count_changes.c

#include <stdio.h>
#include <stdlib.h>
#include <memory.h>

#define MAXSIZE 128

typedef struct {
    long files_changed;
    long insertions;
    long deletions;
    long changes;
} changes;

static long upper_bound = 2500;

void show_usage();
void process_text();
changes process_line( char *line );


int main( int argc, const char **argv ) {
    if (argc > 1)
        show_usage();
    else
        process_text();
    return 0;
}

void show_usage() {
    printf("Reads input on stdin; displays output on stdout.\n");
    printf("Input expected to be the changes summary from a git log, e.g.: 23 files changed, 1312 insertions(+), 4 deletions(-)\n");
}

void process_text() {
    char line[MAXSIZE];
    changes line_changes;
    changes all_changes;
    all_changes.files_changed = all_changes.insertions = all_changes.deletions = all_changes.changes = 0;
    long lined_file_changes = 0;
    long maximum_change = 0;
    line[0] = 0;
    while (fgets(line, MAXSIZE, stdin)) {
        if (strstr(line, "file changed,") || strstr(line, "files changed,")) {  // just in case we get extraneous inputs
            line_changes = process_line(line);
            if(line_changes.changes > upper_bound) {
                printf("Skipping commit with %ld lines changed.\n", line_changes.changes);
            } else {
                all_changes.files_changed += line_changes.files_changed;
                all_changes.insertions += line_changes.insertions;
                all_changes.deletions += line_changes.deletions;
                all_changes.changes += line_changes.changes;
                maximum_change = (maximum_change > line_changes.changes) ? maximum_change : line_changes.changes;
                if (line_changes.changes > 0)
                    lined_file_changes += line_changes.files_changed;
            }
        }
    }
    printf("%ld files changed; %ld lines changed.\n", all_changes.files_changed, all_changes.changes);

    if (all_changes.files_changed > 0 && lined_file_changes > 0) {
        printf("Average commit:  %d lines (across all commits), %d lines (across commits with changed lines).\n", (int)
                (all_changes.changes / all_changes.files_changed), lined_file_changes == 0 ? 0 : (int)(all_changes.changes /
                                                                                                       lined_file_changes));
    } else {
        printf("Cannot compute average commit.\n");
    }
    printf("Maximum-sized commit: %ld lines.\n", maximum_change);
}

changes process_line( char *line ) {
    changes result;
    result.files_changed = result.insertions = result.deletions = result.changes = 0;
    char *substring = strtok(line, ",");        // first substring shows files changed
    char *terminator = strchr(substring, 'f');
    result.files_changed = strtol(substring, &terminator, 10);
    substring = strtok(NULL, ",");              // second substring either shows lines inserted or lines deleted
    if ((terminator = strchr(substring, 'd'))) {  // need to check if this is deletion first, because "deletions" has an 'i' in it
        result.deletions = strtol(substring, &terminator, 10);
    } else if ((terminator = strchr(substring, 'i'))) {
        result.insertions = strtol(substring, &terminator, 10);
        substring = strtok(NULL, ",");
        if ((substring != NULL) && (terminator = strchr(substring, 'd'))) {  // third substring, if it exists, shows lines deleted
            result.deletions = strtol(substring, &terminator, 10);
        }
    } else {
        printf("Surprising substring: %s\n", substring);
    }
    result.changes = result.insertions > result.deletions ? result.insertions : result.deletions;

    return result;
}