srohtml/srohtml.c

#include <locale.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>

/* macros */
#define BODY_START_SIZE 26624 /* 26 x 1024 byte = 26 KiB */
#define CLS_MAX 8
#define DATE_LEN 10
#define FAVICON "data:image/gif;base64,R0lGODlhEAAQAKEAAAAAcP///wAAcAAAcCH5BAEKAAIALAAAAAAQABAAQAIwlBWZxxwAQWjtTQRvlZhTnGSfYT3eZ3WaE4mjGa1UasoH7c6tzYanfqmhJMKXglcAADs="
#define LANG_LEN 2
#define LINE_MAX 1000
#define NOTES_START 8
#define NOTES_MAX_DIGITS 3

typedef struct {
	char *desc;
	char *href;
} note;

typedef struct {
	size_t index;
	size_t size;
	note *notes;
} references;

typedef struct {
	size_t len;
	size_t size;
	char *str;
} string;

/* function declarations */
static size_t add_note(const char *href, const char *desc);
static void cmd(char *line);
static void cmd_close(void);
static void cmd_a(const char *line);
static void cmd_au_de_ti(char **str, const char *line);
static void cmd_dt(const char *line);
static void cmd_h(const char *line);
static void cmd_hr(void);
static void cmd_la(const char *line);
static void cmd_le(void);
static void cmd_ls(void);
static void cmd_no(const char *line);
static void cmd_p(void);
static void cmd_pd(const char *line);
static void cmd_st(const char *line);
static void cmd_tm(const char *line);
static void cmd_ur(const char *line);
static void footer(void);
static void footnotes(void);
static int fread_line(FILE *file, char **str, size_t *len);
static void indent_str(char **str);
static void page_head(void);
static void *safe_malloc(const size_t size);
static void *safe_realloc(void *ptr, const size_t size);
static size_t stread_param(const char *src, char **str);
static char *string_cat(string *dst, const char *src);
static char *string_cat_html(string *dst, const char *src, const int is_attr_val);
static void string_enlarge(string *old, const size_t size);
static int string_putc(string *dst, const char c);

/* variables */
char *author = NULL;
char closure[CLS_MAX]; 
char *description = NULL;
int indentation = 2;
char language[LANG_LEN + 1] = "";
int licence_container = 0;
int space = 0;
string head;
string body;
char pub_date[DATE_LEN + 1] = "";
references sitography;
char *title = NULL; 

/* function implementations */
size_t
add_note(const char *href, const char *desc)
{
	note *new;

	if (sitography.index == sitography.size) {
		sitography.size *= 2;
		sitography.notes = safe_realloc(sitography.notes, sizeof(note) * sitography.size);
	}

	new = sitography.notes + sitography.index;
	new->href = safe_malloc(sizeof(char) * (strlen(href) + 1));
	strcpy(new->href, href);
	new->desc = safe_malloc(sizeof(char) * (strlen(desc) + 1));
	strcpy(new->desc, desc);
	sitography.index++;

	return sitography.index;
}

void
cmd(char *line)
{
	char *end;
	size_t cmdlen;

	if (!(end = strchr(line, ' ')))
		end = line + strlen(line);
	cmdlen = end - ++line;

	if (!strncmp(line, "A", cmdlen))
		cmd_a(line);
	else if (!strncmp(line, "AU", cmdlen))
		cmd_au_de_ti(&author, line);
	else if (!strncmp(line, "DE", cmdlen))
		cmd_au_de_ti(&description, line);
	else if (!strncmp(line, "DT", cmdlen))
		cmd_dt(line);
	else if (!strncmp(line, "H1", cmdlen))
		cmd_h(line);
	else if (!strncmp(line, "H2", cmdlen))
		cmd_h(line);
	else if (!strncmp(line, "H3", cmdlen))
		cmd_h(line);
	else if (!strncmp(line, "H4", cmdlen))
		cmd_h(line);
	else if (!strncmp(line, "H5", cmdlen))
		cmd_h(line);
	else if (!strncmp(line, "H6", cmdlen))
		cmd_h(line);
	else if (!strncmp(line, "HR", cmdlen))
		cmd_hr();
	else if (!strncmp(line, "LA", cmdlen))
		cmd_la(line);
	else if (!strncmp(line, "LE", cmdlen))
		cmd_le();
	else if (!strncmp(line, "LS", cmdlen))
		cmd_ls();
	else if (!strncmp(line, "NO", cmdlen))
		cmd_no(line);
	else if (!strncmp(line, "P", cmdlen))
		cmd_p();
	else if (!strncmp(line, "PD", cmdlen))
		cmd_pd(line);
	else if (!strncmp(line, "ST", cmdlen))
		cmd_st(line);
	else if (!strncmp(line, "TI", cmdlen))
		cmd_au_de_ti(&title, line);
	else if (!strncmp(line, "TM", cmdlen))
		cmd_tm(line);
	else if (!strncmp(line, "UR", cmdlen))
		cmd_ur(line);
}

void
cmd_a(const char *line)
{
	char *href, *in, *title;

	href = NULL;
	in = NULL;
	title = NULL;

	/* move past "A" */
	line += 1;
	/* move past arguments, once found them */
	line += stread_param(line, &href);
	line += stread_param(line, &in);
	line += stread_param(line, &title);

	if (space)
		string_cat(&body, " ");
	string_cat(&body, "<a class=\"");
	if (!strcmp(href, in))
		string_cat(&body, "url");
	else
		string_cat(&body, "nav");
	string_cat(&body, "\" href=\"");
	string_cat(&body, href);
	string_cat(&body, "\"");
	if (strcmp(title, "")) {
		string_cat(&body, " title=\"");
		string_cat_html(&body, title, 1);
		string_cat(&body, "\"");
	}
	string_cat(&body, ">");
	string_cat_html(&body, in, 1);
	string_cat(&body, "</a>");
	
	free(href);
	free(in);
	free(title);
}

void
cmd_au_de_ti(char **str, const char *line)
{
	/* move past command */
	line += 2;
	/* move past ' ', if present */
	if (line[0] == ' ')
		line++;

	if (*str != NULL)
		free(*str);

	*str = safe_malloc(sizeof(char) * (strlen(line) + 1));
	strcpy(*str, line);
}

void
cmd_close(void)
{
	string_cat(&body, closure);
	string_cat(&body, "\n");
	strcpy(closure, "");
	space = 0;
}

void
cmd_dt(const char *line)
{
	char *datetime, *in;

	datetime = NULL;
	in = NULL;

	/* move past "DT" */
	line += 2;
	/* move past arguments, once found them */
	line += stread_param(line, &datetime);
	line += stread_param(line, &in);

	if (space)
		string_cat(&body, " ");
	string_cat(&body, "<time datetime=\"");
	string_cat_html(&body, datetime, 1);
	string_cat(&body, "\">");
	if (!strcmp(in, ""))
		string_cat_html(&body, datetime, 0);
	else
		string_cat_html(&body, in, 0);
	string_cat(&body, "</time>");

	free(datetime);
	free(in);
}

void
cmd_h(const char *line)
{
	string_cat(&body, "\t\t<h");
	string_putc(&body, line[1]);
	if (line[2] == ' ' && line[3] != '\0') {
		string_cat(&body, " id=\"");
		string_cat_html(&body, line + 3, 1);
		string_cat(&body, "\"");
	}
	string_cat(&body, ">");
	strcpy(closure, "</h");
	closure[3] = line[1];
	closure[4] = '>';
	closure[5] = '\0';
}

void
cmd_hr(void)
{
	string_cat(&body, "\t\t<hr>");
}

void
cmd_la(const char *line)
{
	/* move past "LA" */
	line += 2;
	/* move past ' ', if present */
	if (line[0] == ' ')
		line++;

	if (strlen(line) != LANG_LEN)
		return;

	strncpy(language, line, LANG_LEN);
}

void
cmd_le(void)
{
	char *ind;

	ind = NULL;

	indentation--;
	indent_str(&ind);
	string_cat(&body, ind);
	string_cat(&body, "</div>\n");

	free(ind);
}

void
cmd_ls(void)
{
	char *ind;

	ind = NULL;
	licence_container = 1;

	indent_str(&ind);
	string_cat(&body, ind);
	string_cat(&body, "<div class=\"contenitore-licenza\">\n");
	indentation++;

	free(ind);
	ind = NULL;
	indent_str(&ind);
	string_cat(&body, ind);
	string_cat(&body, "<hr>\n");
	
	free(ind);
}

void
cmd_no(const char *line)
{
	char *desc, *href, *title;
	char index[NOTES_MAX_DIGITS + 1];

	desc = NULL;
	href = NULL;
	title = NULL;

	/* move past "NO" */
	line += 2;
	/* move past arguments, once found them */
	line += stread_param(line, &href);
	line += stread_param(line, &desc);
	line += stread_param(line, &title);

	if (space)
		string_cat(&body, " ");
	sprintf(index, "%u", add_note(href, desc));
	string_cat(&body, "<sup><a id=\"nota-");
	string_cat(&body, index);
	string_cat(&body, "\" class=\"nav\" href=\"#sitografia-");
	string_cat(&body, index);
	string_cat(&body, "\"");
	if (strcmp(title, "")) {
		string_cat(&body, " title=\"");
		string_cat_html(&body, title, 1);
		string_cat(&body, "\"");
	}
	string_cat(&body, ">[");
	string_cat(&body, index);
	string_cat(&body, "]</a></sup>");

	free(desc);
	free(href);
	free(title);
}

void
cmd_p(void)
{
	char *ind;

	ind = NULL;

	indent_str(&ind);
	string_cat(&body, ind);
	string_cat(&body, "<p>");
	strcpy(closure, "</p>");

	free(ind);
}

void
cmd_pd(const char *line)
{
	size_t len;

	/* move past "PD" */
	line += 2;

	len = strlen(line);
	/* move past ' ' after PD */
	if (len && line[0] == ' ') {
		line++;
		len--;
		if (len == DATE_LEN) 
			strncpy(pub_date, line, DATE_LEN);
	}
}

void
cmd_st(const char *line)
{
	/* move past "ST" */
	line += 2;
	/* move past space, if present */
	if (line[0] == ' ')
		line++;

	if (space)
		string_cat(&body, " ");
	string_cat(&body, "<strong>");
	string_cat(&body, line);
	string_cat(&body, "</strong>");
}

void
cmd_tm(const char *line)
{
	char *date, *hours, *in;

	/* move past "TM" */
	line += 2;
	/* move past arguments, once found them */
	line += stread_param(line, &date);
	line += stread_param(line, &hours);
	line += stread_param(line, &in);

	if (space)
		string_cat(&body, " ");
	string_cat(&body, "<time datetime=\"");
	string_cat(&body, date);
	if (strcmp(hours, "")) {
		string_cat(&body, "T");
		string_cat(&body, hours);
	}
	string_cat(&body, "\">");
	string_cat_html(&body, in, 0);
	string_cat(&body, "</time>");

	free(date);
	free(hours);
	free(in);
}

void
cmd_ur(const char *line)
{
	/* move past "UR" */
	line += 2;
	/* move past ' ', if present */
	if (line[0] == ' ')
		line++;

	if (space)
		string_cat(&body, " ");
	string_cat(&body, "&lt;<a class=\"url\" href=\"");
	string_cat_html(&body, line, 1);
	string_cat(&body, "\">");
	string_cat_html(&body, line, 0);
	string_cat(&body, "</a>&gt;");
}

void
footer(void)
{
	size_t max;
	struct tm *now;
	char str[16];
	time_t s;
	int tm_mday, tm_mon, tm_year;
	struct tm tm_pd = { 0 };

	max = 16;

	string_cat(&body, "\t\t<footer>\n");

	if (strcmp(pub_date, "")) {
		string_cat(&body, "\t\t\t<p>Pubblicazione: <time datetime=\"");
		string_cat(&body, pub_date);
		string_cat(&body, "\">");

		sscanf(pub_date, "%4i-%2i-%2i", &tm_year, &tm_mon, &tm_mday);
		tm_pd.tm_mday = tm_mday;
		tm_pd.tm_mon = --tm_mon;
		tm_pd.tm_year = tm_year - 1900;
		strftime(str, max, "%x", &tm_pd);
		string_cat(&body, str);
		string_cat(&body, "</time></p>\n");
	}

	s = time(NULL);
	now = localtime(&s);
	
	string_cat(&body, "\t\t\t<p>Revisione: <time datetime=\"");
	strftime(str, max, "%Y-%m-%d", now);
	string_cat(&body, str);

	strftime(str, max, "%R", now);
	string_cat(&body, "T");
	string_cat(&body, str);

	strftime(str, max, "%z", now);
	string_cat(&body, str);
	string_cat(&body, "\">");

	strftime(str, max, "%x", now);
	string_cat_html(&body, str, 0);

	strftime(str, max, " %R", now);
	string_cat_html(&body, str, 0);
	string_cat(&body, "</time></p>\n");
	
	string_cat(&body, "\t\t</footer>\n");
}

void
footnotes(void)
{
	note *cur;
	size_t i, dig_num, x;
	char id[NOTES_MAX_DIGITS + 1];
	char pad[6 * (NOTES_MAX_DIGITS - 1) + 1];

	if (sitography.index == 0)
		return;

	string_cat(&body, "\t\t<hr>\n");
	string_cat(&body, "\t\t<div class=\"contenitore-sitografia\">\n");
	string_cat(&body, "\t\t\t<h3 id=\"sitografia\">Sitografia</h3>\n");

	sprintf(id, "%u", sitography.index);
	dig_num = strlen(id);

	for (i = 0; i < sitography.index; i++) {
		cur = sitography.notes + i;
		sprintf(id, "%u", i + 1);
		string_cat(&body, "\t\t\t<p id=\"sitografia-");
		string_cat(&body, id);
		string_cat(&body, "\"><span>");
		strcpy(pad, "");
		for (x = strlen(id); x < dig_num; x++)
			strcat(pad, "&nbsp;");
		string_cat(&body, pad);
		string_cat(&body, "<a class=\"nav\" href=\"#nota-");
		string_cat(&body, id);
		string_cat(&body, "\">");
		string_cat(&body, id);
		string_cat(&body, "</a></span> ");
		if (strcmp(cur->desc, "")) {
			string_cat_html(&body, cur->desc, 0);
			string_cat(&body, " ");
		}
		string_cat(&body, "&lt;<a class=\"url\" href=\"");
		string_cat(&body, cur->href);
		string_cat(&body, "\">");
		string_cat_html(&body, cur->href, 0);
		string_cat(&body, "</a>&gt;</p>\n");
		free(cur->desc);
		free(cur->href);
	}
	free(sitography.notes);

	string_cat(&body, "\t\t</div>\n");
}

int
fread_line(FILE *file, char **str, size_t *len)
{
	int c;
	char *end;
	size_t i;

	i = 0;
	while ((c = fgetc(file)) && c != '\n' && c != EOF && i < LINE_MAX)
		i++;
	
	*str = safe_malloc(i + 1);

	fseek(file, -i - 1, SEEK_CUR);
	fread(*str, 1, i, file);
	(*str)[i] = '\0';
	*len = i;
	fseek(file, 1, SEEK_CUR);

	return c == EOF ? EOF : c;
}

void
indent_str(char **str)
{
	int i;

	*str = safe_malloc(sizeof(char) * (indentation + 1));
	for (i = 0; i < indentation; i++)
		(*str)[i] = '\t';
	(*str)[indentation] = '\0';
}

int
main (int argc, char *argv[])
{
	FILE *in, *out;
	size_t len;
	char *line;

	if (argc == 2 && (!strcmp(argv[1], "-v") || !strcmp(argv[1], "--version"))) {
		fprintf(stdout, BIN " " VERSION "\n");
		fprintf(stdout, "Copyright (C) 2024 Matteo Bini\n");
		fprintf(stdout, "License: GPLv3+ <http://www.gnu.org/licenses/gpl.html>.\n");
		fprintf(stdout, "This is free software: you are free to change and redistribute it.\n");
		fprintf(stdout, "There is NO WARRANTY, to the extent permitted by law.\n");
		fprintf(stdout, "\n");
		fprintf(stdout, "Written by Matteo Bini.\n");
		return 1;
	}

	setlocale(LC_ALL, "");

	in = fopen(argv[1], "r");
	if (in == NULL) {
		fprintf(stderr, BIN ": Can't open %s.\n", argv[1]);
		return 1;
	}

	head.size = BODY_START_SIZE / 4;
	head.str = safe_malloc(head.size);
	strcpy(head.str, "");
	head.len = 0;

	body.size = BODY_START_SIZE;
	body.str = safe_malloc(body.size);
	strcpy(body.str, "");
	body.len = 0;

	string_cat(&body, "\t<body>\n");

	sitography.index = 0;
	sitography.size = NOTES_START;
	sitography.notes = malloc(sizeof(note) * sitography.size);

	while (fread_line(in, &line, &len) != EOF) {
		if (!len)
			cmd_close();
		else if (len >= 2 && line[0] == '.')
			cmd(line);
		else if (len >= 2 && line[0] == '\\' && line[1] == '"')
			continue;
		else {
			if (space &&
				line[0] != '!' &&
				line[0] != ',' &&
				line[0] != '.' &&
				line[0] != '.' &&
				line[0] != ';' &&
				line[0] != '?')
				string_cat(&body, " ");
			string_cat_html(&body, line, 0);
			space = 1;
		}
		free(line);
	}
	free(line);
	fclose(in);

	if (strcmp(closure, ""))
		cmd_close();

	page_head();
	footnotes();
	footer();

	string_cat(&body, "\t</body>\n</html>\n");

	if (argc < 3)
		out = stdout;
	else
		out = fopen(argv[2], "w");
	if (out == NULL) {
		fprintf(stderr, BIN ": Can't open %s.\n", argv[2]);
		return 1;
	}

	fputs(head.str, out);
	fputs(body.str, out);
	fclose(out);

	return 0;
}

void
page_head(void)
{
	string_cat(&head, "<!doctype html>\n");

	string_cat(&head, "<html lang=\"");
	string_cat_html(&head, language, 1);
	string_cat(&head, "\">\n");

	string_cat(&head, "\t<head>\n");

	string_cat(&head, "\t\t<link rel=\"icon\" href=\"" FAVICON "\">\n");

	string_cat(&head, "\t\t<meta charset=\"utf-8\">\n");

	string_cat(&head, "\t\t<meta name=\"author\" content=\"");
	string_cat_html(&head, author, 1);
	string_cat(&head, "\">\n");

	string_cat(&head, "\t\t<meta name=\"description\" content=\"");
	string_cat_html(&head, description, 1);
	string_cat(&head, "\">\n");

	string_cat(&head, "\t\t<meta name=\"viewport\" content=\"width=device-width,initial-scale=1\">\n");

	string_cat(&head, "\t\t<style>\n" \
"			body {\n" \
"				margin: .5em auto;\n" \
"				max-width: 50em;\n" \
"				width: 90%;\n" \
"			}\n" \
"\n" \
"			p {\n" \
"				line-height: 1.5;\n" \
"			}\n" \
"\n" \
"			a.url {\n" \
"				hyphens: manual;\n" \
"				word-wrap: break-word;\n" \
"			}\n"
	);

	if (licence_container) {
		string_cat(&head, "\n" \
"			.contenitore-licenza {\n" \
"				font-size: .8em;\n" \
"				text-align: justify;\n" \
"			}\n"
		);
	}

	if (sitography.index > 0) {
		string_cat(&head, "\n" \
"			sup {\n" \
"				font-size: .7em;\n" \
"			}\n" \
"\n" \
"			.contenitore-sitografia p {\n" \
"				margin: .5em 0;\n" \
"				text-align: left;\n" \
"			}\n" \
"\n" \
"			.contenitore-sitografia p span {\n" \
"				font-family: monospace;\n" \
"			}\n"
		);
	}
	
	string_cat(&head, "\n" \
"			footer {\n" \
"				margin: 2em auto 1em;\n" \
"			}\n" \
"\n" \
"			footer p {\n" \
"				margin: 0 auto;\n" \
"				text-align: center;\n" \
"			}\n" \
"\n" \
"			@media all and (min-width: 50em) {\n" \
"				p {\n" \
"					hyphens: auto;\n" \
"					text-align: justify;\n" \
"				}\n" \
"			}\n" \
"\n" \
"			@media print {\n" \
"				body {\n" \
"					max-width: 90%;\n" \
"				}\n" \
"\n" \
"				a.nav {\n" \
"					color: inherit;\n" \
"					text-decoration: none;\n" \
"				}\n" \
"			}\n"
	);

	string_cat(&head, "\t\t</style>\n");

	string_cat(&head, "\t\t<title>");
	string_cat_html(&head, title, 0);
	string_cat(&head, "</title>\n");

	string_cat(&head, "\t</head>\n");

	free(author);
	free(description);
	free(title);
}

void *
safe_malloc(const size_t size)
{
	void *ptr; 

	ptr = malloc(size);
	if (ptr == NULL) {
		fprintf(stderr, BIN ": Can't allocate %lu bytes.\n", size);
		exit(1);
	}

	return ptr;
}

void *
safe_realloc(void *ptr, const size_t size)
{
	void *new; 

	new = realloc(ptr, size);
	if (new == NULL) {
		fprintf(stderr, BIN ": Can't allocate %lu bytes.\n", size);
		exit(1);
	}

	return new;
}

size_t
stread_param(const char *src, char **str)
{
	char c;
	char *end;
	size_t i, len;
	int p;

	c = ' ';
	p = 0;
	if (src[0] == ' ') {
		src++;
		p++;
	}
	if (src[0] == '"' || src[0] == '\'') {
		c = src[0];
		src++;
		p += 2;
	}
	end = strchr(src, c);
	if (end)
		len = end - src;
	else
		len = strlen(src);

	*str = safe_malloc(len + 1);
	for (i = 0; i <= len; i++)
		(*str)[i] = '\0';
	strncpy(*str, src, len);

	return len + p;
}

char *
string_cat(string *dst, const char *src)
{
	size_t len;

	len = strlen(src);

	if (len >= (dst->size - dst->len))
		string_enlarge(dst, (dst->size + len) * 2 + 1);

	strcat(dst->str, src);
	dst->len = dst->len + len;

	return dst->str;
}

char *
string_cat_html(string *dst, const char *src, const int is_attr_val)
{
	size_t i, len;

	len = strlen(src);

	for (i = 0; i < len; i++) {
		if (is_attr_val && src[i] == '"')
			string_cat(dst, "&quot;");
		else if (src[i] == '&')
			string_cat(dst, "&amp;");
		else if (src[i] == '<')
			string_cat(dst, "&lt;");
		else if (src[i] == '>')
			string_cat(dst, "&gt;");
		else
			string_putc(dst, src[i]);
	}

	return dst->str;
}

void
string_enlarge(string *old, const size_t size)
{
	old->str = safe_realloc(old->str, sizeof(char) * size);
	old->size = size;
}

int
string_putc(string *dst, const char c)
{
	if ((dst->size - dst->len) <= 1)
		string_enlarge(dst, (dst->size + 1) * 2 + 1);

	dst->str[dst->len] = c;
	dst->str[dst->len + 1] = '\0';
	dst->len++;

	return (int) c;
}