search.c

elinks下lynx是最重要的二个文本浏览器, 在linux下非常实用, elinks也是gentoo安装过程中默认使用的浏览器, 这是elinks源代码

	txt = case_sensitive ? stracpy(*doc_view->search_word)
			     : lowered_string(*doc_view->search_word, l);
	if (!txt) return;

	box = &doc_view->box;
	xoffset = box->x - doc_view->vs->x;
	yoffset = box->y - doc_view->vs->y;

#define maybe_tolower(c) (case_sensitive ? (c) : tolower(c))

	for (; s1 <= s2; s1++) {
		int i;

		if (maybe_tolower(s1[0].c) != txt[0]) {
srch_failed:
			continue;
		}

		for (i = 1; i < l; i++)
			if (maybe_tolower(s1[i].c) != txt[i])
				goto srch_failed;

		for (i = 0; i < l; i++) {
			int j;
			int y = s1[i].y + yoffset;

			if (!row_is_in_box(box, y))
				continue;

			for (j = 0; j < s1[i].n; j++) {
				int sx = s1[i].x + j;
				int x = sx + xoffset;

				if (!col_is_in_box(box, x))
					continue;

				if (!realloc_points(&points, len))
					continue;

				points[len].x = sx;
				points[len++].y = s1[i].y;
			}
		}
	}

#undef maybe_tolower

	mem_free(txt);
	*pt = points;
	*pl = len;
}

#ifdef HAVE_REGEX_H
struct get_searched_regex_context {
	int xoffset;
	int yoffset;
	struct box *box;
	struct point *points;
	int len;
};

static void
get_searched_regex_match(struct regex_match_context *common_ctx, void *data)
{
	struct get_searched_regex_context *ctx = data;
	int i;

	for (i = 0; i < common_ctx->textlen; i++) {
		int j;
		int y = common_ctx->s1[i].y + ctx->yoffset;

		if (!row_is_in_box(ctx->box, y))
			continue;

		for (j = 0; j < common_ctx->s1[i].n; j++) {
			int sx = common_ctx->s1[i].x + j;
			int x = sx + ctx->xoffset;

			if (!col_is_in_box(ctx->box, x))
				continue;

			if (!realloc_points(&ctx->points, ctx->len))
				continue;

			ctx->points[ctx->len].x = sx;
			ctx->points[ctx->len++].y = common_ctx->s1[i].y;
		}
	}
}

static void
get_searched_regex(struct document_view *doc_view, struct point **pt, int *pl,
		   int textlen, struct search *s1, struct search *s2)
{
	struct regex_match_context common_ctx;
	struct get_searched_regex_context ctx;

	ctx.points = NULL;
	ctx.len = 0;
	ctx.box = &doc_view->box;
	ctx.xoffset = ctx.box->x - doc_view->vs->x;
	ctx.yoffset = ctx.box->y - doc_view->vs->y;

	common_ctx.found = 0;
	common_ctx.textlen = textlen;
	common_ctx.y1 = doc_view->vs->y - 1;
	common_ctx.y2 = doc_view->vs->y + ctx.box->height;
	common_ctx.pattern = *doc_view->search_word;
	common_ctx.s1 = s1;
	common_ctx.s2 = s2;

	search_for_pattern(&common_ctx, &ctx, get_searched_regex_match);

	*pt = ctx.points;
	*pl = ctx.len;
}
#endif /* HAVE_REGEX_H */

static void
get_searched(struct document_view *doc_view, struct point **pt, int *pl)
{
	struct search *s1, *s2;
	int l;

	assert(doc_view && doc_view->vs && pt && pl);
	if_assert_failed return;

	if (!has_search_word(doc_view))
		return;

	get_search_data(doc_view->document);
	l = strlen(*doc_view->search_word);
	if (get_range(doc_view->document, doc_view->vs->y,
		      doc_view->box.height, l, &s1, &s2)) {
		*pt = NULL;
		*pl = 0;

		return;
	}

#ifdef HAVE_REGEX_H
	if (get_opt_int("document.browse.search.regex"))
		get_searched_regex(doc_view, pt, pl, l, s1, s2);
	else
#endif
		get_searched_plain(doc_view, pt, pl, l, s1, s2);
}

/* Highlighting of searched strings. */
void
draw_searched(struct terminal *term, struct document_view *doc_view)
{
	struct point *pt = NULL;
	int len = 0;

	assert(term && doc_view);
	if_assert_failed return;

	if (!has_search_word(doc_view))
		return;

	get_searched(doc_view, &pt, &len);
	if (len) {
		int i;
		struct color_pair *color = get_bfu_color(term, "searched");
		int xoffset = doc_view->box.x - doc_view->vs->x;
		int yoffset = doc_view->box.y - doc_view->vs->y;

		for (i = 0; i < len; i++) {
			int x = pt[i].x + xoffset;
			int y = pt[i].y + yoffset;

			/* TODO: We should take in account original colors and
			 * combine them with defined color. */
#if 0
			/* This piece of code shows the old way of handling
			 * colors and screen char attributes. */
			unsigned co = get_char(term, x, y);
			co = ((co >> 3) & 0x0700) | ((co << 3) & 0x3800);
#endif

			draw_char_color(term, x, y, color);
		}
	}

	mem_free_if(pt);
}


enum find_error {
	FIND_ERROR_NONE,
	FIND_ERROR_NO_PREVIOUS_SEARCH,
	FIND_ERROR_HIT_TOP,
	FIND_ERROR_HIT_BOTTOM,
	FIND_ERROR_NOT_FOUND,
	FIND_ERROR_MEMORY,
	FIND_ERROR_REGEX,
};

static enum find_error find_next_do(struct session *ses,
				    struct document_view *doc_view,
				    int direction);

static void print_find_error(struct session *ses, enum find_error find_error);

static enum find_error
search_for_do(struct session *ses, unsigned char *str, int direction,
	      int report_errors)
{
	struct document_view *doc_view;
	enum find_error error;

	assert(ses && str);
	if_assert_failed return FIND_ERROR_NOT_FOUND;

	doc_view = current_frame(ses);

	assert(doc_view);
	if_assert_failed return FIND_ERROR_NOT_FOUND;

	mem_free_set(&ses->search_word, NULL);
	mem_free_set(&ses->last_search_word, NULL);

	if (!*str) return FIND_ERROR_NOT_FOUND;

	/* We only set the last search word because we don.t want find_next()
	 * to try to find next link in search before the search data has been
	 * initialized. find_next() will set ses->search_word for us. */
	ses->last_search_word = stracpy(str);
	if (!ses->last_search_word) return FIND_ERROR_NOT_FOUND;

	ses->search_direction = direction;

	error = find_next_do(ses, doc_view, 1);

	if (report_errors)
		print_find_error(ses, error);

	return error;
}

static void
search_for_back(struct session *ses, unsigned char *str)
{
	assert(ses && str);
	if_assert_failed return;

	search_for_do(ses, str, -1, 1);
}

static void
search_for(struct session *ses, unsigned char *str)
{
	assert(ses && str);
	if_assert_failed return;

	search_for_do(ses, str, 1, 1);
}


static inline int
point_intersect(struct point *p1, int l1, struct point *p2, int l2)
{
#define HASH_SIZE	4096
#define HASH(p) ((((p).y << 6) + (p).x) & (HASH_SIZE - 1))

	int i;
	static char hash[HASH_SIZE];
	static int first_time = 1;

	assert(p2);
	if_assert_failed return 0;

	if (first_time) memset(hash, 0, HASH_SIZE), first_time = 0;

	for (i = 0; i < l1; i++) hash[HASH(p1[i])] = 1;

	for (i = 0; i < l2; i++) {
		int j;

		if (!hash[HASH(p2[i])]) continue;

		for (j = 0; j < l1; j++) {
			if (p1[j].x != p2[i].x) continue;
			if (p1[j].y != p2[i].y) continue;

			for (i = 0; i < l1; i++)
				hash[HASH(p1[i])] = 0;

			return 1;
		}
	}

	for (i = 0; i < l1; i++) hash[HASH(p1[i])] = 0;

	return 0;

#undef HASH
#undef HASH_SIZE
}

static int
find_next_link_in_search(struct document_view *doc_view, int direction)
{
	assert(doc_view && doc_view->vs);
	if_assert_failed return 0;

	if (direction == -2 || direction == 2) {
		direction /= 2;
		if (direction < 0)
			find_link_page_up(doc_view);
		else
			find_link_page_down(doc_view);

		if (doc_view->vs->current_link == -1) return 1;
		goto nt;
	}

	while (doc_view->vs->current_link != -1
	       && next_link_in_view(doc_view, doc_view->vs->current_link + direction,
	                            direction)) {
		struct point *pt = NULL;
		struct link *link;
		int len;

nt:
		link = &doc_view->document->links[doc_view->vs->current_link];
		get_searched(doc_view, &pt, &len);
		if (point_intersect(pt, len, link->points, link->npoints)) {
			mem_free(pt);
			return 0;
		}
		mem_free_if(pt);
	}

	if (direction < 0)
		find_link_page_up(doc_view);
	else
		find_link_page_down(doc_view);

	return 1;
}

static enum find_error
find_next_do(struct session *ses, struct document_view *doc_view, int direction)
{
	int p, min, max, c = 0;
	int step, hit_bottom = 0, hit_top = 0;
	int height;

	assert(ses && ses->tab && ses->tab->term && doc_view && doc_view->vs
	       && direction);
	if_assert_failed return FIND_ERROR_NONE;

	direction *= ses->search_direction;
	p = doc_view->vs->y;
	height = doc_view->box.height;
	step = direction * height;

	if (ses->search_word) {
		if (!find_next_link_in_search(doc_view, direction))
			return FIND_ERROR_NONE;
		p += step;
	}

	if (!ses->search_word) {
		if (!ses->last_search_word) {
			return FIND_ERROR_NO_PREVIOUS_SEARCH;
		}
		ses->search_word = stracpy(ses->last_search_word);
		if (!ses->search_word) return FIND_ERROR_NONE;
	}

	get_search_data(doc_view->document);

	do {
		int in_range = is_in_range(doc_view->document, p, height,
					   ses->search_word, &min, &max);

		if (in_range == -1) return FIND_ERROR_MEMORY;
		if (in_range == -2) return FIND_ERROR_REGEX;
		if (in_range) {
			doc_view->vs->y = p;
			if (max >= min)
				doc_view->vs->x = int_min(int_max(doc_view->vs->x,
								  max - doc_view->box.width),
								  min);

			set_link(doc_view);
			find_next_link_in_search(doc_view, direction * 2);

			if (hit_top)
				return FIND_ERROR_HIT_TOP;

			if (hit_bottom)
				return FIND_ERROR_HIT_BOTTOM;

			return FIND_ERROR_NONE;
		}
		p += step;
		if (p > doc_view->document->height) {
			hit_bottom = 1;
			p = 0;
		}
		if (p < 0) {
			hit_top = 1;
			p = 0;
			while (p < doc_view->document->height) p += height;
			p -= height;
		}
		c += height;
	} while (c < doc_view->document->height + height);

	return FIND_ERROR_NOT_FOUND;
}

static void
print_find_error_not_found(struct session *ses, unsigned char *title,
			   unsigned char *message, unsigned char *search_string)
{
	switch (get_opt_int("document.browse.search.show_not_found")) {
		case 2:
			info_box(ses->tab->term, MSGBOX_FREE_TEXT,
				 title, ALIGN_CENTER,
				 msg_text(ses->tab->term, message,
					  search_string));
			break;

		case 1:
			beep_terminal(ses->tab->term);

		default:
			break;
	}
}

static void
print_find_error(struct session *ses, enum find_error find_error)
{
	int hit_top = 0;
	unsigned char *message = NULL;

	switch (find_error) {
		case FIND_ERROR_HIT_TOP:
			hit_top = 1;
		case FIND_ERROR_HIT_BOTTOM:
			if (!get_opt_bool("document.browse.search"
					  ".show_hit_top_bottom"))
				break;

			message = hit_top
				 ? N_("Search hit top, continuing at bottom.")
				 : N_("Search hit bottom, continuing at top.");
			break;
		case FIND_ERROR_NO_PREVIOUS_SEARCH:
			message = N_("No previous search");
			break;
		case FIND_ERROR_NOT_FOUND:
			print_find_error_not_found(ses, N_("Search"),
						   N_("Search string"
						      " '%s' not found"),
						   ses->search_word);

			break;

		case FIND_ERROR_REGEX:
			print_find_error_not_found(ses, N_("Search"),
						   N_("Could not compile"
						      " regular expression"
						      " '%s'"),
						   ses->search_word);

			break;

		case FIND_ERROR_MEMORY:
			/* Why bother trying to create a msg_box?
			 * We probably don't have the memory... */
		case FIND_ERROR_NONE:
			break;
	}

	if (!message) return;
	info_box(ses->tab->term, 0, N_("Search"), ALIGN_CENTER, message);
}

enum frame_event_status
find_next(struct session *ses, struct document_view *doc_view, int direction)
{
	print_find_error(ses, find_next_do(ses, doc_view, direction));

	/* FIXME: Make this more fine-grained */
	return FRAME_EVENT_REFRESH;
}


/* Link typeahead */

enum typeahead_code {
	TYPEAHEAD_MATCHED,
	TYPEAHEAD_ERROR,
	TYPEAHEAD_ERROR_NO_FURTHER,
	TYPEAHEAD_CANCEL,
};

static void
typeahead_error(struct session *ses, unsigned char *typeahead, int no_further)
{
	unsigned char *message;

	if (no_further)
		message = N_("No further matches for '%s'.");
	else
		message = N_("Could not find a link with the text '%s'.");

	print_find_error_not_found(ses, N_("Typeahead"), message, typeahead);
}

static inline unsigned char *
get_link_typeahead_text(struct link *link)
{
	unsigned char *name = get_link_name(link);

	if (name) return name;
	if (link->where) return link->where;
	if (link->where_img) return link->where_img;

	return "";
}

static int
match_link_text(struct link *link, unsigned char *text, int textlen,
		int case_sensitive)
{
	unsigned char *match = get_link_typeahead_text(link);
	unsigned char *matchpos;

	if (link_is_form(link) || textlen > strlen(match))
		return -1;

	matchpos = case_sensitive ? strstr(match, text)
				  : strcasestr(match, text);

	if (matchpos) {
		return matchpos - match;
	}

	return -1;
}

/* Searches the @document for a link with the given @text. takes the
 * current_link in the view, the link to start searching from @i and the
 * direction to search (1 is forward, -1 is back). */
static inline int
search_link_text(struct document *document, int current_link, int i,