blob: introduce git_blob_filter

Provide a function to filter blobs that allows for more functionality
than the existing `git_blob_filtered_content` function.

2 files changed, 68 insertions, 4 deletions

diff --git a/include/git2/blob.h b/include/git2/blob.h
index 072522d5c..be721edd0 100644
--- a/include/git2/blob.h
+++ b/include/git2/blob.h
@@ -97,6 +97,27 @@ 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),
+} 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.
  *
  * This applies filters as if the blob was being checked out to the
@@ -115,6 +136,24 @@ 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 opts Options to use for filtering the blob
+ * @return 0 on success or an error code
+ */
+GIT_EXTERN(int) git_blob_filter(
+	git_buf *out,
+	git_blob *blob,
+	const char *as_path,
+	git_blob_filter_options *opts);
+
+/**
+ * Get a buffer with the filtered content of a blob.  This is
+ * equivalent to calling `git_blob_filter`, with the only possible
+ * option being the binary check.
+ *
+ * @see git_blob_filter
+ * @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?
  * @return 0 on success or an error code
diff --git a/src/blob.c b/src/blob.c
index 31e6ccf86..efc9fc884 100644
--- a/src/blob.c
+++ b/src/blob.c
@@ -400,25 +400,34 @@ 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 (!(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);
 
@@ -428,6 +437,22 @@ int git_blob_filtered_content(
 	return error;
 }
 
+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);
+}
+
 /* Deprecated functions */
 
 int git_blob_create_frombuffer(