Add lax parsing for commit and tag objectsrb/object-parse-flexibility

This changes the behavior of object parsing for commits and tags
so that even when bad data is found inside the object, we will
continue to try to parse as much of the object as we can.  The
existing functions (`git_object_lookup` for example) will still
delete the partially parsed object before returning an error, but
this also adds a new function `git_object_lookup_lax` that will
still return the error, but will also return the object with the
partial data (if we got far enough along in the parsing process to
even create the base object).

1 files changed, 41 insertions, 4 deletions

diff --git a/src/object.h b/src/object.h
index d187c55b7..47f82f36d 100644
--- a/src/object.h
+++ b/src/object.h
@@ -7,6 +7,9 @@
 #ifndef INCLUDE_object_h__
 #define INCLUDE_object_h__
 
+#include "common.h"
+#include "array.h"
+
 /** Base git object for inheritance */
 struct git_object {
 	git_cached_obj cached;
@@ -17,15 +20,49 @@ struct git_object {
 void git_object__free(void *object);
 
 int git_object__from_odb_object(
-	git_object **object_out,
+	git_object **out,
 	git_repository *repo,
 	git_odb_object *odb_obj,
-	git_otype type);
+	git_otype type,
+	bool lax);
 
 int git_object__resolve_to_type(git_object **obj, git_otype type);
 
-int git_oid__parse(git_oid *oid, const char **buffer_out, const char *buffer_end, const char *header);
-
 void git_oid__writebuf(git_buf *buf, const char *header, const git_oid *oid);
 
+enum {
+	GIT_PARSE_BODY_OPTIONAL = -2,
+	GIT_PARSE_BODY = -1,
+	GIT_PARSE_MODE_OPTIONAL = 0,
+	GIT_PARSE_OID = 1,
+	GIT_PARSE_OID_ARRAY = 2,
+	GIT_PARSE_OTYPE = 3,
+	GIT_PARSE_SIGNATURE = 4,
+	GIT_PARSE_TO_EOL = 5,
+};
+
+typedef git_array_t(git_oid) git_oid_array;
+
+typedef struct {
+	const char *tag;
+	size_t taglen;
+	int type;
+	union {
+		git_oid *id;
+		git_otype *otype;
+		char **text;
+		git_signature **sig;
+		git_oid_array *ids;
+		const char **body;
+	} value;
+} git_object_parse_t;
+
+/* parse tagged lines followed by blank line and message body */
+int git_object__parse_lines(
+	git_otype type,
+	git_object_parse_t *parse,
+	const char *buf,
+	const char *buf_end);
+
 #endif
+