Back to main

---

Lab Report 5 - 3/8/2022

In this lab report, I am going to demonstrate the difference between our group’s Markdown Parse code and the example’s by using the commonmark-spec tests. I am specifically including two tests where the two implementations return different results to talk about which of the two returns the correct one.

Our group’s implementation is here: https://github.com/Alexander-Kourjanski/markdown-parse

The example implementation is here: https://github.com/Alexander-Kourjanski/markdown-parse

Here is the table of content:

• Find the diff-erence
• Test 1
  • The Difference
  • How to Fix The Bug
• Test 2
  • The Difference
  • How to Fix The Bug

Find the diff-erence

I found the tests with difference result by comparing the outputs of them:

First, I used a simple shell script that loops through all the test files and call the markdown parse on each of the file:

for file in test-files/*.md;
do
    echo $file
    java MarkdownParse $file
done

That script is stored in the script.sh file.

Then, I used the > symbol to redirect the output of each markdown parse to he corresponding results.txt file with this command:

bash script.sh > results.txt

I ran the script on each of the implementation repositories.

And lastly I used the diff command to show the difference between the outputs:

(There are more differences than just those two…)

Test 1

The diff command shows that there is a difference on line 230 of the results.txt files. That correspond to test file 201 from the commonmark-spec tests, which is:

[foo]: <bar>(baz)

[foo]

The Difference

In the output of the diff command, I found:

The first line represents the line number of difference - 230.

The second and the fourth line represent the output from our implementation and the example implementation respectively, which, as you see, are different.

How to Fix The Bug

According the the preview of VS Code, the output should have been empty ([]) as there is no link present in the file. Thus, our group’s implementation is correct.

The bug in the example code is that it did not verify whether the open parenthesis is next to the close square bracket or not. Since the link is only valid if they are close to each other, there should be a verification of the relative position of the open parenthesis to the close square bracket after these lines of code:

Snippet of code taken from the MarkdownParse.java provided in the example implementation.

Test 2

The diff command shows that there is a difference on line 270 of the results.txt files. That correspond to test file 22 from the commonmark-spec tests, which is:

[foo](/bar\* "ti\*tle")

The Difference

In the output of the diff command, I found:

The first line represents the line number of difference - 270.

The second and the fourth line represent the output from our implementation and the example implementation respectively, which, again, are different.

How to Fix The Bug

Here, according to the preview of VS Code, the expected output should have been ["bar*"] since the link on the text foo points to the address bar*. None of our implementations correctly identified that case.

For our implementation, the bug is that we did not check whether quotation marks exist inside the open parenthesis or not - apparently, when a link is in the format:

[linkname](text1 "text2")

linkname will be a link that only points to text1.

To address that case, we should add code after

Snippet of code taken from our impplementation of MarkdownParse.java.

to check whether a pair of quotation marks exist inside the pair of parenthesis, along with a space that separate the quoted text from the original text. For example, we can use markdown.indexOf(" \"", openParen) to find the position of
the next quotation mark (if exists) and see if it is between the parenthesis, etc.