threads / patch / 12793

v3pretty.c: add %x00 format specifier.

Subject: [PATCH v3] pretty.c: add %x00 format specifier.

## tl;dr

4 messages between Mar 21, 2008 and Mar 23, 2008. Diffs are folded; open one to read it.

replies: 3people: 2as markdown or json

Govind Salinas· Mar 21, 2008, 15:05 UTC · lore

This adds a %x00 format which parses the 00 as hex encoding for a byte and prints the resulting byte. This can be used to add null bytes or other bytes that can make machine parsing easier. It is also necessary to use fwrite to print out the data since printf will terminate if you feed it a null.

Junio supplied the hex decoding.
Signed-off-by: Govind Salinas <blix@sophiasuchtig.com>
---
 Documentation/pretty-formats.txt |    1 +
 log-tree.c                       |    6 ++++--
 pretty.c                         |   11 +++++++++++
 3 files changed, 16 insertions(+), 2 deletions(-)
Show changes to 3 files +16 −2

Documentation/pretty-formats.txt, log-tree.c, pretty.c

diff --git a/Documentation/pretty-formats.txt b/Documentation/pretty-formats.txt
index 0193c3c..e8bea3e 100644
--- a/Documentation/pretty-formats.txt
+++ b/Documentation/pretty-formats.txt
@@ -123,3 +123,4 @@ The placeholders are:
 - '%Creset': reset color
 - '%m': left, right or boundary mark
 - '%n': newline
+- '%x00': print a byte from a hex code
diff --git a/log-tree.c b/log-tree.c
index 608f697..3bb6e49 100644
--- a/log-tree.c
+++ b/log-tree.c
@@ -308,8 +308,10 @@ void show_log(struct rev_info *opt, const char *sep)
 	if (opt->show_log_size)
 		printf("log size %i\n", (int)msgbuf.len);

-	if (msgbuf.len)
-		printf("%s%s%s", msgbuf.buf, extra, sep);
+	if (msgbuf.len) {
+		fwrite(msgbuf.buf, sizeof(char), msgbuf.len, stdout);
+		printf("%s%s", extra, sep);
+	}
 	strbuf_release(&msgbuf);
 }

diff --git a/pretty.c b/pretty.c
index 703f521..cb4944c 100644
--- a/pretty.c
+++ b/pretty.c
@@ -457,6 +457,7 @@ static size_t format_commit_item(struct strbuf
*sb, const char *placeholder,
 	const struct commit *commit = c->commit;
 	const char *msg = commit->buffer;
 	struct commit_list *p;
+	int h1, h2;

 	/* these are independent of the commit */
 	switch (placeholder[0]) {
@@ -478,6 +479,16 @@ static size_t format_commit_item(struct strbuf
*sb, const char *placeholder,
 	case 'n':		/* newline */
 		strbuf_addch(sb, '\n');
 		return 1;
+	case 'x':
+		/* %x00 == NUL, %x0a == LF, etc. */
+		if (0 <= (h1 = hexval_table[0xff & placeholder[1]]) &&
+		    h1 <= 16 &&
+		    0 <= (h2 = hexval_table[0xff & placeholder[2]]) &&
+		    h2 <= 16) {
+			strbuf_addch(sb, (h1<<4)|h2);
+			return 3;
+		} else
+			return 0;
 	}

 	/* these depend on the commit */

-- 
1.5.4.4.552.g7113.dirty
Johannes Schindelin· Mar 21, 2008, 15:47 UTC · re: Govind Salinas · lore

Re: [PATCH v3] pretty.c: add %x00 format specifier.

Hi,
On Fri, 21 Mar 2008, Govind Salinas wrote:
Show 7 quoted lines
> This adds a %x00 format which parses the 00 as hex encoding for a byte 
> and prints the resulting byte.  This can be used to add null bytes or 
> other bytes that can make machine parsing easier.  It is also necessary 
> to use fwrite to print out the data since printf will terminate if you 
> feed it a null.
> 
> Junio supplied the hex decoding.

This is all nice and well, but if I understood you correctly, your original aim was to have a pretty-formatted list of the revisions, but NUL-delimited for easy parsing.

Which you can do with "git log --pretty=format:%s -z" right now.

Ciao, Dscho

Govind Salinas· Mar 21, 2008, 17:45 UTC · re: Johannes Schindelin · lore

Re: [PATCH v3] pretty.c: add %x00 format specifier.

On Fri, Mar 21, 2008 at 10:47 AM, Johannes Schindelin <Johannes.Schindelin@gmx.de> wrote:

Show 23 quoted lines
> Hi,
>
>
>  On Fri, 21 Mar 2008, Govind Salinas wrote:
>
>  > This adds a %x00 format which parses the 00 as hex encoding for a byte
>  > and prints the resulting byte.  This can be used to add null bytes or
>  > other bytes that can make machine parsing easier.  It is also necessary
>  > to use fwrite to print out the data since printf will terminate if you
>  > feed it a null.
>  >
>  > Junio supplied the hex decoding.
>
>  This is all nice and well, but if I understood you correctly, your
>  original aim was to have a pretty-formatted list of the revisions, but
>  NUL-delimited for easy parsing.
>
>  Which you can do with "git log --pretty=format:%s -z" right now.
>
>  Ciao,
>  Dscho
>
>

Unfortunately that almost work for me. I am in the following situation If I stick to just using the --pretty=format options then I would be fine but I am also adding --stat and -p to git-log, so now I have 3 multiline hunks of data and I need to tell where each one ends deterministically. I *could* just look for "---" and "diff --git" at every line but I would prefer to use a delimiter.

Alternatively, if -z would add a null before the diffstat was printed, that would probably work for me as well. Even so, using the same delimiter is nicer than parsing part with newlines and part with nulls, which I would have to do if I just used -z. Think "%H%x00%ae%x00%an%x00%s%x00%b%x00 -z -p --stat". there everything splits nicely with nulls. The alternative (with -z adding a null before the diffstat is "%H%n%ae%n%an%n%s%n%b -z -p --stat" which would work, but I would have to parse the last 3 fields differently.

Perhaps there is a better way to get all this information. But I do want all the information to come from the same call.

Thanks, Govind.

Govind Salinas· Mar 23, 2008, 01:22 UTC · re: Govind Salinas · lore

Re: [PATCH v3] pretty.c: add %x00 format specifier.

On Fri, Mar 21, 2008 at 12:45 PM, Govind Salinas <govind@sophiasuchtig.com> wrote:

Show 49 quoted lines
>
> On Fri, Mar 21, 2008 at 10:47 AM, Johannes Schindelin
>  <Johannes.Schindelin@gmx.de> wrote:
>  > Hi,
>  >
>  >
>  >  On Fri, 21 Mar 2008, Govind Salinas wrote:
>  >
>  >  > This adds a %x00 format which parses the 00 as hex encoding for a byte
>  >  > and prints the resulting byte.  This can be used to add null bytes or
>  >  > other bytes that can make machine parsing easier.  It is also necessary
>  >  > to use fwrite to print out the data since printf will terminate if you
>  >  > feed it a null.
>  >  >
>  >  > Junio supplied the hex decoding.
>  >
>  >  This is all nice and well, but if I understood you correctly, your
>  >  original aim was to have a pretty-formatted list of the revisions, but
>  >  NUL-delimited for easy parsing.
>  >
>  >  Which you can do with "git log --pretty=format:%s -z" right now.
>  >
>  >  Ciao,
>  >  Dscho
>  >
>  >
>
>  Unfortunately that almost work for me.  I am in the following situation
>  If I stick to just using the --pretty=format options then I would be fine
>  but I am also adding --stat and -p to git-log, so now I have 3 multiline
>  hunks of data and I need to tell where each one ends deterministically.
>  I *could* just look for "---" and "diff  --git" at every line but I would prefer
>  to use a delimiter.
>
>  Alternatively, if -z would add a null before the diffstat was printed, that
>  would probably work for me as well.  Even so, using the same delimiter
>  is nicer than parsing part with newlines and part with nulls, which I would
>  have to do if I just used -z.  Think
>  "%H%x00%ae%x00%an%x00%s%x00%b%x00 -z -p --stat". there
>  everything splits nicely with nulls.  The alternative (with -z adding a null
>  before the diffstat is "%H%n%ae%n%an%n%s%n%b -z -p --stat" which
>  would work, but I would have to parse the last 3 fields differently.
>
>  Perhaps there is a better way to get all this information.  But I do want
>  all the information to come from the same call.
>
>  Thanks,
>  Govind.
>
Ok, how about this instead?  Perhaps it is more palatable.
Show changes to log-tree.c +1 −1
diff --git a/log-tree.c b/log-tree.c
index 608f697..5f55683 100644
--- a/log-tree.c
+++ b/log-tree.c
@@ -338,7 +338,7 @@ int log_tree_diff_flush(struct rev_info *opt)
                        int pch = DIFF_FORMAT_DIFFSTAT | DIFF_FORMAT_PATCH;
                        if ((pch & opt->diffopt.output_format) == pch)
                                printf("---");
-                       putchar('\n');
+                       putchar(opt->diffopt.line_termination);
                }
        }
        diff_flush(&opt->diffopt);

← back to recent threads