Teach mailinfo to ignore everything before -- >8 -- mark
authorJunio C Hamano <gitster@pobox.com>
Sun, 23 Aug 2009 08:56:19 +0000 (01:56 -0700)
committerJunio C Hamano <gitster@pobox.com>
Thu, 27 Aug 2009 05:21:32 +0000 (22:21 -0700)
This teaches mailinfo the scissors -- >8 -- mark; the command ignores
everything before it in the message body.

For lefties among us, we also support -- 8< -- ;-)

Signed-off-by: Junio C Hamano <gitster@pobox.com>
builtin-mailinfo.c
t/t5100-mailinfo.sh
t/t5100/info0014 [new file with mode: 0644]
t/t5100/msg0014 [new file with mode: 0644]
t/t5100/patch0014 [new file with mode: 0644]
t/t5100/sample.mbox
index 50326d911c2ea317886b70f456c1ac56405a37bb..a6f6b123b46e35b468d91342368c9cd9385c8bd1 100644 (file)
@@ -712,6 +712,56 @@ static inline int patchbreak(const struct strbuf *line)
        return 0;
 }
 
+static int is_scissors_line(const struct strbuf *line)
+{
+       size_t i, len = line->len;
+       int scissors = 0, gap = 0;
+       int first_nonblank = -1;
+       int last_nonblank = 0, visible, perforation = 0, in_perforation = 0;
+       const char *buf = line->buf;
+
+       for (i = 0; i < len; i++) {
+               if (isspace(buf[i])) {
+                       if (in_perforation) {
+                               perforation++;
+                               gap++;
+                       }
+                       continue;
+               }
+               last_nonblank = i;
+               if (first_nonblank < 0)
+                       first_nonblank = i;
+               if (buf[i] == '-') {
+                       in_perforation = 1;
+                       perforation++;
+                       continue;
+               }
+               if (i + 1 < len &&
+                   (!memcmp(buf + i, ">8", 2) || !memcmp(buf + i, "8<", 2))) {
+                       in_perforation = 1;
+                       perforation += 2;
+                       scissors += 2;
+                       i++;
+                       continue;
+               }
+               in_perforation = 0;
+       }
+
+       /*
+        * The mark must be at least 8 bytes long (e.g. "-- >8 --").
+        * Even though there can be arbitrary cruft on the same line
+        * (e.g. "cut here"), in order to avoid misidentification, the
+        * perforation must occupy more than a third of the visible
+        * width of the line, and dashes and scissors must occupy more
+        * than half of the perforation.
+        */
+
+       visible = last_nonblank - first_nonblank + 1;
+       return (scissors && 8 <= visible &&
+               visible < perforation * 3 &&
+               gap * 2 < perforation);
+}
+
 static int handle_commit_msg(struct strbuf *line)
 {
        static int still_looking = 1;
@@ -723,7 +773,8 @@ static int handle_commit_msg(struct strbuf *line)
                strbuf_ltrim(line);
                if (!line->len)
                        return 0;
-               if ((still_looking = check_header(line, s_hdr_data, 0)) != 0)
+               still_looking = check_header(line, s_hdr_data, 0);
+               if (still_looking)
                        return 0;
        }
 
@@ -731,6 +782,24 @@ static int handle_commit_msg(struct strbuf *line)
        if (metainfo_charset)
                convert_to_utf8(line, charset.buf);
 
+       if (is_scissors_line(line)) {
+               int i;
+               rewind(cmitmsg);
+               ftruncate(fileno(cmitmsg), 0);
+               still_looking = 1;
+
+               /*
+                * We may have already read "secondary headers"; purge
+                * them to give ourselves a clean restart.
+                */
+               for (i = 0; header[i]; i++) {
+                       if (s_hdr_data[i])
+                               strbuf_release(s_hdr_data[i]);
+                       s_hdr_data[i] = NULL;
+               }
+               return 0;
+       }
+
        if (patchbreak(line)) {
                fclose(cmitmsg);
                cmitmsg = NULL;
index e70ea94a1368dc045469808d30c717aa2b8bb158..e84855665972d39f01f9eb37e1331b4a13142058 100755 (executable)
@@ -11,7 +11,7 @@ test_expect_success 'split sample box' \
        'git mailsplit -o. "$TEST_DIRECTORY"/t5100/sample.mbox >last &&
        last=`cat last` &&
        echo total is $last &&
-       test `cat last` = 13'
+       test `cat last` = 14'
 
 for mail in `echo 00*`
 do
diff --git a/t/t5100/info0014 b/t/t5100/info0014
new file mode 100644 (file)
index 0000000..ab9c8d0
--- /dev/null
@@ -0,0 +1,5 @@
+Author: Junio C Hamano
+Email: gitster@pobox.com
+Subject: Teach mailinfo to ignore everything before -- >8 -- mark
+Date: Thu, 20 Aug 2009 17:18:22 -0700
+
diff --git a/t/t5100/msg0014 b/t/t5100/msg0014
new file mode 100644 (file)
index 0000000..259c6a4
--- /dev/null
@@ -0,0 +1,4 @@
+This teaches mailinfo the scissors -- >8 -- mark; the command ignores
+everything before it in the message body.
+
+Signed-off-by: Junio C Hamano <gitster@pobox.com>
diff --git a/t/t5100/patch0014 b/t/t5100/patch0014
new file mode 100644 (file)
index 0000000..124efd2
--- /dev/null
@@ -0,0 +1,64 @@
+---
+ builtin-mailinfo.c |   37 ++++++++++++++++++++++++++++++++++++-
+ 1 files changed, 36 insertions(+), 1 deletions(-)
+
+diff --git a/builtin-mailinfo.c b/builtin-mailinfo.c
+index b0b5d8f..461c47e 100644
+--- a/builtin-mailinfo.c
++++ b/builtin-mailinfo.c
+@@ -712,6 +712,34 @@ static inline int patchbreak(const struct strbuf *line)
+       return 0;
+ }
++static int scissors(const struct strbuf *line)
++{
++      size_t i, len = line->len;
++      int scissors_dashes_seen = 0;
++      const char *buf = line->buf;
++
++      for (i = 0; i < len; i++) {
++              if (isspace(buf[i]))
++                      continue;
++              if (buf[i] == '-') {
++                      scissors_dashes_seen |= 02;
++                      continue;
++              }
++              if (i + 1 < len && !memcmp(buf + i, ">8", 2)) {
++                      scissors_dashes_seen |= 01;
++                      i++;
++                      continue;
++              }
++              if (i + 7 < len && !memcmp(buf + i, "cut here", 8)) {
++                      i += 7;
++                      continue;
++              }
++              /* everything else --- not scissors */
++              break;
++      }
++      return scissors_dashes_seen == 03;
++}
++
+ static int handle_commit_msg(struct strbuf *line)
+ {
+       static int still_looking = 1;
+@@ -723,10 +751,17 @@ static int handle_commit_msg(struct strbuf *line)
+               strbuf_ltrim(line);
+               if (!line->len)
+                       return 0;
+-              if ((still_looking = check_header(line, s_hdr_data, 0)) != 0)
++              still_looking = check_header(line, s_hdr_data, 0);
++              if (still_looking)
+                       return 0;
+       }
++      if (scissors(line)) {
++              fseek(cmitmsg, 0L, SEEK_SET);
++              still_looking = 1;
++              return 0;
++      }
++
+       /* normalize the log message to UTF-8. */
+       if (metainfo_charset)
+               convert_to_utf8(line, charset.buf);
+-- 
+1.6.4.1
index c3074ac5736fe266e901b4da3b060208cd71406e..13fa4ae03b917148e1c9c771ba0672c001a53046 100644 (file)
@@ -561,3 +561,92 @@ From: <a.u.thor@example.com> (A U Thor)
 Date: Fri, 9 Jun 2006 00:44:16 -0700
 Subject: [PATCH] a patch
 
+From nobody Mon Sep 17 00:00:00 2001
+From: Junio Hamano <junkio@cox.net>
+Date: Thu, 20 Aug 2009 17:18:22 -0700
+Subject: Why doesn't git-am does not like >8 scissors mark?
+
+Subject: [PATCH] BLAH ONE
+
+In real life, we will see a discussion that inspired this patch
+discussing related and unrelated things around >8 scissors mark
+in this part of the message.
+
+Subject: [PATCH] BLAH TWO
+
+And then we will see the scissors.
+
+ This line is not a scissors mark -- >8 -- but talks about it.
+ - - >8 - - please remove everything above this line - - >8 - -
+
+Subject: [PATCH] Teach mailinfo to ignore everything before -- >8 -- mark
+From: Junio C Hamano <gitster@pobox.com>
+
+This teaches mailinfo the scissors -- >8 -- mark; the command ignores
+everything before it in the message body.
+
+Signed-off-by: Junio C Hamano <gitster@pobox.com>
+---
+ builtin-mailinfo.c |   37 ++++++++++++++++++++++++++++++++++++-
+ 1 files changed, 36 insertions(+), 1 deletions(-)
+
+diff --git a/builtin-mailinfo.c b/builtin-mailinfo.c
+index b0b5d8f..461c47e 100644
+--- a/builtin-mailinfo.c
++++ b/builtin-mailinfo.c
+@@ -712,6 +712,34 @@ static inline int patchbreak(const struct strbuf *line)
+       return 0;
+ }
++static int scissors(const struct strbuf *line)
++{
++      size_t i, len = line->len;
++      int scissors_dashes_seen = 0;
++      const char *buf = line->buf;
++
++      for (i = 0; i < len; i++) {
++              if (isspace(buf[i]))
++                      continue;
++              if (buf[i] == '-') {
++                      scissors_dashes_seen |= 02;
++                      continue;
++              }
++              if (i + 1 < len && !memcmp(buf + i, ">8", 2)) {
++                      scissors_dashes_seen |= 01;
++                      i++;
++                      continue;
++              }
++              if (i + 7 < len && !memcmp(buf + i, "cut here", 8)) {
++                      i += 7;
++                      continue;
++              }
++              /* everything else --- not scissors */
++              break;
++      }
++      return scissors_dashes_seen == 03;
++}
++
+ static int handle_commit_msg(struct strbuf *line)
+ {
+       static int still_looking = 1;
+@@ -723,10 +751,17 @@ static int handle_commit_msg(struct strbuf *line)
+               strbuf_ltrim(line);
+               if (!line->len)
+                       return 0;
+-              if ((still_looking = check_header(line, s_hdr_data, 0)) != 0)
++              still_looking = check_header(line, s_hdr_data, 0);
++              if (still_looking)
+                       return 0;
+       }
++      if (scissors(line)) {
++              fseek(cmitmsg, 0L, SEEK_SET);
++              still_looking = 1;
++              return 0;
++      }
++
+       /* normalize the log message to UTF-8. */
+       if (metainfo_charset)
+               convert_to_utf8(line, charset.buf);
+-- 
+1.6.4.1