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

/* input/output streams */
FILE *IN;
FILE *OUT;

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

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

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

void
emit_sanitized_string(char *str)
{
    while (*str != '\0') {
        emit_sanitized_char(*str);
        str++;
    }
}

/* read n bytes from IN */
void
read_n_bytes(size_t n, char *buf)
{
    size_t i;
    for (i = 0; i < n; i++)
        buf[i] = fgetc(IN);
    buf[i] = '\0';
}

/* remove occurrences of c from the right side of buf */
void
rstrip(char *buf, char c)
{
    size_t i = strlen(buf) - 1;

    for(;;) {
        if (buf[i] == c)
            buf[i] = '\0';
        else if (i <= 0 || buf[i] != c)
            break;
        i--;
    }
}

/* headings */
int
heading_level(void)
{
    int heading_level = 1;

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

    return heading_level;
}

void
emit_heading(void)
{
    int level = heading_level();

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

    char c;
    while ((c = fgetc(IN)) != '\n')
        emit_sanitized_char(c);

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

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


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

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

bool
internal_implicit_link_p(void)
{
    int count = 0;
    bool ret = true;

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

        count--;
    }

    fseek(IN, count, SEEK_CUR);

    return ret;
}

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

void
get_link(char *out)
{
    char c;

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

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

void
emit_external_link(void)
{
    char link[1024];
    get_link(link);

    char linkname[1024];
    get_linkname(linkname);

    fprintf(OUT, "<a href=\"%s\" rel=\"external\">%s</a>", link, linkname);
}

void
emit_internal_implicit_link(void)
{
    fseek(IN, -1, SEEK_CUR);

    char link[1024];
    char page[1024];
    get_link(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(OUT, "<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(void)
{
    fseek(IN, -1, SEEK_CUR);

    char page[1024];
    get_link(page);

    char linkname[1024];
    get_linkname(linkname);

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

void
emit_link(void)
{
    if (external_link_p())
        emit_external_link();
    else if (internal_implicit_link_p())
        emit_internal_implicit_link();
    else
        emit_internal_explicit_link();
}


/* code blocks */
void
emit_code_block(char firstc)
{
    char nextc = fgetc(IN);
    fseek(IN, -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(OUT, "<pre>");
    else
        fprintf(OUT, "<code>");

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

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

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

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

/* blockquotes */
void
emit_blockquote(void)
{
    fprintf(OUT, "<blockquote>");

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

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

void
emit_image(void)
{
    char buf[80];
    char *bufptr = buf;

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

    *bufptr = '\0';

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

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

void
emit_table(void)
{
    char buf[256] = {0};
    char *bufptr = buf;

    fprintf(OUT, "<table border=\"1\"><tbody>\n");

    for(;;) {
        read_n_bytes(4, buf);
        if (!strcmp("===\n", buf)) {
            fseek(IN, -1L, SEEK_CUR);
            break;
        }
        else {
            fseek(IN, -4L, SEEK_CUR);
            memset(buf, '\0', 256);
            fprintf(OUT, "<tr>\n");
            char c;
            while ((c = fgetc(IN)) != '\n') {
                if (c == '|') {
                    rstrip(buf, ' ');
                    fprintf(OUT, "<td>");
                    emit_sanitized_string(buf);
                    fprintf(OUT, "</td>\n");
                    memset(buf, '\0', 256);
                    bufptr = buf;
                }
                else {
                    *(bufptr++) = c;
                }
            }
            rstrip(buf, ' ');
            fprintf(OUT, "<td>");
            emit_sanitized_string(buf);
            fprintf(OUT, "</td>\n");
            fprintf(OUT, "</tr>\n");
            bufptr = buf;
        }
    }

    fprintf(OUT, "</tbody></table>\n");
}

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

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

    while ((c = fgetc(IN)) != EOF) {
        if (c == '\\') { /* escaped character */
            emit_sanitized_char(fgetc(IN));
        }
        else if (c == '*' && lastc == '\n') { /* heading */
            emit_heading();
        }
        else if (c == '{') { /* link */
            if (!last_was_paragraph && lastc == '\n')
                fprintf(OUT, "\n<p>");
            emit_link();
        }
        else if (c == '$') { /* maybe image */
            char buf[5];
            read_n_bytes(4, buf);

            if (!strcmp(buf, "img ")) {
                emit_image();
            }
            else {
                fseek(IN, -4L, SEEK_CUR);
                emit_sanitized_char(c);
            }
        }
        else if (c == '_') {
            fprintf(OUT, "%s", em_tags[in_em]);
            in_em = !in_em;
        }
        else if (c == '`') { /* code block */
            if (fgetc(IN) == '`')
                emit_code_block(lastc);
            else
                fseek(IN, -1L, SEEK_CUR);
        }
        else if (c == '>' && lastc == '\n') { /* quote */
            emit_blockquote();
        }
        else if (c == '=' && lastc == '\n') { /* maybe table */
            char buf[4];
            read_n_bytes(3, buf);

            if (!strcmp(buf, "==\n")) {
                emit_table();
            }
            else {
                fseek(IN, -3L, SEEK_CUR);
                emit_sanitized_char(c);
            }
        }
        else if (lastc == '\n' && c == '\n') { /* new paragraph */
            fprintf(OUT, "\n<p>");
            last_was_paragraph = true;
            lastc = c;
            continue;
        }
        else if (c == '\n') { /* just a new line, same paragraph */
            fputc(' ', OUT);
        }
        else {
            emit_sanitized_char(c);
        }
        lastc = c;
        last_was_paragraph = false;
    }
}


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 < 2)
        die("No filename provided, exiting.");

    IN = fopen(argv[1], "r");

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

    fclose(IN);
    fclose(OUT);
}