nihdoc

nihdoc.c (5700B) [raw]

---

 /*
  * nihdoc.c -- Not Invented Here Doc
  *             -   -        -    ---
  *
  *              ~akarle, MIT License
  *
  * "because markdown isn't in base!"
  */
 #include <stdio.h>
 #ifndef __PLAN_9
 #include <unistd.h>
 #endif
 #include <stdbool.h>
 
 /* Global Constants and Enums */
 char *FMT_STRS[] = {
 	['_'] = "em",
 	['*'] = "strong",
 	['`'] = "code",
 };
 
 enum Block {
 	NONE,
 	HEADER,
 	HEADER_PARSE,
 	PARAGRAPH,
 	CODE,
 	LIST,
 	LIST_PARSE,
 };
 
 enum Link {
 	NOL,
 	URL_PARSE,
 	DESC_PARSE,
 	OPT_URL,
 };
 
 
 /* Start Global State */
 enum Block in = NONE;
 enum Link in_link = NOL;
 int hlvl = 0;
 bool fmts[256] = {false}; /* indexed by _ * ` */
 bool escape = false;
 char lnkdes[2048] = {0};
 int lnkidx = 0;
 int indent = 0;
 int previndent = 0;
 int listdepth = 0;
 int lastc = '0';
 bool ol = false;
 bool linestarted = false;
 bool blockquote = false;
 
 
 /* Helper functions */
 void
 putesc(int c)
 {
 	switch (c) {
 	case '<':
 		printf("&lt;");
 		break;
 	case '>':
 		printf("&gt;");
 		break;
 	case '&':
 		printf("&amp;");
 		break;
 	default:
 		putchar(c);
 	}
 }
 
 void
 newlist(void)
 {
 	in = LIST;
 	previndent = indent;
 	printf("<%s>\n<li>\n", ol ? "ol" : "ul");
 	listdepth++;
 }
 
 int
 endlist(void)
 {
 	in = LIST;
 	previndent = indent;
 	printf("</li>\n</%s>\n", ol ? "ol" : "ul");
 	return --listdepth;
 }
 
 /*
  * All inline types should start the paragraph if no other major type present.
  * This function does just that, IFF we aren't already in a major type.
  */
 void
 maybe_startp(void)
 {
 	if (in == NONE) {
 		in = PARAGRAPH;
 		printf("<p>\n");
 	}
 }
 
 void
 handle_lf(void)
 {
 	indent = 0;
 	linestarted = false;
 
 	/* single line types (one lf to close) */
 	if (in == HEADER) {
 		in = NONE;
 		printf("</h%d>\n", hlvl);
 	}
 
 	/* multi-line types (two lf to close) */
 	if (lastc == '\n' || (lastc == '>' && blockquote)) {
 		switch (in) {
 		case PARAGRAPH:
 			printf("</p>\n");
 			break;
 		case CODE:
 			printf("</code></pre>\n");
 			break;
 		case LIST:
 			previndent = 0;
 			while (endlist())
 				;
 			break;
 		default:
 			break; /* no op */
 		}
 		in = NONE;
 		if (blockquote && lastc == '\n') {
 			printf("</blockquote>\n");
 			blockquote = false;
 		}
 	}
 }
 
 /* Returns whether inline styles should be allowed at this moment. */
 bool
 fmt_disabled(int c)
 {
 	/* `` blocks all but the next `, likewise CODE makes all disabled */
 	if (in == CODE || in_link == URL_PARSE)
 		return true;
 	else
 		return fmts['`'] && c != '`';
 }
 
 void
 toggle_format(int c)
 {
 	if (!fmt_disabled(c)) {
 		maybe_startp();
 		printf("<%s%s>", fmts[c] ? "/" : "", FMT_STRS[c]);
 		fmts[c] ^= true;
 	} else
 		putesc(c);
 }
 
 
 /*
  * The main state machine [1], abstracted into a function to allow
  * playback support for buffered types (i.e. link descriptions)
  *
  * [1]: home grown spaghetti code
  */
 void
 handlec(int c)
 {
 	/*
 	 * Store link descriptions as we go, skipping the regular loop since
 	 * we'll play them back later via recursion on handlec
 	 */
 	if (in_link == DESC_PARSE && c != ']') {
 		lnkdes[lnkidx++] = c;
 		return;
 	}
 
 	/* Any character other than a '(' terminates a link at ']' */
 	if (in_link == OPT_URL && c != '(') {
 		in_link = NOL;
 		lnkdes[lnkidx] = '\0';
 		printf("<a href=\"%s\">%s</a>", lnkdes, lnkdes);
 	}
 
 	/* Handle Escapes before any other bit of the main switch */
 	if (escape) {
 		maybe_startp();
 		putesc(c);
 		escape = false;
 		return;
 	}
 
 	/* Store the indentation and return without printing */
 	if (!linestarted && c == ' ') {
 		indent++;
 		return;
 	}
 
 	switch (c) {
 	case '\\':
 		escape = true;
 		break;
 	case '#':
 		if (in == NONE) {
 			in = HEADER_PARSE;
 			hlvl = 1;
 		} else if (in == HEADER_PARSE)
 			hlvl++;
 		else
 			putesc(c);
 		break;
 	case ' ':
 		if (in == HEADER_PARSE) {
 			printf("<h%d>", hlvl);
 			in = HEADER;
 		} else if (in == LIST_PARSE) {
 			if (!listdepth) {
 				newlist();
 			} else {
 				if (previndent < indent) {
 					newlist();
 				} else if (previndent > indent) {
 					endlist();
 					printf("</li>\n<li>\n");
 				} else {
 					in = LIST;
 					printf("</li>\n<li>\n");
 				}
 			}
 		} else
 			putesc(c);
 		break;
 	case '*':
 	case '`':
 	case '_':
 		toggle_format(c);
 		break;
 	case '\t':
 		if (in == NONE) {
 			in = CODE;
 			printf("<pre><code>");
 		} else if (lastc != '\n')
 			putesc(c);
 		break;
 	case '>':
 		if (in == NONE) {
 			in = PARAGRAPH;
 			printf("%s", blockquote ? "<p>\n" : "<blockquote>\n<p>\n");
 			blockquote = true;
 		} else if (lastc != '\n')
 			putesc(c);
 		break;
 	case '[':
 		if (in_link == NOL && !fmt_disabled(c)) {
 			maybe_startp();
 			in_link = DESC_PARSE;
 			lnkidx = 0;
 		} else
 			putesc(c);
 		break;
 	case ']':
 		if (in_link == DESC_PARSE)
 			in_link = OPT_URL;
 		else
 			putesc(c);
 		break;
 	case '(':
 		if (in_link == OPT_URL) {
 			in_link = URL_PARSE;
 			printf("<a href=\"");
 		} else
 			putesc(c);
 		break;
 	case ')':
 		if (in_link == URL_PARSE) {
 			in_link = NOL;
 			printf("\">");
 			for (int i = 0; i < lnkidx; i++)
 				handlec(lnkdes[i]);
 			printf("</a>");
 		} else
 			putesc(c);
 		break;
 	case '1':
 	case '2':
 	case '3':
 	case '4':
 	case '5':
 	case '6':
 	case '7':
 	case '8':
 	case '9':
 	case '.':
 	case '-':
 		if (in == NONE || !linestarted) {
 			ol = c != '-';
 			in = LIST_PARSE;
 		} else if (in != LIST_PARSE)
 			putesc(c);
 		break;
 	case '\n':
 		handle_lf();
 		if (in != NONE)
 			putesc(c);
 		break;
 	default:
 		maybe_startp();
 		putesc(c);
 		break;
 	}
 	lastc = c;
 	linestarted = c != '\n';
 }
 
 
 /* nihdoc: a text -> HTML parser */
 int
 main(int argc, char *argv[])
 {
 	int c;
 
 #ifdef __OpenBSD__
 	pledge("stdio", "stdio");
 #endif
 
 	if (argc > 1) {
 		fprintf(stderr, "error: %s takes no arguments\n", argv[0]);
 		return 1;
 	}
 
 	while ((c = getchar()) != EOF)
 		handlec(c);
 
 	/* pretend there's a final LF to close any blocks */
 	handle_lf();
 
 	return 0;
 }