Add ident filter

This adds the ident filter (that knows how to replace $Id$) and
tweaks the filter APIs and code so that git_filter_source objects
actually have the updated OID of the object being filtered when
it is a known value.

9 files changed, 170 insertions, 16 deletions

diff --git a/src/blob.c b/src/blob.c
index e6bba033a..97fd6f70d 100644
--- a/src/blob.c
+++ b/src/blob.c
@@ -195,7 +195,7 @@ int git_blob__create_from_paths(
 		if (try_load_filters)
 			/* Load the filters for writing this file to the ODB */
 			error = git_filter_list_load(
-				&fl, repo, hint_path, GIT_FILTER_TO_ODB);
+				&fl, repo, NULL, hint_path, GIT_FILTER_TO_ODB);
 
 		if (error < 0)
 			/* well, that didn't work */;
@@ -331,19 +331,19 @@ int git_blob_is_binary(git_blob *blob)
 int git_blob_filtered_content(
 	git_buffer *out,
 	git_blob *blob,
-	const char *as_path,
+	const char *path,
 	int check_for_binary_data)
 {
 	int error = 0;
 	git_filter_list *fl = NULL;
 
-	assert(blob && as_path && out);
+	assert(blob && path && out);
 
 	if (check_for_binary_data && git_blob_is_binary(blob))
 		return 0;
 
 	if (!(error = git_filter_list_load(
-			&fl, git_blob_owner(blob), as_path, GIT_FILTER_TO_WORKTREE))) {
+			&fl, git_blob_owner(blob), blob, path, GIT_FILTER_TO_WORKTREE))) {
 
 		error = git_filter_list_apply_to_blob(out, fl, blob);
 
diff --git a/src/checkout.c b/src/checkout.c
index 1def58b0a..7e79c9b5e 100644
--- a/src/checkout.c
+++ b/src/checkout.c
@@ -718,7 +718,7 @@ static int blob_content_to_file(
 
 	if (!opts->disable_filters && !git_blob_is_binary(blob))
 		error = git_filter_list_load(
-			&fl, git_blob_owner(blob), path, GIT_FILTER_TO_WORKTREE);
+			&fl, git_blob_owner(blob), blob, path, GIT_FILTER_TO_WORKTREE);
 
 	if (!error)
 		error = git_filter_list_apply_to_blob(&out, fl, blob);
diff --git a/src/crlf.c b/src/crlf.c
index 99c154f70..f61a870da 100644
--- a/src/crlf.c
+++ b/src/crlf.c
@@ -324,11 +324,6 @@ static void crlf_cleanup(
 	git__free(payload);
 }
 
-static void crlf_shutdown(git_filter *self)
-{
-	git__free(self);
-}
-
 git_filter *git_crlf_filter_new(void)
 {
 	struct crlf_filter *f = git__calloc(1, sizeof(struct crlf_filter));
@@ -336,7 +331,7 @@ git_filter *git_crlf_filter_new(void)
 	f->f.version = GIT_FILTER_VERSION;
 	f->f.attributes = "crlf eol text";
 	f->f.initialize = NULL;
-	f->f.shutdown = crlf_shutdown;
+	f->f.shutdown = git_filter_free;
 	f->f.check    = crlf_check;
 	f->f.apply    = crlf_apply;
 	f->f.cleanup  = crlf_cleanup;
diff --git a/src/diff.c b/src/diff.c
index b1cde36bc..4d9ace183 100644
--- a/src/diff.c
+++ b/src/diff.c
@@ -570,7 +570,7 @@ int git_diff__oid_for_file(
 	} else {
 		git_filter_list *fl = NULL;
 
-		result = git_filter_list_load(&fl, repo, path, GIT_FILTER_TO_ODB);
+		result = git_filter_list_load(&fl, repo, NULL, path, GIT_FILTER_TO_ODB);
 		if (!result) {
 			int fd = git_futils_open_ro(full_path.ptr);
 			if (fd < 0)
diff --git a/src/diff_file.c b/src/diff_file.c
index e0e244b65..d02787c75 100644
--- a/src/diff_file.c
+++ b/src/diff_file.c
@@ -311,7 +311,7 @@ static int diff_file_content_load_workdir_file(
 		goto cleanup;
 
 	if ((error = git_filter_list_load(
-			&fl, fc->repo, fc->file->path, GIT_FILTER_TO_ODB)) < 0)
+			&fl, fc->repo, NULL, fc->file->path, GIT_FILTER_TO_ODB)) < 0)
 		goto cleanup;
 
 	/* if there are no filters, try to mmap the file */
diff --git a/src/filter.c b/src/filter.c
index 79ccac0cf..f20611471 100644
--- a/src/filter.c
+++ b/src/filter.c
@@ -103,7 +103,23 @@ static int filter_registry_initialize(void)
 
 	git__on_shutdown(filter_registry_shutdown);
 
-	return git_filter_register(GIT_FILTER_CRLF, git_crlf_filter_new(), 0);
+	/* try to register both default filters */
+	{
+		git_filter *crlf = git_crlf_filter_new();
+		git_filter *ident = git_ident_filter_new();
+
+		if (crlf && git_filter_register(
+				GIT_FILTER_CRLF, crlf, GIT_FILTER_CRLF_PRIORITY) < 0)
+			crlf = NULL;
+		if (ident && git_filter_register(
+				GIT_FILTER_IDENT, ident, GIT_FILTER_IDENT_PRIORITY) < 0)
+			ident = NULL;
+
+		if (!crlf || !ident)
+			return -1;
+	}
+
+	return 0;
 
 cleanup:
 	git_vector_free(&reg->filters);
@@ -132,7 +148,7 @@ static int filter_def_scan_attrs(
 
 		if (scan > start) {
 			(*nattr)++;
-			if (has_eq || *scan == '-' || *scan == '+' || *scan == '!')
+			if (has_eq || *start == '-' || *start == '+' || *start == '!')
 				(*nmatch)++;
 
 			if (has_eq)
@@ -312,6 +328,11 @@ git_filter *git_filter_lookup(const char *name)
 	return fdef->filter;
 }
 
+void git_filter_free(git_filter *filter)
+{
+	git__free(filter);
+}
+
 git_repository *git_filter_source_repo(const git_filter_source *src)
 {
 	return src->repo;
@@ -410,6 +431,7 @@ int git_filter_list_new(
 int git_filter_list_load(
 	git_filter_list **filters,
 	git_repository *repo,
+	git_blob *blob, /* can be NULL */
 	const char *path,
 	git_filter_mode_t mode)
 {
@@ -426,6 +448,8 @@ int git_filter_list_load(
 	src.repo = repo;
 	src.path = path;
 	src.mode = mode;
+	if (blob)
+		git_oid_cpy(&src.oid, git_blob_id(blob));
 
 	git_vector_foreach(&git__filter_registry->filters, idx, fdef) {
 		const char **values = NULL;
@@ -630,5 +654,8 @@ int git_filter_list_apply_to_blob(
 		(char *)git_blob_rawcontent(blob), git_blob_rawsize(blob), 0
 	};
 
+	if (filters)
+		git_oid_cpy(&filters->source.oid, git_blob_id(blob));
+
 	return git_filter_list_apply_to_data(out, filters, &in);
 }
diff --git a/src/filter.h b/src/filter.h
index 1bde1e306..d0ace0f9a 100644
--- a/src/filter.h
+++ b/src/filter.h
@@ -19,10 +19,13 @@ typedef enum {
 	GIT_CRLF_AUTO,
 } git_crlf_t;
 
+extern void git_filter_free(git_filter *filter);
+
 /*
  * Available filters
  */
 
 extern git_filter *git_crlf_filter_new(void);
+extern git_filter *git_ident_filter_new(void);
 
 #endif
diff --git a/src/ident.c b/src/ident.c
new file mode 100644
index 000000000..aedb973f9
--- /dev/null
+++ b/src/ident.c
@@ -0,0 +1,128 @@
+/*
+ * Copyright (C) the libgit2 contributors. All rights reserved.
+ *
+ * This file is part of libgit2, distributed under the GNU GPL v2 with
+ * a Linking Exception. For full terms see the included COPYING file.
+ */
+
+#include "git2/sys/filter.h"
+#include "filter.h"
+#include "buffer.h"
+
+int ident_find_id(
+	const char **id_start, const char **id_end, const char *start, size_t len)
+{
+	const char *found;
+
+	while (len > 0 && (found = memchr(start, '$', len)) != NULL) {
+		size_t remaining = len - (size_t)(found - start);
+		if (remaining < 3)
+			return GIT_ENOTFOUND;
+		if (found[1] == 'I' && found[2] == 'd')
+			break;
+		start = found + 1;
+		len = remaining - 1;
+	}
+
+	if (len < 3)
+		return GIT_ENOTFOUND;
+	*id_start = found;
+
+	if ((found = memchr(found + 3, '$', len - 3)) == NULL)
+		return GIT_ENOTFOUND;
+
+	*id_end = found + 1;
+	return 0;
+}
+
+static int ident_insert_id(
+	git_buffer *to, const git_buffer *from, const git_filter_source *src)
+{
+	char oid[GIT_OID_HEXSZ+1];
+	const char *id_start, *id_end, *from_end = from->ptr + from->size;
+	size_t need_size;
+	git_buf to_buf = GIT_BUF_FROM_BUFFER(to);
+
+	/* replace $Id$ with blob id */
+
+	if (!git_filter_source_id(src))
+		return GIT_ENOTFOUND;
+
+	git_oid_tostr(oid, sizeof(oid), git_filter_source_id(src));
+
+	if (ident_find_id(&id_start, &id_end, from->ptr, from->size) < 0)
+		return GIT_ENOTFOUND;
+
+	need_size = (size_t)(id_start - from->ptr) +
+		5 /* "$Id: " */ + GIT_OID_HEXSZ + 1 /* "$" */ +
+		(size_t)(from_end - id_end);
+
+	if (git_buf_grow(&to_buf, need_size) < 0)
+		return -1;
+
+	git_buf_set(&to_buf, from->ptr, (size_t)(id_start - from->ptr));
+	git_buf_put(&to_buf, "$Id: ", 5);
+	git_buf_put(&to_buf, oid, GIT_OID_HEXSZ);
+	git_buf_putc(&to_buf, '$');
+	git_buf_put(&to_buf, id_end, (size_t)(from_end - id_end));
+
+	if (git_buf_oom(&to_buf))
+		return -1;
+
+	git_buffer_from_buf(to, &to_buf);
+	return 0;
+}
+
+static int ident_remove_id(
+	git_buffer *to, const git_buffer *from)
+{
+	const char *id_start, *id_end, *from_end = from->ptr + from->size;
+	size_t need_size;
+	git_buf to_buf = GIT_BUF_FROM_BUFFER(to);
+
+	if (ident_find_id(&id_start, &id_end, from->ptr, from->size) < 0)
+		return GIT_ENOTFOUND;
+
+	need_size = (size_t)(id_start - from->ptr) +
+		4 /* "$Id$" */ + (size_t)(from_end - id_end);
+
+	if (git_buf_grow(&to_buf, need_size) < 0)
+		return -1;
+
+	git_buf_set(&to_buf, from->ptr, (size_t)(id_start - from->ptr));
+	git_buf_put(&to_buf, "$Id$", 4);
+	git_buf_put(&to_buf, id_end, (size_t)(from_end - id_end));
+
+	if (git_buf_oom(&to_buf))
+		return -1;
+
+	git_buffer_from_buf(to, &to_buf);
+	return 0;
+}
+
+static int ident_apply(
+	git_filter        *self,
+	void              **payload,
+	git_buffer        *to,
+	const git_buffer  *from,
+	const git_filter_source *src)
+{
+	GIT_UNUSED(self); GIT_UNUSED(payload);
+
+	if (git_filter_source_mode(src) == GIT_FILTER_SMUDGE)
+		return ident_insert_id(to, from, src);
+	else
+		return ident_remove_id(to, from);
+}
+
+git_filter *git_ident_filter_new(void)
+{
+	git_filter *f = git__calloc(1, sizeof(git_filter));
+
+	f->version = GIT_FILTER_VERSION;
+	f->attributes = "+ident"; /* apply to files with ident attribute set */
+	f->shutdown = git_filter_free;
+	f->apply    = ident_apply;
+
+	return f;
+}
diff --git a/src/repository.c b/src/repository.c
index 94700e4e3..76e8228b7 100644
--- a/src/repository.c
+++ b/src/repository.c
@@ -1671,7 +1671,8 @@ int git_repository_hashfile(
 
 	/* passing empty string for "as_path" indicated --no-filters */
 	if (strlen(as_path) > 0) {
-		error = git_filter_list_load(&fl, repo, as_path, GIT_FILTER_TO_ODB);
+		error = git_filter_list_load(
+			&fl, repo, NULL, as_path, GIT_FILTER_TO_ODB);
 		if (error < 0)
 			return error;
 	} else {