html_tbl.c

来自「this is the file used to browse web」· C语言 代码 · 共 1,263 行 · 第 1/3 页

#include "links.h"

#define AL_TR		-1

#define VAL_TR		-1
#define VAL_TOP		0
#define VAL_MIDDLE	1
#define VAL_BOTTOM	2

#define W_AUTO		-1
#define W_REL		-2

#define F_VOID		0
#define F_ABOVE		1
#define F_BELOW		2
#define F_HSIDES	3
#define F_LHS		4
#define F_RHS		8
#define F_VSIDES	12
#define F_BOX		15

#define R_NONE		0
#define R_ROWS		1
#define R_COLS		2
#define R_ALL		3
#define R_GROUPS	4

void get_align(char *attr, int *a)
{
	char *al;
	if ((al = get_attr_val(attr, "align"))) {
		if (!(strcasecmp(al, "left"))) *a = AL_LEFT;
		if (!(strcasecmp(al, "right"))) *a = AL_RIGHT;
		if (!(strcasecmp(al, "center"))) *a = AL_CENTER;
		if (!(strcasecmp(al, "justify"))) *a = AL_BLOCK;
		if (!(strcasecmp(al, "char"))) *a = AL_RIGHT; /* NOT IMPLEMENTED */
		mem_free(al);
	}
}

void get_valign(char *attr, int *a)
{
	char *al;
	if ((al = get_attr_val(attr, "valign"))) {
		if (!(strcasecmp(al, "top"))) *a = VAL_TOP;
		if (!(strcasecmp(al, "middle"))) *a = VAL_MIDDLE;
		if (!(strcasecmp(al, "bottom"))) *a = VAL_BOTTOM;
		if (!(strcasecmp(al, "baseline"))) *a = VAL_TOP; /* NOT IMPLEMENTED */
		mem_free(al);
	}
}

void get_c_width(char *attr, int *w, int sh)
{
	char *al;
	if ((al = get_attr_val(attr, "width"))) {
		if (*al && al[strlen(al) - 1] == '*') {
			char *en;
			unsigned long n;
			al[strlen(al) - 1] = 0;
			n = strtoul(al, &en, 10);
			if (n < 10000 && !*en) *w = W_REL - n;
		} else {
			int p = get_width(attr, "width", sh);
			if (p >= 0) *w = p;
		}
		mem_free(al);
	}
}

#define INIT_X		8
#define INIT_Y		8

struct table_cell {
	int used;
	int spanned;
	int mx, my;
	unsigned char *start;
	unsigned char *end;
	int align;
	int valign;
	int b;
	struct rgb bgcolor;
	int group;
	int colspan;
	int rowspan;
	int min_width;
	int max_width;
	int x_width;
	int height;
	int xpos, ypos, xw, yw;
	int link_num;
};

struct table_column {
	int group;
	int align;
	int valign;
	int width;
};

struct table {
	struct part *p;
	int x, y;
	int rx, ry;
	int border, cellpd, vcellpd, cellsp;
	int frame, rules, width, wf;
	int *min_c, *max_c;
	int *w_c;
	int rw;
	int min_t, max_t;
	struct table_cell *cells;
	int c, rc;
	struct table_column *cols;
	int xc;
	int *xcols;
	int *r_heights;
	int rh;
	int link_num;
};

#define CELL(t, x, y) (&(t)->cells[(y) * (t)->rx + (x)])

unsigned char frame_table[81] = {
	0x00, 0xb3, 0xba,	0xc4, 0xc0, 0xd3,	0xcd, 0xd4, 0xc8,
	0xc4, 0xd9, 0xbd,	0xc4, 0xc1, 0xd0,	0xcd, 0xd4, 0xc8,
	0xcd, 0xbe, 0xbc,	0xcd, 0xbe, 0xbc,	0xcd, 0xcf, 0xca,

	0xb3, 0xb3, 0xba,	0xda, 0xc3, 0xd3,	0xd5, 0xc6, 0xc8,
	0xbf, 0xb4, 0xbd,	0xc2, 0xc5, 0xd0,	0xd5, 0xc6, 0xc8,
	0xb8, 0xb5, 0xbc,	0xb8, 0xb5, 0xbc,	0xd1, 0xd8, 0xca,

	0xba, 0xba, 0xba,	0xd6, 0xd6, 0xc7,	0xc9, 0xc9, 0xcc,
	0xb7, 0xb7, 0xb6,	0xd2, 0xd2, 0xd7,	0xc9, 0xc9, 0xcc,
	0xbb, 0xbb, 0xb9,	0xbb, 0xbb, 0xb9,	0xcb, 0xcb, 0xce,
};

unsigned char hline_table[3] = { 0x20, 0xc4, 0xcd };
unsigned char vline_table[3] = { 0x20, 0xb3, 0xba };

struct table *new_table()
{
	struct table *t;
	t = mem_alloc(sizeof(struct table));
	memset(t, 0, sizeof(struct table));
	t->p = NULL;
	t->x = t->y = 0;
	t->rx = INIT_X;
	t->ry = INIT_Y;
	t->cells = mem_alloc(INIT_X * INIT_Y * sizeof(struct table_cell));
	memset(t->cells, 0, INIT_X * INIT_Y * sizeof(struct table_cell));
	t->c = 0;
	t->rc = INIT_X;
	t->cols = mem_alloc(INIT_X * sizeof(struct table_column));
	memset(t->cols, 0, INIT_X * sizeof(struct table_column));
	t->xcols = DUMMY;
	t->xc = 0;
	t->r_heights = DUMMY;
	return t;
}

void free_table(struct table *t)
{
	if (t->min_c) mem_free(t->min_c);
	if (t->max_c) mem_free(t->max_c);
	if (t->w_c) mem_free(t->w_c);
	mem_free(t->r_heights);
	mem_free(t->cols);
	mem_free(t->xcols);
	mem_free(t->cells);
	mem_free(t);
}

void expand_cells(struct table *t, int x, int y)
{
	int i, j;
	if (x >= t->x) {
		if (t->x) {
			for (i = 0; i < t->y; i++) if (CELL(t, t->x - 1, i)->colspan == -1) {
				for (j = t->x; j <= x; j++) {
					CELL(t, j, i)->used = 1;
					CELL(t, j, i)->spanned = 1;
					CELL(t, j, i)->rowspan = CELL(t, t->x - 1, i)->rowspan;
					CELL(t, j, i)->colspan = -1;
					CELL(t, j, i)->mx = CELL(t, t->x - 1, i)->mx;
					CELL(t, j, i)->my = CELL(t, t->x - 1, i)->my;
				}
			}
		}
		t->x = x + 1;
	}
	if (y >= t->y) {
		if (t->y) {
			for (i = 0; i < t->x; i++) if (CELL(t, i, t->y - 1)->rowspan == -1) {
				for (j = t->y; j <= y; j++) {
					CELL(t, i, j)->used = 1;
					CELL(t, i, j)->spanned = 1;
					CELL(t, i, j)->rowspan = -1;
					CELL(t, i, j)->colspan = CELL(t, i, t->y - 1)->colspan;
					CELL(t, i, j)->mx = CELL(t, i, t->y - 1)->mx;
					CELL(t, i, j)->my = CELL(t, i, t->y - 1)->my;
				}
			}
		}
		t->y = y + 1;
	}
}

struct table_cell *new_cell(struct table *t, int x, int y)
{
	struct table nt;
	int i, j;
	if (x < t->x && y < t->y) goto ret;
	rep:
	if (x < t->rx && y < t->ry) {
		expand_cells(t, x, y);
		goto ret;
	}
	nt.rx = t->rx;
	nt.ry = t->ry;
	while (x >= nt.rx) {
		if ((unsigned)nt.rx > MAXINT / 2) overalloc();
		nt.rx *= 2;
	}
	while (y >= nt.ry) {
		if ((unsigned)nt.ry > MAXINT / 2) overalloc();
		nt.ry *= 2;
	}
	if ((unsigned)nt.rx * (unsigned)nt.ry / (unsigned)nt.rx != (unsigned)nt.ry) overalloc();
	if ((unsigned)nt.rx * (unsigned)nt.ry > MAXINT / sizeof(struct table_cell)) overalloc();
	nt.cells = mem_alloc(nt.rx * nt.ry * sizeof(struct table_cell));
	memset(nt.cells, 0, nt.rx * nt.ry * sizeof(struct table_cell));
	for (i = 0; i < t->x; i++)
		for (j = 0; j < t->y; j++)
			memcpy(CELL(&nt, i, j), CELL(t, i, j), sizeof(struct table_cell));
	mem_free(t->cells);
	t->cells = nt.cells;
	t->rx = nt.rx;
	t->ry = nt.ry;
	goto rep;

	ret:
	return CELL(t, x, y);
}

void new_columns(struct table *t, int span, int width, int align, int valign, int group)
{
	if ((unsigned)t->c + (unsigned)span > MAXINT) overalloc();
	if (t->c + span > t->rc) {
		int n = t->rc;
		struct table_column *nc;
		while (t->c + span > n) {
			if ((unsigned)n > MAXINT / 2) overalloc();
			n *= 2;
		}
		if ((unsigned)n > MAXINT / sizeof(struct table_column)) overalloc();
		nc = mem_realloc(t->cols, n * sizeof(struct table_column));
		t->rc = n;
		t->cols = nc;
	}
	while (span--) {
		t->cols[t->c].align = align;
		t->cols[t->c].valign = valign;
		t->cols[t->c].width = width;
		t->cols[t->c++].group = group;
		group = 0;
	}
}

void set_td_width(struct table *t, int x, int width, int f)
{
	if (x >= t->xc) {
		int n = t->xc ? t->xc : 1;
		int i;
		int *nc;
		while (x >= n) {
			if ((unsigned)n > MAXINT / 2) overalloc();
			n *= 2;
		}
		if ((unsigned)n > MAXINT / sizeof(int)) overalloc();
		nc = mem_realloc(t->xcols, n * sizeof(int));
		for (i = t->xc; i < n; i++) nc[i] = W_AUTO;
		t->xc = n;
		t->xcols = nc;
	}
	if (t->xcols[x] == W_AUTO || f) {
		set:
		t->xcols[x] = width;
		return;
	}
	if (width == W_AUTO) return;
	if (width < 0 && t->xcols[x] >= 0) goto set;
	if (width >= 0 && t->xcols[x] < 0) return;
	t->xcols[x] = (t->xcols[x] + width) / 2;
}

unsigned char *skip_table(unsigned char *html, unsigned char *eof)
{
	int level = 1;
	unsigned char *name;
	int namelen;
	r:
	while (html < eof && (*html != '<' || parse_element(html, eof, &name, &namelen, NULL, &html))) html++;
	if (html >= eof) return eof;
	if (namelen == 5 && !casecmp(name, "TABLE", 5)) level++;
	if (namelen == 6 && !casecmp(name, "/TABLE", 6)) if (!--level) return html;
	goto r;
}

struct s_e {
	unsigned char *s, *e;
};

struct table *parse_table(unsigned char *html, unsigned char *eof, unsigned char **end, struct rgb *bgcolor, int sh, struct s_e **bad_html, int *bhp)
{
	int qqq;
	struct table *t;
	struct table_cell *cell;
	unsigned char *t_name, *t_attr, *en;
	int t_namelen;
	int x = 0, y = -1;
	int p = 0;
	unsigned char *lbhp = NULL;
	int l_al = AL_LEFT;
	int l_val = VAL_MIDDLE;
	int csp, rsp;
	int group = 0;
	int i, j, k;
	struct rgb l_col;
	int c_al = AL_TR, c_val = VAL_TR, c_width = W_AUTO, c_span = 0;
	memcpy(&l_col, bgcolor, sizeof(struct rgb));
	*end = html;
	if (bad_html) {
		*bad_html = DUMMY;
		*bhp = 0;
	}
	if (!(t = new_table())) return NULL;
	se:
	en = html;
	see:
	html = en;
	if (bad_html && !p && !lbhp) {
		if (!(*bhp & (ALLOC_GR-1))) {
			if ((unsigned)*bhp > MAXINT / sizeof(struct s_e) - ALLOC_GR) overalloc();
			*bad_html = mem_realloc(*bad_html, (*bhp + ALLOC_GR) * sizeof(struct s_e));
		}
		lbhp = (*bad_html)[(*bhp)++].s = html;
	}
	while (html < eof && *html != '<') html++;
	if (html >= eof) {
		if (p) CELL(t, x, y)->end = html;
		if (lbhp) (*bad_html)[*bhp-1].e = html;
		goto scan_done;
	}
	if (html + 2 <= eof && (html[1] == '!' || html[1] == '?')) {
		html = skip_comment(html, eof);
		goto se;
	}
	if (parse_element(html, eof, &t_name, &t_namelen, &t_attr, &en)) {
		html++;
		goto se;
	}
	if (t_namelen == 5 && !casecmp(t_name, "TABLE", 5)) {
		en = skip_table(en, eof);
		goto see;
	}
	if (t_namelen == 6 && !casecmp(t_name, "/TABLE", 6)) {
		if (c_span) new_columns(t, c_span, c_width, c_al, c_val, 1);
		if (p) CELL(t, x, y)->end = html;
		if (lbhp) (*bad_html)[*bhp-1].e = html;
		goto scan_done;
	}
	if (t_namelen == 8 && !casecmp(t_name, "COLGROUP", 8)) {
		if (c_span) new_columns(t, c_span, c_width, c_al, c_val, 1);
		if (lbhp) (*bad_html)[*bhp-1].e = html, lbhp = NULL;
		c_al = AL_TR;
		c_val = VAL_TR;
		c_width = W_AUTO;
		get_align(t_attr, &c_al);
		get_valign(t_attr, &c_val);
		get_c_width(t_attr, &c_width, sh);
		if ((c_span = get_num(t_attr, "span")) == -1) c_span = 1;
		goto see;
	}
	if (t_namelen == 9 && !casecmp(t_name, "/COLGROUP", 9)) {
		if (c_span) new_columns(t, c_span, c_width, c_al, c_val, 1);
		if (lbhp) (*bad_html)[*bhp-1].e = html, lbhp = NULL;
		c_span = 0;
		c_al = AL_TR;
		c_val = VAL_TR;
		c_width = W_AUTO;
		goto see;
	}
	if (t_namelen == 3 && !casecmp(t_name, "COL", 3)) {
		int sp, wi, al, val;
		if (lbhp) (*bad_html)[*bhp-1].e = html, lbhp = NULL;
		if ((sp = get_num(t_attr, "span")) == -1) sp = 1;
		wi = c_width;
		al = c_al;
		val = c_val;
		get_align(t_attr, &al);
		get_valign(t_attr, &val);
		get_c_width(t_attr, &wi, sh);
		new_columns(t, sp, wi, al, val, !!c_span);
		c_span = 0;
		goto see;
	}
	if (t_namelen == 3 && (!casecmp(t_name, "/TR", 3) || !casecmp(t_name, "/TD", 3) || !casecmp(t_name, "/TH", 3))) {
		if (c_span) new_columns(t, c_span, c_width, c_al, c_val, 1);
		if (p) CELL(t, x, y)->end = html, p = 0;
		if (lbhp) (*bad_html)[*bhp-1].e = html, lbhp = NULL;
	}
	if (t_namelen == 2 && !casecmp(t_name, "TR", 2)) {
		if (c_span) new_columns(t, c_span, c_width, c_al, c_val, 1);
		if (p) CELL(t, x, y)->end = html, p = 0;
		if (lbhp) (*bad_html)[*bhp-1].e = html, lbhp = NULL;
		if (group) group--;
		l_al = AL_LEFT;
		l_val = VAL_MIDDLE;
		memcpy(&l_col, bgcolor, sizeof(struct rgb));
		get_align(t_attr, &l_al);