JPH0365728A - Solution system for referring to label in translation system - Google Patents

Abstract

PURPOSE:To shorten average reference time of a label by registering the label defined for the first time in a source program to the hash chain of a hash table corresponding to its hash value, and registering other same labels in a same name chain starting from the same label in the hash table. CONSTITUTION:Every time the label is defined in the source program, a label table 3 generating means retrieves the hash chain 31 starting from the entry of the hash table 2 corresponding to the hash value related to that label and obtained by a hash value calculating means 13, and if the same label does not exist, it registers the label of this time in this hash chain. When the same label exists, it registers the label of this time in the same name chain starting from this existing same label, and on the other hand, every time the label is referred to in the source program, the label table retrieving means 12 retrieves the hash chain 31 starting from the entry of the hash table 2 corresponding to the hash value obtained by the hash value calculating means 13 and related to that label and the same name chain derived from the hash chain at need. Thus, the average reference time of the label is shortened.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Field of Application] The present invention relates to a name reference resolution method in a translation system.

[Conventional technology]

In a source program translation system, a name name defined in a source program is registered in a name name table, and then reference resolution processing for the name name is performed. Usually, this kind of name reference resolution method in a translation system is realized by searching a name name table using a hashing technique. What is the hacking technique?
A hash table is prepared separately from the name table, the hash value is calculated from the name when the name table is generated, and the name name with the same hash value is linked to a chain starting from the same entry in the hash table (a hash chain). This is a technique in which the hash chain is searched for by registering the name in a name tag (called a name tag), and then searching for the name tag by following the hash chain when resolving the name name reference. In the conventional name name reference resolution method, even if the names are the same, the number of names defined in the source program is registered in the same hash chain of the name table, and then referenced. had been adopted.

[Problem to be solved by the invention]

Although name reference resolution processing is possible with the conventional name name reference resolution method described above, the more the same name name is defined in the source program, the more the number of name names in the corresponding hash chain decreases. This has the disadvantage that the average reference time for famous landmarks increases.

The present invention aims to reduce the average reference time for such landmarks.

[Means for Solving the Problems] In order to achieve the above object, the present invention includes a name table creation means 2, a bubble table search means, and a hash value calculation means, and uses a hash table and a name table. In a translation system that performs reference resolution of name marks, the name table creation means calculates a hash value corresponding to the name name calculated by the hash value calculation means each time a name name is defined in the source program. Search for the hash chain starting from the entry in the hash table, and if the same name does not exist, register the current name in that hash chain, and if the same name exists, register the same name that existed. The current name name is registered in the same name chain starting from the name name, and on the other hand, the name name table search means searches for the name name obtained by the hash value calculation means every time the name name is referenced in the source program. The hash chain starting from the hash table entry corresponding to the hash value and, if necessary, the same name chain derived from the hash chain are searched.

[Effect]

In the name reference resolution method in the translation system of the present invention, among the names defined in the source program and having the same hash value, different names have the same hash value. It is registered in the hash chain starting from the stable entry, one of the same name marks is registered in the hash chain, and the other name marks are registered in the same name chain starting from the one name mark, and when referenced, that name is registered in the hash chain. chain and, if necessary, a chain with the same name.

〔Example〕

Next, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 shows the constituent countries of an embodiment of the present invention, where 1 is a translation system, 2 is a hash table, and 3 is a name table.

The hash table 2 has entries corresponding to the hash values "OJ~"m", and each entry stores a pointer pointing to the first name of the corresponding hash chain if it exists. If such a hash chain does not exist, a NULL value is stored.

A plurality of name names are registered in the name name table 3.

In the figure, names al, a2, a3 are added to the hash chain 31 starting from the hash (direct) OJ of the hash table 2.
．． a4 is registered, the name tag bl is registered in the hashish chain 33 starting from the hashish value "2", and the hashish value "m" is registered.
Name marks cl and c2 are registered in the hash chain 34 starting from ``, and furthermore, in the same name chain 32 starting from the name mark a3 registered in the hash chain 31, the same name mark a3 as the name mark a3 in the hash chain 31 is registered. is registered.

Further, the translation system 1 is a system that translates a source program (not shown) to generate a target program, etc.
Functional units related to the name reference resolution method of the present invention include name name table creation means 11 name name table search means 12 and hash value calculation means 13.

The hashish value calculation means 13 is a means for calculating the hashish value of a name tag, and is called by the name name table creation means 11 and the name name table search means 12.

The name table creation means 11 is a means that is called by the translation system 1 every time a name is defined in the source program. When called, the name table creation means 11 searches for a hash chain starting from the entry in the hash table 2 that corresponds to the hash value for the name name calculated by the hash value calculation means 13, and if no identical name name exists, registers the current name mark in the hash chain, and if the same name tag exists, performs processing such as registering the current name mark in the same name chain starting from the existing same name mark.

The name table search means 12 is a means that is called by the translation system 1 every time a name is referenced in the source program. When called, the name table search means 12 creates a hash chain starting from the entry in the hash table 2 corresponding to the hash value for the name name calculated by the hash value calculation means 13 and, if necessary, derives from the hash chain. Processes such as searching for chains with the same name are performed.

The operation of this embodiment configured as described above will be explained below.

The translation system 1 calls the name name table creation means 11 every time a name name is defined in the source program.

The name table creation means 11 first calls the hash value calculation means 13 to obtain the hash value of the current name name, and then calculates the name name of the hash chain starting from the hash value 1 to 2 entry corresponding to this hash value. We will compare the names that should be registered this time in order. If the same name tag is already registered in that hash chain,
Add this name mark to the same name chain that starts from the same name mark registered above instead of the hash chain, and only if there is no same name mark registered yet, add this name mark to that hash chain. to add. That is, in FIG. 1, the name mark a3. on the hash chain 31. If name mark a3 on the same name chain 32 is not registered together, name mark a
When the definition of 3 appears, first the hash chain 31
If the name mark a3 is registered above and the same name mark a3 appears thereafter, the name mark a3 is registered in the same name chain 32 starting from the name mark a3 on the hash chain 31.

After the translation system 1 has finished processing all the names defined in the source program, it then calls the name table search means 12 every time a name is referenced in the source program.

The name table search means 12 first calls the hash value calculation means 13 to obtain the hash value of the current name name, and then searches the name name of the hash chain starting from the entry in the hash table 2 corresponding to this hash value. Search in order to obtain landmarks to be referenced and resolved. Also, if this name tag has a same name chain like the same name chain 32 in Fig. 1,
The name table search means 12 further searches the name table of the same name chain in order and determines the name to be referenced and resolved.

〔Effect of the invention〕

As explained above, in the name reference resolution method in the translation system of the present invention, a name defined for the first time in a source program is registered in the hash chain starting from the hash table entry corresponding to its hash value. However, other identical names are registered in the same name chain starting from the above intermediate name in the hash table, so even if many identical names are defined in the source program, the names in the hash chain are This has the effect of reducing the average reference time for landmarks without increasing the number of landmarks.

[Brief explanation of drawings]

FIG. 1 shows the constituent countries of one embodiment of the present invention. In the figure, 1... Translation system 2... Hash table 3... Name name table 11... Name name table creation means 12... Name table search means 13... Hash value calculation means 31 .33.34...Hatush chain 32...Chain with the same name

Claims (1)

[Scope of Claim] A translation system that includes a name name table creation means, a name name table search means, and a hash value calculation means, and performs name name reference resolution using the hash table and the name name table, comprising: the name name table; Each time a name is defined in the source program, the creation means searches for a hash chain starting from an entry in the hash table that corresponds to the hash value for the name obtained by the hash value calculation means, If the name does not exist, the current name is registered in that hash chain, and if the same name exists, the current name is registered in the same name chain starting from the existing same name, and the name name table search is performed. The means includes a hash chain starting from an entry in the hash table corresponding to the hash value for the name calculated by the hash value calculation means, and the hash as necessary, each time the name is referenced in the source program. A name reference resolution method in a translation system, characterized in that the same-name chain derived from the chain is searched.