codex/format.c

#include <ctype.h>
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include <stdbool.h>
#include <string.h>

/* util */
void
die(char *msg)
{
    fprintf(stderr, "%s\n", msg);
    exit(1);
}

/* header/footer */
void
write_file(FILE *output, char *input)
{
    FILE *f = fopen(input, "r");
    char c;
    while ((c = fgetc(f)) != EOF)
        fputc(c, output);
    fclose(f);
}

/* remove dangerous characters */
void
emit_sanitized_char(char c, FILE *output)
{
    switch (c) {
    case '<':
        fprintf(output, "&lt;");
        break;
    case '>':
        fprintf(output, "&gt;");
        break;
    case '&':
        fprintf(output, "&amp;");
        break;
    case '\'':
        fprintf(output, "&#39;");
        break;
    case '\"':
        fprintf(output, "&quot;");
        break;
    default:
        fputc(c, output);
    }
}

/* headings */
int
heading_level(FILE *input)
{
    int heading_level = 1;

    while (fgetc(input) == '*')
        heading_level++;

    return heading_level;
}

void
emit_heading(FILE *input, FILE *output)
{
    int level = heading_level(input);

    fprintf(output, "<h%d>", level);

    char c;
    while ((c = fgetc(input)) != '\n')
        emit_sanitized_char(c, output);

    fprintf(output, "</h%d>", level);

    /* We want the newline we used to break the loop to be picked up by the
     * formatter again. */
    fseek(input, -1, SEEK_CUR);
}


/* links */
bool
external_link_p(FILE *input)
{
    char buf[5] = {0};
    fgets(buf, 5, input);
    fseek(input, -4, SEEK_CUR);

    return !strcmp(buf, "http");
}

bool
internal_implicit_link_p(FILE *input)
{
    int count = 0;
    bool ret = true;

    while (1) {
        char c = fgetc(input);
        if (c == EOF) {
            die("End of file encountered while parsing link.");
        }
        else if (c == '}') {
            break;
        }
        else if (c == ' ') {
            ret = false;
            break;
        }

        count--;
    }

    fseek(input, count, SEEK_CUR);

    return ret;
}

void
get_linkname(FILE *input, char *out)
{
    char c;
    while ((c = fgetc(input)) != '}') {
        if (c == EOF)
            die("End of file encountered while parsing link name.");
        else if (c == '\n')
            c = ' ';
        *(out++) = c;
    }
    *out = '\0';
}

void
get_link(FILE *input, char *out)
{
    char c;

    /* some links add the initial { for some reason */
    if (fgetc(input) != '{')
        fseek(input, -1L, SEEK_CUR);

    while (((c = fgetc(input)) != ' ') && c != '\n' && c != '}') {
        if (c == EOF)
            die("End of file encountered while parsing link.");
        *(out)++ = c;
    }
    *out = '\0';
}

void
emit_external_link(FILE *input, FILE *output)
{
    char link[1024];
    get_link(input, link);

    char linkname[1024];
    get_linkname(input, linkname);

    fprintf(output, "<a href=\"%s\">[%s]</a>", link, linkname);
}

void
emit_internal_implicit_link(FILE *input, FILE *output)
{
    fseek(input, -1, SEEK_CUR);

    char link[1024];
    char page[1024];
    get_link(input, link);

    memcpy(page, link, 1024);

    for (size_t i = 0; link[i] != '\0'; i++)
        if (link[i] >= 'A' && link[i] <= 'Z') link[i] = link[i] + 32;

    for (size_t i = 0; page[i] != '\0'; i++)
        if (page[i] == '-') page[i] = ' ';

    fprintf(output, "<a href=\"%s.html\">{%s}</a>", link, page);
}

bool
explicit_link_extension(char *link)
{
    while (*link != '\0') {
        if (*link++ == '.')
            return true;
    }

    return false;
}

void
emit_internal_explicit_link(FILE *input, FILE *output)
{
    fseek(input, -1, SEEK_CUR);

    char page[1024];
    get_link(input, page);

    char linkname[1024];
    get_linkname(input, linkname);

    if (explicit_link_extension(page))
        fprintf(output, "<a href=\"%s\">{%s}</a>", page, linkname);
    else
        fprintf(output, "<a href=\"%s.html\">{%s}</a>", page, linkname);
}

void
emit_link(FILE *input, FILE *output)
{
    if (external_link_p(input))
        emit_external_link(input, output);
    else if (internal_implicit_link_p(input))
        emit_internal_implicit_link(input, output);
    else
        emit_internal_explicit_link(input, output);
}


/* code blocks */
void
emit_code_block(char firstc, FILE *input, FILE *output)
{
    char nextc = fgetc(input);
    fseek(input, -1L, SEEK_CUR);

    bool pre = firstc == '\n' && nextc =='\n';

    /* if this code block is on a separate line, create a <pre>. otherwise, it
     * is inlined in a paragraph and we want a <code>. */
    if (pre)
        fprintf(output, "<pre>");
    else
        fprintf(output, "<code>");

    for (;;) {
        char c = fgetc(input);

        if (c == '`') {
            if (fgetc(input) == '`')
                break;
            else
                fseek(input, -1L, SEEK_CUR);
        }

        if (c == EOF)
            die("Unexpected EOF while converting code block.");
        emit_sanitized_char(c, output);
    }

    if (pre)
        fprintf(output, "</pre>");
    else
        fprintf(output, "</code>");
}

/* blockquotes */
void
emit_blockquote(FILE *input, FILE *output)
{
    fprintf(output, "<blockquote>");

    char c = fgetc(input), lastc = 'a';
    for(;;) {
        if (lastc == '\n' && c != '>')
            break;
        else if (lastc == '>' && c == ' ')
            ;
        else if (lastc == '\n' && c == '>')
            fputc(' ', output);
        else if (c != '\n')
            emit_sanitized_char(c, output);
        lastc = c;
        c = fgetc(input);
    }
    fseek(input, -1L, SEEK_CUR);

    fprintf(output, "</blockquote>\n");
}

void
emit_image(FILE *input, FILE *output)
{
    char buf[80];
    char *bufptr = buf;

    for(;;) {
        char c = fgetc(input);
        if (isspace(c))
            break;
        else if (c == EOF)
            die("EOF encountered while parsing image filename.");
        else
            *(bufptr++) = c;
    }

    *bufptr = '\0';

    fprintf(output, "<img src=\"../resources/img/%s\" />", buf);

    fseek(input, -1L, SEEK_CUR);
}

/* main formatter */
void
format(FILE *input, FILE *output)
{
    char c, lastc = '\n';

    char *em_tags[] = {"<em>", "</em>"};
    int in_em = 0; /* boolean used as index, therefore it's an int */

    while ((c = fgetc(input)) != EOF) {
        if (c == '\\') { /* escaped character */
            emit_sanitized_char(fgetc(input), output);
        }
        else if (c == '*' && lastc == '\n') { /* heading */
            emit_heading(input, output);
        }
        else if (c == '{') { /* link */
            if (lastc == '\n')
                fprintf(output, "<p>");
            emit_link(input, output);
        }
        else if (c == '$') { /* maybe image */
            char buf[5];
            size_t i;
            for (i = 0; i < 4; i++)
                buf[i] = fgetc(input);
            buf[i] = '\0';

            if (!strcmp(buf, "img ")) {
                emit_image(input, output);
            }
            else {
                fseek(input, -4L, SEEK_CUR);
                emit_sanitized_char(c, output);
            }
        }
        else if (c == '_') {
            fprintf(output, "%s", em_tags[in_em]);
            in_em = !in_em;
        }
        else if (c == '`') { /* code block */
            if (fgetc(input) == '`')
                emit_code_block(lastc, input, output);
            else
                fseek(input, -1L, SEEK_CUR);
        }
        else if (c == '>' && lastc == '\n') { /* quote */
            emit_blockquote(input, output);
        }
        else if (lastc == '\n' && c == '\n') { /* new paragraph */
            fputc('\n', output);
            fprintf(output, "<p>");
        }
        else if (c == '\n') { /* just a new line, same paragraph */
            fputc(' ', output);
        }
        else {
            emit_sanitized_char(c, output);
        }
        lastc = c;
    }
}


void
basename(char *in, char *out, char delim)
{
    for (size_t i = 0; in[i] != delim; i++)
        out[i] = in[i];
}

/* entrypoint */
int
main(int argc, char *argv[])
{
    if (!argc)
        die("No filename provided, exiting.");

    FILE *in = fopen(argv[1], "r");

    FILE *out;
    if (argc == 2) {
        char bn[80] = {0};
        basename(argv[1], bn, '.');

        char out_name[85] = {0};
        sprintf(out_name, "%s.html", bn);

        out = fopen(out_name, "w");
    }
    else {
        out = fopen(argv[2], "w");
    }

    write_file(out, "resources/header.html");
    format(in, out);
    write_file(out, "resources/footer.html");

    fclose(in);
    fclose(out);
}