X-Git-Url: https://code.citadel.org/?a=blobdiff_plain;f=libcitadel%2Flib%2Fhtml_to_ascii.c;h=45beb2cbb079b97eeaa2041df905e8676ce3145a;hb=21185d642546b081924a541bad2adcaad9416fca;hp=ccf074a48574e8dbb4b15e71cb38da1d15eae401;hpb=56d69e5d8434e98835a2582c59b771ba69475431;p=citadel.git

diff --git a/libcitadel/lib/html_to_ascii.c b/libcitadel/lib/html_to_ascii.c
index ccf074a48..45beb2cbb 100644
--- a/libcitadel/lib/html_to_ascii.c
+++ b/libcitadel/lib/html_to_ascii.c
@@ -1,8 +1,8 @@
 /*
  * Functions which handle translation between HTML and plain text
- * Copyright (c) 2000-2010 by the citadel.org team
+ * Copyright (c) 2000-2018 by the citadel.org team
  *
- * This program is free software; you can redistribute it and/or modify
+ * This program is open source software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
  * the Free Software Foundation; either version 3 of the License, or
  * (at your option) any later version.
@@ -47,15 +47,14 @@
  *
  * inputmsg      = pointer to raw HTML message
  * screenwidth   = desired output screenwidth
- * do_citaformat = set to 1 to indent newlines with spaces
  */
-char *html_to_ascii(char *inputmsg, int msglen, int screenwidth, int do_citaformat) {
+char *html_to_ascii(const char *inputmsg, int msglen, int screenwidth) {
 	char inbuf[SIZ];
 	int inbuf_len = 0;
 	char outbuf[SIZ];
 	char tag[1024];
 	int done_reading = 0;
-	char *inptr;
+	const char *inptr;
 	char *outptr;
 	size_t outptr_buffer_size;
 	size_t output_len = 0;
@@ -67,6 +66,7 @@ char *html_to_ascii(char *inputmsg, int msglen, int screenwidth, int do_citaform
 	int bytes_processed = 0;
 	char nl[128];
 
+	tag[0] = '\0';
 	strcpy(nl, "\n");
 	inptr = inputmsg;
 	strcpy(inbuf, "");
@@ -197,6 +197,9 @@ char *html_to_ascii(char *inputmsg, int msglen, int screenwidth, int do_citaform
 					strcat(outbuf, nl);
 				}
 
+#if 0
+	These seemed like a good idea at the time, but it just makes a mess.
+
 				else if (
 					(!strcasecmp(tag, "B"))
 					|| (!strcasecmp(tag, "/B"))
@@ -204,7 +207,6 @@ char *html_to_ascii(char *inputmsg, int msglen, int screenwidth, int do_citaform
 					|| (!strcasecmp(tag, "/STRONG"))
 				) {
 					strcat(outbuf, "*");
-					
 				}
 
 				else if (
@@ -214,7 +216,6 @@ char *html_to_ascii(char *inputmsg, int msglen, int screenwidth, int do_citaform
 					|| (!strcasecmp(tag, "/EM"))
 				) {
 					strcat(outbuf, "/");
-					
 				}
 
 				else if (
@@ -222,8 +223,8 @@ char *html_to_ascii(char *inputmsg, int msglen, int screenwidth, int do_citaform
 					|| (!strcasecmp(tag, "/U"))
 				) {
 					strcat(outbuf, "_");
-					
 				}
+#endif
 
 				else if (!strcasecmp(tag, "BR")) {
 					strcat(outbuf, nl);
@@ -273,7 +274,7 @@ char *html_to_ascii(char *inputmsg, int msglen, int screenwidth, int do_citaform
 				tag[strlen(tag)] = ch;
 			}
 				
-			else if (!nest) {
+			else if ((!nest) && (styletag == 0)) {
 				outbuf[strlen(outbuf)+1] = 0;
 				outbuf[strlen(outbuf)] = ch;
 			}
@@ -469,9 +470,23 @@ char *html_to_ascii(char *inputmsg, int msglen, int screenwidth, int do_citaform
 				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], "–", 7)) {
+				outbuf[i] = '-';
+				strcpy(&outbuf[i+1], &outbuf[i+7]);
+			}
+
 			/* two-digit decimal equivalents */
-			else if ((!strncmp(&outbuf[i], "&#", 2))
-			      && (outbuf[i+4] == ';') ) {
+			else if (outbuf[i] == '&'       &&
+				 outbuf[i + 1] == '#'   &&
+				 isdigit(outbuf[i + 2]) && 
+				 isdigit(outbuf[i + 3]) &&
+				 (outbuf[i+4] == ';') ) 
+			{
 				scanch = 0;
 				sscanf(&outbuf[i+2], "%02d", &scanch);
 				outbuf[i] = scanch;
@@ -479,14 +494,34 @@ char *html_to_ascii(char *inputmsg, int msglen, int screenwidth, int do_citaform
 			}
 
 			/* three-digit decimal equivalents */
-			else if ((!strncmp(&outbuf[i], "&#", 2))
-			      && (outbuf[i+5] == ';') ) {
+			else if (outbuf[i] == '&'       &&
+				 outbuf[i + 1] == '#'   &&
+				 isdigit(outbuf[i + 2]) && 
+				 isdigit(outbuf[i + 3]) && 
+				 isdigit(outbuf[i + 4]) &&
+				 (outbuf[i + 5] == ';') ) 
+			{
 				scanch = 0;
 				sscanf(&outbuf[i+2], "%03d", &scanch);
 				outbuf[i] = scanch;
 				strcpy(&outbuf[i+1], &outbuf[i+6]);
 			}
 
+			/* four-digit decimal equivalents */
+			else if (outbuf[i] == '&'       &&
+				 outbuf[i + 1] == '#'   &&
+				 isdigit(outbuf[i + 2]) && 
+				 isdigit(outbuf[i + 3]) && 
+				 isdigit(outbuf[i + 4]) &&
+				 isdigit(outbuf[i + 5]) &&
+				 (outbuf[i + 6] == ';') ) 
+			{
+				scanch = 0;
+				sscanf(&outbuf[i+2], "%04d", &scanch);
+				outbuf[i] = scanch;
+				strcpy(&outbuf[i+1], &outbuf[i+7]);
+			}
+
 		}
 
 		/* Make sure the output buffer is big enough */
@@ -508,11 +543,6 @@ char *html_to_ascii(char *inputmsg, int msglen, int screenwidth, int do_citaform
 					strncpy(&outptr[output_len], outbuf, i+1);
 					output_len += (i+1);
 
-					if (do_citaformat) {
-						strcpy(&outptr[output_len], " ");
-						++output_len;
-					}
-
 					strcpy(outbuf, &outbuf[i+1]);
 					i = 0;
 					did_out = 1;
@@ -532,10 +562,6 @@ char *html_to_ascii(char *inputmsg, int msglen, int screenwidth, int do_citaform
 				output_len += rb;
 				strcpy(&outptr[output_len], nl);
 				output_len += strlen(nl);
-				if (do_citaformat) {
-					strcpy(&outptr[output_len], " ");
-					++output_len;
-				}
 				strcpy(outbuf, &outbuf[rb+1]);
 			} else {
 				strncpy(&outptr[output_len], outbuf,
@@ -543,10 +569,6 @@ char *html_to_ascii(char *inputmsg, int msglen, int screenwidth, int do_citaform
 				output_len += (screenwidth-2);
 				strcpy(&outptr[output_len], nl);
 				output_len += strlen(nl);
-				if (do_citaformat) {
-					strcpy(&outptr[output_len], " ");
-					++output_len;
-				}
 				strcpy(outbuf, &outbuf[screenwidth-2]);
 			}
 		}