X-Git-Url: https://git.distorted.org.uk/~mdw/disorder/blobdiff_plain/5818980adea84ce4cd92a1b4549e9806c9d56c58..82c01b317cd1892f4376c68be88a74f971493428:/lib/mime.c

diff --git a/lib/mime.c b/lib/mime.c
index 87fc43a..8776647 100644
--- a/lib/mime.c
+++ b/lib/mime.c
@@ -34,6 +34,7 @@
 #include "vector.h"
 #include "hex.h"
 #include "log.h"
+#include "base64.h"
 
 /** @brief Match whitespace characters */
 static int whitespace(int c) {
@@ -72,7 +73,7 @@ static int tspecial(int c) {
   }
 }
 
-/** @brief Match RFC2616 seprator characters */
+/** @brief Match RFC2616 separator characters */
 static int http_separator(int c) {
   switch(c) {
   case '(':
@@ -106,7 +107,9 @@ static int iscrlf(const char *ptr) {
 }
 
 /** @brief Skip whitespace
+ * @param s Pointer into string
  * @param rfc822_comments If true, skip RFC822 nested comments
+ * @return Pointer into string after whitespace
  */
 static const char *skipwhite(const char *s, int rfc822_comments) {
   int c, depth;
@@ -206,8 +209,10 @@ static const char *parsetoken(const char *s, char **valuep,
  * @param s Start of field
  * @param typep Where to store type
  * @param parameternamep Where to store parameter name
- * @param parameternvaluep Wher to store parameter value
+ * @param parametervaluep Wher to store parameter value
  * @return 0 on success, non-0 on error
+ *
+ * See <a href="http://tools.ietf.org/html/rfc2045#section-5">RFC 2045 s5</a>.
  */
 int mime_content_type(const char *s,
 		      char **typep,
@@ -253,7 +258,12 @@ int mime_content_type(const char *s,
  * @param s Start of message
  * @param callback Called for each header field
  * @param u Passed to callback
- * @return Pointer to decoded body (might be in original string)
+ * @return Pointer to decoded body (might be in original string), or NULL on error
+ *
+ * This does an RFC 822-style parse and honors Content-Transfer-Encoding as
+ * described in <a href="http://tools.ietf.org/html/rfc2045#section-6">RFC 2045
+ * s6</a>.  @p callback is called for each header field encountered, in order,
+ * with ASCII characters in the header name forced to lower case.
  */
 const char *mime_parse(const char *s,
 		       int (*callback)(const char *name, const char *value,
@@ -290,6 +300,7 @@ const char *mime_parse(const char *s,
   return s;
 }
 
+/** @brief Match the boundary string */
 static int isboundary(const char *ptr, const char *boundary, size_t bl) {
   return (ptr[0] == '-'
 	  && ptr[1] == '-'
@@ -300,6 +311,7 @@ static int isboundary(const char *ptr, const char *boundary, size_t bl) {
 		  && (iscrlf(ptr + bl + 4) || *(ptr + bl + 4) == 0))));
 }
 
+/** @brief Match the final boundary string */
 static int isfinal(const char *ptr, const char *boundary, size_t bl) {
   return (ptr[0] == '-'
 	  && ptr[1] == '-'
@@ -311,10 +323,14 @@ static int isfinal(const char *ptr, const char *boundary, size_t bl) {
 
 /** @brief Parse a multipart MIME body
  * @param s Start of message
- * @param callback CAllback for each part
+ * @param callback Callback for each part
  * @param boundary Boundary string
  * @param u Passed to callback
  * @return 0 on success, non-0 on error
+ * 
+ * See <a href="http://tools.ietf.org/html/rfc2046#section-5.1">RFC 2046
+ * s5.1</a>.  @p callback is called for each part (not yet decoded in any way)
+ * in succession; you should probably call mime_parse() for each part.
  */
 int mime_multipart(const char *s,
 		   int (*callback)(const char *s, void *u),
@@ -346,10 +362,13 @@ int mime_multipart(const char *s,
 
 /** @brief Parse an RFC2388-style content-disposition field
  * @param s Start of field
- * @param typep Where to store type
+ * @param dispositionp Where to store disposition
  * @param parameternamep Where to store parameter name
- * @param parameternvaluep Wher to store parameter value
+ * @param parametervaluep Wher to store parameter value
  * @return 0 on success, non-0 on error
+ *
+ * See <a href="http://tools.ietf.org/html/rfc2388#section-3">RFC 2388 s3</a>
+ * and <a href="http://tools.ietf.org/html/rfc2183">RFC 2183</a>.
  */
 int mime_rfc2388_content_disposition(const char *s,
 				     char **dispositionp,
@@ -388,6 +407,9 @@ int mime_rfc2388_content_disposition(const char *s,
 /** @brief Convert MIME quoted-printable
  * @param s Quoted-printable data
  * @return Decoded data
+ *
+ * See <a href="http://tools.ietf.org/html/rfc2045#section-6.7">RFC 2045
+ * s6.7</a>.
  */
 char *mime_qp(const char *s) {
   struct dynstr d;
@@ -431,91 +453,12 @@ char *mime_qp(const char *s) {
   return d.vec;
 }
 
-static const char mime_base64_table[] =
-  "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
-
-/** @brief Convert MIME base64
- * @param s base64 data
- * @return Decoded data
- */
-char *mime_base64(const char *s, size_t *nsp) {
-  struct dynstr d;
-  const char *t;
-  int b[4], n, c;
-
-  dynstr_init(&d);
-  n = 0;
-  while((c = (unsigned char)*s++)) {
-    if((t = strchr(mime_base64_table, c))) {
-      b[n++] = t - mime_base64_table;
-      if(n == 4) {
-	dynstr_append(&d, (b[0] << 2) + (b[1] >> 4));
-	dynstr_append(&d, (b[1] << 4) + (b[2] >> 2));
-	dynstr_append(&d, (b[2] << 6) + b[3]);
-	n = 0;
-      }
-    } else if(c == '=') {
-      if(n >= 2) {
-	dynstr_append(&d, (b[0] << 2) + (b[1] >> 4));
-	if(n == 3)
-	  dynstr_append(&d, (b[1] << 4) + (b[2] >> 2));
-      }
-      break;
-    }
-  }
-  if(nsp)
-    *nsp = d.nvec;
-  dynstr_terminate(&d);
-  return d.vec;
-}
-
-/** @brief Convert a binary string to base64
- * @param s Bytes to convert
- * @param ns Number of bytes to convert
- * @return Encoded data
- *
- * This function does not attempt to split up lines.
- */
-char *mime_to_base64(const uint8_t *s, size_t ns) {
-  struct dynstr d[1];
-
-  dynstr_init(d);
-  while(ns >= 3) {
-    /* Input bytes with output bits: AAAAAABB BBBBCCCC CCDDDDDD */
-    /* Output bytes with input bits: 000000 001111 111122 222222 */
-    dynstr_append(d, mime_base64_table[s[0] >> 2]);
-    dynstr_append(d, mime_base64_table[((s[0] & 3) << 4)
-				       + (s[1] >> 4)]);
-    dynstr_append(d, mime_base64_table[((s[1] & 15) << 2)
-				       + (s[2] >> 6)]);
-    dynstr_append(d, mime_base64_table[s[2] & 63]);
-    ns -= 3;
-    s += 3;
-  }
-  if(ns > 0) {
-    dynstr_append(d, mime_base64_table[s[0] >> 2]);
-    switch(ns) {
-    case 1:
-      dynstr_append(d, mime_base64_table[(s[0] & 3) << 4]);
-      dynstr_append(d, '=');
-      dynstr_append(d, '=');
-      break;
-    case 2:
-      dynstr_append(d, mime_base64_table[((s[0] & 3) << 4)
-					 + (s[1] >> 4)]);
-      dynstr_append(d, mime_base64_table[(s[1] & 15) << 2]);
-      dynstr_append(d, '=');
-      break;
-    }
-  }
-  dynstr_terminate(d);
-  return d->vec;
-}
-
 /** @brief Parse a RFC2109 Cookie: header
  * @param s Header field value
  * @param cd Where to store result
  * @return 0 on success, non-0 on error
+ *
+ * See <a href="http://tools.ietf.org/html/rfc2109">RFC 2109</a>.
  */
 int parse_cookie(const char *s,
 		 struct cookiedata *cd) {
@@ -588,6 +531,132 @@ const struct cookie *find_cookie(const struct cookiedata *cd,
   return 0;
 }
 
+/** @brief RFC822 quoting
+ * @param s String to quote
+ * @param force If non-0, always quote
+ * @return Possibly quoted string
+ */
+char *quote822(const char *s, int force) {
+  const char *t;
+  struct dynstr d[1];
+  int c;
+
+  if(!force) {
+    /* See if we need to quote */
+    for(t = s; (c = (unsigned char)*t); ++t) {
+      if(tspecial(c) || http_separator(c) || whitespace(c))
+	break;
+    }
+    if(*t)
+      force = 1;
+  }
+
+  if(!force)
+    return xstrdup(s);
+
+  dynstr_init(d);
+  dynstr_append(d, '"');
+  for(t = s; (c = (unsigned char)*t); ++t) {
+    if(c == '"' || c == '\\')
+      dynstr_append(d, '\\');
+    dynstr_append(d, c);
+  }
+  dynstr_append(d, '"');
+  dynstr_terminate(d);
+  return d->vec;
+}
+
+/** @brief Return true if @p ptr points at trailing space */
+static int is_trailing_space(const char *ptr) {
+  if(*ptr == ' ' || *ptr == '\t') {
+    while(*ptr == ' ' || *ptr == '\t')
+      ++ptr;
+    return *ptr == '\n' || *ptr == 0;
+  } else
+    return 0;
+}
+
+/** @brief Encoding text as quoted-printable
+ * @param text String to encode
+ * @return Encoded string
+ *
+ * See <a href="http://tools.ietf.org/html/rfc2045#section-6.7">RFC2045
+ * s6.7</a>.
+ */
+char *mime_to_qp(const char *text) {
+  struct dynstr d[1];
+  int linelength = 0;			/* length of current line */
+  char buffer[10];
+
+  dynstr_init(d);
+  /* The rules are:
+   * 1. Anything except newline can be replaced with =%02X
+   * 2. Newline, 33-60 and 62-126 stand for themselves (i.e. not '=')
+   * 3. Non-trailing space/tab stand for themselves.
+   * 4. Output lines are limited to 76 chars, with =<newline> being used
+   *    as a soft line break
+   * 5. Newlines aren't counted towards the 76 char limit.
+   */
+  while(*text) {
+    const int c = (unsigned char)*text;
+    if(c == '\n') {
+      /* Newline stands as itself */
+      dynstr_append(d, '\n');
+      linelength = 0;
+    } else if((c >= 33 && c <= 126 && c != '=')
+	      || ((c == ' ' || c == '\t')
+		  && !is_trailing_space(text))) {
+      /* Things that can stand for themselves  */
+      dynstr_append(d, c);
+      ++linelength;
+    } else {
+      /* Anything else that needs encoding */
+      snprintf(buffer, sizeof buffer, "=%02X", c);
+      dynstr_append_string(d, buffer);
+      linelength += 3;
+    }
+    ++text;
+    if(linelength > 73 && *text && *text != '\n') {
+      /* Next character might overflow 76 character limit if encoded, so we
+       * insert a soft break */
+      dynstr_append_string(d, "=\n");
+      linelength = 0;
+    }
+  }
+  /* Ensure there is a final newline */
+  if(linelength)
+    dynstr_append(d, '\n');
+  /* That's all */
+  dynstr_terminate(d);
+  return d->vec;
+}
+
+/** @brief Encode text
+ * @param text Underlying UTF-8 text
+ * @param charsetp Where to store charset string
+ * @param encodingp Where to store encoding string
+ * @return Encoded text (might be @ref text)
+ */
+const char *mime_encode_text(const char *text,
+			     const char **charsetp,
+			     const char **encodingp) {
+  const char *ptr;
+
+  /* See if there are in fact any non-ASCII characters */
+  for(ptr = text; *ptr; ++ptr)
+    if((unsigned char)*ptr >= 128)
+      break;
+  if(!*ptr) {
+    /* Plain old ASCII, no encoding required */
+    *charsetp = "us-ascii";
+    *encodingp = "7bit";
+    return text;
+  }
+  *charsetp = "utf-8";
+  *encodingp = "quoted-printable";
+  return mime_to_qp(text);
+}
+
 /*
 Local Variables:
 c-basic-offset:2