rohrpost

commit 138f1bf78d0f6f4cea849b9ef87268811c4b8b90
parent 5c204376213ec37399d68aca4abd0abfea2366bd
Author: Christoph Lohmann <20h@r-36.net>
Date:   Tue, 21 Aug 2012 16:22:45 +0200

Add charset guessing. This is lame but does it for sure.

Diffstat:
bin/rpopen
 | 
2
+-
bin/rpsyncmail
 | 
2
+-
meta.c
 | 
2
+-
mime.c
 | 
65
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++---
mime.h
 | 
1
+
scan.c
 | 
8
++++----
util.c
 | 
2
+-
view.c
 | 
4
++--

8 files changed, 73 insertions(+), 13 deletions(-)
diff --git a/bin/rpopen b/bin/rpopen
@@ -15,7 +15,7 @@ fi
 
 msgid="$*"
 
-url=`rpview $msgid | awk '/URL:/ {print $2}' | tr -d '\r'`
+url=`rpview $msgid | awk '/^URL:/ {print $2}' | tr -d '\r'`
 for i in $url
 do
 	echo $i
diff --git a/bin/rpsyncmail b/bin/rpsyncmail
@@ -1,6 +1,6 @@
 #!/bin/sh
 
-if [ "$1" = "mb" ];
+if [ "$1" = "mb" ] || [ $# -eq 0 ];
 then
 	dsync -R -u chrissi mirror chrissi@r-36.net
 else
diff --git a/meta.c b/meta.c
@@ -73,7 +73,7 @@ meta_headers2meta(mime_t *mime)
 
 	rhdr = llist_new();
 	forllist(meta->hdrs, header) {
-		enc = mime_decodeheader(header->data);
+		enc = mime_guessheader(header->data);
 		if (enc == NULL)
 			continue;
 
diff --git a/mime.c b/mime.c
@@ -295,6 +295,65 @@ mime_decodeheader(char *value)
 	return ret;
 }
 
+char *cstries[] = {
+	"utf-8",
+	"iso-8859-1",
+	"windows-1252",
+	"Shift_JIS",
+	"Big5",
+	"iso-8859-15",
+	NULL
+};
+
+char *
+mime_guesscharset(char *str)
+{
+	int i, eq;
+	char *itry;
+
+	for (i = 0; i < nelem(cstries); i++) {
+		itry = mime_iconv(str, cstries[i], cstries[i]);
+		if (itry == NULL)
+			continue;
+		eq = strcmp(str, itry);
+		free(itry);
+		if (!eq)
+			break;
+	}
+
+	return cstries[i];
+}
+
+char *
+mime_guessheader(char *value)
+{
+	char *nvalue, *gcs;
+
+	gcs = NULL;
+
+	nvalue = value;
+	if (!strisascii(value)) {
+		/*
+		 * Guessing begins. Some major MUA developers did not read any
+		 * RFCs.
+		 */
+
+		gcs = mime_guesscharset(value);
+		if (gcs != NULL) {
+			nvalue = mime_iconv(value, gcs, "utf-8");
+			if (nvalue == NULL) {
+				nvalue = value;
+				gcs = NULL;
+			}
+		}
+	}
+
+	value  = mime_decodeheader(nvalue);
+	if (gcs != NULL)
+		free(nvalue);
+	return value;
+}
+
 char *
 mime_decodeparam(char *value)
 {
@@ -1173,7 +1232,7 @@ mime_filename(mime_t *mime)
 		if (hdrp != NULL) {
 			name = llist_ciget(hdrp, "filename");
 			if (name != NULL && name->data != NULL) {
-				filename = mime_decodeheader(
+				filename = mime_guessheader(
 						(char *)name->data);
 			}
 		}
@@ -1192,7 +1251,7 @@ mime_filename(mime_t *mime)
 		if (hdrp != NULL) {
 			name = llist_ciget(hdrp, "name");
 			if (name != NULL && name->data != NULL) {
-				filename = mime_decodeheader(
+				filename = mime_guessheader(
 						(char *)name->data);
 			}
 		}
@@ -1221,7 +1280,7 @@ mime_mkfilename(char *id, mime_t *mime)
 	 */
 	hdr = llist_ciget(mime->hdrs, "Content-Description");
 	if (hdr != NULL && hdr->data != NULL) {
-		filename = mime_decodeheader((char *)hdr->data);
+		filename = mime_guessheader((char *)hdr->data);
 		if (filename != NULL)
 			return filename;
 	}
diff --git a/mime.h b/mime.h
@@ -36,6 +36,7 @@ char *mime_iconv(char *str, char *from, char *to);
 char *mime_decodeheaderext(char *value);
 int mime_isextws(char *str, int len);
 char *mime_decodeheader(char *value);
+char *mime_guessheader(char *value);
 char *mime_decodeparam(char *value);
 char *mime_encodestring(char *value);
 char *mime_encodeheader(char *header, char *value);
diff --git a/scan.c b/scan.c
@@ -93,7 +93,7 @@ scanmain(int argc, char *argv[])
 		frome = llist_ciget(mime->hdrs, "from");
 
 		if (subjecte != NULL) {
-			subject = mime_decodeheader((char *)subjecte->data);
+			subject = mime_guessheader((char *)subjecte->data);
 		} else {
 			subject = "<empty>";
 		}
@@ -110,7 +110,7 @@ scanmain(int argc, char *argv[])
 		}
 
 		if (frome != NULL) {
-			from = mime_decodeheader((char *)frome->data);
+			from = mime_guessheader((char *)frome->data);
 		} else {
 			from = "<empty>";
 		}
@@ -165,7 +165,7 @@ scanmain(int argc, char *argv[])
 		id = (char *)ide->data;
 
 		if (subjecte != NULL) {
-			subject = mime_decodeheader((char *)subjecte->data);
+			subject = mime_guessheader((char *)subjecte->data);
 		} else {
 			subject = "<empty>";
 		}
@@ -182,7 +182,7 @@ scanmain(int argc, char *argv[])
 		}
 
 		if (frome != NULL) {
-			from = mime_decodeheader((char *)frome->data);
+			from = mime_guessheader((char *)frome->data);
 		} else {
 			from = "<empty>";
 		}
diff --git a/util.c b/util.c
@@ -71,7 +71,7 @@ utilmain(int argc, char *argv[])
 		ress = mime_encodestring(works);
 		printf("%s", ress);
 	} else if (status & DECODE) {
-		ress = mime_decodeheader(works);
+		ress = mime_guessheader(works);
 		printf("%s", ress);
 	} else {
 		free(works);
diff --git a/view.c b/view.c
@@ -141,7 +141,7 @@ view_printpart(char *id, mime_t *mime, llist_t *dhdrs, llist_t *partl,
 		didprint = 0;
 		if (!strcasecmp(dhdrs->first->key, "all")) {
 			forllist(mime->hdrs, helem) {
-				hvalue = mime_decodeheader((char *)helem->data);
+				hvalue = mime_guessheader((char *)helem->data);
 				if (hvalue == NULL)
 					continue;
 				if (options & PRINT_VALUE) {
@@ -157,7 +157,7 @@ view_printpart(char *id, mime_t *mime, llist_t *dhdrs, llist_t *partl,
 				hlist = llist_cifind(mime->hdrs, delem->key);
 				if (hlist != NULL) {
 					forllist(hlist, helem) {
-						hvalue = mime_decodeheader(
+						hvalue = mime_guessheader(
 								(char *)helem->data);
 						if (hvalue == NULL)
 							continue;