When generating manpages, delete outdated targets first.
[gitweb.git] / builtin-mailinfo.c
index 489c2c58c01514ac3d967d1c3f46f1243f853580..b558754142b26399df78472b0bb124144dc194da 100644 (file)
@@ -237,8 +237,6 @@ static int eatspace(char *line)
 
 static char *cleanup_subject(char *subject)
 {
-       if (keep_subject)
-               return subject;
        for (;;) {
                char *p;
                int len, remove;
@@ -425,6 +423,7 @@ static int read_one_header_line(char *line, int sz, FILE *in)
                        if (addlen >= sz - len)
                                addlen = sz - len - 1;
                        memcpy(line + len, continuation, addlen);
+                       line[len] = '\n';
                        len += addlen;
                }
        }
@@ -499,15 +498,42 @@ static int decode_b_segment(char *in, char *ot, char *ep)
        return 0;
 }
 
+/*
+ * When there is no known charset, guess.
+ *
+ * Right now we assume that if the target is UTF-8 (the default),
+ * and it already looks like UTF-8 (which includes US-ASCII as its
+ * subset, of course) then that is what it is and there is nothing
+ * to do.
+ *
+ * Otherwise, we default to assuming it is Latin1 for historical
+ * reasons.
+ */
+static const char *guess_charset(const char *line, const char *target_charset)
+{
+       if (is_encoding_utf8(target_charset)) {
+               if (is_utf8(line))
+                       return NULL;
+       }
+       return "latin1";
+}
+
 static void convert_to_utf8(char *line, const char *charset)
 {
-       static const char latin_one[] = "latin1";
-       const char *input_charset = *charset ? charset : latin_one;
-       char *out = reencode_string(line, metainfo_charset, input_charset);
+       char *out;
+
+       if (!charset || !*charset) {
+               charset = guess_charset(line, metainfo_charset);
+               if (!charset)
+                       return;
+       }
 
+       if (!strcmp(metainfo_charset, charset))
+               return;
+       out = reencode_string(line, metainfo_charset, charset);
        if (!out)
                die("cannot convert from %s to %s\n",
-                   input_charset, metainfo_charset);
+                   charset, metainfo_charset);
        strcpy(line, out);
        free(out);
 }
@@ -819,6 +845,22 @@ static void handle_body(void)
        return;
 }
 
+static void output_header_lines(FILE *fout, const char *hdr, char *data)
+{
+       while (1) {
+               char *ep = strchr(data, '\n');
+               int len;
+               if (!ep)
+                       len = strlen(data);
+               else
+                       len = ep - data;
+               fprintf(fout, "%s: %.*s\n", hdr, len, data);
+               if (!ep)
+                       break;
+               data = ep + 1;
+       }
+}
+
 static void handle_info(void)
 {
        char *sub;
@@ -836,9 +878,13 @@ static void handle_info(void)
                        continue;
 
                if (!memcmp(header[i], "Subject", 7)) {
-                       sub = cleanup_subject(hdr);
-                       cleanup_space(sub);
-                       fprintf(fout, "Subject: %s\n", sub);
+                       if (keep_subject)
+                               sub = hdr;
+                       else {
+                               sub = cleanup_subject(hdr);
+                               cleanup_space(sub);
+                       }
+                       output_header_lines(fout, "Subject", sub);
                } else if (!memcmp(header[i], "From", 4)) {
                        handle_from(hdr);
                        fprintf(fout, "Author: %s\n", name);