add url encoding

CHANGES

@@ -4,6 +4,7 @@
   implement fmt_uuencoded and scan_uuencoded.
   implement fmt_base64, scan_base64.
   implement fmt_quotedprintable, scan_quotedprintable
+  implement fmt_urlencoded, scan_urlencoded
 
 0.11:
   fix fmt_long (didn't count the '-'), which in turn broke

Makefile

@@ -62,8 +62,7 @@ $(BUFFER_OBJS) $(MMAP_OBJS) $(TEXTCODE_OBJS)
 	ar cr $@ $^
 	-ranlib $@
 
-t: t.o socket.a stralloc.a buffer.a scan.a uint.a mmap.a open.a fmt.a \
+t: t.o libowfat.a
-str.a byte.a textcode.a
 	$(DIET) $(CC) -g -o $@ $^
 
 .PHONY: clean tar install rename

t.c

@@ -20,10 +20,10 @@ int main(int argc,char* argv[]) {
   char buf[100];
   char buf2[100];
   unsigned int len,len2;
-  buf[fmt_quotedprintable(buf,"zunächst einmal vielen Dank für die MUTT FAQ",44)]=0;
+  buf[fmt_urlencoded(buf,"http://localhost/~fefe",22)]=0;
   buffer_puts(buffer_1,buf);
   buffer_putsflush(buffer_1,"\n");
-  if ((buf[len2=scan_quotedprintable(buf,buf2,&len)])!=0) {
+  if ((buf[len2=scan_urlencoded(buf,buf2,&len)])!=0) {
     buffer_putsflush(buffer_2,"parse error!\n");
     return 1;
   }

textcode.h

@@ -6,7 +6,7 @@
 unsigned int fmt_uuencoded(char* dest,const char* src,unsigned int len);
 unsigned int fmt_base64(char* dest,const char* src,unsigned int len);
 unsigned int fmt_quotedprintable(char* dest,const char* src,unsigned int len);
-unsigned int fmt_urlencode(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);
 
 /* These read one line from src, decoded it, and write the result to
@@ -15,7 +15,7 @@ unsigned int fmt_yenc(char* dest,const char* src,unsigned int len);
 unsigned int scan_uuencoded(const char *src,char *dest,unsigned int *destlen);
 unsigned int scan_base64(const char *src,char *dest,unsigned int *destlen);
 unsigned int scan_quotedprintable(const char *src,char *dest,unsigned int *destlen);
-unsigned int scan_urlencode(const char *src,char *dest,unsigned int *destlen);
+unsigned int scan_urlencoded(const char *src,char *dest,unsigned int *destlen);
 unsigned int scan_yenc(const char *src,char *dest,unsigned int *destlen);
 
 extern const char base64[64];

textcode/fmt_urlencoded.c

@@ -0,0 +1,27 @@
+#include "fmt.h"
+#include "textcode.h"
+#include "str.h"
+#include "haveinline.h"
+
+static inline int tohex(char c) {
+  return c>9?c-10+'A':c+'0';
+}
+
+unsigned int fmt_urlencoded(char* dest,const char* src,unsigned int len) {
+  const char unsafe[]=" %<>\"#{}|\\^~[]`;/?:@=&";
+  register const unsigned char* s=(const unsigned char*) src;
+  unsigned long written=0,i;
+  for (i=0; i<len; ++i) {
+    if (s[i]&0x80 || unsafe[str_chr(unsafe,s[i])]==s[i]) {
+      if (dest) {
+	dest[written]='%';
+	dest[written+1]=tohex(s[i]>>4);
+	dest[written+2]=tohex(s[i]&15);
+      }
+      written+=3;
+    } else {
+      if (dest) dest[written]=s[i]; ++written;
+    }
+  }
+  return written;
+}

textcode/scan_urlencoded.c

@@ -0,0 +1,31 @@
+#include "fmt.h"
+#include "textcode.h"
+#include "haveinline.h"
+
+static inline int fromhex(char c) {
+  if (c>='0' && c<='9') return c-'0';
+  if (c>='A' && c<='F') return c-'A'+10;
+  if (c>='a' && c<='f') return c-'a'+10;
+  return -1;
+}
+
+unsigned int scan_urlencoded(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]=='%') {
+      int j=fromhex(s[i+1]);
+      if (j<0) break;
+      dest[written]=j<<4;
+      j=fromhex(s[i+2]);
+      if (j<0) break;
+      dest[written]|=j;
+      i+=2;
+    } else {
+      dest[written]=s[i];
+    }
+    ++written;
+  }
+  *destlen=written;
+  return i;
+}