camel/camel-mime-filter-html.c

   1 /* -*- Mode: C; tab-width: 8; indent-tabs-mode: t; c-basic-offset: 8 -*- */
   2 /*
   3  *  Copyright (C) 2001 Ximian Inc.
   4  *
   5  *  Authors: Michael Zucchi <notzed@ximian.com>
   6  *
   7  * This program is free software; you can redistribute it and/or
   8  * modify it under the terms of version 2 of the GNU Lesser General Public
   9  * License as published by the Free Software Foundation.
  10  *
  11  * This program is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14  * General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU Lesser General Public
  17  * License along with this program; if not, write to the
  18  * Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  19  * Boston, MA 02110-1301, USA.
  20  */
  21
  22 #include <fcntl.h>
  23 #include <stdarg.h>
  24 #include <stdio.h>
  25 #include <stdlib.h>
  26 #include <string.h>
  27 #include <unistd.h>
  28 #include <sys/stat.h>
  29 #include <sys/types.h>
  30
  31 #include "camel-html-parser.h"
  32 #include "camel-mime-filter-html.h"
  33
  34 #define d(x)
  35
  36 static void camel_mime_filter_html_class_init (CamelMimeFilterHTMLClass *klass);
  37 static void camel_mime_filter_html_init       (CamelObject *o);
  38 static void camel_mime_filter_html_finalize   (CamelObject *o);
  39
  40 static CamelMimeFilterClass *camel_mime_filter_html_parent;
  41
  42 struct _CamelMimeFilterHTMLPrivate {
  43         CamelHTMLParser *ctxt;
  44 };
  45
  46 /* ********************************************************************** */
  47
  48 #if 0
  49
  50 /* well we odnt use this stuff yet */
  51
  52 #define ARRAY_LEN(x) (sizeof(x)/sizeof((x)[0]))
  53
  54 static struct {
  55         char *element;
  56         char *remap;
  57 } map_start[] = {
  58         { "p", "\n\n" },
  59         { "br", "\n" },
  60         { "h1", "\n" }, { "h2", "\n" }, { "h3", "\n" }, { "h4", "\n" }, { "h5", "\n" }, { "h6", "\n" },
  61 };
  62
  63
  64 static struct {
  65         char *element;
  66         char *remap;
  67 } map_end[] = {
  68         { "h1", "\n" }, { "h2", "\n" }, { "h3", "\n" }, { "h4", "\n" }, { "h5", "\n" }, { "h6", "\n" },
  69 };
  70 #endif
  71
  72
  73 /* ********************************************************************** */
  74
  75
  76 CamelType
  77 camel_mime_filter_html_get_type (void)
  78 {
  79         static CamelType type = CAMEL_INVALID_TYPE;
  80
  81         if (type == CAMEL_INVALID_TYPE) {
  82                 type = camel_type_register (camel_mime_filter_get_type (), "CamelMimeFilterHTML",
  83                                             sizeof (CamelMimeFilterHTML),
  84                                             sizeof (CamelMimeFilterHTMLClass),
  85                                             (CamelObjectClassInitFunc) camel_mime_filter_html_class_init,
  86                                             NULL,
  87                                             (CamelObjectInitFunc) camel_mime_filter_html_init,
  88                                             (CamelObjectFinalizeFunc) camel_mime_filter_html_finalize);
  89         }
  90
  91         return type;
  92 }
  93
  94 static void
  95 camel_mime_filter_html_finalize(CamelObject *o)
  96 {
  97         CamelMimeFilterHTML *f = (CamelMimeFilterHTML *)o;
  98
  99         camel_object_unref((CamelObject *)f->priv->ctxt);
 100         g_free(f->priv);
 101 }
 102
 103 static void
 104 camel_mime_filter_html_init       (CamelObject *o)
 105 {
 106         CamelMimeFilterHTML *f = (CamelMimeFilterHTML *)o;
 107
 108         f->priv = g_malloc0(sizeof(*f->priv));
 109         f->priv->ctxt = camel_html_parser_new();
 110 }
 111
 112 static void
 113 run(CamelMimeFilter *mf, char *in, size_t inlen, size_t prespace, char **out, size_t *outlenptr, size_t *outprespace, int last)
 114 {
 115         CamelMimeFilterHTML *f = (CamelMimeFilterHTML *) mf;
 116         camel_html_parser_t state;
 117         char *outp;
 118
 119         d(printf("converting html:\n%.*s\n", (int)inlen, in));
 120
 121         /* We should generally shrink the data, but this'll do */
 122         camel_mime_filter_set_size (mf, inlen * 2 + 256, FALSE);
 123         outp = mf->outbuf;
 124
 125         camel_html_parser_set_data (f->priv->ctxt, in, inlen, last);
 126         do {
 127                 const char *data;
 128                 int len;
 129
 130                 state = camel_html_parser_step(f->priv->ctxt, &data, &len);
 131
 132                 switch(state) {
 133                 case CAMEL_HTML_PARSER_DATA:
 134                 case CAMEL_HTML_PARSER_ENT:
 135                         memcpy(outp, data, len);
 136                         outp += len;
 137                         break;
 138                 case CAMEL_HTML_PARSER_ELEMENT:
 139                         /* FIXME: do some whitespace processing here */
 140                         break;
 141                 default:
 142                         /* ignore everything else */
 143                         break;
 144                 }
 145         } while (state != CAMEL_HTML_PARSER_EOF && state != CAMEL_HTML_PARSER_EOD);
 146
 147         *out = mf->outbuf;
 148         *outlenptr = outp - mf->outbuf;
 149         *outprespace = mf->outbuf - mf->outreal;
 150
 151         d(printf("converted html end:\n%.*s\n", (int)*outlenptr, *out));
 152 }
 153
 154 static void
 155 complete(CamelMimeFilter *mf, char *in, size_t len, size_t prespace, char **out, size_t *outlenptr, size_t *outprespace)
 156 {
 157         run(mf, in, len, prespace, out, outlenptr, outprespace, TRUE);
 158 }
 159
 160 static void
 161 filter(CamelMimeFilter *mf, char *in, size_t len, size_t prespace, char **out, size_t *outlenptr, size_t *outprespace)
 162 {
 163         run(mf, in, len, prespace, out, outlenptr, outprespace, FALSE);
 164 }
 165
 166 static void
 167 reset(CamelMimeFilter *mf)
 168 {
 169         CamelMimeFilterHTML *f = (CamelMimeFilterHTML *)mf;
 170
 171         camel_object_unref((CamelObject *)f->priv->ctxt);
 172         f->priv->ctxt = camel_html_parser_new();
 173 }
 174
 175 static void
 176 camel_mime_filter_html_class_init (CamelMimeFilterHTMLClass *klass)
 177 {
 178         CamelMimeFilterClass *filter_class = (CamelMimeFilterClass *) klass;
 179
 180         camel_mime_filter_html_parent = CAMEL_MIME_FILTER_CLASS (camel_type_get_global_classfuncs (camel_mime_filter_get_type ()));
 181
 182         filter_class->reset = reset;
 183         filter_class->filter = filter;
 184         filter_class->complete = complete;
 185 }
 186
 187
 188 /**
 189  * camel_mime_filter_html_new:
 190  *
 191  * Create a new #CamelMimeFilterHTML object.
 192  *
 193  * Returns a new #CamelMimeFilterHTML object
 194  **/
 195 CamelMimeFilterHTML *
 196 camel_mime_filter_html_new (void)
 197 {
 198         CamelMimeFilterHTML *new = CAMEL_MIME_FILTER_HTML ( camel_object_new (camel_mime_filter_html_get_type ()));
 199         return new;
 200 }