JP2017142712A - Call graph difference extraction method, call graph difference extraction program, and information processing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To compare two call graphs and efficiently extract an edge difference when an edge difference is detected.SOLUTION: An information processing device 10 classifies a plurality of first edges 1e-1g into categories for respective distances in a first call graph 1 by distance from a first node 1a that is a starting point, and classifies a plurality of second edges 2f-2i in a second call graph 2 by distance from a second node 2a that is a starting point. A difference candidate first edge where second edges of the same category having a common call relation do not exist and a difference candidate second edge where first edges of the same category having a common call relation do not exist are specified. A difference first edge, among the difference candidate first edges, where difference candidate second edges having a common call relation do not exist, and a difference candidate second edge, among the difference candidate second edges, where difference candidate first edges having a common call relation do not exist are extracted.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a call graph difference extraction method, a call graph difference extraction program, and an information processing apparatus.

  A program to be executed by a computer may define a call for another program as a subroutine. In a large-scale software, such a calling relationship between programs is complicatedly intertwined. Therefore, the call relationship between programs may be represented by a graph so that the call relationship between programs included in software can be visually grasped. Such a graph representing a call relationship between programs is called a call graph.

  The call graph is a directed graph composed of a set of nodes representing programs and a set of edges representing call relationships between programs. In the call graph, an edge is represented by a line having directionality from a node corresponding to a calling source program to a node corresponding to a calling destination program.

  Such a call graph can be used to grasp the influence of other programs due to the modification of a part of the software program. For example, by comparing a call graph indicating a call relationship of software before correction with a call graph indicating a call relationship of software after correction, it is possible to grasp that the caller of the subroutine has been changed. At this time, if the subroutine includes a process that depends on the caller, it is understood that the subroutine program needs to be corrected.

  As a technique for grasping the structure of a program using a graph, for example, there is a hierarchical graph analysis method for simply displaying the structure of a computer program in a directed graph. In addition, when a failure occurs in an application, an application failure cause identification work support system that uses a call graph to identify a location where processing that is the root cause of the failure is performed is also considered. Furthermore, a source code equivalence verification apparatus that performs structure comparison using a structure graph obtained by analyzing the source code when verifying the equivalence of the source code is also considered.

JP-A-6-187138 JP 2007-241426 A JP 2014-126985 A

  However, when two call graphs are compared and edge differences are detected, if both of the two call graphs have a large number of edges, it is difficult to complete the comparison process within a practical time. For example, when each edge in one call graph is compared with all the edges in the other call graph in a brute force manner, edge identity determination processing is performed for the number of times obtained by multiplying the number of edges in each call graph. . In a large-scale software such as an OS (Operating System), the number of edges included in a call graph may reach 1 million, and the number of repetitions of edge identity determination processing becomes enormous. As a result, it takes time to compare the two call graphs.

  In one aspect, the object is to enable efficient extraction of edge differences.

In one proposal, a call graph difference extraction method is provided in which a computer performs the following processing.
In this call graph difference extraction method, when a starting point program module is designated, the computer displays a calling relationship among a plurality of first program modules in the first software including the starting point program module by using the plurality of first program modules. Based on the first call graph represented by the plurality of first edges connecting the plurality of first nodes shown, the call relation is traced from the first node corresponding to the origin program module to reach each of the plurality of first edges. The plurality of first edges in the first call graph are classified into categories for each distance. In addition, the computer uses a plurality of second edges connecting a plurality of second nodes indicating a plurality of second program modules to establish a calling relationship between the plurality of second program modules in the second software including the starting program module. Based on the expressed second call graph, the plurality of second edges are classified into categories for each distance according to the distance from the second node corresponding to the starting program module to the call relationship and reaching each of the plurality of second edges. Classify. Next, the computer targets each of the plurality of first edges, and among the second edges belonging to the same category as the target first edge, the program module pair of the caller and the callee is common to the target first edge. If the second edge does not exist, the target first edge is specified as the difference candidate first edge. In addition, the computer targets each of the plurality of second edges, and among the first edges belonging to the same category as the target second edge, the set of program modules of the caller and the callee is the same as the target second edge. When the first edge does not exist, the target second edge is specified as the difference candidate second edge. Further, the computer extracts a first difference edge that does not have a second difference candidate edge that has a common program module set of the caller and the callee among the first difference candidates. Further, the computer extracts a difference second edge that does not include a difference candidate first edge that has a common set of program modules of the caller and the callee among the difference candidate second edges.

  According to one aspect, the difference between edges can be efficiently extracted.

It is a figure which shows the structural example of the information processing apparatus which concerns on 1st Embodiment. It is a figure which shows the example of a production | generation of a call graph. It is a figure which shows an example of the data structure of the edge of a call graph. It is a figure which shows the example of the distance of the edge on a call graph. It is a figure which shows the example of the call graph of the source code after an update. It is a figure which shows one structural example of the hardware of the computer used for 2nd Embodiment. It is a block diagram which shows the function of the computer for call graph comparison. It is a figure which shows the example of the data in a 1st, 2nd call graph memory | storage part. It is a figure which shows the example of the data in the edge list memory | storage part classified by 1st, 2nd distance. It is a figure which shows the example of the data in a 1st, 2nd difference edge memory | storage part. It is a figure which shows the example of the data in a 1st, 2nd presence edge memory | storage part. It is a figure which shows the example of the data in a difference extraction result memory | storage part. It is a flowchart which shows the procedure of a difference edge extraction process. It is a flowchart which shows an example of the procedure of edge classification processing classified by distance. It is a flowchart which shows an example of the procedure of the edge comparison process between the same distances. It is a flowchart which shows an example of the procedure of a distance difference edge presence confirmation process. It is a figure which shows the comparative example of the 1st reference method and the method which concerns on 2nd Embodiment. It is a figure which shows the comparative example of the 2nd reference method and the method which concerns on 2nd Embodiment.

Hereinafter, the present embodiment will be described with reference to the drawings. Each embodiment can be implemented by combining a plurality of embodiments within a consistent range.
[First Embodiment]
First, the first embodiment will be described.

FIG. 1 is a diagram illustrating a configuration example of an information processing apparatus according to the first embodiment. The information processing apparatus 10 includes a storage unit 11 and a calculation unit 12.
The storage unit 11 stores the first call graph 1 and the second call graph 2. The first call graph 1 includes a plurality of first edges that connect a plurality of first nodes 1a to 1d indicating a plurality of first program modules with respect to a calling relationship between the plurality of first program modules in the first software. It is the directed graph represented by 1e-1g. The second call graph 2 includes a plurality of second edges that connect a plurality of second nodes 2a to 2e indicating a plurality of second program modules with respect to a calling relationship between the plurality of second program modules in the second software. It is a directed graph represented by 2f-2i. It is assumed that the identifier of the first software is “1” and the identifier of the second software is “2”.

The calculation unit 12 includes a classification unit 12a, a specification unit 12b, and an extraction unit 12c.
When the start program module included in both the first software and the second software is designated, the classification unit 12a includes the first edges 1e to 1g included in the first call graph 1 and the second call graph 2, respectively. And the second edges 2f to 2i are classified into a plurality of categories. For example, the classification unit 12a identifies the first node 1a corresponding to the starting program module based on the first call graph 1. Then, the classifying unit 12a classifies the plurality of first edges 1e to 1g into categories for each distance according to the distance from the identified first node 1a to the respective first edges 1e to 1g following the calling relationship. To do. For example, the classification unit 12a divides the plurality of first edges 1e to 1g into categories and stores them in the first distance-specific edge list 3. The edge registered in the first distance-specific edge list 3 is represented by a pair (program name pair) of the name of the call source program module and the name of the call destination program module in the corresponding call relationship. .

  Further, the classification unit 12a identifies the second node 2a corresponding to the starting point program module based on the second call graph 2. Then, the classification unit 12a classifies the plurality of second edges 2f to 2i into categories for each distance based on the distance from the identified second node 2a to the respective second edges 2f to 2i following the calling relationship. To do. For example, the classification unit 12a divides the plurality of second edges 2f to 2i into categories and stores them in the second distance-specific edge list 4. The edge registered in the second distance edge list 4 is represented by a pair (program name pair) of the name of the call source program module and the name of the call destination program module in the corresponding call relationship. .

  The specifying unit 12b targets each of the plurality of first edges 1e to 1g. Then, the specifying unit 12b, when the second edge that belongs to the same category as the target first edge and the set of program modules of the caller and the callee does not have a second edge common to the target first edge, The target first edge is specified as the difference candidate first edge.

  The specifying unit 12b targets each of the plurality of second edges 2f to 2i. Then, the specifying unit 12b, when the first edge belonging to the same category as the target second edge does not have a first edge common to the target second edge in the set of program modules of the caller and the callee, The target second edge is specified as the difference candidate second edge.

  For example, the specifying unit 12b compares the first edge and the second edge in the same category in the first distance-based edge list 3 and the second distance-specific edge list 4. Then, the specifying unit 12b stores, in the first difference edge list 5, the first edge that does not include the second edge represented by the same program name pair in the same category as the difference candidate first edge. Further, the specifying unit 12b stores, in the second difference edge list 6, the second edge that does not include the first edge represented by the same program name pair in the same category as the difference candidate second edge.

  The extraction unit 12c extracts a difference first edge that does not include a difference candidate second edge that has a common set of program modules of the caller and the callee among the difference candidate first edges. Further, the extraction unit 12c extracts a difference second edge that does not include a difference candidate first edge having a common program module set of the caller and the callee among the difference candidate second edges.

  For example, the extraction unit 12 c compares the difference candidate first edge in the first difference edge list 5 with the difference candidate second edge in the second difference edge list 6 brute force. Then, the extraction unit 12c stores the difference candidate first edge having a different program name pair from any of the comparison partners in the final difference edge list 7 in association with the identifier of the first software. Further, the extraction unit 12c stores the difference candidate first edge having a different program name pair from any of the comparison partners in the final difference edge list 7 in association with the identifier of the second software.

  According to such an information processing apparatus 10, when the starting point program module is designated, the edges of the first call graph 1 and the second call graph 2 are classified into categories according to the distance from the starting point nodes 1a and 2a. Is done. Next, the difference candidate first edge and the difference candidate second edge are specified by the classification unit 12a by comparing the first edge and the second edge of the same category. Then, the extraction unit 12c causes the difference first edge and the difference second edge to exist only in one of the first call graph 1 and the second call graph 2 from the difference candidate first edge and the difference candidate second edge. Is extracted.

  In the first embodiment, the candidate for edge to be extracted is narrowed down by the specifying unit 12b, and the difference first edge and the difference second edge are extracted from the candidate edges, so that the difference edge can be extracted efficiently. As a result, the time required for extracting the difference between call graphs is shortened.

  Note that the calculation unit 12 can also improve the efficiency of the extraction process by the extraction unit 12c using the hash values of the difference candidate first edge and the difference candidate second edge. In this case, the specifying unit 12b generates a first existence edge list in which the hash value of the difference candidate first edge is registered and a second existence edge list in which the hash value of the difference candidate second edge is registered. The hash value can be obtained by calculating a program name pair of the difference candidate first edge or the difference candidate second edge with a predetermined hash function. The extraction unit 12c extracts the difference candidate first edge from which the hash value different from the hash value registered in the second existence edge list is obtained as the difference first edge. In addition, the extraction unit 12c extracts the difference candidate second edge from which the hash value different from the hash value registered in the first existing edge list is obtained as the difference second edge.

By using a hash value in this way, a very long program name pair can be compressed into short data, and efficient processing using a small amount of memory becomes possible.
Further, the calculation unit 12 sorts (sorts) the first edge and the second edge within the classified category, thereby further improving the efficiency of the identification process between the difference candidate first edge and the difference candidate second edge. Can do. In that case, the specifying unit 12b sorts the first edge and the second edge on a predetermined basis for each classified category. Next, the specifying unit 12b uses the first edge group and the second edge group classified into the same distance category as comparison targets. Further, the specifying unit 12b selects the target first edge in order from the top of the first edge group, and selects the target second edge in order from the top of the second edge group. Then, the specifying unit 12b determines whether a set of program modules of the call source and the call destination at the target first edge and the target second edge is common. When not common, the specifying unit 12b specifies, as the difference candidate first edge or the difference candidate second edge, the higher one of the target first edge and the target second edge when sorted according to a predetermined criterion. To do.

  When specifying the difference candidate first edge, the specifying unit 12b changes the target first edge to the next first edge in the first edge group. In addition, when the difference candidate second edge is specified, the specifying unit 12b changes the target second edge to the next second edge in the second edge group. When determining that the program module set of the call source and the call destination is common, the specifying unit 12b changes the target first edge to the next first edge in the first edge group, and sets the target second edge. To the next second edge in the second edge group.

  Thus, by comparing the identities after sorting the first edge and the second edge within the category, it is possible to reduce the number of comparison processes compared to the case where comparison is made brute force. As a result, processing efficiency can be improved.

  In addition, the calculating part 12 is realizable with the processor which the information processing apparatus 10 has, for example. Moreover, the memory | storage part 11 is realizable with the memory or storage apparatus which the information processing apparatus 10 has, for example. The classification unit 12a, the specifying unit 12b, and the extraction unit 12c in the calculation unit 12 can be realized by causing the calculation unit 12 to execute a program, for example. At that time, the first distance-specific edge list 3, the second distance-specific edge list 4, the first differential edge list 5, the second differential edge list 6, and the final differential edge list 7 are stored in, for example, a memory included in the information processing apparatus 10 or It is temporarily stored in the storage device.

The above is the description of the first embodiment.
[Second Embodiment]
Next, a second embodiment will be described. In the second embodiment, the extraction of the difference between the edges of the call graph created from each of two pieces of software created by object-oriented programming is efficiently performed using a computer. In the call graph of two software created by object-oriented programming, a plurality of methods constituting the software are represented by nodes, and the call relationship between methods is represented by edges. That is, the method is an example of a program module shown in the first embodiment.

FIG. 2 is a diagram illustrating an example of generating a call graph. As shown in FIG. 2, by analyzing the source code 31, a call graph 32 showing a calling relationship between methods is obtained.
For example, when the source code 31 is analyzed, it can be seen that the methods "Tranceiver.transmit ()", "Transformer.encode ()", "Protocol.makeHeader ()", and "Tranceiver.send ()" exist. . Therefore, nodes 32 a to 32 d corresponding to the methods are generated in the call graph 32. When the source code 31 is analyzed, the method “Tranceiver.transmit ()” calls the other three methods “Transformer.encode ()”, “Protocol.makeHeader ()”, and “Tranceiver.send ()”. I understand that. Therefore, edges 32e to 32g connecting the node 32a corresponding to the method “Tranceiver.transmit ()” and the other three nodes 32b to 32d are generated. The edges 32e to 32g are arrows pointing from the node 32a corresponding to the call source method to the nodes 32b to 32d corresponding to the call destination method.

  In this manner, a call graph 32 is generated that represents a method by the nodes 32a to 32d and represents a call relationship between the methods by the edges 32e to 32g. Each edge 32e to 32g of the call graph 32 can be expressed by a pair of character strings.

  FIG. 3 is a diagram illustrating an example of a data structure of an edge of a call graph. Each edge 32e to 32g includes a fully qualified name character string of the caller method and a fully qualified name character string of the callee method. The character string on the left is the fully qualified name of the caller method, and the character string on the right is the fully qualified name of the callee method. An edge is specified by such a pair of character strings. In other words, it can be determined that edges having the same pair of character strings are edges indicating the same calling relationship.

  Note that the fully qualified names of the methods indicating the nodes 32a to 32d are included as character strings representing the edges 32e to 32g. Therefore, if there is information on the edges 32e to 32g, the entire call graph 32 including the nodes 32a to 32d can be generated.

In addition, the plurality of edges can be compared in size and in size by using character strings included in the data structure. For example, as shown in FIG. 3, there are two edges (N1, N2) and (N3, N4) represented by a combination of a character string indicating a caller method and a character string indicating a callee method. . Here, N1, N2, N3, and N4 are character strings. At this time, whether or not each edge is the same can be determined under the following conditions.
(N1, N2) = (N3, N4) ⇔N1 = N3 and N2 = N4
Two edges can be compared in size as follows.
(N1, N2) <(N3, N4) ⇔N1 <N3 or {N1 = N3 and N2 <N4}
(N1, N2) ≦ (N3, N4) ⇔N1 ≦ N3 or {N1 = N3 and N2 ≦ N4}
The magnitude relationship between character strings is, for example, in alphabetical order. In that case, “A” is the smallest value and “Z” is the largest value. This comparison method can be used to sort edges in a unique order.

  Such a call graph 32 is used during software development. For example, edge differences may be extracted from a plurality of call graphs generated from a plurality of source codes having different versions in the software development process. Extracting edge differences between call graphs is useful for grasping the range of influence of a modification when a part of a large-scale software is modified. For example, in developing software, software including calling a source code set created by a third party such as OSS (Open Source Software) may be created. Or the source code set created by a third party may be modified or functions may be added to complete the software.

  In the software created in this way, when the source code set to be called is updated, the method that the developer called or modified in the previous version (the method that the developer focuses on) has an effect on the update. Check if you are receiving. Hereinafter, a node corresponding to a method focused on by the developer is referred to as a starting node. From each of the two call graphs, it is possible to determine whether or not the influence of the update affects the starting point method, for example, by tracking the calling relationship from the starting point node and comparing the tracked calling relationship and extracting the difference. You can also specify other methods to scrutinize.

  In the second embodiment, such call-related difference extraction can be performed at high speed by computer processing. Therefore, in the second embodiment, the edges in the call graph are classified by the distance from the starting node.

  FIG. 4 is a diagram illustrating an example of edge distance on the call graph. The call graph 40 shown in FIG. 4 includes nodes 41a to 41k and edges 42a to 42k. When it is possible to reach another node by tracing d edges (d is an integer of 0 or more) from the starting node on the call graph 40 in the same direction, the distance between the two nodes is defined as d. The direction (search direction) following the edge at this time is the direction from the caller to the callee (callee direction), and conversely, the direction from the callee to the caller (caller direction). is there. The distance from the starting node to the starting node is defined as 0.

  And the distance of the edge with respect to the origin node can be defined using the distance between nodes. For example, an edge connecting a node located at a distance (d−1) from a starting node and a node located at a distance d defines the distance from the starting node as d.

  In the example of FIG. 4, the node 41g corresponding to “method g” is the starting node. The distance from the starting node to the node 41d corresponding to “method d” is “1”. Then, the distance of the edge 42h connecting the node 41g as the starting node and the node 41d corresponding to the node “method d” is “1”.

  The distance from the starting node to the node 41f corresponding to “method f” is “2”. Then, the distance of the edge 42d connecting the node 41d at the distance “1” from the starting node and the node 41f corresponding to the node “method f” is “2”.

  Here, it is assumed that the source code that is the generation source of the call graph 40 illustrated in FIG. 4 is updated. In this case, by generating a call graph for the updated source code, the call relationship between the methods in the updated source code is represented by the call graph.

  FIG. 5 is a diagram illustrating an example of a call graph of the updated source code. In the example of FIG. 5, when the call graph 40a of the updated source code is compared with the call graph 40 before the update shown in FIG. 4, the node 41k is deleted and the node 41l is added. An edge 42d connecting the node 41d and the node 41f is deleted, and an edge 42l connecting the node 41d and the node 41l and an edge 42m connecting the node 41l and the node 41f are added.

Hereinafter, a computer that performs high-speed edge difference extraction processing between the two call graphs 40 and 40a illustrated in FIGS. 4 and 5 will be described in detail.
<Hardware configuration>
FIG. 6 is a diagram illustrating a configuration example of hardware of a computer used in the second embodiment. The computer 100 is entirely controlled by a processor 101. A memory 102 and a plurality of peripheral devices are connected to the processor 101 via a bus 109. The processor 101 may be a multiprocessor. The processor 101 is, for example, a CPU (Central Processing Unit), an MPU (Micro Processing Unit), or a DSP (Digital Signal Processor). At least a part of the functions realized by the processor 101 executing the program may be realized by an electronic circuit such as an ASIC (Application Specific Integrated Circuit) or a PLD (Programmable Logic Device).

  The memory 102 is used as a main storage device of the computer 100. The memory 102 temporarily stores at least part of an OS program and application programs to be executed by the processor 101. The memory 102 stores various data necessary for processing by the processor 101. As the memory 102, for example, a volatile semiconductor storage device such as a RAM (Random Access Memory) is used.

  Peripheral devices connected to the bus 109 include a storage device 103, a graphic processing device 104, an input interface 105, an optical drive device 106, a device connection interface 107, and a network interface 108.

  The storage device 103 writes and reads data electrically or magnetically with respect to a built-in storage medium. The storage device 103 is used as an auxiliary storage device of a computer. The storage device 103 stores an OS program, application programs, and various data. For example, an HDD (Hard Disk Drive) or an SSD (Solid State Drive) can be used as the storage device 103.

  A monitor 21 is connected to the graphic processing device 104. The graphic processing device 104 displays an image on the screen of the monitor 21 in accordance with an instruction from the processor 101. Examples of the monitor 21 include a display device using a CRT (Cathode Ray Tube) and a liquid crystal display device.

  A keyboard 22 and a mouse 23 are connected to the input interface 105. The input interface 105 transmits signals sent from the keyboard 22 and the mouse 23 to the processor 101. The mouse 23 is an example of a pointing device, and other pointing devices can also be used. Examples of other pointing devices include a touch panel, a tablet, a touch pad, and a trackball.

  The optical drive device 106 reads data recorded on the optical disc 24 using laser light or the like. The optical disc 24 is a portable recording medium on which data is recorded so that it can be read by reflection of light. The optical disc 24 includes a DVD (Digital Versatile Disc), a DVD-RAM, a CD-ROM (Compact Disc Read Only Memory), a CD-R (Recordable) / RW (ReWritable), and the like.

  The device connection interface 107 is a communication interface for connecting peripheral devices to the computer 100. For example, the memory device 25 and the memory reader / writer 26 can be connected to the device connection interface 107. The memory device 25 is a recording medium equipped with a communication function with the device connection interface 107. The memory reader / writer 26 is a device that writes data to the memory card 27 or reads data from the memory card 27. The memory card 27 is a card type recording medium.

  The network interface 108 is connected to the network 20. The network interface 108 transmits and receives data to and from other computers or communication devices via the network 20.

  With the hardware configuration described above, the processing functions of the second embodiment can be realized. The information processing apparatus 10 shown in the first embodiment can also be realized by hardware similar to the computer 100 shown in FIG.

  The computer 100 implements the processing functions of the second embodiment by executing a program recorded on a computer-readable recording medium, for example. A program describing the processing content to be executed by the computer 100 can be recorded in various recording media. For example, a program to be executed by the computer 100 can be stored in the storage device 103. The processor 101 loads at least a part of the program in the storage apparatus 103 into the memory 102 and executes the program. A program to be executed by the computer 100 can be recorded on a portable recording medium such as the optical disc 24, the memory device 25, and the memory card 27. The program stored in the portable recording medium becomes executable after being installed in the storage apparatus 103 under the control of the processor 101, for example. The processor 101 can also read and execute a program directly from a portable recording medium.

<Processing function>
The computer 100 compares call graphs created from each of the two source code sets and detects an edge difference. The functions of the computer 100 for the call graph comparison process are shown in FIG.

  FIG. 7 is a block diagram illustrating functions of a computer for call graph comparison. The computer 100 includes a first call graph storage unit 111, a second call graph storage unit 112, a request reception unit 120, an edge classification unit 130 by distance, a first edge list storage unit 141 by distance, and a second edge list storage unit by distance 142, same distance edge comparison unit 150, first difference edge storage unit 161, second difference edge storage unit 162, first existence edge storage unit 163, second existence edge storage unit 164, difference edge extraction unit 170, difference extraction A result storage unit 180 and a result response unit 190 are included. Of these, the edge-by-distance classifying unit 130 is an example of the classifying unit 12a according to the first embodiment. The same distance edge comparison unit 150 is an example of the specifying unit 12b in the first embodiment. Furthermore, the differential edge extraction unit 170 is an example of the extraction unit 12c in the first embodiment.

  The first call graph storage unit 111 stores one of the two call graphs to be compared. For example, the first call graph storage unit 111 stores a call graph indicating a call relationship between methods in the software of the version number before modification.

  The second call graph storage unit 112 stores another one of the two call graphs to be compared. For example, the second call graph storage unit 112 stores a call graph indicating a call relationship between methods in the modified version of software.

  The request receiving unit 120 receives a difference extraction request input from a developer. The difference extraction request includes information specifying a method of interest and information specifying a search direction. The request reception unit 120 transmits the received difference extraction request to the edge classification unit 130 by distance.

  The edge classification unit by distance classifies the edges of each of the two call graphs stored in the first call graph storage unit 111 and the second call graph storage unit 112 according to the distance from the node corresponding to the method of interest. To do. Specifically, the edge-by-distance classifying unit 130 generates a category for each distance from the focused method. The distance-specific edge classification unit 130 reads the call graph stored in the first call graph storage unit 111, determines the category to which the edge in the call graph belongs, and determines each edge as the first distance-specific edge for each category. Stored in the list storage unit 141. Similarly, the edge-by-distance classification unit 130 reads the call graph stored in the second call graph storage unit 112, determines the category to which the edge in the call graph belongs, and sets each edge to the second distance for each category. It is stored in another edge list storage unit 142.

  The first distance-specific edge list storage unit 141 classifies and stores the edges included in the call graph in the first call graph storage unit 111 into categories according to the distance from the method of interest.

  The second distance-specific edge list storage unit 142 classifies and stores the edges included in the call graph in the second call graph storage unit 112 into categories according to the distance from the method of interest.

  The same distance edge comparison unit 150 compares edges classified into the same category in the first distance-specific edge list storage unit 141 and the second distance-specific edge list storage unit 142, and determines the presence or absence of the same edge. The same distance edge comparison unit 150 stores edges that are stored in a certain category in the first distance-specific edge list storage unit 141 but are not stored in the same category in the second distance-specific edge list storage unit 142. And stored in the first differential edge storage unit 161. The same distance edge comparison unit 150 stores edges that are stored in a certain category in the second distance edge list storage unit 142 but are not stored in the same category in the first distance edge list storage unit 141. , And stored in the second differential edge storage unit 162.

  Further, the same distance edge comparison unit 150 calculates the hash value of the edge stored in the first difference edge storage unit 161 and stores the obtained hash value in the first existence edge storage unit 163. Further, the same distance edge comparison unit 150 calculates the hash value of the edge stored in the second difference edge storage unit 162 and stores the obtained hash value in the second existence edge storage unit 164.

  The first differential edge storage unit 161 stores edges that are stored in a certain category in the first distance-specific edge list storage unit 141 but are not stored in the same category in the second distance-specific edge list storage unit 142. To do.

  The second difference edge storage unit 162 stores edges that are stored in a certain category in the second distance-specific edge list storage unit 142 but are not stored in the same category in the first distance-specific edge list storage unit 141. To do.

The first existence edge storage unit 163 stores the hash value of the edge stored in the first differential edge storage unit 161.
The second existence edge storage unit 164 stores the hash value of the edge stored in the second differential edge storage unit 162.

  The difference edge extraction unit 170 extracts edges included in only one of the two call graphs to be compared from the first difference edge storage unit 161 and the second difference edge storage unit 162. Specifically, the differential edge extraction unit 170 determines whether or not the same hash value as the hash value of the edge stored in the first differential edge storage unit 161 is stored in the second existence edge storage unit 164. to decide. If the corresponding hash value is not stored, the difference edge extraction unit 170 extracts the edge that is the generation source of the hash value from the first difference edge storage unit 161 and stores it in the difference extraction result storage unit 180. Further, the difference edge extraction unit 170 determines whether or not the same hash value as the edge hash value stored in the second difference edge storage unit 162 is stored in the first existence edge storage unit 163. If the corresponding hash value is not stored, the difference edge extraction unit 170 extracts the edge that is the generation source of the hash value from the second difference edge storage unit 162 and stores it in the difference extraction result storage unit 180.

  The difference extraction result storage unit 180 stores an extraction result of edges included only in the call graph stored in one of the first call graph storage unit 111 and the second call graph storage unit 112.

The result response unit 190 outputs edge information stored in the difference extraction result storage unit 180 as a response to the difference extraction request.
In addition, the line which connects between each element shown in FIG. 7 shows a part of communication path, and communication paths other than the illustrated communication path can also be set. Moreover, the function of each element shown in FIG. 7 can be realized, for example, by causing a computer to execute a program module corresponding to the element.

<Data structure>
Next, data stored in each storage unit in the computer 100 will be described in detail with reference to FIGS.

  FIG. 8 is a diagram illustrating an example of data in the first and second call graph storage units. The first call graph storage unit 111 stores call graph data 111 a indicating the call graph 40. The call graph data 111a includes two sets of character strings indicating each of the plurality of edges 42a to 42k included in the call graph 40. The second call graph storage unit 112 stores call graph data 112a indicating the call graph 40a. The call graph data 112a includes two sets of character strings indicating the plurality of edges 42a to 42c, 42e to 42k, 42l, and 42m included in the call graph 40a.

  FIG. 9 is a diagram illustrating an example of data in the first and second distance-specific edge list storage units. The first distance-specific edge list storage unit 141 stores a first distance-specific edge list 141a indicating the classification result of the edges in the call graph 40 based on the distance from the starting node. The second distance-specific edge list storage unit 142 stores a second distance-specific edge list 142a indicating the classification result of the edges in the call graph 40a according to the distance from the starting node.

  In the first distance-specific edge list 141a and the second distance-specific edge list 142a, the storage area is divided into blocks categorized by distance. In each block, an edge whose distance from the starting node is the same as the distance set in the block is registered. For example, in the “distance 1” column, an edge whose distance from the starting node is “1” is registered.

  FIG. 10 is a diagram illustrating an example of data in the first and second differential edge storage units. The first differential edge storage unit 161 stores a first differential edge list 161a. The first difference edge list 161a is registered in a certain distance column in the first distance-specific edge list 141a, but is not registered in the same distance column in the second distance-specific edge list 142a. Indicates a set. The second differential edge storage unit 162 stores a second differential edge list 162a. The second difference edge list 162a is registered in a certain distance column in the second distance-specific edge list 142a, but is not registered in the same distance column in the first distance-specific edge list 141a. Indicates a set.

  FIG. 11 is a diagram illustrating an example of data in the first and second existence edge storage units. The first existing edge storage unit 163 stores a first existing edge list 163a. The first existence edge list 163a is data indicating hash values of a plurality of edges in the first differential edge list 161a. The second existing edge storage unit 164 stores a second existing edge list 164a. The second existence edge list 164a is data indicating hash values of a plurality of edges in the second differential edge list 162a.

  FIG. 12 is a diagram illustrating an example of data in the difference extraction result storage unit. The difference extraction result storage unit 180 stores a final difference edge list 181 indicating a set of edges included only in one of the two call graphs 40 and 40a. In the final difference edge list 181, for each edge, two sets of character strings representing the edge and an identification number of the call graph that includes the edge are set. For example, the identification number of the call graph 40 is “1”, and the identification number of the call graph 40 a is “2”. In this case, the identification number “1” of the call graph 40 is assigned to the edge 42k (a pair of character strings (“g”, “k”) that exists in the call graph 40 but does not exist in the call graph 40a. .

<Processing procedure>
The computer 100 performs differential edge extraction processing of the two call graphs 40 and 40a using the above data. Hereinafter, the differential edge extraction process will be described in detail.

  FIG. 13 is a flowchart illustrating a procedure of differential edge extraction processing. In the following, the process illustrated in FIG. 13 will be described in order of step number. The difference edge extraction process is executed when a difference extraction request is input.

  [Step S101] Upon receiving the difference extraction request, the request reception unit 120 transmits the difference extraction request to the edge-based edge classification unit 130. Thereafter, the processes of step S102 and step S103 are executed in parallel.

  [Step S <b> 102] The edge-by-distance classifying unit 130 activates two threads upon obtaining the difference extraction request. Then, the edge classification unit 130 by distance executes the edge classification process by distance for the call graph data 111a in the first call graph storage unit 111 by the first thread. As a result, the first distance-specific edge list 141 a is stored in the first distance-specific edge list storage unit 141.

  [Step S103] The edge-by-distance classification unit 130 executes edge-by-distance classification processing on the call graph data 112a in the second call graph storage unit 112 by the second thread. As a result, the second distance-specific edge list 142 a is stored in the second distance-specific edge list storage unit 142.

  The distance-based edge classification processing in steps S102 and S103 is processing for classifying the edges in the call graph by the distance from the starting node. Details of the distance-based edge classification processing will be described later (see FIG. 14). When the processes in both step S102 and step S103 are completed, the process proceeds to step S104.

  [Step S104] The edge-by-distance classifying unit 130 executes edge comparison processing between the same distances. In the same distance edge comparison processing, edge sets in the same distance category from the starting node in each of the first distance edge list 141a and the second distance edge list 142a are compared to determine the presence or absence of the same edge. Is done. By the edge comparison process between the same distances, the first difference edge list 161a is stored in the first difference edge storage unit 161, and the second difference edge list 162a is stored in the second difference edge storage unit 162. Further, by the same distance edge comparison processing, the first existence edge list 163a is stored in the first existence edge storage unit 163, and the second existence edge list 164a is stored in the second existence edge storage unit 164. Details of edge comparison processing between the same distances will be described later (see FIG. 15). When the edge comparison process between the same distances is completed, the processes of step S105 and step S106 are executed in parallel.

  [Step S105] The differential edge extraction unit 170 starts two threads. Then, the difference edge extraction unit 170 executes the distance difference edge existence confirmation process for the first difference edge list 161a by the first thread. In this distance difference edge existence confirmation process, edges that are included in the first differential edge list 161a but are not included in the second differential edge list 162a are specified.

  [Step S106] The difference edge extraction unit 170 executes the distance difference edge existence confirmation process for the second difference edge list 162a by the second thread. In this distance difference edge existence confirmation process, an edge that is included in the second differential edge list 162a but is not included in the first differential edge list 161a is specified.

  Details of the difference edge existence confirmation processing in steps S105 and S106 will be described later (see FIG. 16). The final difference edge list 181 is stored in the difference extraction result storage unit 180 by the distance difference edge existence confirmation processing in steps S105 and S106. When the processes in both step S105 and step S106 are completed, the process proceeds to step S107.

[Step S107] The result reply unit 190 outputs the final difference edge list 181 indicating the result of the difference extraction process as a response to the difference extraction request.
The difference edge between the two call graphs 40 and 40a can be extracted by the procedure as described above. Hereinafter, the distance-based edge classification processing (steps S102 and S103), the same distance edge comparison processing (step S104), and the distance difference edge presence confirmation processing (steps S105 and S106) will be described in detail with reference to FIGS. Explained.

  FIG. 14 is a flowchart illustrating an example of a procedure of distance-based edge classification processing. In the following, the process shown in FIG. 14 will be described in order of step numbers, taking as an example the case where the call graph data 111a in the first call graph storage unit 111 is to be processed.

  [Step S111] The edge-by-distance classifying unit 130 generates a current node list in which the node corresponding to the method of interest specified in the difference extraction request is set as the starting node, and only the identifier of the starting node is registered as an element. The identifier of the node registered in the current node list is, for example, the method name of the method corresponding to that node. The generated current node list is stored in the memory 102, for example.

[Step S112] The edge classification unit 130 sets “1” as the distance d to be compared.
[Step S113] The edge-by-distance classifying unit 130 generates a next node list whose contents are empty. The generated next node list is stored in the memory 102, for example.

  [Step S114] The edge classification unit 130 by distance sets all edges included in the call graph data 111a to an unsearched state. For example, when the flag indicating “searched” is set for the edge in the call graph data 111a, the edge classification unit 130 by distance deletes the flag.

[Step S115] The edge-based edge classification unit 130 selects one element registered in the current node list as the current node.
[Step S116] The edge-by-distance classifying unit 130 selects one of the edges connected to the search direction from the current node, to which the “searched” flag has not been assigned, as a search target. The search direction is specified in the difference extraction request. If the search direction is the callee direction, the edge classification unit by distance 130 selects one edge having the current node as the caller from the call graph data 111a. If the search direction is the caller direction, the distance-based edge classification unit 130 selects one edge having the current node as the callee from the call graph data 111a.

  [Step S117] The edge-by-distance classifying unit 130 stores the selected edge in the block of the distance d in the first edge-by-distance list 141a. Then, the edge-by-distance classifying unit 130 sets a “searched” flag to the selected edge of the call graph data 111a.

  [Step S118] The edge-by-distance classification unit 130 adds, to the next node list, nodes other than the current node among the two nodes connected by the selected edge to the next node list. If the search direction is the call destination direction, the call destination node of the selected edge is added to the next node list. If the search direction is the caller direction, the caller node of the selected edge is added to the next node list.

  [Step S119] The edge-by-distance classifying unit 130 determines whether or not all edges connected in the search direction from the current node have been searched. If all edges are to be searched, the process proceeds to step S120. If there is an edge that is not a search target, the process proceeds to step S116.

  [Step S120] The edge-by-distance classification unit 130 determines whether all nodes in the current node list have been processed. If the processing of all the nodes has been completed, the edge-based edge classification processing ends. If there is an unprocessed node, the process proceeds to step S121.

  [Step S121] The edge-by-distance classifying unit 130 matches the contents of the current node list with the contents of the next node list. For example, the distance-based edge classification unit 130 deletes all the contents of the current node list, and then registers all the nodes registered in the next node list in the current node list.

[Step S122] The edge-based edge classification unit 130 deletes all the nodes in the next node list and empties the next node list. Thereafter, the process proceeds to step S115.
In this way, the edges 42a to 42k included in the call graph 40 are categorized by the distance and registered in the first distance-specific edge list 141a. Similarly, if the edge-based edge classification processing in FIG. 14 is executed with the call graph data 112a in the second call graph storage unit 112 as a processing target, the edges 42a to 42c, 42e to 42j, and 42l included in the call graph 40a. , 42m are classified by distance. Then, the edges 42a to 42c, 42e to 42j, 42l, and 42m are registered in the second distance-specific edge list 142a while being divided into categories for each distance. Thereafter, the same distance edge comparison processing is executed by the same distance edge comparison unit 150.

FIG. 15 is a flowchart illustrating an example of the same distance edge comparison process. In the following, the process illustrated in FIG. 15 will be described in order of step number.
[Step S131] The same distance edge comparison section 150 sets 1 to the distance d.

  [Step S132] The edge comparison unit 150 between the same distances sorts the edges stored in the block of the distance d in the first distance-specific edge list 141a according to a predetermined criterion. For example, the same distance edge comparison unit 150 sorts the edges categorized at the same distance in alphabetical order based on a pair of character strings representing the edges. Then, the same distance edge comparison unit 150 sets the sorted edge arrangement as the first list.

  [Step S133] The edge comparison unit 150 within the same distance sorts the edges stored in the block of the distance d in the second distance-specific edge list 142a according to the same criteria as in Step S132. Then, the same distance edge comparison unit 150 sets the sorted edge arrangement as the second list.

  [Step S134] The same distance edge comparison unit 150 sets the first edge of the first list as the point of interest in the first list. Similarly, the same distance edge comparison unit 150 sets the first edge of the second list as a point of interest in the second list.

  [Step S135] The same distance edge comparison unit 150 determines whether the edge of the point of interest in the first list is the same as the edge of the point of interest in the second list. If the edges are the same, the process proceeds to step S136. If the edges are not identical, the process proceeds to step S137.

  [Step S136] The same distance edge comparison unit 150 moves the point of interest in the first list to the edge next to the edge of the current point of interest in the first list. The same distance edge comparison unit 150 moves the point of interest in the second list to the next edge of the edge of the current point of interest in the second list. Thereafter, the process proceeds to step S144.

  [Step S137] The same distance edge comparison unit 150 compares the edges of the points of interest in the first list and the edges of the points of interest in the second list based on the sort conditions in steps S132 and S133. For example, the same distance edge comparison unit 150 determines that the higher edge in the sort condition is the smaller edge. If the edge of the target point in the first list is smaller, the process proceeds to step S138. If the edge of the point of interest in the second list is smaller, the process proceeds to step S141.

[Step S138] The same-distance edge comparison unit 150 adds the edge of the target point of the first list to the first difference edge list 161a.
[Step S139] The edge comparison unit 150 between the same distances adds the hash value of the edge of the point of interest in the first list to the first existing edge list 163a.

  [Step S140] The same distance edge comparison unit 150 moves the point of interest in the first list to the edge next to the edge of the current point of interest in the first list. Thereafter, the process proceeds to step S144.

[Step S141] The same distance edge comparison unit 150 adds the edge of the target point of the second list to the second difference edge list 162a.
[Step S142] The same-distance edge comparison unit 150 adds the hash value of the edge of the target point in the second list to the second existence edge list 164a.

[Step S143] The same distance edge comparison unit 150 moves the point of interest in the second list to the next edge of the edge of the current point of interest in the second list.
[Step S144] The same distance edge comparison unit 150 determines whether all edges have been compared in both the first list and the second list. If all edges have been compared, the process proceeds to step S145. If at least one of the first list and the second list includes an uncompared edge (an edge that is not a point of interest), the process proceeds to step S135.

  In addition, when all the edges have been compared in one of the first list and the second list, the same or larger comparison partner does not exist for the remaining edges as comparison targets. In this case, the edge remaining as the comparison target is determined as “NO” in the identity determination in step S135. In the size determination in step S137, it is determined that the remaining edge as a comparison target is smaller.

  [Step S145] The same distance edge comparison unit 150 determines whether or not processing has been performed for all distances d existing as categories. When the process is completed for all the distances d, the edge comparison process between the same distances ends. If there is an unprocessed distance, the process proceeds to step S146.

[Step S146] The same distance edge comparison unit 150 adds 1 to the value of the distance d, and advances the process to Step S132.
In this way, the first differential edge list 161a, the second differential edge list 162a, the first existing edge list 163a, and the second existing edge list 164a are generated. Thereafter, the difference edge extraction unit 170 executes a difference distance edge existence confirmation process (steps S105 and S106).

  FIG. 16 is a flowchart illustrating an example of the procedure of the difference distance edge presence confirmation process. In the following, the process shown in FIG. 16 will be described in order of step number, taking as an example the case where the first differential edge list 161a in the first differential edge storage unit 161 is used as an edge extraction source.

[Step S151] The differential edge extraction unit 170 selects one unprocessed edge from the first differential edge list 161a.
[Step S152] The differential edge extraction unit 170 calculates a hash value of the selected edge, and determines whether the hash value exists in the second existence edge list 164a. If the hash value exists, the process proceeds to step S154. If the hash value does not exist, the process proceeds to step S153.

  [Step S153] The difference edge extraction unit 170 adds the identifier of the source code from which the edge is extracted to the selected edge, and adds it to the final difference edge list 181 in the difference extraction result storage unit 180.

  [Step S154] The differential edge extraction unit 170 determines whether or not all the edges in the first differential edge list 161a have been processed. When the processing for all the edges is completed, the difference distance edge presence confirmation processing ends. If there is an unprocessed edge, the process proceeds to step S151.

  In this way, it is confirmed whether or not the same edge exists in the other call graph between the edges with different distances from the origin node, and only the nonexistent edges are stored in the final difference edge list 181. Then, the edge in the final difference edge list 181 is output as a response to the difference extraction request.

<Effects of Second Embodiment>
As described above, according to the second embodiment, the existence of the same edge for each edge is determined by classifying the distance from the starting node and extracting the difference between the edges having the same distance first. Confirmation is easy. As a result, the processing is speeded up.

  Here, the processing time of the difference extraction processing shown in the second embodiment is compared with the processing time of other difference extraction processing. The following two reference methods are assumed as a difference extraction method to be compared.

  [First Reference Method] In this method, each edge in one call graph is compared with all the edges in the other call graph. In this approach, it is detected that the matched edge is in both call graphs and the unmatched edge is in only one call graph.

  [Second Reference Method] This is a method of sorting the edge groups in each call graph, comparing the sorted lists of edges, and extracting different edges. In this method, after sorting the edges for each call graph, the two sorted edge lists are compared while moving the point of interest sequentially from the top. If they match, the point of interest is moved to the next edge in both lists, and if they do not match, the point of interest is moved to the next edge only in a list that has a smaller edge. A matching edge is found in both call graphs, and a non-matching edge is determined to have a smaller edge in only one call graph. This second reference method is obtained by excluding distance-based edge classification processing from the method of the second embodiment.

Hereinafter, with reference to FIGS. 17 and 18, the time of difference extraction processing between each of the first and second reference techniques and the technique according to the second embodiment is compared. In the following description, the required time is calculated using the variables M, N, k, r, and n. The meaning of each variable is as follows.
M, N: Number of edges included in each call graph k: Radius of target range (maximum distance between calling method and calling method)
r: Ratio of difference between two source code sets that are the basis of two call graphs (number of differences / (number of differences + number of differences))
n: Number of methods calling (or called) one method << Comparison with the first reference method >>
In the first reference method, matching between edges is determined brute force. Here, the time required for one comparison of two edges is set as a unit time, and how many times the unit time is completed is calculated as the unit time. In the case of the first reference method, the time rt0 required for difference extraction is expressed by the following equation.
rt0 = MN (1)
On the other hand, the required time rt2 when the difference extraction process is performed by the method shown in the second embodiment can be expressed by the following equation.
rt2 = Mlog ₂ (M / k) + Nlog ₂ (N / k) + M + N + rM + rN
... (2)
The first term “Mlog ₂ (M / k)” on the right side of Equation (2) indicates the time required to sort the edges belonging to one category (distance) of one of the two call graphs. Yes. A time of “(M / k) log ₂ (M / k)” is required to sort one category, and “Mlog ₂ (M / k)” is obtained by multiplying the time by the number of categories k. The second term “Nlog ₂ (N / k)” indicates the time required to sort the edges belonging to one category of the other call graph.

  The third term “M” and the fourth term “N” are times required for comparison between edges of the same category. The time required for comparison for one category is “(M / k) + (N / k)”, and “M + N” is obtained by multiplying the time by the number of categories k.

  The fifth term “rM” and the sixth term “rN” are the time required to compare the difference edge, for which the same edge is not found in the comparison between edges of the same category, with the edge of a different category.

  Here, if there is only a small difference between the two source code sets that are the basis of the two call graphs, the following approximation can be made using the number n of methods that call (or are called) one method: It is possible ("~" below is an approximate symbol).

In this case, rt0 and rt2 are as follows.
rt0~M ² ~n ^2k ··· (4)
rt2 to 2M (log ₂ (M / k) + 1 + r) to 2n ^k (klog ₂ d-log ₂ k + 1 + r) (5)
In many cases, it is n-5. In addition, the minor version upgrade of OS of various systems is about r to 0.01. When the improvement rate is calculated under such conditions, it is as shown in FIG.

FIG. 17 is a diagram illustrating a comparative example between the first reference method and the method according to the second embodiment. FIG. 17 shows the improvement rate of the difference extraction time according to the maximum value of the distance from the starting node. The improvement rate is calculated by the following formula.
Improvement rate = (rt0−rt2) / rt0) (6)
In FIG. 17, the improvement rate is expressed as a percentage (value of equation (6) × 100). In this example, it can be seen that when k = 2 or more, the speed can be increased by 49% or more.

<< Comparison with the second reference method >>
In the case of the second reference method, the time rt1 required for difference extraction can be expressed by the following equation.
rt1 = Mlog ₂ M + Nlog ₂ N + M + N (7)
The first term “Mlog ₂ M” on the right side of Equation (7) represents the time required to sort the edges of one call graph. The second term “Nlog ₂ N” represents the time required to sort the edges of the other call graph. The third term “M” and the fourth term “N” are times for comparing edges in the sorted list.

When the approximation shown in Expression (3) is performed, rt1 is as follows.
rt1 to 2M (log ₂ M + 1) to 2n ^k (klog ₂ d + 1) (8)
Here, when the improvement rate is calculated under the conditions of n to 5 and r to 0.01, it is as shown in FIG.

FIG. 18 is a diagram illustrating a comparative example between the second reference method and the method according to the second embodiment. FIG. 18 shows the improvement rate of the difference extraction time according to the maximum value of the distance from the starting node. The improvement rate is calculated by the following formula.
Improvement rate = (rt1-rt2) / rt1) (9)
In FIG. 18, the improvement rate is expressed as a percentage (value of equation (9) × 100). In this example, it can be seen that when k = 2 or more, the speed can be increased by about 20%.

  Thus, in the second embodiment, high-speed differential extraction processing is possible by performing classification of edges according to the distance from the starting node and comparison after sorting edges belonging to the same distance. It has become.

<< Other effects >>
In the second embodiment, the computer 100 calculates the hash value of the edge extracted as the difference as a result of the edge comparison between the same distances, and generates the first existence edge list 163a and the second existence edge list 164a. Yes. Then, the first existence edge list 163a and the second existence edge list 164a are set as search targets when the existence of different distance edges is confirmed. In this way, the processing is streamlined by performing the edge existence confirmation using the hash value. For example, when comparing a pair of two character strings representing the fully qualified name of a method indicating an edge, the comparison target character string is long, so that the memory consumption increases and the comparison also takes time. If the comparison is performed by converting the hash value into MD5 (Message Digest 5) or the like, the character string to be compared is shortened, the memory capacity is reduced, and the processing time is shortened.

For example, in the OS for portable information terminal devices such as smartphones and tablet terminals, the following names are used as fully qualified names of methods. In addition, the character used for the name shown below shall be a half-width character.
・ Google.protobuf.compiler.StaticDescriptorInitializer_google_2fprotobuf_2fcompiler_2fplugin_2eproto.StaticDescriptorInitializer_google_2fprotobuf_2fcompiler_2fplugin_2eproto (172 characters)
-__Gnu_pbds.detail.left_child_next_sibling_heap_node_point_const_iterator_.left_child_next_sibling_heap_node_point_const_iterator_ (129 characters)
・ Extensions.FileBrowserPrivateLogoutUserForReauthenticationFunction.FileBrowserPrivateLogoutUserForReauthenticationFunction (122 characters)
・ Line_map.tree_base.tree_decl_common.tree_decl_with_vis.tree_function_decl.VEC_call_site_record_base_unordered_remove (116 characters)
The OS has about 600,000 methods. The longest fully qualified name is 279 characters, the shortest is 5 characters, the average is 46.2 characters, and the standard deviation is 15.6 characters. The edge data structure includes a pair of such long method names.

  When a character string indicating an edge is converted into, for example, an MD5 hash value, the length becomes 128 bits. 128 bits are equivalent to about 18 characters when encoded with ASCII code (7 bits). Then, the length (about 93 characters) of the above-mentioned fully qualified name with the average length is sufficiently longer than the hash value (even if the standard deviation is taken into consideration).

Therefore, efficient processing with a small amount of data can be performed by performing edge comparison processing for the same distance using the hash value of the edge.
The above is the description of the second embodiment.

[Other Embodiments]
In the second embodiment, the distance difference edge existence confirmation is performed using the first existence edge list 163a and the second existence edge list 164a, but the first existence edge list 163a, the second existence edge list 164a, It is also possible to perform edge existence confirmation at different distances without using. In that case, of the edges in the first differential edge list 161 a, an edge that does not have the same edge in the second differential edge list 162 a is extracted and registered in the final differential edge list 181. Similarly, of the edges in the second differential edge list 162a, an edge that does not have the same edge in the first differential edge list 161a is extracted and registered in the final differential edge list 181.

  As described above, when the first differential edge list 161a and the second differential edge list 162a are to be searched, the edge is an associative array using a pair of character strings of each edge as a key (array subscript). It may be held. At this time, the value (value) of the associative array corresponding to the key may be a blank or a predetermined value such as “0”. In other words, when an associative array is read using a pair of character strings of the processing target edge as a key, if the corresponding value is read, the same edge as that edge exists, and if an error occurs, that edge The same edge does not exist. In this way, by storing two sets of character strings indicating edges in the associative array, whether or not an edge that exists in one of the first differential edge list 161a and the second differential edge list 162a exists in the other. Such a search can be carried out efficiently.

  It is also possible to hold the first existence edge list 163a and the second existence edge list 164a using the hash value of each edge as an array key. As a result, the presence check of the edge based on the first existence edge list 163a and the second existence edge list 164a can be performed at high speed.

  As mentioned above, although embodiment was illustrated, the structure of each part shown by embodiment can be substituted by the other thing which has the same function. Moreover, other arbitrary structures and processes may be added. Further, any two or more configurations (features) of the above-described embodiments may be combined.

DESCRIPTION OF SYMBOLS 1 1st call graph 1a-1d 1st node 1e-1g 1st edge 2 2nd call graph 2a-2e 2nd node 2f-2i 2nd edge 3 Edge list classified by 1st distance 4 Edge list classified by 2nd distance 5 1st 1 differential edge list 6 second differential edge list 7 final differential edge list 10 information processing apparatus 11 storage unit 12 calculation unit 12a classification unit 12b identification unit 12c extraction unit

Claims (5)

Computer
When a starting point program module is specified, a calling relationship between a plurality of first program modules in the first software including the starting point program module is connected between a plurality of first nodes indicating the plurality of first program modules. Based on the first call graph represented by the plurality of first edges, the plurality of the plurality of the plurality of the plurality of the plurality of the plurality of the plurality of the plurality of the plurality of the plurality of first edges by the distance from the first node corresponding to the starting program module to the respective first edges. Categorize the first edge of
A calling relationship between a plurality of second program modules in the second software including the starting program module is represented by a plurality of second edges connecting a plurality of second nodes indicating the plurality of second program modules. Based on the second call graph, the plurality of second edges are classified into categories for each distance according to the distance from the second node corresponding to the origin program module to the respective second edges following the calling relationship. Classified into
Each of the plurality of first edges is a second edge that belongs to the same category as the target first edge, and a set of program modules of a caller and a callee is a second edge that is common to the target first edge If the target edge does not exist, the target first edge is specified as the difference candidate first edge,
A first edge that is a target of each of the plurality of second edges, and a set of program modules of a caller and a callee is common to the target second edge, among the first edges belonging to the same category as the target second edge If the target second edge does not exist, the target second edge is specified as the difference candidate second edge,
From the difference candidate first edge, extract a difference first edge that does not have the difference candidate second edge where a set of program modules of a caller and a callee is common,
Extracting the difference second edge in which the pair of program modules of the caller and the callee is common, and the difference candidate first edge does not exist among the difference candidate second edges,
Call graph difference extraction method. The computer further comprises:
Generating a first existing edge list in which a hash value of the difference candidate first edge is registered and a second existing edge list in which a hash value of the difference candidate second edge is registered;
In the extraction of the difference first edge, the difference candidate first edge from which a hash value different from the hash value registered in the second existence edge list is obtained is extracted as the difference first edge;
In the extraction of the difference second edge, the difference candidate second edge from which a hash value different from the hash value registered in the first existence edge list is obtained is extracted as the difference second edge.
The call graph difference extraction method according to claim 1. In the identification of the difference candidate first edge and the difference candidate second edge, the first edge and the second edge are sorted according to a predetermined criterion for each classified category, and are classified into categories of the same distance. Selecting the target first edge in order from the top of the first edge group, selecting the target second edge in order from the top of the second edge group, and comparing the edge group and the second edge group, When the set of program modules of the call source and the call destination at the target first edge and the target second edge is not common, the target first edge and the target second edge are sorted according to the predetermined criterion The higher one is specified as the difference candidate first edge or the difference candidate second edge,
The call graph difference extraction method according to claim 1 or 2. On the computer,
When a starting point program module is specified, a calling relationship between a plurality of first program modules in the first software including the starting point program module is connected between a plurality of first nodes indicating the plurality of first program modules. Based on the first call graph represented by the plurality of first edges, the plurality of the plurality of the plurality of the plurality of the plurality of the plurality of the plurality of the plurality of the plurality of the plurality of first edges by the distance from the first node corresponding to the starting program module to the respective first edges. Categorize the first edge of
A calling relationship between a plurality of second program modules in the second software including the starting program module is represented by a plurality of second edges connecting a plurality of second nodes indicating the plurality of second program modules. Based on the second call graph, the plurality of second edges are classified into categories for each distance according to the distance from the second node corresponding to the origin program module to the respective second edges following the calling relationship. Classified into
Each of the plurality of first edges is a second edge that belongs to the same category as the target first edge, and a set of program modules of a caller and a callee is a second edge that is common to the target first edge If the target edge does not exist, the target first edge is specified as the difference candidate first edge,
A first edge that is a target of each of the plurality of second edges, and a set of program modules of a caller and a callee is common to the target second edge, among the first edges belonging to the same category as the target second edge If the target second edge does not exist, the target second edge is specified as the difference candidate second edge,
From the difference candidate first edge, extract a difference first edge that does not have the difference candidate second edge where a set of program modules of a caller and a callee is common,
Extracting the difference second edge in which the pair of program modules of the caller and the callee is common, and the difference candidate first edge does not exist among the difference candidate second edges,
Call graph difference extraction program that executes processing. A first call graph representing a call relationship between a plurality of first program modules in the first software by a plurality of first edges connecting a plurality of first nodes indicating the plurality of first program modules; A second call graph representing a calling relationship between a plurality of second program modules in the second software by a plurality of second edges connecting the plurality of second nodes indicating the plurality of second program modules; A storage unit for storing;
When a starting point program module included in both the first software and the second software is specified, the calling relationship is traced from the first node corresponding to the starting point program module based on the first call graph. The plurality of first edges are classified into categories for each distance according to distances to reach each of the plurality of first edges, and are called from the second node corresponding to the origin program module based on the second call graph. A classification unit that classifies the plurality of second edges into categories for each distance according to distances that follow the relationship and reach each of the plurality of second edges;
Each of the plurality of first edges is a second edge that belongs to the same category as the target first edge, and a set of program modules of a caller and a callee is a second edge that is common to the target first edge If the target first edge is specified as a difference candidate first edge, each of the plurality of second edges is targeted, and the first edge belonging to the same category as the target second edge A specific unit that identifies the target second edge as a difference candidate second edge when the set of program modules with the callee does not have a first edge common to the target second edge;
The difference candidate first edge is extracted from the difference candidate first edge, and the difference candidate second edge in which the pair of program modules of the caller and the callee is common does not exist, and of the difference candidate second edge, An extraction unit that extracts a difference second edge in which a pair of program modules of a caller and a callee is common, and the difference candidate first edge does not exist;
An information processing apparatus.

Images