commit 05fd31691c7f3c33867fb0acd5008fa2f555b461
from: Stefan Sperling <stsp@stsp.name>
date: Tue Mar 19 13:47:25 2024 UTC

add support for reading blobs to object_open_io.c

commit - 4c9b88110ffd0deb6eed9861f320246d8354bae2
commit + 05fd31691c7f3c33867fb0acd5008fa2f555b461
blob - babca5450a093226d85ce667e921aa5d399584b8
blob + f786101c27c52e49d41dab5ca56ae883ecf272a5
--- lib/object_open_io.c
+++ lib/object_open_io.c
@@ -26,6 +26,7 @@
 #include <stdlib.h>
 #include <string.h>
 #include <unistd.h>
+#include <zlib.h>
 
 #include "got_error.h"
 #include "got_object.h"
@@ -38,6 +39,8 @@
 #include "got_lib_object_parse.h"
 #include "got_lib_pack.h"
 #include "got_lib_repository.h"
+#include "got_lib_inflate.h"
+#include "got_lib_hash.h"
 
 const struct got_error *
 got_object_open_packed(struct got_object **obj, struct got_object_id *id,
@@ -607,13 +610,253 @@ got_object_tree_open(struct got_tree_object **tree,
 {
 	return open_tree(tree, repo, got_object_get_id(obj), 1);
 }
+
+static const struct got_error *
+read_packed_blob(uint8_t **outbuf, size_t *size, size_t *hdrlen,
+    int outfd, struct got_pack *pack, struct got_packidx *packidx, int idx,
+    struct got_object_id *id, struct got_repository *repo)
+{
+	const struct got_error *err = NULL;
+	struct got_object *obj;
+	FILE *outfile = NULL, *basefile = NULL, *accumfile = NULL;
+	uint64_t blob_size;
+
+	*hdrlen = 0;
+
+	err = got_object_open_from_packfile(&obj, id, pack, packidx, idx,
+	    repo);
+	if (err)
+		return err;
 
+	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
+		err = got_pack_get_max_delta_object_size(&blob_size, obj,
+		    pack);
+		if (err)
+			goto done;
+	} else
+		blob_size = obj->size;
+
+	if (blob_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
+		err = got_packfile_extract_object_to_mem(outbuf, size,
+		    obj, pack); 
+	} else {
+		/*
+		 * XXX This uses 3 file extra descriptors for no good reason.
+		 * We should have got_packfile_extract_object_to_fd().
+		 */
+		err = wrap_fd(&outfile, outfd);
+		if (err)
+			goto done;
+		err = wrap_fd(&basefile, pack->basefd);
+		if (err)
+			goto done;
+		err = wrap_fd(&accumfile, pack->accumfd);
+		if (err)
+			goto done;
+		err = got_packfile_extract_object(pack, obj, outfile, basefile,
+		    accumfile);
+		if (err)
+			goto done;
+		*size = obj->size;
+	}
+
+	/* XXX verify checksum? */
+done:
+	got_object_close(obj);
+	if (outfile && fclose(outfile) == EOF && err == NULL)
+		err = got_error_from_errno("fclose");
+	if (basefile && fclose(basefile) == EOF && err == NULL)
+		err = got_error_from_errno("fclose");
+	if (accumfile && fclose(accumfile) == EOF && err == NULL)
+		err = got_error_from_errno("fclose");
+	return err;
+}
+
+static const struct got_error *
+read_blob(uint8_t **outbuf, size_t *size, size_t *hdrlen, int outfd, int infd,
+    struct got_object_id *id, struct got_repository *repo)
+{
+	const struct got_error *err = NULL;
+	struct got_object *obj = NULL;
+	FILE *f = NULL;
+	struct got_object_id expected_id;
+	struct got_inflate_checksum csum;
+	struct got_hash ctx;
+
+	got_hash_init(&ctx, GOT_HASH_SHA1);
+	memset(&csum, 0, sizeof(csum));
+	csum.output_ctx = &ctx;
+
+	memcpy(&expected_id, id, sizeof(expected_id));
+
+	err = got_object_read_header(&obj, infd);
+	if (err)
+		goto done;
+
+	if (lseek(infd, SEEK_SET, 0) == -1) {
+		err = got_error_from_errno("lseek");
+		goto done;
+	}
+
+	f = fdopen(infd, "rb");
+	if (f == NULL) {
+		err = got_error_from_errno("fdopen");
+		goto done;
+	}
+	infd = -1;
+
+	if (obj->size + obj->hdrlen <= GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
+		err = got_inflate_to_mem(outbuf, size, NULL, &csum, f);
+		if (err)
+			goto done;
+	} else {
+		err = got_inflate_to_fd(size, f, &csum, outfd);
+		if (err)
+			goto done;
+	}
+
+	if (*size < obj->hdrlen) {
+		err = got_error(GOT_ERR_BAD_OBJ_HDR);
+		goto done;
+	}
+
+	*hdrlen = obj->hdrlen;
+
+	got_hash_final_object_id(&ctx, id);
+	if (got_object_id_cmp(&expected_id, id) != 0) {
+		err = got_error_checksum(&expected_id);
+		goto done;
+	}
+done:
+	if (f && fclose(f) == EOF && err == NULL)
+		err = got_error_from_errno("fclose");
+	if (infd != -1 && close(infd) == -1 && err == NULL)
+		err = got_error_from_errno("close");
+
+	return err;
+}
+
+static const struct got_error *
+open_blob(struct got_blob_object **blob, struct got_repository *repo,
+    struct got_object_id *id, size_t blocksize, int outfd)
+{
+	const struct got_error *err = NULL;
+	struct got_packidx *packidx = NULL;
+	int idx, dfd = -1;
+	char *path_packfile = NULL;
+	uint8_t *outbuf;
+	size_t size, hdrlen;
+	struct stat sb;
+
+	*blob = calloc(1, sizeof(**blob));
+	if (*blob == NULL)
+		return got_error_from_errno("calloc");
+
+	(*blob)->read_buf = malloc(blocksize);
+	if ((*blob)->read_buf == NULL) {
+		err = got_error_from_errno("malloc");
+		goto done;
+	}
+
+	if (ftruncate(outfd, 0L) == -1) {
+		err = got_error_from_errno("ftruncate");
+		goto done;
+	}
+	if (lseek(outfd, SEEK_SET, 0) == -1) {
+		err = got_error_from_errno("lseek");
+		goto done;
+	}
+
+	err = got_repo_search_packidx(&packidx, &idx, repo, id);
+	if (err == NULL) {
+		struct got_pack *pack = NULL;
+
+		err = got_packidx_get_packfile_path(&path_packfile,
+		    packidx->path_packidx);
+		if (err)
+			goto done;
+
+		pack = got_repo_get_cached_pack(repo, path_packfile);
+		if (pack == NULL) {
+			err = got_repo_cache_pack(&pack, repo, path_packfile,
+			    packidx);
+			if (err)
+				goto done;
+		}
+		err = read_packed_blob(&outbuf, &size, &hdrlen, outfd,
+		    pack, packidx, idx, id, repo);
+	} else if (err->code == GOT_ERR_NO_OBJ) {
+		int infd;
+
+		err = got_object_open_loose_fd(&infd, id, repo);
+		if (err)
+			goto done;
+		err = read_blob(&outbuf, &size, &hdrlen, outfd, infd,
+		    id, repo);
+	}
+	if (err)
+		goto done;
+
+	if (hdrlen > size) {
+		err = got_error(GOT_ERR_BAD_OBJ_HDR);
+		goto done;
+	}
+
+	if (outbuf) {
+		(*blob)->f = fmemopen(outbuf, size, "rb");
+		if ((*blob)->f == NULL) {
+			err = got_error_from_errno("fmemopen");
+			free(outbuf);
+			goto done;
+		}
+		(*blob)->data = outbuf;
+	} else {
+		if (fstat(outfd, &sb) == -1) {
+			err = got_error_from_errno("fstat");
+			goto done;
+		}
+
+		if (sb.st_size != size) {
+			err = got_error(GOT_ERR_PRIVSEP_LEN);
+			goto done;
+		}
+
+		dfd = dup(outfd);
+		if (dfd == -1) {
+			err = got_error_from_errno("dup");
+			goto done;
+		}
+
+		(*blob)->f = fdopen(dfd, "rb");
+		if ((*blob)->f == NULL) {
+			err = got_error_from_errno("fdopen");
+			close(dfd);
+			dfd = -1;
+			goto done;
+		}
+	}
+
+	(*blob)->hdrlen = hdrlen;
+	(*blob)->blocksize = blocksize;
+	memcpy(&(*blob)->id, id, sizeof(*id));
+
+done:
+	free(path_packfile);
+	if (err) {
+		if (*blob) {
+			got_object_blob_close(*blob);
+			*blob = NULL;
+		}
+	}
+	return err;
+}
+
 const struct got_error *
 got_object_open_as_blob(struct got_blob_object **blob,
     struct got_repository *repo, struct got_object_id *id, size_t blocksize,
     int outfd)
 {
-	return got_error(GOT_ERR_NOT_IMPL);
+	return open_blob(blob, repo, id, blocksize, outfd);
 }
 
 const struct got_error *
@@ -621,7 +864,7 @@ got_object_blob_open(struct got_blob_object **blob,
     struct got_repository *repo, struct got_object *obj, size_t blocksize,
     int outfd)
 {
-	return got_error(GOT_ERR_NOT_IMPL);
+	return open_blob(blob, repo, got_object_get_id(obj), blocksize, outfd);
 }
 
 static const struct got_error *