[1/1] gitignore: ignore comments on the same line as a pattern

Commit Message

Linus Arver via GitGitGadget Oct. 2, 2019, 3:13 a.m. UTC

From: Chris Zehner <[e-mail]>

Signed-off-by: Chris Zehner <[e-mail]>
---
 Documentation/gitignore.txt |  8 ++++++--
 dir.c                       | 33 +++++++++++++++++++++++++++++++++
 2 files changed, 39 insertions(+), 2 deletions(-)

Comments

Junio C Hamano Oct. 2, 2019, 7:02 a.m. UTC | #1

"Chris Zehner via GitGitGadget" <[e-mail]> writes:

> -   that begin with a hash.
> +   Put a backslash ("`\`") in front of each hash for patterns
> +   containing a hash.
> +
> + - A # after a pattern will be treated as the start of a comment.
> +   Put a backslash ("`\`") in front of each hash for patterns
> +   containing a hash.

Besides being backward incompatible, this looks somewhat misdesigned
by lacking a way to escape a backslash (i.e. what should a project
do if they want a patttern with backslash followed by a hash
literally in there?).

> diff --git a/dir.c b/dir.c
> index cab9c2a458..aeefe142bc 100644
> --- a/dir.c
> +++ b/dir.c
> @@ -658,6 +658,38 @@ void clear_pattern_list(struct pattern_list *pl)
>  	memset(pl, 0, sizeof(*pl));
>  }
>  
> +static void trim_trailing_comments(char *buf)
> +{
> +	char *p, *last_hash = NULL;
> +	int escape_seq = 0;
> +
> +	for (p = buf; *p; p++)
> +	{

Style (see Documentation/CodingGuidelines).  The opening parenthesis
of a control structure like if/for/switch are placed on the same
line, i.e.

	for (p = buf; *p; p++) {

Do we even need a separate 'p', instead of scanning with 'buf' itself?

> +		if (!*p)
> +			return;

What happens when an entry ends with '\' followed by an EOL?  IOW,
what if escape_seq is true here?

> +		switch (*p) {
> +		case '#':
> +			if (escape_seq)
> +			{
> +				escape_seq = 0;
> +				p++;
> +				break;
> +			}
> +			if (!last_hash)
> +				last_hash = p;
> +			break;
> +		case '\\':
> +			escape_seq = 1;
> +			break;

Patch

diff --git a/Documentation/gitignore.txt b/Documentation/gitignore.txt
index d47b1ae296..1778cf1d38 100644
--- a/Documentation/gitignore.txt
+++ b/Documentation/gitignore.txt
@@ -74,8 +74,12 @@  PATTERN FORMAT
    for readability.
 
  - A line starting with # serves as a comment.
-   Put a backslash ("`\`") in front of the first hash for patterns
-   that begin with a hash.
+   Put a backslash ("`\`") in front of each hash for patterns
+   containing a hash.
+
+ - A # after a pattern will be treated as the start of a comment.
+   Put a backslash ("`\`") in front of each hash for patterns
+   containing a hash.
 
  - Trailing spaces are ignored unless they are quoted with backslash
    ("`\`").
diff --git a/dir.c b/dir.c
index cab9c2a458..aeefe142bc 100644
--- a/dir.c
+++ b/dir.c
@@ -658,6 +658,38 @@  void clear_pattern_list(struct pattern_list *pl)
 	memset(pl, 0, sizeof(*pl));
 }
 
+static void trim_trailing_comments(char *buf)
+{
+	char *p, *last_hash = NULL;
+	int escape_seq = 0;
+
+	for (p = buf; *p; p++)
+	{
+		if (!*p)
+			return;
+		switch (*p) {
+		case '#':
+			if (escape_seq)
+			{
+				escape_seq = 0;
+				p++;
+				break;
+			}
+			if (!last_hash)
+				last_hash = p;
+			break;
+		case '\\':
+			escape_seq = 1;
+			break;
+		default:
+			escape_seq = 0;
+		}
+	}
+
+	if (last_hash)
+		*last_hash = '\0';
+}
+
 static void trim_trailing_spaces(char *buf)
 {
 	char *p, *last_space = NULL;
@@ -859,6 +891,7 @@  static int add_patterns_from_buffer(char *buf, size_t size,
 		if (buf[i] == '\n') {
 			if (entry != buf + i && entry[0] != '#') {
 				buf[i - (i && buf[i-1] == '\r')] = 0;
+				trim_trailing_comments(entry);
 				trim_trailing_spaces(entry);
 				add_pattern(entry, base, baselen, pl, lineno);
 			}