Getting rid of all comments in C

Question

Right now this code doesn't remove inline comments, how do I change it so it also removes inline comments?

FILE *output;
output = fopen("preprocess_output.c", "w");

while (fgets(line, LINE_LENGTH, file) != NULL)
{
    for (int i = 0; i < strlen(line); i++)
    {
        if (line[i]  ==  '/' && line[i + 1]  ==  '/')
        {
            comment_lines++;
        }
        else
        {
            fprintf(output, line);
        }
        if (line[i] != '\n' && line[i] != '\t')
        {
            non_blank++;
            break;
        }   
    }
}

Which kind of inline comments would you want removed? Those starting and ending on the same line? — Yunnosch, Mar 03 '19 at 10:49
Think about how you, yourself, identify comments and teach it to the computer. — machine_1, Mar 03 '19 at 10:49
Look at some piece of code with comments you want removed. Underline them with a pen. Try to realise how you recognised them. Describe it - to yourself then to us in your question. That would be one step towards "demonstrating own research effort", which you are supposed to do. — Yunnosch, Mar 03 '19 at 10:50
so right now im writing to a new file and im supposed to remove all comments in that new file, but it doesn't remove the comments that are on the same line as other code — S.Gammon, Mar 03 '19 at 10:50
Please give examples. I think we are actually thinking of different thinkgs when reading "inline comments". Please show them. Define them. Show the desired result. — Yunnosch, Mar 03 '19 at 10:51
You should loop on chars, not on lines. Just check if "//" then skip until '\n', if "/*" then skip until "*/" — caxapexac, Mar 03 '19 at 10:52
To phrase it differently: Please answer the following question in a way that it cannot be misunderstood (and I will intentionally try...): What is an inline comment? — Yunnosch, Mar 03 '19 at 10:52
int add(int a, int b) { return a + b; // An inline comment. <------ it does not remove this comment } — S.Gammon, Mar 03 '19 at 10:54
What is wrong about this approach "Look at the first two letters in a line. If they are not // print the line. If the line is not empty go to next line." ? Does anything about that sound inappropriate? — Yunnosch, Mar 03 '19 at 10:56
i'm just not very good at C, started it a couple months ago so not everything makes sense to me — S.Gammon, Mar 03 '19 at 10:57
In that case: https://ericlippert.com/2014/03/21/find-a-simpler-problem/ or https://ericlippert.com/2014/03/05/how-to-debug-small-programs/ — Yunnosch, Mar 03 '19 at 10:58
You need to truncate the line at the point where you find the // by placing a `'/0'` there. This raises another thing: `fgets` leaves a newline at the end of each line which must be removed. — Weather Vane, Mar 03 '19 at 10:59
Read my question about the approach. It is English, not C. Do you spot anything weird about that? Look at a line with an inline comment, read my English approach. What is wrong? — Yunnosch, Mar 03 '19 at 10:59
the thing is i know what i done wrong as I'm checking if a line begins with / followed by a / then it removes the line, im just not sure how to change it so it also removes comments which don't start at the beginning of a line — S.Gammon, Mar 03 '19 at 11:03
Cleanest approach is to make a StateMachine (FSA/DFA). It only needs to recognise `*` `/`and`\n`. Maybe later also `"`,`'`and backslash. Maintaining state across lines would allow you to recognise multiline `/*... */` comments as well. — wildplasser, Mar 03 '19 at 12:22
See comments in https://stackoverflow.com/questions/50500652/c-program-to-remove-comments-from-file which provide some idea as to how complex it is to properly remove comments. — Richard Chambers, Mar 03 '19 at 13:19
The preprocessor removes comments, among other things. Not sure why you want to remove comments, but you could generate the intermediate source file without comments by using the `-E` (or `/E`) option on your compiler. — cdarke, Mar 03 '19 at 14:14
@S.Gammon: you can accept one of the answers by clicking on the grey checkmark below its score. — chqrlie, Apr 13 '19 at 12:03

chqrlie · Answer 1 · 2019-03-07T07:25:30.473

Here is a small program that strips C comments in almost all cases.

/* strip C comments by chqrlie */

#include <errno.h>
#include <stdio.h>
#include <string.h>

/* read the next byte from the C source file, handing escaped newlines */
int getcpp(FILE *fp, int *lineno_p) {
    int ch;
    while ((ch = getc(fp)) == '\\') {
        if ((ch = getc(fp)) != '\n') {
            ungetc(ch, fp);
            return '\\';
        }
        *lineno_p += 1;
    }
    if (ch == '\n')
        *lineno_p += 1;
    return ch;
}

int main(int argc, char *argv[]) {
    FILE *fp = stdin, *ft = stdout;
    const char *filename = "<stdin>";
    int ch, lineno;

    if (argc > 1) {
        if ((fp = fopen(filename = argv[1], "r")) == NULL) {
            fprintf(stderr, "Cannot open input file %s: %s\n",
                    filename, strerror(errno));
            return 1;
        }
    }
    if (argc > 2) {
        if ((ft = fopen(argv[2], "w")) == NULL) {
            fprintf(stderr, "Cannot open output file %s: %s\n",
                    argv[2], strerror(errno));
            return 1;
        }
    }
    lineno = 1;
    while ((ch = getcpp(fp, &lineno)) != EOF) {
        int startline = lineno;
        if (ch == '/') {
            if ((ch = getcpp(fp, &lineno)) == '/') {
                /* single-line comment */
                while ((ch = getcpp(fp, &lineno)) != EOF && ch != '\n')
                    continue;
                if (ch == EOF) {
                    fprintf(stderr, "%s:%d: unterminated single line comment\n",
                            filename, startline);
                    break;
                }
                putc('\n', ft);  /* replace comment with newline */
                continue;
            }
            if (ch == '*') {
                /* multi-line comment */
                int lastc = 0;
                while ((ch = getcpp(fp, &lineno)) != EOF) {
                    if (ch == '/' && lastc == '*') {
                        break;
                    }
                    lastc = ch;
                }
                if (ch == EOF) {
                    fprintf(stderr, "%s:%d: unterminated comment\n",
                            filename, startline);
                    break;
                }
                putc(' ', ft);  /* replace comment with single space */
                continue;
            }
            putc('/', ft);
            /* keep parsing to handle n/"a//"[i] */
        }
        if (ch == '\'' || ch == '"') {
            int sep = ch;
            const char *const_type = (ch == '"') ? "string" : "character";

            putc(sep, ft);
            while ((ch = getcpp(fp, &lineno)) != EOF) {
                putc(ch, ft);
                if (ch == sep)
                    break;;
                if (ch == '\\') {
                    if ((ch = getcpp(fp, &lineno)) == EOF)
                        break;
                    putc(ch, ft);
                }
                if (ch == '\n') {
                    fprintf(stderr, "%s:%d: unescaped newline in %s constant\n",
                            filename, lineno - 1, const_type);
                    /* This is a syntax error but keep going as if constant was terminated */
                    break;
                }
            }
            if (ch == EOF) {
                fprintf(stderr, "%s:%d: unterminated %s constant\n",
                        filename, startline, const_type);
                break;
            }
            continue;
        }
        putc(ch, ft);
    }
    if (fp != stdin)
        fclose(fp);
    if (ft != stdout)
        fclose(ft);
    return 0;
}

Since you get a full answer for free, try and learn how the above code handles strings and escaped newlines. There are still some corner cases that are not supported, can you find them?

one such corner case is the code does not parse trigraphs, an obsolescent feature that may be used to hide \ characters.

eyalm · Answer 2 · 2019-03-03T13:14:17.310

-1

in the following solution, there is a single pass over the line. If a comment was found (//), we terminate and print it. supporting (/* */) requires more work.

while (fgets(line, LINE_LENGTH, file) != NULL)
{
    size_t len = strlen(line);
    size_t i;

    for (i=0; i<len; i++)
    {
        if (line[i]=='/' && line[i + 1]=='/')
        {
            line[i] = '\0';
            break;
        }
    }
    fprintf(output, "%s", line);
}

note to two points in addition to the logic:

when printing using printf, always use a format string. If the line contains % it might do unexpected things.
do not put strlen in the condition of a loop. It generates a lot of unnecessary loops to calculate the length.

edited Mar 03 '19 at 13:14

answered Mar 03 '19 at 11:13

eyalm

3,366
19
21

i appreciate your help but this code that you suggested removes the whole line, along with the code that's on the same line – S.Gammon Mar 03 '19 at 11:21
1

`*line = '\0';` ==> `line[i] = '\0';`. This also removes any trailing newline so you should remove the trailing newline from *every* line anyway. – Weather Vane Mar 03 '19 at 11:22
I'm afraid your code is incorrect: you must replace the `//` with `\n\0` for the line to have a newline and avoid concatenation with the next line. – chqrlie Mar 03 '19 at 11:25
It is also advisable to use type `size_t` for `i` and `len`. – chqrlie Mar 03 '19 at 11:27
1

Finally, how do you handle overlong lines that are read in chunks by `fgets()`? The OP does not handle these either, but you should at least hint at the potential problem. It is easier to handle this problem with a loop that reads one byte at a time. – chqrlie Mar 03 '19 at 11:28

Getting rid of all comments in C

2 Answers2