[PATCH 0/2] fast-export: simplification

[PATCH 0/2] fast-export: simplification

[Felipe Contreras]

Hi,

No functional changes, but get_tags_and_duplicates() is quite complex as it is,
and can be simplified by spliting code into a separate function.

Felipe Contreras (2):
  fast-export: make extra_refs global
  fast-export: refactor get_tags_and_duplicates()

 builtin/fast-export.c | 87 ++++++++++++++++++++++++++++-----------------------
 1 file changed, 47 insertions(+), 40 deletions(-)

-- 
1.8.4-337-g7358a66-dirty

[PATCH 1/2] fast-export: make extra_refs global

[Felipe Contreras]

There's no need to pass it around everywhere. This would make easier
further refactoring that makes use of this variable.

Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
---
 builtin/fast-export.c | 21 ++++++++++-----------
 1 file changed, 10 insertions(+), 11 deletions(-)

diff --git a/builtin/fast-export.c b/builtin/fast-export.c
index 8e19058..957392c 100644
--- a/builtin/fast-export.c
+++ b/builtin/fast-export.c
@@ -30,6 +30,7 @@ static int fake_missing_tagger;
 static int use_done_feature;
 static int no_data;
 static int full_tree;
+static struct string_list extra_refs = STRING_LIST_INIT_NODUP;
 
 static int parse_opt_signed_tag_mode(const struct option *opt,
 				     const char *arg, int unset)
@@ -484,8 +485,7 @@ static void handle_tag(const char *name, struct tag *tag)
 	       (int)message_size, (int)message_size, message ? message : "");
 }
 
-static void get_tags_and_duplicates(struct rev_cmdline_info *info,
-				    struct string_list *extra_refs)
+static void get_tags_and_duplicates(struct rev_cmdline_info *info)
 {
 	struct tag *tag;
 	int i;
@@ -512,7 +512,7 @@ static void get_tags_and_duplicates(struct rev_cmdline_info *info,
 			/* handle nested tags */
 			while (tag && tag->object.type == OBJ_TAG) {
 				parse_object(tag->object.sha1);
-				string_list_append(extra_refs, full_name)->util = tag;
+				string_list_append(&extra_refs, full_name)->util = tag;
 				tag = (struct tag *)tag->tagged;
 			}
 			if (!tag)
@@ -542,20 +542,20 @@ static void get_tags_and_duplicates(struct rev_cmdline_info *info,
 		 * sure it gets properly updated eventually.
 		 */
 		if (commit->util || commit->object.flags & SHOWN)
-			string_list_append(extra_refs, full_name)->util = commit;
+			string_list_append(&extra_refs, full_name)->util = commit;
 		if (!commit->util)
 			commit->util = full_name;
 	}
 }
 
-static void handle_tags_and_duplicates(struct string_list *extra_refs)
+static void handle_tags_and_duplicates(void)
 {
 	struct commit *commit;
 	int i;
 
-	for (i = extra_refs->nr - 1; i >= 0; i--) {
-		const char *name = extra_refs->items[i].string;
-		struct object *object = extra_refs->items[i].util;
+	for (i = extra_refs.nr - 1; i >= 0; i--) {
+		const char *name = extra_refs.items[i].string;
+		struct object *object = extra_refs.items[i].util;
 		switch (object->type) {
 		case OBJ_TAG:
 			handle_tag(name, (struct tag *)object);
@@ -657,7 +657,6 @@ int cmd_fast_export(int argc, const char **argv, const char *prefix)
 {
 	struct rev_info revs;
 	struct object_array commits = OBJECT_ARRAY_INIT;
-	struct string_list extra_refs = STRING_LIST_INIT_NODUP;
 	struct commit *commit;
 	char *export_filename = NULL, *import_filename = NULL;
 	uint32_t lastimportid;
@@ -709,7 +708,7 @@ int cmd_fast_export(int argc, const char **argv, const char *prefix)
 	if (import_filename && revs.prune_data.nr)
 		full_tree = 1;
 
-	get_tags_and_duplicates(&revs.cmdline, &extra_refs);
+	get_tags_and_duplicates(&revs.cmdline);
 
 	if (prepare_revision_walk(&revs))
 		die("revision walk setup failed");
@@ -725,7 +724,7 @@ int cmd_fast_export(int argc, const char **argv, const char *prefix)
 		}
 	}
 
-	handle_tags_and_duplicates(&extra_refs);
+	handle_tags_and_duplicates();
 
 	if (export_filename && lastimportid != last_idnum)
 		export_marks(export_filename);
-- 
1.8.4-337-g7358a66-dirty

[PATCH 2/2] fast-export: refactor get_tags_and_duplicates()

[Felipe Contreras]

Split into a separate helper function get_commit() so that the part that
finds the relevant commit, and the part that does something with it
(handle tag object, etc.) are in different places.

No functional changes.

Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
---
 builtin/fast-export.c | 68 ++++++++++++++++++++++++++++-----------------------
 1 file changed, 38 insertions(+), 30 deletions(-)

diff --git a/builtin/fast-export.c b/builtin/fast-export.c
index 957392c..03e1090 100644
--- a/builtin/fast-export.c
+++ b/builtin/fast-export.c
@@ -485,9 +485,32 @@ static void handle_tag(const char *name, struct tag *tag)
 	       (int)message_size, (int)message_size, message ? message : "");
 }
 
+static struct commit *get_commit(struct rev_cmdline_entry *e, char *full_name)
+{
+	switch (e->item->type) {
+	case OBJ_COMMIT:
+		return (struct commit *)e->item;
+	case OBJ_TAG: {
+		struct tag *tag = (struct tag *)e->item;
+
+		/* handle nested tags */
+		while (tag && tag->object.type == OBJ_TAG) {
+			parse_object(tag->object.sha1);
+			string_list_append(&extra_refs, full_name)->util = tag;
+			tag = (struct tag *)tag->tagged;
+		}
+		if (!tag)
+			die("Tag %s points nowhere?", e->name);
+		return (struct commit *)tag;
+		break;
+	}
+	default:
+		return NULL;
+	}
+}
+
 static void get_tags_and_duplicates(struct rev_cmdline_info *info)
 {
-	struct tag *tag;
 	int i;
 
 	for (i = 0; i < info->nr; i++) {
@@ -502,41 +525,26 @@ static void get_tags_and_duplicates(struct rev_cmdline_info *info)
 		if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
 			continue;
 
-		switch (e->item->type) {
-		case OBJ_COMMIT:
-			commit = (struct commit *)e->item;
-			break;
-		case OBJ_TAG:
-			tag = (struct tag *)e->item;
-
-			/* handle nested tags */
-			while (tag && tag->object.type == OBJ_TAG) {
-				parse_object(tag->object.sha1);
-				string_list_append(&extra_refs, full_name)->util = tag;
-				tag = (struct tag *)tag->tagged;
-			}
-			if (!tag)
-				die ("Tag %s points nowhere?", e->name);
-			switch(tag->object.type) {
-			case OBJ_COMMIT:
-				commit = (struct commit *)tag;
-				break;
-			case OBJ_BLOB:
-				export_blob(tag->object.sha1);
-				continue;
-			default: /* OBJ_TAG (nested tags) is already handled */
-				warning("Tag points to object of unexpected type %s, skipping.",
-					typename(tag->object.type));
-				continue;
-			}
-			break;
-		default:
+		commit = get_commit(e, full_name);
+		if (!commit) {
 			warning("%s: Unexpected object of type %s, skipping.",
 				e->name,
 				typename(e->item->type));
 			continue;
 		}
 
+		switch(commit->object.type) {
+		case OBJ_COMMIT:
+			break;
+		case OBJ_BLOB:
+			export_blob(commit->object.sha1);
+			continue;
+		default: /* OBJ_TAG (nested tags) is already handled */
+			warning("Tag points to object of unexpected type %s, skipping.",
+				typename(commit->object.type));
+			continue;
+		}
+
 		/*
 		 * This ref will not be updated through a commit, lets make
 		 * sure it gets properly updated eventually.
-- 
1.8.4-337-g7358a66-dirty

Re: [PATCH 2/2] fast-export: refactor get_tags_and_duplicates()

[Junio C Hamano]

Felipe Contreras <felipe.contreras@gmail.com> writes:

> Split into a separate helper function get_commit() so that the part that
> finds the relevant commit, and the part that does something with it
> (handle tag object, etc.) are in different places.
>
> No functional changes.

Actually, the old code used to use commit unchecked if e->item->type
said it is OBJ_COMMIT but e->item was somehow NULL.  The new code
checks this case and skips with a warning(), which I think is an
improvement, if not a bugfix (it only makes it easier to diagnose a
bug in the code that populates rev_cmdline_entry).

Thanks; will queue.

> Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
> ---
>  builtin/fast-export.c | 68 ++++++++++++++++++++++++++++-----------------------
>  1 file changed, 38 insertions(+), 30 deletions(-)
>
> diff --git a/builtin/fast-export.c b/builtin/fast-export.c
> index 957392c..03e1090 100644
> --- a/builtin/fast-export.c
> +++ b/builtin/fast-export.c
> @@ -485,9 +485,32 @@ static void handle_tag(const char *name, struct tag *tag)
>  	       (int)message_size, (int)message_size, message ? message : "");
>  }
>  
> +static struct commit *get_commit(struct rev_cmdline_entry *e, char *full_name)
> +{
> +	switch (e->item->type) {
> +	case OBJ_COMMIT:
> +		return (struct commit *)e->item;
> +	case OBJ_TAG: {
> +		struct tag *tag = (struct tag *)e->item;
> +
> +		/* handle nested tags */
> +		while (tag && tag->object.type == OBJ_TAG) {
> +			parse_object(tag->object.sha1);
> +			string_list_append(&extra_refs, full_name)->util = tag;
> +			tag = (struct tag *)tag->tagged;
> +		}
> +		if (!tag)
> +			die("Tag %s points nowhere?", e->name);
> +		return (struct commit *)tag;
> +		break;
> +	}
> +	default:
> +		return NULL;
> +	}
> +}
> +
>  static void get_tags_and_duplicates(struct rev_cmdline_info *info)
>  {
> -	struct tag *tag;
>  	int i;
>  
>  	for (i = 0; i < info->nr; i++) {
> @@ -502,41 +525,26 @@ static void get_tags_and_duplicates(struct rev_cmdline_info *info)
>  		if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
>  			continue;
>  
> -		switch (e->item->type) {
> -		case OBJ_COMMIT:
> -			commit = (struct commit *)e->item;
> -			break;
> -		case OBJ_TAG:
> -			tag = (struct tag *)e->item;
> -
> -			/* handle nested tags */
> -			while (tag && tag->object.type == OBJ_TAG) {
> -				parse_object(tag->object.sha1);
> -				string_list_append(&extra_refs, full_name)->util = tag;
> -				tag = (struct tag *)tag->tagged;
> -			}
> -			if (!tag)
> -				die ("Tag %s points nowhere?", e->name);
> -			switch(tag->object.type) {
> -			case OBJ_COMMIT:
> -				commit = (struct commit *)tag;
> -				break;
> -			case OBJ_BLOB:
> -				export_blob(tag->object.sha1);
> -				continue;
> -			default: /* OBJ_TAG (nested tags) is already handled */
> -				warning("Tag points to object of unexpected type %s, skipping.",
> -					typename(tag->object.type));
> -				continue;
> -			}
> -			break;
> -		default:
> +		commit = get_commit(e, full_name);
> +		if (!commit) {
>  			warning("%s: Unexpected object of type %s, skipping.",
>  				e->name,
>  				typename(e->item->type));
>  			continue;
>  		}
>  
> +		switch(commit->object.type) {
> +		case OBJ_COMMIT:
> +			break;
> +		case OBJ_BLOB:
> +			export_blob(commit->object.sha1);
> +			continue;
> +		default: /* OBJ_TAG (nested tags) is already handled */
> +			warning("Tag points to object of unexpected type %s, skipping.",
> +				typename(commit->object.type));
> +			continue;
> +		}
> +
>  		/*
>  		 * This ref will not be updated through a commit, lets make
>  		 * sure it gets properly updated eventually.

Re: [PATCH 2/2] fast-export: refactor get_tags_and_duplicates()

[Junio C Hamano]

Junio C Hamano <gitster@pobox.com> writes:

> Felipe Contreras <felipe.contreras@gmail.com> writes:
>
>> Split into a separate helper function get_commit() so that the part that
>> finds the relevant commit, and the part that does something with it
>> (handle tag object, etc.) are in different places.
>>
>> No functional changes.
>
> Actually, the old code used to use commit unchecked if e->item->type
> said it is OBJ_COMMIT but e->item was somehow NULL.  The new code
> checks ...

Nah, I am an idiot.  We will segfault where we check e->item->type
before reaching that far down the code either way, so this is really
a no-op but makes it easier to fix (if we wanted to---I do not think
it matters that much).