lib-mail: Include blockquotes as quotes in plain text

<blockquote>This is now preserved</blockquote>
which is fine

renders as

> this is now preserved
which is fine

diff --git a/src/lib-mail/mail-html2text.c b/src/lib-mail/mail-html2text.c
index 0297269fd609da70b5566970e242fc60faca089a..3d9f9beb4f8455dca6f229571c59aab6a1afb6ab 100644 (file)
--- a/src/lib-mail/mail-html2text.c
+++ b/src/lib-mail/mail-html2text.c
@@ -41,6 +41,7 @@ struct mail_html2text {
        enum html_state state;
        buffer_t *input;
        unsigned int quote_level;
+       bool add_newline;
 };
 
 static struct {
@@ -86,18 +87,18 @@ parse_tag_name(struct mail_html2text *ht,
                return 8 + 1;
        }
 
-       if ((ht->flags & MAIL_HTML2TEXT_FLAG_SKIP_QUOTED) != 0) {
-               if (size >= 10 && i_memcasecmp(data, "blockquote", 10) == 0 &&
-                   (HTML_WHITESPACE(data[10]) || data[10] == '>')) {
-                       ht->quote_level++;
-                       ht->state = HTML_STATE_TAG;
-                       return 1;
-               } else if (ht->quote_level > 0 &&
-                          size >= 12 && i_memcasecmp(data, "/blockquote>", 12) == 0) {
-                       ht->quote_level--;
-                       ht->state = HTML_STATE_TAG;
-                       return 1;
-               }
+       if (size >= 10 && i_memcasecmp(data, "blockquote", 10) == 0 &&
+           (HTML_WHITESPACE(data[10]) || data[10] == '>')) {
+               ht->quote_level++;
+               ht->state = HTML_STATE_TAG;
+               return 1;
+       } else if (ht->quote_level > 0 &&
+                  size >= 12 && i_memcasecmp(data, "/blockquote>", 12) == 0) {
+               ht->quote_level--;
+               if ((ht->flags & MAIL_HTML2TEXT_FLAG_SKIP_QUOTED) == 0)
+                       ht->add_newline = TRUE;
+               ht->state = HTML_STATE_TAG;
+               return 1;
        }
        if (size < 12) {
                /* can we see the whole tag name? */
@@ -191,15 +192,16 @@ parse_data(struct mail_html2text *ht,
                                if (ret == 0)
                                        return i;
                                i += ret - 1;
-                       } else if (ht->quote_level == 0) {
-                               if (c == '&') {
-                                       ret = parse_entity(data+i+1, size-i-1, output);
-                                       if (ret == 0)
-                                               return i;
-                                       i += ret - 1;
-                               } else {
-                                       buffer_append_c(output, c);
-                               }
+                       } else if (ht->quote_level > 0 &&
+                                  (ht->flags & MAIL_HTML2TEXT_FLAG_SKIP_QUOTED) != 0) {
+                                       break;
+                       } else if (c == '&') {
+                               ret = parse_entity(data+i+1, size-i-1, output);
+                               if (ret == 0)
+                                       return i;
+                               i += ret - 1;
+                       } else {
+                               buffer_append_c(output, c);
                        }
                        break;
                case HTML_STATE_TAG:
@@ -209,6 +211,13 @@ parse_data(struct mail_html2text *ht,
                                ht->state = HTML_STATE_TAG_SQUOTED;
                        else if (c == '>') {
                                ht->state = HTML_STATE_TEXT;
+                               if (ht->quote_level > 0 &&
+                                   (ht->flags & MAIL_HTML2TEXT_FLAG_SKIP_QUOTED) == 0) {
+                                       buffer_append(output, "\n>", 2);
+                               } else if (ht->add_newline) {
+                                       buffer_append_c(output, '\n');
+                               }
+                               ht->add_newline = FALSE;
                                mail_html2text_add_space(output);
                        }
                        break;
@@ -284,7 +293,8 @@ parse_data(struct mail_html2text *ht,
                                        break;
                                }
                        }
-                       if (ht->quote_level == 0)
+                       if (ht->quote_level == 0 ||
+                           (ht->flags & MAIL_HTML2TEXT_FLAG_SKIP_QUOTED) == 0)
                                buffer_append_c(output, c);
                        break;
                }

diff --git a/src/lib-mail/test-mail-html2text.c b/src/lib-mail/test-mail-html2text.c
index cbb0a6630f88a624a3f3eda66f0f772f11e6cd9b..73e93f70b2482d1fabf3b03d202b887d9bad1402 100644 (file)
--- a/src/lib-mail/test-mail-html2text.c
+++ b/src/lib-mail/test-mail-html2text.c
@@ -44,9 +44,17 @@ static const struct {
        { "&#deee;", "" }, // invalid codepoint
 };
 
-static const char *test_blockquote_input =
-       "a<blockquote>b<blockquote><blockquote>c</blockquote>d</blockquote>e</blockquote>f";
-static const char *test_blockquote_output = "a b c d e f";
+static const char *test_blockquote_input[] = {
+       "a<blockquote>b<blockquote><blockquote>c</blockquote>d</blockquote>e</blockquote>f",
+       "a&amp;<blockquote>b&amp;<blockquote>&amp;<blockquote>&amp;c</blockquote>d&amp;</blockquote>&amp;e</blockquote>f&amp;",
+       NULL
+};
+
+static const char *test_blockquote_output[] = {
+       "a\n> b\n> \n> c\n> d\n> e\nf",
+       "a&\n> b&\n> &\n> &c\n> d&\n> &e\nf&",
+       NULL
+};
 
 static void test_mail_html2text(void)
 {
@@ -67,11 +75,14 @@ static void test_mail_html2text(void)
        }
 
        /* test without skipping quoted */
-       ht = mail_html2text_init(0);
-       mail_html2text_more(ht, (const void *)test_blockquote_input,
-                           strlen(test_blockquote_input), str);
-       test_assert(strcmp(str_c(str), test_blockquote_output) == 0);
-       mail_html2text_deinit(&ht);
+       for (unsigned int i = 0; test_blockquote_input[i] != NULL; i++) {
+               str_truncate(str, 0);
+               ht = mail_html2text_init(0);
+               mail_html2text_more(ht, (const void *)test_blockquote_input[i],
+                                   strlen(test_blockquote_input[i]), str);
+               test_assert_idx(strcmp(str_c(str), test_blockquote_output[i]) == 0, i);
+               mail_html2text_deinit(&ht);
+       }
 
        test_end();
 }