* $Id$
*
* Functions which handle translation between HTML and plain text
- * Copyright (c) 2000-2001 by Art Cancro and others. This program is
+ * Copyright (c) 2000-2005 by Art Cancro and others. This program is
* released under the terms of the GNU General Public License.
*/
-#ifdef DLL_EXPORT
-#define IN_LIBCIT
-#endif
-
#include "sysdep.h"
#include <stdlib.h>
#include <unistd.h>
* screenwidth = desired output screenwidth
* do_citaformat = set to 1 to indent newlines with spaces
*/
-char *html_to_ascii(char *inputmsg, int screenwidth, int do_citaformat) {
+char *html_to_ascii(char *inputmsg, int msglen, int screenwidth, int do_citaformat) {
char inbuf[SIZ];
char outbuf[SIZ];
char tag[1024];
char *outptr;
size_t outptr_buffer_size;
size_t output_len = 0;
- int i, j, ch, did_out, rb;
+ int i, j, ch, did_out, rb, scanch;
int nest = 0; /* Bracket nesting level */
+ int blockquote = 0; /* BLOCKQUOTE nesting level */
+ int bytes_processed = 0;
inptr = inputmsg;
strcpy(inbuf, "");
strcpy(outbuf, "");
+ if (msglen == 0) msglen = strlen(inputmsg);
outptr_buffer_size = strlen(inptr) + SIZ;
outptr = malloc(outptr_buffer_size);
done_reading = 1;
}
+ ++bytes_processed;
+ if (bytes_processed > msglen) {
+ done_reading = 1;
+ }
+
}
/* Do some parsing */
if (inbuf[i]==10) inbuf[i]=32;
if (inbuf[i]==13) inbuf[i]=32;
if (inbuf[i]==9) inbuf[i]=32;
+ /*** we like foreign characters now.
if ((inbuf[i]<32) || (inbuf[i]>126)) {
inbuf[i] = '?';
- /* strcpy(&inbuf[i], &inbuf[i+1]); */
- }
+ } */
}
for (i=0; i<strlen(inbuf); ++i) {
while ((inbuf[i]==32)&&(inbuf[i+1]==32))
strcpy(tag, "");
}
- else if (ch == '>') {
+ else if (ch == '>') { /* We have a tag. */
if (nest > 0) --nest;
+
+ /* Unqualify the tag (truncate at first space) */
+ if (strchr(tag, ' ') != NULL) {
+ strcpy(strchr(tag, ' '), "");
+ }
if (!strcasecmp(tag, "P")) {
strcat(outbuf, "\n\n");
strcat(outbuf, "\n\n");
}
+ if (!strcasecmp(tag, "LI")) {
+ strcat(outbuf, "\n * ");
+ }
+
+ else if (!strcasecmp(tag, "/UL")) {
+ strcat(outbuf, "\n\n");
+ }
+
else if (!strcasecmp(tag, "H1")) {
strcat(outbuf, "\n\n");
}
strcat(outbuf, "\n");
}
+ else if (!strcasecmp(tag, "BLOCKQUOTE")) {
+ strcat(outbuf, "\n\n <<\n");
+ ++blockquote;
+ }
+
+ else if (!strcasecmp(tag, "/BLOCKQUOTE")) {
+ strcat(outbuf, "\n >>\n\n");
+ --blockquote;
+ }
+
}
else if ((nest > 0) && (strlen(tag)<(sizeof(tag)-1))) {
/* Convert &; tags to the forbidden characters */
if (strlen(outbuf)>0) for (i=0; i<strlen(outbuf); ++i) {
+ /* Character entity references */
if (!strncasecmp(&outbuf[i], " ", 6)) {
outbuf[i] = ' ';
strcpy(&outbuf[i+1], &outbuf[i+6]);
}
+ if (!strncasecmp(&outbuf[i], " ", 6)) {
+ outbuf[i] = ' ';
+ strcpy(&outbuf[i+1], &outbuf[i+6]);
+ }
+
+ if (!strncasecmp(&outbuf[i], " ", 6)) {
+ outbuf[i] = ' ';
+ strcpy(&outbuf[i+1], &outbuf[i+6]);
+ }
+
+ if (!strncasecmp(&outbuf[i], " ", 8)) {
+ outbuf[i] = ' ';
+ strcpy(&outbuf[i+1], &outbuf[i+8]);
+ }
+
else if (!strncasecmp(&outbuf[i], "<", 4)) {
outbuf[i] = '<';
strcpy(&outbuf[i+1], &outbuf[i+4]);
strcpy(&outbuf[i+1], &outbuf[i+6]);
}
+ else if (!strncasecmp(&outbuf[i], "‘", 7)) {
+ outbuf[i] = '`';
+ strcpy(&outbuf[i+1], &outbuf[i+7]);
+ }
+
+ else if (!strncasecmp(&outbuf[i], "’", 7)) {
+ outbuf[i] = '\'';
+ strcpy(&outbuf[i+1], &outbuf[i+7]);
+ }
+
else if (!strncasecmp(&outbuf[i], "©", 6)) {
outbuf[i] = '(';
outbuf[i+1] = 'c';
strcpy(&outbuf[i+3], &outbuf[i+6]);
}
+ else if (!strncasecmp(&outbuf[i], "…", 8)) {
+ outbuf[i] = '.';
+ outbuf[i+1] = '.';
+ outbuf[i+2] = '.';
+ strcpy(&outbuf[i+3], &outbuf[i+8]);
+ }
+
+ else if (!strncasecmp(&outbuf[i], "™", 7)) {
+ outbuf[i] = '(';
+ outbuf[i+1] = 't';
+ outbuf[i+2] = 'm';
+ outbuf[i+3] = ')';
+ strcpy(&outbuf[i+4], &outbuf[i+7]);
+ }
+
else if (!strncasecmp(&outbuf[i], "®", 5)) {
outbuf[i] = '(';
outbuf[i+1] = 'r';
strcpy(&outbuf[i+3], &outbuf[i+5]);
}
+ else if (!strncasecmp(&outbuf[i], "¼", 8)) {
+ outbuf[i] = '1';
+ outbuf[i+1] = '/';
+ outbuf[i+2] = '4';
+ strcpy(&outbuf[i+3], &outbuf[i+8]);
+ }
+
+ else if (!strncasecmp(&outbuf[i], "½", 8)) {
+ outbuf[i] = '1';
+ outbuf[i+1] = '/';
+ outbuf[i+2] = '2';
+ strcpy(&outbuf[i+3], &outbuf[i+8]);
+ }
+
+ else if (!strncasecmp(&outbuf[i], "¾", 8)) {
+ outbuf[i] = '3';
+ outbuf[i+1] = '/';
+ outbuf[i+2] = '4';
+ strcpy(&outbuf[i+3], &outbuf[i+8]);
+ }
+
+ else if (!strncasecmp(&outbuf[i], "–", 7)) {
+ outbuf[i] = '-';
+ outbuf[i+1] = '-';
+ strcpy(&outbuf[i+2], &outbuf[i+7]);
+ }
+
+ else if (!strncasecmp(&outbuf[i], "—", 7)) {
+ outbuf[i] = '-';
+ outbuf[i+1] = '-';
+ outbuf[i+2] = '-';
+ strcpy(&outbuf[i+3], &outbuf[i+7]);
+ }
+
+ /* two-digit decimal equivalents */
+ else if ((!strncmp(&outbuf[i], "&#", 2))
+ && (outbuf[i+4] == ';') ) {
+ scanch = 0;
+ sscanf(&outbuf[i+2], "%02d", &scanch);
+ outbuf[i] = scanch;
+ strcpy(&outbuf[i+1], &outbuf[i+5]);
+ }
+
+ /* three-digit decimal equivalents */
+ else if ((!strncmp(&outbuf[i], "&#", 2))
+ && (outbuf[i+5] == ';') ) {
+ scanch = 0;
+ sscanf(&outbuf[i+2], "%03d", &scanch);
+ outbuf[i] = scanch;
+ strcpy(&outbuf[i+1], &outbuf[i+6]);
+ }
+
}
/* Make sure the output buffer is big enough */
/* Output any lines terminated with hard line breaks */
do {
did_out = 0;
- if (strlen(outbuf)>0)
+ if (strlen(outbuf)>0) {
for (i = 0; i<strlen(outbuf); ++i) {
if ( (i<(screenwidth-2)) && (outbuf[i]=='\n')) {
did_out = 1;
}
}
+ }
} while (did_out);
/* Add soft line breaks */
- if (strlen(outbuf) > (screenwidth - 2)) {
+ if (strlen(outbuf) > (screenwidth - 2 )) {
rb = (-1);
for (i=0; i<(screenwidth-2); ++i) {
if (outbuf[i]==32) rb = i;