add html encoders

add vd (mldonkey support program) as test for socket API

CHANGES

@@ -10,6 +10,7 @@
   make install forgot to install ndelay.h
   fix typos in several man pages (Hynek Schlawack)
   add stralloc versions of textcode API (Kai Ruemmler)
+  add html to textcode ('<' to '&lt;' etc)
 
 0.14:
   avoid bus errors in byte_copy

test/vd.c

@@ -0,0 +1,36 @@
+#include "socket.h"
+#include "buffer.h"
+#include <unistd.h>
+#include <stdio.h>
+
+main() {
+  int s=socket_tcp4();
+  char line[1024];
+  char buf[4096];
+  int l;
+  int header=1;
+  buffer filein;
+  buffer_init(&filein,read,s,buf,sizeof buf);
+  if (socket_connect4(s,"\x7f\x00\x00\x01",4000)) {
+    perror("connect");
+    return 1;
+  }
+  write(s,"vd\nq\n",5);
+  for (;;) {
+    line[0]=0;
+    if ((l=buffer_getline(&filein,line,(sizeof line)-1))==0 && line[l]!='\n')
+      break;
+    else {
+      line[l+1]=0;
+      if (!header) {
+	if (strcmp(line,"\e[7mMLdonkey command-line:\e[2;37;0m\n") &&
+	    strcmp(line,"\e[2;37;0m\e[7mMLdonkey command-line:\e[2;37;0m\n") &&
+	    strncmp(line,"> ",2))
+	  buffer_put(buffer_1,line,l+1);
+      }
+      if (!strcmp(line,"Use \e[31m?\e[2;37;0m for help\n")) header=0;
+      if (!strcmp(line,"Use ? for help\n")) header=0;
+    }
+  }
+  buffer_flush(buffer_1);
+}

textcode.h

@@ -9,6 +9,8 @@ unsigned int fmt_quotedprintable(char* dest,const char* src,unsigned int len);
 unsigned int fmt_urlencoded(char* dest,const char* src,unsigned int len);
 unsigned int fmt_yenc(char* dest,const char* src,unsigned int len);
 unsigned int fmt_hexdump(char* dest,const char* src,unsigned int len);
+/* this changes '<' to '&lt;' and '&' to '&amp;' */
+unsigned int fmt_html(char* dest,const char* src,unsigned int len);
 
 /* These read one line from src, decoded it, and write the result to
  * dest.  The number of decoded bytes is written to destlen.  dest
@@ -19,6 +21,7 @@ unsigned int scan_quotedprintable(const char *src,char *dest,unsigned int *destl
 unsigned int scan_urlencoded(const char *src,char *dest,unsigned int *destlen);
 unsigned int scan_yenc(const char *src,char *dest,unsigned int *destlen);
 unsigned int scan_hexdump(const char *src,char *dest,unsigned int *destlen);
+unsigned int scan_html(const char *src,char *dest,unsigned int *destlen);
 
 #ifdef STRALLOC_H
 /* These take len bytes from src and write them in encoded form to sa.

textcode/fmt_html.c

@@ -0,0 +1,24 @@
+#include "fmt.h"
+#include "textcode.h"
+#include "str.h"
+#include "haveinline.h"
+
+unsigned int fmt_html(char* dest,const char* src,unsigned int len) {
+  register const unsigned char* s=(const unsigned char*) src;
+  unsigned long written=0,i;
+  const char* seq;
+  for (i=0; i<len; ++i) {
+    switch (s[i]) {
+    case '&': seq="&amp;"; goto doit;
+    case '<': seq="&lt;"; goto doit;
+    case '>': seq="&gt;"; goto doit;
+    case '\n':
+	seq="<br>";
+      doit:
+	written+=fmt_str(dest?dest+written:0,"&gt;");
+	break;
+    default: if (dest) dest[written]=s[i]; ++written; break;
+    }
+  }
+  return written;
+}

textcode/scan_html.c

@@ -0,0 +1,36 @@
+#include "fmt.h"
+#include "textcode.h"
+#include "haveinline.h"
+#include "case.h"
+
+unsigned int scan_html(const char *src,char *dest,unsigned int *destlen) {
+  register const unsigned char* s=(const unsigned char*) src;
+  unsigned long written=0,i;
+  for (i=0; s[i]; ++i) {
+    if (s[i]=='&') {
+      if (case_starts(s+i+1,"amp;")) {
+	dest[written]='&';
+	i+=4;
+      } else if (case_starts(s+i+1,"lt;")) {
+	dest[written]='<';
+	i+=3;
+      } else if (case_starts(s+i+1,"gt;")) {
+	dest[written]='>';
+	i+=3;
+      }
+    } else if (s[i]=='<') {
+      if (case_starts(s+i+1,"br>")) {
+	dest[written]='\n';
+	i+=3;
+      } else if (case_starts(s+i+1,"p>")) {
+	dest[written]='\n'; ++written;
+	dest[written]='\n';
+	i+=3;
+      }
+    } else
+      dest[written]=s[i];
+    ++written;
+  }
+  *destlen=written;
+  return i;
+}