forms.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

/* HTML forms parser */

#ifdef HAVE_CONFIG_H
#include "config.h"
#endif

#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#include "elinks.h"

#include "bfu/listmenu.h"
#include "bfu/menu.h"
#include "document/html/parser/forms.h"
#include "document/html/parser/link.h"
#include "document/html/parser/stack.h"
#include "document/html/parser/parse.h"
#include "document/html/parser.h"
#include "document/html/renderer.h"
#include "document/forms.h"
#include "intl/charsets.h"
#include "protocol/protocol.h"
#include "protocol/uri.h"
#include "util/conv.h"
#include "util/error.h"
#include "util/memdebug.h"
#include "util/memlist.h"
#include "util/memory.h"
#include "util/string.h"

/* Unsafe macros */
#include "document/html/internal.h"



void
html_form(struct html_context *html_context, unsigned char *a,
          unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	unsigned char *al;
	struct form *form;

	html_context->was_br = 1;

	form = init_form();
	if (!form) return;

	form->method = FORM_METHOD_GET;
	form->form_num = a - html_context->startf;

	al = get_attr_val(a, "method", html_context->options);
	if (al) {
		if (!strcasecmp(al, "post")) {
			unsigned char *enctype;

			enctype  = get_attr_val(a, "enctype",
			                        html_context->options);

			form->method = FORM_METHOD_POST;
			if (enctype) {
				if (!strcasecmp(enctype, "multipart/form-data"))
					form->method = FORM_METHOD_POST_MP;
				else if (!strcasecmp(enctype, "text/plain"))
					form->method = FORM_METHOD_POST_TEXT_PLAIN;
				mem_free(enctype);
			}
		}
		mem_free(al);
	}

	al = get_attr_val(a, "name", html_context->options);
	if (al) form->name = al;

	al = get_attr_val(a, "action", html_context->options);
	/* The HTML specification at
	 * http://www.w3.org/TR/REC-html40/interact/forms.html#h-17.3 states
	 * that the behavior of an empty action attribute should be undefined.
	 * Mozilla handles action="" as action="<current-URI>" which seems
	 * reasonable. (bug 615) */
	if (al && *al) {
		form->action = join_urls(html_context->base_href, trim_chars(al, ' ', NULL));
		mem_free(al);

	} else {
		enum uri_component components = URI_ORIGINAL;

		mem_free_if(al);

		/* We have to do following for GET method, because we would end
		 * up with two '?' otherwise. */
		if (form->method == FORM_METHOD_GET)
			components = URI_FORM_GET;

		form->action = get_uri_string(html_context->base_href, components);

		/* No action URI should contain post data */
		assert(!form->action || !strchr(form->action, POST_CHAR));

		/* GET method URIs should not have '?'. */
		assert(!form->action
			|| form->method != FORM_METHOD_GET
			|| !strchr(form->action, '?'));
	}

	al = get_target(html_context->options, a);
	form->target = al ? al : stracpy(html_context->base_target);

	html_context->special_f(html_context, SP_FORM, form);
}


static int
get_form_mode(struct html_context *html_context, unsigned char *attr)
{
	if (has_attr(attr, "disabled", html_context->options))
		return FORM_MODE_DISABLED;

	if (has_attr(attr, "readonly", html_context->options))
		return FORM_MODE_READONLY;

	return FORM_MODE_NORMAL;
}

static struct form_control *
init_form_control(enum form_type type, unsigned char *attr,
                  struct html_context *html_context)
{
	struct form_control *fc;

	fc = mem_calloc(1, sizeof(*fc));
	if (!fc) return NULL;

	fc->type = type;
	fc->position = attr - html_context->startf;
	fc->mode = get_form_mode(html_context, attr);

	return fc;
}

void
html_button(struct html_context *html_context, unsigned char *a,
            unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	unsigned char *al;
	struct form_control *fc;
	enum form_type type = FC_SUBMIT;

	html_focusable(html_context, a);

	al = get_attr_val(a, "type", html_context->options);
	if (!al) goto no_type_attr;

	if (!strcasecmp(al, "button")) {
		type = FC_BUTTON;
	} else if (!strcasecmp(al, "reset")) {
		type = FC_RESET;
	} else if (strcasecmp(al, "submit")) {
		/* unknown type */
		mem_free(al);
		return;
	}
	mem_free(al);

no_type_attr:
	fc = init_form_control(type, a, html_context);
	if (!fc) return;

	fc->name = get_attr_val(a, "name", html_context->options);
	fc->default_value = get_attr_val(a, "value", html_context->options);
	if (!fc->default_value) {
		if (fc->type == FC_SUBMIT)
			fc->default_value = stracpy("Submit");
		else if (fc->type == FC_RESET)
			fc->default_value = stracpy("Reset");
		else if (fc->type == FC_BUTTON)
			fc->default_value = stracpy("Button");
	}
	if (!fc->default_value)
		fc->default_value = stracpy("");

	html_context->special_f(html_context, SP_CONTROL, fc);
	format.form = fc;
	format.style.attr |= AT_BOLD;
}

static void
html_input_format(struct html_context *html_context, unsigned char *a,
	   	  struct form_control *fc)
{
	put_chrs(html_context, " ", 1);
	html_stack_dup(html_context, ELEMENT_KILLABLE);
	html_focusable(html_context, a);
	format.form = fc;
	if (format.title) mem_free(format.title);
	format.title = get_attr_val(a, "title", html_context->options);
	switch (fc->type) {
		case FC_TEXT:
		case FC_PASSWORD:
		case FC_FILE:
		{
			int i;

			format.style.attr |= AT_BOLD;
			for (i = 0; i < fc->size; i++)
				put_chrs(html_context, "_", 1);
			break;
		}
		case FC_CHECKBOX:
			format.style.attr |= AT_BOLD;
			put_chrs(html_context, "[&nbsp;]", 8);
			break;
		case FC_RADIO:
			format.style.attr |= AT_BOLD;
			put_chrs(html_context, "(&nbsp;)", 8);
			break;
		case FC_IMAGE:
		{
			unsigned char *al;

			mem_free_set(&format.image, NULL);
			al = get_url_val(a, "src", html_context->options);
			if (!al)
				al = get_url_val(a, "dynsrc",
				                 html_context->options);
			if (al) {
				format.image = join_urls(html_context->base_href, al);
				mem_free(al);
			}
			format.style.attr |= AT_BOLD;
			put_chrs(html_context, "[&nbsp;", 7);
			if (fc->alt)
				put_chrs(html_context, fc->alt, strlen(fc->alt));
			else if (fc->name)
				put_chrs(html_context, fc->name, strlen(fc->name));
			else
				put_chrs(html_context, "Submit", 6);

			put_chrs(html_context, "&nbsp;]", 7);
			break;
		}
		case FC_SUBMIT:
		case FC_RESET:
		case FC_BUTTON:
			format.style.attr |= AT_BOLD;
			put_chrs(html_context, "[&nbsp;", 7);
			if (fc->default_value)
				put_chrs(html_context, fc->default_value, strlen(fc->default_value));
			put_chrs(html_context, "&nbsp;]", 7);
			break;
		case FC_TEXTAREA:
		case FC_SELECT:
		case FC_HIDDEN:
			INTERNAL("bad control type");
	}
	kill_html_stack_item(html_context, &html_top);
	put_chrs(html_context, " ", 1);
}

void
html_input(struct html_context *html_context, unsigned char *a,
           unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	unsigned char *al;
	struct form_control *fc;

	fc = init_form_control(FC_TEXT, a, html_context);
	if (!fc) return;

	al = get_attr_val(a, "type", html_context->options);
	if (al) {
		if (!strcasecmp(al, "text")) fc->type = FC_TEXT;
		else if (!strcasecmp(al, "hidden")) fc->type = FC_HIDDEN;
		else if (!strcasecmp(al, "button")) fc->type = FC_BUTTON;
		else if (!strcasecmp(al, "checkbox")) fc->type = FC_CHECKBOX;
		else if (!strcasecmp(al, "radio")) fc->type = FC_RADIO;
		else if (!strcasecmp(al, "password")) fc->type = FC_PASSWORD;
		else if (!strcasecmp(al, "submit")) fc->type = FC_SUBMIT;
		else if (!strcasecmp(al, "reset")) fc->type = FC_RESET;
		else if (!strcasecmp(al, "file")) fc->type = FC_FILE;
		else if (!strcasecmp(al, "image")) fc->type = FC_IMAGE;
		/* else unknown type, let it default to FC_TEXT. */
		mem_free(al);
	}

	if (fc->type != FC_FILE)
		fc->default_value = get_attr_val(a, "value",
		                                 html_context->options);
	if (!fc->default_value) {
		if (fc->type == FC_CHECKBOX)
			fc->default_value = stracpy("on");
		else if (fc->type == FC_SUBMIT)
			fc->default_value = stracpy("Submit");
		else if (fc->type == FC_RESET)
			fc->default_value = stracpy("Reset");
		else if (fc->type == FC_BUTTON)
			fc->default_value = stracpy("Button");
	}
	if (!fc->default_value)
		fc->default_value = stracpy("");

	fc->name = get_attr_val(a, "name", html_context->options);

	fc->size = get_num(a, "size", html_context->options);
	if (fc->size == -1)
		fc->size = html_context->options->default_form_input_size;
	fc->size++;
	if (fc->size > html_context->options->box.width)
		fc->size = html_context->options->box.width;
	fc->maxlength = get_num(a, "maxlength", html_context->options);
	if (fc->maxlength == -1) fc->maxlength = INT_MAX;
	if (fc->type == FC_CHECKBOX || fc->type == FC_RADIO)
		fc->default_state = has_attr(a, "checked",
		                             html_context->options);
	if (fc->type == FC_IMAGE)
		fc->alt = get_attr_val(a, "alt", html_context->options);

	if (fc->type != FC_HIDDEN) {
		html_input_format(html_context, a, fc);
	}

	html_context->special_f(html_context, SP_CONTROL, fc);
}

static struct list_menu lnk_menu;

static void
do_html_select(unsigned char *attr, unsigned char *html,
	       unsigned char *eof, unsigned char **end,
	       struct html_context *html_context)
{
	struct conv_table *ct = html_context->special_f(html_context, SP_TABLE, NULL);
	struct form_control *fc;
	struct string lbl = NULL_STRING, orig_lbl = NULL_STRING;
	unsigned char **values = NULL;
	unsigned char **labels;
	unsigned char *name, *t_attr, *en;
	int namelen;
	int nnmi = 0;
	int order = 0;
	int preselect = -1;
	int group = 0;
	int i, max_width;
	int closing_tag;

	html_focusable(html_context, attr);
	init_menu(&lnk_menu);

se:
        en = html;

see:
        html = en;
	while (html < eof && *html != '<') html++;

	if (html >= eof) {

abort:
		*end = html;