/*
 * subst.c - Substitution rules
 *
 * Copyright 2012 by Werner Almesberger
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 */


#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <ctype.h>
#include <sys/types.h>
#include <regex.h>
#include <assert.h>

#include "util.h"
#include "lang.h"
#include "subst.h"


/* ----- Rule set construction --------------------------------------------- */


static struct subst *alloc_subst(enum subst_type type)
{
	struct subst *sub;

	sub = alloc_type(struct subst);
	sub->type = type;
	sub->next = NULL;
	return sub;
}


struct subst *subst_match(const char *src, const char *re)
{
	char error[1000];
	struct subst *sub;
	char *tmp;
	int err, len;

	sub = alloc_subst(st_match);
	sub->u.match.src = src;
	len = strlen(re);
	tmp = alloc_size(len+3);
	tmp[0] = '^';
	memcpy(tmp+1, re, len);
	tmp[len+1] = '$';
	tmp[len+2] = 0;
	err = regcomp(&sub->u.match.re, tmp, REG_EXTENDED);
	free(tmp);
	if (err) {
		regerror(err, &sub->u.match.re, error, sizeof(error));
		yyerrorf("%s", error);
	}
	return sub;
}


static void end_chunk(struct chunk ***last, const char *start, const char *s)
{
	struct chunk *c;

	if (s == start)
		return;

	c = alloc_type(struct chunk);
	c->type = ct_string;
	c->u.s = stralloc_n(start, s-start);;
	c->unit = NULL;
	c->next = NULL;
	**last = c;
	*last = &c->next;
}


static const char *parse_var(struct chunk *c, const char *s)
{
	const char *t;
	int braced;

	if (!*s)
		yyerror("trailing dollar sign");

	braced = *s == '{';
	if (braced)
		s++;

	t = s;
	while (*t) {
		if (braced && *t == '}')
			break;
		if (s == t && *t == '$') {
			t++;
			break;
		}
		if (!isalnum(*t))
			break;
		t++;
	}
	if (s == t)
		yyerror("invalid variable name");
	if (braced && !*t)
		yyerror("unterminated variable name");
	if (isdigit(*s)) {
		if (t != s+1 || *s == '0')
			yyerror("invalid variable name");
		c->type = ct_sub;
		c->u.sub = *s-'0';
	} else if (isalnum(*s)) {
		char *tmp;

		c->type = ct_var;
		tmp = stralloc_n(s, t-s);
		c->u.var = unique(tmp);
		free(tmp);
	} else {
		c->type = ct_sub;
		c->u.sub = 0;
	}

	if (*t != '#') {
		if (braced) {
			assert(*t == '}');
			t++;
		}
		return t;
	}

	s = ++t;
	while (*t) {
		if (braced && *t == '}')
			break;
		if (!braced && t != s)
			break;
		t++;
	}
	if (s == t)
		yyerror("invalid unit");
	c->unit = stralloc_n(s, t-s);
	if (braced) {
		if (!*t)
			yyerror("unterminated unit");
		t++;
	}
	return t;
}


static struct chunk *parse_pattern(const char *s)
{
	struct chunk *res = NULL, **last = &res;
	struct chunk *c;
	const char *start = s;

	while (*s) {
		if (*s == '\\') {
			if (!s[1])
				yyerror("trailing backslash");
			end_chunk(&last, start, s);
			start = s+1;
			s += 2;
			continue;
		}
		if (*s != '$') {
			s++;
			continue;
		}

		end_chunk(&last, start, s);
		c = alloc_type(struct chunk);
		c->unit = NULL;
		c->next = NULL;
		*last = c;
		last = &c->next;
		start = s = parse_var(c, s+1);
	}
	end_chunk(&last, start, s);
	return res;
}


struct subst *subst_assign(const char *dst, const char *pat)
{
	struct subst *sub;

	sub = alloc_subst(st_assign);
	sub->u.assign.dst = dst;
	sub->u.assign.pat = parse_pattern(pat);
	return sub;
}


struct subst *subst_end(void)
{
	return alloc_subst(st_end);
}


struct subst *subst_break(const char *block)
{
	struct subst *sub;

	sub = alloc_subst(st_break);
	sub->u.tmp = block;
	return sub;
}


struct subst *subst_again(const char *block)
{
	struct subst *sub;

	sub = alloc_subst(st_again);
	sub->u.tmp = block;
	return sub;
}


/* ----- Jump resolution --------------------------------------------------- */


struct parent {
	const struct subst *sub;
	const struct parent *parent;
};


static const struct subst *resolve_jump(const char *name,
    const struct parent *parent)
{
	while (parent) {
		assert(parent->sub->type == st_match);
		if (!strcmp(name, parent->sub->u.match.src))
			return parent->sub;
		parent = parent->parent;
	}
	yyerrorf("cannot find \"%s\"", name);
}


static void recurse_fin(struct subst *sub, const struct parent *parent)
{
	struct parent next = {
		.sub = sub,
		.parent = parent,
	};

	while (sub) {
		switch (sub->type) {
		case st_match:
			recurse_fin(sub->u.match.block, &next);
		case st_assign:
		case st_end:
			break;
		case st_break:
		case st_again:
			sub->u.jump = resolve_jump(sub->u.tmp, parent);
			break;
		default:
			abort();
		}
		sub = sub->next;
	}
}


void subst_finalize(struct subst *sub)
{
	recurse_fin(sub, NULL);
}


/* ----- Dumping ----------------------------------------------------------- */


#define	INDENT	4


static void dump_chunks(FILE *file, const struct chunk *c)
{
	while (c) {
		switch (c->type) {
		case ct_string:
			fprintf(file, "%s", c->u.s);
			break;
		case ct_var:
			fprintf(file, "${%s", c->u.var);
			break;
		case ct_sub:
			if (c->u.sub)
				fprintf(file, "${%d", c->u.sub);
			else
				fprintf(file, "${$");
			break;
		default:
			abort();
		}
		if (c->type != ct_string) {
			if (c->unit)
				fprintf(file, "#%s", c->unit);
			fprintf(file, "}");
		}
		c = c->next;
	}
}


static void recurse_dump(FILE *file, const struct subst *sub, int level)
{
	while (sub) {
		fprintf(file, "%*s", INDENT*level, "");
		switch (sub->type) {
		case st_match:
			fprintf(file, "%s=RE {\n", sub->u.match.src);
			recurse_dump(file, sub->u.match.block, level+1);
			fprintf(file, "%*s}\n", INDENT*level, "");
			break;
		case st_assign:
			fprintf(file, "%s=", sub->u.assign.dst);
			dump_chunks(file, sub->u.assign.pat);
			fprintf(file, "\n");
			break;
		case st_end:
			fprintf(file, "end\n");
			break;
		case st_break:
			fprintf(file, "break %s\n", sub->u.jump->u.match.src);
			break;
		case st_again:
			fprintf(file, "again %s\n", sub->u.jump->u.match.src);
			break;
		default:
			abort();
		}
		sub = sub->next;
	}
}


void subst_dump(FILE *file, const struct subst *sub)
{
	recurse_dump(file, sub, 0);
}