One approach to understanding a system's source code is to count things, such as the number of statements, functions, classes, expressions, and LOC (Lines Of Code). The program srcfacts produces a report of these counts.
Writing code in a program to directly parse C++, i.e., identify the syntactic parts of the code, is very difficult. So, the srcfacts program does not parse the C++ source code. It uses another tool, srcML, to convert the C++ source code into an XML format. The srcfacts program inputs this XML format, parses the XML while collecting counts of parts of the input code, and produces the report. The input has to be in the srcML format, e.g., demo.xml, however it can be compressed or part of an archive, e.g., demo.xml.zip, linux.xml.gz.
The srcfacts.cpp program is one large, main() function using almost no design features. It is extremely fast; for example, it can produce a report on the entire Linux kernel (verison 6.13, 60,392 source-code files, an 4.7 GB file in the srcML format) in under 8 seconds. However, in its current form, it is not easy to debug the XML parsing, add additional program element counts, or even understand what is going on. It includes code for parsing all parts of XML, but it would not be easy to adapt the XML parsing code for another purpose, e.g., another report. The only way to reuse this code is by copy/paste reuse. Overall, it has the following:
| Design Characteristic | Level |
|---|---|
| Scalability | Very High |
| Performance | Very High |
| Portability | Medium |
| Usability | Medium |
| Modularity | Very Low |
| Extensibility | None |
| Reusability | None |
| Maintainability | None |
Your task is to take the srcfacts code, improve the overall design, and extract the XML parsing part of the code. When completed, a set of functions in the files xml_parser.hpp and xml_parser.cpp will handle as much XML parsing as possible, while the main program, srcfacts.cpp, generates the report using the XML parser functions for parsing.
The steps (in order) with the associated (git) tag are:
Tag v1a: Without changing the design, add the code necessary for the report to include the number of return statements. First see how the other counts are collected, then use it as a guide. Do not make any other changes to the project. The heading in the table must be "Returns". Due Jan 28
Tag v1b: Move the refillContent() function into the separate refillContent.hpp and refillContent.cpp files. The build is already configured for these files. Due Jan 31
Tag v1c: Apply the specific design changes from the Coding Practices Task List (an issue in your repository). As you verify that the code follows the practices given, check them off. Once you have completed all of them, close the issue. Due Feb 4
Tag v1d: Extract a set of (free) functions to handle the low-level details of the XML parsing. Extract one function at a time, i.e., each commit can only include a single extracted function. Due TBA
Tag v1e: Add the necessary code to count the number of line comments and literal strings. The headings in the table must be "Line Comments" and "Strings". Due TBA
Your repository for this project is through GitHub Classroom, with a link on the Brightspace course page.
v1a must be completed before Step v1b.g++ and clang. An easy way to verify is with GitHub Codespaces.classes or structs created for this project. You are extracting a set of free functions. Any class you create to solve this problem now would not solve the primary problem of how the srcFacts code uses an XML parser.main() handles XML errors (invalid XML format) with a direct return 1; so the program exits with the status of 1. To exit the program from within any function, use exit(1); which requires the include file stdlib.h.git clone and git checkout.sudo apt-get install libarchive-dev. To install it on macOS, use brew install libarchive. The GitHub Codespaces image already has libarchive-dev installed.General Guidelines