Merge pull request libgit2#5189 from libgit2/ethomson/attrs_from_head

Optionally read `.gitattributes` from HEAD

Author: ethomson

include/git2/attr.h

@@ -119,13 +119,20 @@ GIT_EXTERN(git_attr_value_t) git_attr_value(const char *attr);
 #define GIT_ATTR_CHECK_INDEX_ONLY		2
 
 /**
- * Check attribute flags: Using the system attributes file.
+ * Check attribute flags: controlling extended attribute behavior.
  *
  * Normally, attribute checks include looking in the /etc (or system
  * equivalent) directory for a `gitattributes` file.  Passing this
  * flag will cause attribute checks to ignore that file.
+ * equivalent) directory for a `gitattributes` file.  Passing the
+ * `GIT_ATTR_CHECK_NO_SYSTEM` flag will cause attribute checks to
+ * ignore that file.
+ *
+ * Passing the `GIT_ATTR_CHECK_INCLUDE_HEAD` flag will use attributes
+ * from a `.gitattributes` file in the repository at the HEAD revision.
  */
-#define GIT_ATTR_CHECK_NO_SYSTEM		(1 << 2)
+#define GIT_ATTR_CHECK_NO_SYSTEM        (1 << 2)
+#define GIT_ATTR_CHECK_INCLUDE_HEAD     (1 << 3)
 
 /**
  * Look up the value of one git attribute for path.

include/git2/blob.h

@@ -96,6 +96,39 @@ GIT_EXTERN(const void *) git_blob_rawcontent(const git_blob *blob);
  */
 GIT_EXTERN(git_off_t) git_blob_rawsize(const git_blob *blob);
 
+/**
+ * Flags to control the functionality of `git_blob_filter`.
+ */
+typedef enum {
+	/** When set, filters will not be applied to binary files. */
+	GIT_BLOB_FILTER_CHECK_FOR_BINARY = (1 << 0),
+
+	/**
+	 * When set, filters will not load configuration from the
+	 * system-wide `gitattributes` in `/etc` (or system equivalent).
+	 */
+	GIT_BLOB_FILTER_NO_SYSTEM_ATTRIBUTES = (1 << 1),
+
+	/**
+	 * When set, filters will be loaded from a `.gitattributes` file
+	 * in the HEAD commit.
+	 */
+	GIT_BLOB_FILTER_ATTTRIBUTES_FROM_HEAD = (1 << 2),
+} git_blob_filter_flag_t;
+
+/**
+ * The options used when applying filter options to a file.
+ */
+typedef struct {
+	int version;
+
+	/** Flags to control the filtering process */
+	git_blob_filter_flag_t flags;
+} git_blob_filter_options;
+
+#define GIT_BLOB_FILTER_OPTIONS_VERSION 1
+#define GIT_BLOB_FILTER_OPTIONS_INIT {GIT_BLOB_FILTER_OPTIONS_VERSION, GIT_BLOB_FILTER_CHECK_FOR_BINARY}
+
 /**
  * Get a buffer with the filtered content of a blob.
  *
@@ -115,15 +148,14 @@ GIT_EXTERN(git_off_t) git_blob_rawsize(const git_blob *blob);
  * @param out The git_buf to be filled in
  * @param blob Pointer to the blob
  * @param as_path Path used for file attribute lookups, etc.
- * @param check_for_binary_data Should this test if blob content contains
- *        NUL bytes / looks like binary data before applying filters?
+ * @param opts Options to use for filtering the blob
  * @return 0 on success or an error code
  */
-GIT_EXTERN(int) git_blob_filtered_content(
+GIT_EXTERN(int) git_blob_filter(
 	git_buf *out,
 	git_blob *blob,
 	const char *as_path,
-	int check_for_binary_data);
+	git_blob_filter_options *opts);
 
 /**
  * Read a file from the working folder of a repository

include/git2/deprecated.h

@@ -90,6 +90,13 @@ GIT_EXTERN(int) git_blob_create_fromstream_commit(
 GIT_EXTERN(int) git_blob_create_frombuffer(
 	git_oid *id, git_repository *repo, const void *buffer, size_t len);
 
+/** Deprecated in favor of @see git_blob_filter */
+GIT_EXTERN(int) git_blob_filtered_content(
+	git_buf *out,
+	git_blob *blob,
+	const char *as_path,
+	int check_for_binary_data);
+
 /**@}*/
 
 /** @name Deprecated Buffer Functions

include/git2/filter.h

@@ -40,7 +40,15 @@ typedef enum {
  */
 typedef enum {
 	GIT_FILTER_DEFAULT = 0u,
+
+	/** Don't error for `safecrlf` violations, allow them to continue. */
 	GIT_FILTER_ALLOW_UNSAFE = (1u << 0),
+
+	/** Don't load `/etc/gitattributes` (or the system equivalent) */
+	GIT_FILTER_NO_SYSTEM_ATTRIBUTES = (1u << 1),
+
+	/** Load attributes from `.gitattributes` in the root of HEAD */
+	GIT_FILTER_ATTRIBUTES_FROM_HEAD = (1u << 2),
 } git_filter_flag_t;
 
 /**

src/attr.c

@@ -305,7 +305,10 @@ static int system_attr_file(
 	return 0;
 }
 
-static int attr_setup(git_repository *repo, git_attr_session *attr_session)
+static int attr_setup(
+	git_repository *repo,
+	git_attr_session *attr_session,
+	uint32_t flags)
 {
 	git_buf path = GIT_BUF_INIT;
 	git_index *idx = NULL;
@@ -352,6 +355,11 @@ static int attr_setup(git_repository *repo, git_attr_session *attr_session)
 				       NULL, GIT_ATTR_FILE, true)) < 0)
 			goto out;
 
+	if ((flags & GIT_ATTR_CHECK_INCLUDE_HEAD) != 0 &&
+	    (error = preload_attr_file(repo, attr_session, GIT_ATTR_FILE__FROM_HEAD,
+				       NULL, GIT_ATTR_FILE, true)) < 0)
+		goto out;
+
 	if (attr_session)
 		attr_session->init_setup = 1;
 
@@ -428,6 +436,9 @@ static int attr_decide_sources(
 		break;
 	}
 
+	if ((flags & GIT_ATTR_CHECK_INCLUDE_HEAD) != 0)
+		srcs[count++] = GIT_ATTR_FILE__FROM_HEAD;
+
 	return count;
 }
 
@@ -460,7 +471,7 @@ static int push_attr_file(
 static int push_one_attr(void *ref, const char *path)
 {
 	attr_walk_up_info *info = (attr_walk_up_info *)ref;
-	git_attr_file_source src[2];
+	git_attr_file_source src[GIT_ATTR_FILE_NUM_SOURCES];
 	int error = 0, n_src, i;
 	bool allow_macros;
 
@@ -499,7 +510,7 @@ static int collect_attr_files(
 	const char *workdir = git_repository_workdir(repo);
 	attr_walk_up_info info = { NULL };
 
-	if ((error = attr_setup(repo, attr_session)) < 0)
+	if ((error = attr_setup(repo, attr_session, flags)) < 0)
 		return error;
 
 	/* Resolve path in a non-bare repo */

src/attr_file.c

@@ -109,6 +109,8 @@ int git_attr_file__load(
 	bool allow_macros)
 {
 	int error = 0;
+	git_tree *tree = NULL;
+	git_tree_entry *tree_entry = NULL;
 	git_blob *blob = NULL;
 	git_buf content = GIT_BUF_INIT;
 	const char *content_str;
@@ -117,6 +119,8 @@ int git_attr_file__load(
 	bool nonexistent = false;
 	int bom_offset;
 	git_bom_t bom;
+	git_oid id;
+	git_off_t blobsize;
 
 	*out = NULL;
 
@@ -125,9 +129,6 @@ int git_attr_file__load(
 		/* in-memory attribute file doesn't need data */
 		break;
 	case GIT_ATTR_FILE__FROM_INDEX: {
-		git_oid id;
-		git_off_t blobsize;
-
 		if ((error = attr_file_oid_from_index(&id, repo, entry->path)) < 0 ||
 			(error = git_blob_lookup(&blob, repo, &id)) < 0)
 			return error;
@@ -157,6 +158,25 @@ int git_attr_file__load(
 
 		break;
 	}
+	case GIT_ATTR_FILE__FROM_HEAD: {
+		if ((error = git_repository_head_tree(&tree, repo)) < 0 ||
+		    (error = git_tree_entry_bypath(&tree_entry, tree, entry->path)) < 0 ||
+		    (error = git_blob_lookup(&blob, repo, git_tree_entry_id(tree_entry))) < 0)
+			goto cleanup;
+
+		/*
+		 * Do not assume that data straight from the ODB is NULL-terminated;
+		 * copy the contents of a file to a buffer to work on.
+		 */
+		blobsize = git_blob_rawsize(blob);
+
+		GIT_ERROR_CHECK_BLOBSIZE(blobsize);
+		if ((error = git_buf_put(&content,
+			git_blob_rawcontent(blob), (size_t)blobsize)) < 0)
+			goto cleanup;
+
+		break;
+	}
 	default:
 		git_error_set(GIT_ERROR_INVALID, "unknown file source %d", source);
 		return -1;
@@ -188,6 +208,8 @@ int git_attr_file__load(
 		file->nonexistent = 1;
 	else if (source == GIT_ATTR_FILE__FROM_INDEX)
 		git_oid_cpy(&file->cache_data.oid, git_blob_id(blob));
+	else if (source == GIT_ATTR_FILE__FROM_HEAD)
+		git_oid_cpy(&file->cache_data.oid, git_tree_id(tree));
 	else if (source == GIT_ATTR_FILE__FROM_FILE)
 		git_futils_filestamp_set_from_stat(&file->cache_data.stamp, &st);
 	/* else always cacheable */
@@ -196,6 +218,8 @@ int git_attr_file__load(
 
 cleanup:
 	git_blob_free(blob);
+	git_tree_entry_free(tree_entry);
+	git_tree_free(tree);
 	git_buf_dispose(&content);
 
 	return error;
@@ -236,6 +260,19 @@ int git_attr_file__out_of_date(
 		return (git_oid__cmp(&file->cache_data.oid, &id) != 0);
 	}
 
+	case GIT_ATTR_FILE__FROM_HEAD: {
+		git_tree *tree;
+		int error;
+
+		if ((error = git_repository_head_tree(&tree, repo)) < 0)
+			return error;
+
+		error = git_oid__cmp(&file->cache_data.oid, git_tree_id(tree));
+
+		git_tree_free(tree);
+		return error;
+	}
+
 	default:
 		git_error_set(GIT_ERROR_INVALID, "invalid file type %d", file->source);
 		return -1;

src/attr_file.h

@@ -40,8 +40,9 @@ typedef enum {
 	GIT_ATTR_FILE__IN_MEMORY   = 0,
 	GIT_ATTR_FILE__FROM_FILE   = 1,
 	GIT_ATTR_FILE__FROM_INDEX  = 2,
+	GIT_ATTR_FILE__FROM_HEAD   = 3,
 
-	GIT_ATTR_FILE_NUM_SOURCES  = 3
+	GIT_ATTR_FILE_NUM_SOURCES  = 4
 } git_attr_file_source;
 
 extern const char *git_attr__true;

src/blob.c

@@ -400,25 +400,40 @@ int git_blob_is_binary(const git_blob *blob)
 	return git_buf_text_is_binary(&content);
 }
 
-int git_blob_filtered_content(
+int git_blob_filter(
 	git_buf *out,
 	git_blob *blob,
 	const char *path,
-	int check_for_binary_data)
+	git_blob_filter_options *given_opts)
 {
 	int error = 0;
 	git_filter_list *fl = NULL;
+	git_blob_filter_options opts = GIT_BLOB_FILTER_OPTIONS_INIT;
+	git_filter_flag_t flags = GIT_FILTER_DEFAULT;
 
 	assert(blob && path && out);
 
 	git_buf_sanitize(out);
 
-	if (check_for_binary_data && git_blob_is_binary(blob))
+	GIT_ERROR_CHECK_VERSION(
+		given_opts, GIT_BLOB_FILTER_OPTIONS_VERSION, "git_blob_filter_options");
+
+	if (given_opts != NULL)
+		memcpy(&opts, given_opts, sizeof(git_blob_filter_options));
+
+	if ((opts.flags & GIT_BLOB_FILTER_CHECK_FOR_BINARY) != 0 &&
+	    git_blob_is_binary(blob))
 		return 0;
 
+	if ((opts.flags & GIT_BLOB_FILTER_NO_SYSTEM_ATTRIBUTES) != 0)
+		flags |= GIT_FILTER_NO_SYSTEM_ATTRIBUTES;
+
+	if ((opts.flags & GIT_BLOB_FILTER_ATTTRIBUTES_FROM_HEAD) != 0)
+		flags |= GIT_FILTER_ATTRIBUTES_FROM_HEAD;
+
 	if (!(error = git_filter_list_load(
 			&fl, git_blob_owner(blob), blob, path,
-			GIT_FILTER_TO_WORKTREE, GIT_FILTER_DEFAULT))) {
+			GIT_FILTER_TO_WORKTREE, flags))) {
 
 		error = git_filter_list_apply_to_blob(out, fl, blob);
 
@@ -460,3 +475,19 @@ int git_blob_create_fromstream_commit(
 {
 	return git_blob_create_from_stream_commit(out, stream);
 }
+
+int git_blob_filtered_content(
+	git_buf *out,
+	git_blob *blob,
+	const char *path,
+	int check_for_binary_data)
+{
+	git_blob_filter_options opts = GIT_BLOB_FILTER_OPTIONS_INIT;
+
+	if (check_for_binary_data)
+		opts.flags |= GIT_BLOB_FILTER_CHECK_FOR_BINARY;
+	else
+		opts.flags &= ~GIT_BLOB_FILTER_CHECK_FOR_BINARY;
+
+	return git_blob_filter(out, blob, path, &opts);
+}

src/filter.c

@@ -428,13 +428,21 @@ static int filter_list_check_attributes(
 	git_filter_def *fdef,
 	const git_filter_source *src)
 {
-	int error;
-	size_t i;
 	const char **strs = git__calloc(fdef->nattrs, sizeof(const char *));
+	uint32_t flags = 0;
+	size_t i;
+	int error;
+
 	GIT_ERROR_CHECK_ALLOC(strs);
 
+	if ((src->flags & GIT_FILTER_NO_SYSTEM_ATTRIBUTES) != 0)
+		flags |= GIT_ATTR_CHECK_NO_SYSTEM;
+
+	if ((src->flags & GIT_FILTER_ATTRIBUTES_FROM_HEAD) != 0)
+		flags |= GIT_ATTR_CHECK_INCLUDE_HEAD;
+
 	error = git_attr_get_many_with_session(
-		strs, repo, attr_session, 0, src->path, fdef->nattrs, fdef->attrs);
+		strs, repo, attr_session, flags, src->path, fdef->nattrs, fdef->attrs);
 
 	/* if no values were found but no matches are needed, it's okay! */
 	if (error == GIT_ENOTFOUND && !fdef->nmatches) {