Merge pull request libgit2#6011 from libgit2/ethomson/filter_apply

filter: filter drivers stop taking git_buf as user input

Author: ethomson

include/git2/sys/filter.h

@@ -167,17 +167,18 @@ typedef void GIT_CALLBACK(git_filter_shutdown_fn)(git_filter *self);
  *
  * The `payload` will be a pointer to a reference payload for the filter.
  * This will start as NULL, but `check` can assign to this pointer for
- * later use by the `apply` callback.  Note that the value should be heap
- * allocated (not stack), so that it doesn't go away before the `apply`
+ * later use by the `stream` callback.  Note that the value should be heap
+ * allocated (not stack), so that it doesn't go away before the `stream`
  * callback can use it.  If a filter allocates and assigns a value to the
  * `payload`, it will need a `cleanup` callback to free the payload.
  */
 typedef int GIT_CALLBACK(git_filter_check_fn)(
-	git_filter  *self,
-	void       **payload, /* points to NULL ptr on entry, may be set */
+	git_filter              *self,
+	void                   **payload, /* NULL on entry, may be set */
 	const git_filter_source *src,
-	const char **attr_values);
+	const char             **attr_values);
 
+#ifndef GIT_DEPRECATE_HARD
 /**
  * Callback to actually perform the data filtering
  *
@@ -189,32 +190,45 @@ typedef int GIT_CALLBACK(git_filter_check_fn)(
  *
  * The `payload` value will refer to any payload that was set by the
  * `check` callback.  It may be read from or written to as needed.
+ *
+ * @deprecated use git_filter_stream_fn
  */
 typedef int GIT_CALLBACK(git_filter_apply_fn)(
-	git_filter    *self,
-	void         **payload, /* may be read and/or set */
-	git_buf       *to,
-	const git_buf *from,
+	git_filter              *self,
+	void                   **payload, /* may be read and/or set */
+	git_buf                 *to,
+	const git_buf           *from,
 	const git_filter_source *src);
+#endif
 
+/**
+ * Callback to perform the data filtering.
+ *
+ * Specified as `filter.stream`, this is a callback that filters data
+ * in a streaming manner.  This function will provide a
+ * `git_writestream` that will the original data will be written to;
+ * with that data, the `git_writestream` will then perform the filter
+ * translation and stream the filtered data out to the `next` location.
+ */
 typedef int GIT_CALLBACK(git_filter_stream_fn)(
-	git_writestream **out,
-	git_filter *self,
-	void **payload,
+	git_writestream        **out,
+	git_filter              *self,
+	void                   **payload,
 	const git_filter_source *src,
-	git_writestream *next);
+	git_writestream         *next);
 
 /**
  * Callback to clean up after filtering has been applied
  *
  * Specified as `filter.cleanup`, this is an optional callback invoked
- * after the filter has been applied.  If the `check` or `apply` callbacks
- * allocated a `payload` to keep per-source filter state, use this
- * callback to free that payload and release resources as required.
+ * after the filter has been applied.  If the `check`, `apply`, or
+ * `stream` callbacks allocated a `payload` to keep per-source filter
+ * state, use this callback to free that payload and release resources
+ * as required.
  */
 typedef void GIT_CALLBACK(git_filter_cleanup_fn)(
-	git_filter *self,
-	void       *payload);
+	git_filter              *self,
+	void                    *payload);
 
 /**
  * Filter structure used to register custom filters.
@@ -248,21 +262,28 @@ struct git_filter {
 	/**
 	 * Called to determine whether the filter should be invoked for a
 	 * given file.  If this function returns `GIT_PASSTHROUGH` then the
-	 * `apply` function will not be invoked and the contents will be passed
-	 * through unmodified.
+	 * `stream` or `apply` functions will not be invoked and the
+	 * contents will be passed through unmodified.
 	 */
 	git_filter_check_fn    check;
 
+#ifdef GIT_DEPRECATE_HARD
+	void *reserved;
+#else
 	/**
-	 * Called to actually apply the filter to file contents.  If this
-	 * function returns `GIT_PASSTHROUGH` then the contents will be passed
-	 * through unmodified.
+	 * Provided for backward compatibility; this will apply the
+	 * filter to the given contents in a `git_buf`.  Callers should
+	 * provide a `stream` function instead.
 	 */
 	git_filter_apply_fn    apply;
+#endif
 
 	/**
-	 * Called to apply the filter in a streaming manner.  If this is not
-	 * specified then the system will call `apply` with the whole buffer.
+	 * Called to apply the filter, this function will provide a
+	 * `git_writestream` that will the original data will be
+	 * written to; with that data, the `git_writestream` will then
+	 * perform the filter translation and stream the filtered data
+	 * out to the `next` location.
 	 */
 	git_filter_stream_fn   stream;
 
@@ -289,9 +310,9 @@ GIT_EXTERN(int) git_filter_init(git_filter *filter, unsigned int version);
  * As mentioned elsewhere, the initialize callback will not be invoked
  * immediately.  It is deferred until the filter is used in some way.
  *
- * A filter's attribute checks and `check` and `apply` callbacks will be
- * issued in order of `priority` on smudge (to workdir), and in reverse
- * order of `priority` on clean (to odb).
+ * A filter's attribute checks and `check` and `stream` (or `apply`)
+ * callbacks will be issued in order of `priority` on smudge (to
+ * workdir), and in reverse order of `priority` on clean (to odb).
  *
  * Two filters are preregistered with libgit2:
  * - GIT_FILTER_CRLF with priority 0

src/crlf.c

@@ -386,6 +386,17 @@ static int crlf_apply(
 		return crlf_apply_to_odb(*payload, to, from, src);
 }
 
+static int crlf_stream(
+	git_writestream **out,
+	git_filter *self,
+	void **payload,
+	const git_filter_source *src,
+	git_writestream *next)
+{
+	return git_filter_buffered_stream_new(out,
+		self, crlf_apply, NULL, payload, src, next);
+}
+
 static void crlf_cleanup(
 	git_filter *self,
 	void       *payload)
@@ -405,7 +416,7 @@ git_filter *git_crlf_filter_new(void)
 	f->f.initialize = NULL;
 	f->f.shutdown = git_filter_free;
 	f->f.check    = crlf_check;
-	f->f.apply    = crlf_apply;
+	f->f.stream   = crlf_stream;
 	f->f.cleanup  = crlf_cleanup;
 
 	return (git_filter *)f;

src/filter.c

@@ -839,9 +839,10 @@ int git_filter_list_apply_to_blob(
 	return error;
 }
 
-struct proxy_stream {
+struct buffered_stream {
 	git_writestream parent;
 	git_filter *filter;
+	int (*write_fn)(git_filter *, void **, git_buf *, const git_buf *, const git_filter_source *);
 	const git_filter_source *source;
 	void **payload;
 	git_buf input;
@@ -850,92 +851,120 @@ struct proxy_stream {
 	git_writestream *target;
 };
 
-static int proxy_stream_write(
+static int buffered_stream_write(
 	git_writestream *s, const char *buffer, size_t len)
 {
-	struct proxy_stream *proxy_stream = (struct proxy_stream *)s;
-	GIT_ASSERT_ARG(proxy_stream);
+	struct buffered_stream *buffered_stream = (struct buffered_stream *)s;
+	GIT_ASSERT_ARG(buffered_stream);
 
-	return git_buf_put(&proxy_stream->input, buffer, len);
+	return git_buf_put(&buffered_stream->input, buffer, len);
 }
 
-static int proxy_stream_close(git_writestream *s)
+static int buffered_stream_close(git_writestream *s)
 {
-	struct proxy_stream *proxy_stream = (struct proxy_stream *)s;
+	struct buffered_stream *buffered_stream = (struct buffered_stream *)s;
 	git_buf *writebuf;
 	git_error_state error_state = {0};
 	int error;
 
-	GIT_ASSERT_ARG(proxy_stream);
+	GIT_ASSERT_ARG(buffered_stream);
 
-	error = proxy_stream->filter->apply(
-		proxy_stream->filter,
-		proxy_stream->payload,
-		proxy_stream->output,
-		&proxy_stream->input,
-		proxy_stream->source);
+	error = buffered_stream->write_fn(
+		buffered_stream->filter,
+		buffered_stream->payload,
+		buffered_stream->output,
+		&buffered_stream->input,
+		buffered_stream->source);
 
 	if (error == GIT_PASSTHROUGH) {
-		writebuf = &proxy_stream->input;
+		writebuf = &buffered_stream->input;
 	} else if (error == 0) {
-		if ((error = git_buf_sanitize(proxy_stream->output)) < 0)
+		if ((error = git_buf_sanitize(buffered_stream->output)) < 0)
 			return error;
 
-		writebuf = proxy_stream->output;
+		writebuf = buffered_stream->output;
 	} else {
 		/* close stream before erroring out taking care
 		 * to preserve the original error */
 		git_error_state_capture(&error_state, error);
-		proxy_stream->target->close(proxy_stream->target);
+		buffered_stream->target->close(buffered_stream->target);
 		git_error_state_restore(&error_state);
 		return error;
 	}
 
-	if ((error = proxy_stream->target->write(
-			proxy_stream->target, writebuf->ptr, writebuf->size)) == 0)
-		error = proxy_stream->target->close(proxy_stream->target);
+	if ((error = buffered_stream->target->write(
+			buffered_stream->target, writebuf->ptr, writebuf->size)) == 0)
+		error = buffered_stream->target->close(buffered_stream->target);
 
 	return error;
 }
 
-static void proxy_stream_free(git_writestream *s)
+static void buffered_stream_free(git_writestream *s)
 {
-	struct proxy_stream *proxy_stream = (struct proxy_stream *)s;
+	struct buffered_stream *buffered_stream = (struct buffered_stream *)s;
 
-	if (proxy_stream) {
-		git_buf_dispose(&proxy_stream->input);
-		git_buf_dispose(&proxy_stream->temp_buf);
-		git__free(proxy_stream);
+	if (buffered_stream) {
+		git_buf_dispose(&buffered_stream->input);
+		git_buf_dispose(&buffered_stream->temp_buf);
+		git__free(buffered_stream);
 	}
 }
 
-static int proxy_stream_init(
+int git_filter_buffered_stream_new(
 	git_writestream **out,
 	git_filter *filter,
+	int (*write_fn)(git_filter *, void **, git_buf *, const git_buf *, const git_filter_source *),
 	git_buf *temp_buf,
 	void **payload,
 	const git_filter_source *source,
 	git_writestream *target)
 {
-	struct proxy_stream *proxy_stream = git__calloc(1, sizeof(struct proxy_stream));
-	GIT_ERROR_CHECK_ALLOC(proxy_stream);
-
-	proxy_stream->parent.write = proxy_stream_write;
-	proxy_stream->parent.close = proxy_stream_close;
-	proxy_stream->parent.free = proxy_stream_free;
-	proxy_stream->filter = filter;
-	proxy_stream->payload = payload;
-	proxy_stream->source = source;
-	proxy_stream->target = target;
-	proxy_stream->output = temp_buf ? temp_buf : &proxy_stream->temp_buf;
+	struct buffered_stream *buffered_stream = git__calloc(1, sizeof(struct buffered_stream));
+	GIT_ERROR_CHECK_ALLOC(buffered_stream);
+
+	buffered_stream->parent.write = buffered_stream_write;
+	buffered_stream->parent.close = buffered_stream_close;
+	buffered_stream->parent.free = buffered_stream_free;
+	buffered_stream->filter = filter;
+	buffered_stream->write_fn = write_fn;
+	buffered_stream->output = temp_buf ? temp_buf : &buffered_stream->temp_buf;
+	buffered_stream->payload = payload;
+	buffered_stream->source = source;
+	buffered_stream->target = target;
 
 	if (temp_buf)
 		git_buf_clear(temp_buf);
 
-	*out = (git_writestream *)proxy_stream;
+	*out = (git_writestream *)buffered_stream;
 	return 0;
 }
 
+static int setup_stream(
+	git_writestream **out,
+	git_filter_entry *fe,
+	git_filter_list *filters,
+	git_writestream *last_stream)
+{
+#ifndef GIT_DEPRECATE_HARD
+	GIT_ASSERT(fe->filter->stream || fe->filter->apply);
+
+	/*
+	 * If necessary, create a stream that proxies the traditional
+	 * application.
+	 */
+	if (!fe->filter->stream) {
+		/* Create a stream that proxies the one-shot apply */
+		return git_filter_buffered_stream_new(out,
+			fe->filter, fe->filter->apply, filters->temp_buf,
+			&fe->payload, &filters->source, last_stream);
+	}
+#endif
+
+	GIT_ASSERT(fe->filter->stream);
+	return fe->filter->stream(out, fe->filter,
+		&fe->payload, &filters->source, last_stream);
+}
+
 static int stream_list_init(
 	git_writestream **out,
 	git_vector *streams,
@@ -957,22 +986,11 @@ static int stream_list_init(
 	for (i = 0; i < git_array_size(filters->filters); ++i) {
 		size_t filter_idx = (filters->source.mode == GIT_FILTER_TO_WORKTREE) ?
 			git_array_size(filters->filters) - 1 - i : i;
+
 		git_filter_entry *fe = git_array_get(filters->filters, filter_idx);
 		git_writestream *filter_stream;
 
-		GIT_ASSERT(fe->filter->stream || fe->filter->apply);
-
-		/* If necessary, create a stream that proxies the traditional
-		 * application.
-		 */
-		if (fe->filter->stream)
-			error = fe->filter->stream(&filter_stream, fe->filter,
-				&fe->payload, &filters->source, last_stream);
-		else
-			/* Create a stream that proxies the one-shot apply */
-			error = proxy_stream_init(&filter_stream, fe->filter,
-				filters->temp_buf, &fe->payload, &filters->source,
-				last_stream);
+		error = setup_stream(&filter_stream, fe, filters, last_stream);
 
 		if (error < 0)
 			goto out;

src/filter.h

@@ -11,6 +11,7 @@
 
 #include "attr_file.h"
 #include "git2/filter.h"
+#include "git2/sys/filter.h"
 
 /* Amount of file to examine for NUL byte when checking binary-ness */
 #define GIT_FILTER_BYTES_TO_CHECK_NUL 8000
@@ -51,4 +52,13 @@ extern int git_filter_list__convert_buf(
 extern git_filter *git_crlf_filter_new(void);
 extern git_filter *git_ident_filter_new(void);
 
+extern int git_filter_buffered_stream_new(
+	git_writestream **out,
+	git_filter *filter,
+	int (*write_fn)(git_filter *, void **, git_buf *, const git_buf *, const git_filter_source *),
+	git_buf *temp_buf,
+	void **payload,
+	const git_filter_source *source,
+	git_writestream *target);
+
 #endif