, when the search character string is “start” and includes two or more of the same characters, only one with the smaller shift amount is registered. There is only one “t” column in the actual skip table.

  The experimental results of the high-speed matching method will be described with reference to FIG. In this experiment, random numbers were generated to generate text and search strings, and skip table creation and text search were repeated. Table (a) shows an experiment in which the maximum text length and the maximum search character string length are changed. In the present invention 1, text is referred to in the creation of the skip table, and in the present invention 2, the text is not referred to. When the text is long, it is much faster than the BK method. Table (b) shows an experiment with only the maximum search character string length changed. If the search string is long, it is much faster than the BK method. Table (c) shows an experiment in which the maximum text length is set to a short constant value and the maximum search character string length is changed within a short range. Compared to the BK method, it is rather slow. It can be seen that the present invention is effective when the text and the search character string are long.

  As described above, in the first embodiment of the present invention, the fast matching method uses the shift amount corresponding to the characters in the text and the search character string among the characters of the used character set as the search character string reverse order number, Create a skip table with the shift amount corresponding to characters not in the search string and characters not in the search string as (number of search characters + 1), and use the right-most character of the substring in the text to be compared with the search string as a skip table Since the shift amount is obtained by subtracting, the next partial character string in the text to be compared with the search character string is determined, and the partial character string is compared with the search character string, the same operation is repeated for the same character in the text. The skip table can be created in a short time, the preprocessing time is shortened, and the total time with the search time is shortened.

  In the second embodiment of the present invention, the shift amount corresponding to the character to the right of the rightmost wild card in the search character string is set as the reverse number of the search character string, and the leftmost character including the rightmost wild card is included. Create a skip table with the shift amount corresponding to the character as the number of reverse search numbers of the rightmost wildcard search character, and use the skip table with the character immediately to the right of the substring in the text to be compared with the search character string as an argument Subtracts the shift amount, determines the next partial character string in the text to be compared with the search character string, and compares the partial character string with the search character string, enabling a fuzzy search using wildcards. It is.

  FIG. 6 is a conceptual diagram of an apparatus that executes the high-speed matching method according to the second embodiment of the present invention. The character string search unit 10 is an arithmetic unit that searches a search character string from text while changing a decrement width with reference to a skip table. The configuration of the apparatus that executes the high-speed matching method in the second embodiment is the same as the apparatus of the first embodiment shown in FIG. The difference is the part related to the skip table and the search character string. FIG. 7 is a flowchart showing a skip table creation procedure of the fast matching method according to the second embodiment of the present invention. FIG. 8 is a flowchart showing the search procedure of the fast matching method. FIG. 9 is a diagram illustrating a skip table and a search example of the fast matching method.

  A processing procedure of the high-speed matching method in the second embodiment will be described. Some characters in the search string are wildcards that match all characters. The shift amount of the character on the right side of the wild card in the search character string is the number of reverse search character string numbers. The shift amount of the rightmost character is 1, and the shift amount of the next character is 2. If there is no wild card, the shift amount of the first character is the number of search character strings. If the rightmost wild card in the search character string is third from the right, the wild card shift amount is three. The shift amount between the character on the left side of the rightmost wild card and the wild card is the same as the shift amount of the rightmost wild card.

Furthermore, the decrement width when comparing the text and the search string in order is changed. The width for reducing the subscripts of the text and the search character string is normally 1. However, if a wild card is included, it is skipped without comparing the wild card. If there is one wildcard to the left of a character in the search character string, the decrement width corresponding to that character is 2. If there are two consecutive, it is 3. If the search character string is (* tring), the decrement width corresponding to each character is (121111). Similarly,
(s * ring, 112111), (st * ing, 111211), (str * ng, 111121), (stri * g, 111112),
(strin *, 111111), (** ring, 113111), (s ** ing, 111311), (st ** ng, 111131),
(str ** g, 111113), (stri **, 111111), (s * r * ng, 112121), (s * r ** g, 112113)
It becomes.

  An algorithm of the fast matching method in the second embodiment will be described. This is an algorithm that gives a text and a search character string, checks whether the search character string appears in the text, and if it appears, finds all the positions. First, a used character set is defined and stored in an array, and the number of used characters is obtained. The character set used is the number of characters that make up the text. For example, if the text consists of lowercase letters of the alphabet, there are 26 characters from 'a' to 'z'. Even if the text and the search character string are unknown, it is usually possible to limit the characters constituting the text. When entering a search string, place a mark on the wildcard. For example, the wild card is “*”. The position where the wild card appears last is obtained from the search character string, and the decrement width is obtained and stored in the array. The length of the input text and the search character string is obtained. If the search character string length is longer than the text length, the search character string does not appear in the text. In this case, the process ends without doing anything.

  In the preprocessing, a skip table is created. Take out one character at a time from the used character set and perform the following operations until the characters are exhausted. Check if the retrieved character is included in the search string. If it is included in the search character string, the shift amount corresponding to the character is stored in the array. The shift amount is the smaller of the number of search character string reverse order numbers of the character (the rightmost character if there are two or more characters) or the number of search character string reverse order numbers of the rightmost wild card. If it is not included in the search character string, the shift amount is the number of reverse search numbers of the rightmost wild card search character string. The array subscript uses the internal code of the extracted character. That is, the shift amount can be obtained by subtracting the skip table with the character code.

The skip table creation procedure will be described with reference to FIG. Step 21 extracts characters one by one from the character set used, and step 22 checks whether they are in the search character string. If not, in step 25, the wild card shift amount is registered.
If there is, step 23 checks whether it is right from the rightmost wild card. If it is to the left of the rightmost wild card, in step 25, the wild card shift amount is registered. If it is to the right of the rightmost wild card, the shift amount is registered in the skip table. In step 26, it is checked whether the used character set has been completed. If not, the process returns to step 21 to process the next used character.

  In post-processing, a search is performed. In the actual search, the search start position in the text is determined. One character is extracted forward from this position in the text, one character is extracted forward from the last character of the search character string, and the characters are compared one by one. That is, the subscripts in the text and the search character string decrease. For this reason, since there is no search character string in the portion less than the number of first search character string characters in the text, the initial value of the search start position is set to (number of search character string characters-1). However, the head position is 0.

  The following operation is repeated while the subscript k representing the initial search value moving in the text is shorter than the text length. Initialize the subscript j that moves through the search string to point to the last character. The initial value of the subscript i for moving through the text is set as the search initial value k. The pointer pointing to the character that draws the skip table is at the position (k + 1). While the subscript j can be moved in the search character string and the i-th of the text is equal to the j-th of the search character string, each subscript is decreased by a decrement. If the subscript j of the search character string is negative when exiting this loop, it means that all characters constituting the search character string are matched, and the character string has been found. It is checked whether the initial search value k in the text after finding the character string is shorter than the text length. If shorter, the next search initial value candidate is calculated. If it is not short, a series of search processing is terminated.

  The search procedure of the fast matching method will be described with reference to FIG. The basic procedure is the same as that of the first embodiment shown in FIG. In step 31, the search character string is aligned with the beginning of the text, and the pointer is set to the next character. In step 32, the search character string is compared with the partial character string of the text while skipping the wild card. If they match at step 33, a match result is output at step 34. If the text does not end in step 35, then in step 36, the skip table is subtracted using the character pointed to by the pointer as an argument, and the shift amount considering the wild card is obtained. In step 37, the search character string is shifted and the pointer is moved. Repeat this until the text ends.

An example of fuzzy search will be described using the example shown in FIG. text
"stringstryzgsearchstringsfindstrings"
Consider a case where all the search character strings “str ** gs” are searched from among the search strings. Use lower-case alphabetic character set. The skip table next []
next ['a'] ← 3 next ['b'] ← 3 next ['c'] ← 3 next ['d'] ← 3 next ['e'] ← 3
next ['f'] ← 3 next ['g'] ← 2 next ['h'] ← 3 next ['i'] ← 3 next ['j'] ← 3
next ['k'] ← 3 next ['l'] ← 3 next ['m'] ← 3 next ['n'] ← 3 next ['o'] ← 3
next ['p'] ← 3 next ['q'] ← 3 next ['r'] ← 3 next ['s'] ← 1 next ['t'] ← 3
next ['u'] ← 3 next ['v'] ← 3 next ['w'] ← 3 next ['x'] ← 3 next ['y'] ← 3
next ['z'] ← 3
It becomes.

  When the first seven characters of the text are compared with the search character string, since the text is “strings”, a match is output. Since the character at the pointer destination at this time is “t”, the shift amount is 3 when the skip table is drawn. Shift the search string three characters to the right and compare with the text. Since the text does not match with “ingstry”, the text is shifted to the right by the shift amount of 3 characters of the character “z” at the pointer destination and compared with the text. Since the text is “stryzgs”, a match is output. The character “e” at the pointer destination is shifted to the right by the shift amount of 3 characters and compared with the text. Since the text does not match with “yzgsear”, it is shifted to the right by the shift amount of 3 characters of the character “c” at the pointer destination and compared with the text. Since the text does not match “searchs”, it is shifted to the right by the shift amount of 3 characters of the character “t” at the pointer destination and compared with the text.

  Since the text does not match with “rchstri”, the character “n” at the pointer destination is shifted to the right by the shift amount of 3 characters and compared with the text. Since the text is “strings”, a match is output. The character “f” at the pointer destination is shifted to the right by the shift amount of 3 characters and compared with the text. Since the text does not match with “ingsfin”, it is shifted to the right by the shift amount of 3 characters of the character “d” at the pointer destination and compared with the text. Since the text does not match with “sfindst”, the character “r” at the pointer destination is shifted to the right by the shift amount of 3 characters and compared with the text. Since the text is inconsistent with “ndstrin”, it is shifted to the right by the shift amount of the character “g” at the pointer destination and compared with the text. Since the text is “strings”, a match is output.

  As described above, in the second embodiment of the present invention, the fast matching method uses the shift amount corresponding to the character to the right of the rightmost wild card in the search character string as the search character reverse number, and the rightmost Create a skip table with the shift amount corresponding to the left character including the wild card as the number of reverse search numbers of the rightmost wild card, and compare the search string with the substring in the text Since the rightmost character on the right side is used as an argument to calculate the shift amount by subtracting the skip table, the next partial character string in the text to be compared with the search character string is determined, and the partial character string is compared with the search character string. A fuzzy search can be performed at high speed using wildcards.

  The fast matching method of the present invention is optimal as a method for retrieving a specific pattern from data. Furthermore, it can be applied to image processing such as real-time search for virus patterns and illegal packets on the network, person identification, shape identification, scene identification, and action identification.

The conceptual diagram of the apparatus which performs the high-speed matching method in Example 1 of this invention, The flowchart which shows the skip table creation procedure of the high-speed matching method in Example 1 of this invention, The flowchart which shows the search procedure of the high-speed matching method in Example 1 of this invention, Explanatory drawing of the skip table and search example of the high-speed matching method in Example 1 of this invention, Experimental result table of high-speed matching method in Example 1 of the present invention, The conceptual diagram of the apparatus which performs the high-speed matching method in Example 2 of this invention, The flowchart which shows the skip table creation procedure of the high-speed matching method in Example 2 of this invention, The flowchart which shows the search procedure of the high-speed matching method in Example 2 of this invention, The figure which shows the skip table and search example of the high-speed matching method in Example 2 of this invention, Explanatory drawing of the conventional character string search method, Explanatory drawing of the conventional character string search method, A flowchart showing a skip table creation procedure of a conventional fast matching method, A flowchart showing a conventional high-speed matching method search procedure, It is explanatory drawing of the example of a search by the conventional high-speed matching method.

Explanation of symbols

1 Input Device 2 Used Character Set Storage Unit 3 Text Storage Unit 4 Search Character String Storage Unit 5 Skip Table Creation Unit 6 Skip Table Storage Unit 7 Character String Search Unit 8 Search Result Storage Unit 9 Output Device
10 String search part

Claims (1)

An input device for inputting various data and commands, a used character set storage unit for storing a used character set input from the input device, and a text storage unit for storing a search target text input from the input device A search character string storage unit that stores a search character string including a wildcard that is a specific character by inputting from the input device, while referring to the used character set, the search target text, and the search character string A skip table creation unit that creates a skip table, a skip table storage unit that stores the skip table, and the search character string is searched from the search target text while referring to the skip table while changing a decrement. Search result record that stores the position of the character string search part and the partial character string of the text that matches the search character string And a fast matching method executed in a character string search device comprising an output device for outputting a search result, storing the used character set in the used character set storage unit, and in the skip table creating unit, One character is extracted as a check character from the character set to be used, and it is checked whether the check character is in the search character string. If the check character is not in the search character string, even if it is present, If it is not right than the right wild card, the order number indicating the position of the rightmost wild card in the search character string from the right end is registered in the skip table as the shift amount corresponding to the check character. If the inspection character is on the right side of the rightmost wild card in the search character string, the inspection character in the search character string is The order number indicating whether the character is in the th position is registered in the skip table as a shift amount corresponding to the check character, and this process is repeated until the use character set is completed. In the character string search unit, a character string that is at the beginning of the search target text and has the same number as the search character string is set as the first comparison partial character string, and a pointer is set next to the comparison partial character string. The comparison partial character string is compared with the search character string while skipping the wild card, and if all the characters in the corresponding positions of the comparison partial character string and the search character string are matched, the search character string is found and mismatched. If there is a character that is not the end of the search target text character, the skip table is used for the character indicated by the pointer in the search target text. A shift amount is obtained by pulling a bull, the position of the comparison partial character string is shifted by the shift amount, and the pointer is updated to be set next to the new comparison partial character string. A high-speed matching method that repeats until the end of .

Images