wroclaw-git-backups/nix
1
0
Fork 0
mirror of https://github.com/NixOS/nix synced 2025-06-30 03:23:16 +02:00
Code Activity

Fix format

Browse source
This commit is contained in:
Leandro Reina 2024-11-20 18:24:17 +01:00
parent 79d41062d0
commit 70ffcc83d7
2 changed files with 307 additions and 287 deletions
Download patch file Download diff file Expand all files Collapse all files

23
src/libfetchers-tests/git-utils.cc
View file

@ -110,7 +110,6 @@ TEST_F(GitUtilsTest, sink_hardlink)
} }
}; };
namespace lfs { namespace lfs {
TEST_F(GitUtilsTest, parseGitRemoteUrl) TEST_F(GitUtilsTest, parseGitRemoteUrl)
@ -196,7 +195,8 @@ TEST_F(GitUtilsTest, parseGitRemoteUrl)
EXPECT_EQ(result.path, ""); EXPECT_EQ(result.path, "");
} }
} }
TEST_F(GitUtilsTest, gitUrlToHttp) { TEST_F(GitUtilsTest, gitUrlToHttp)
{
{ {
const GitUrl url = parseGitUrl("git@github.com:user/repo.git"); const GitUrl url = parseGitUrl("git@github.com:user/repo.git");
EXPECT_EQ(url.toHttp(), "https://github.com/user/repo.git"); EXPECT_EQ(url.toHttp(), "https://github.com/user/repo.git");
@ -219,7 +219,8 @@ TEST_F(GitUtilsTest, gitUrlToHttp) {
} }
} }
TEST_F(GitUtilsTest, gitUrlToSsh) { TEST_F(GitUtilsTest, gitUrlToSsh)
{
{ {
const GitUrl url = parseGitUrl("https://example.com/user/repo.git"); const GitUrl url = parseGitUrl("https://example.com/user/repo.git");
const auto [host, path] = url.toSsh(); const auto [host, path] = url.toSsh();
@ -234,9 +235,11 @@ TEST_F(GitUtilsTest, gitUrlToSsh) {
} }
} }
class FetchAttributeTest : public ::testing::Test { class FetchAttributeTest : public ::testing::Test
{
protected: protected:
void SetUp() override { void SetUp() override
{
// test literal (non-wildcard) matches too // test literal (non-wildcard) matches too
std::string content1 = "litfile filter=lfs diff=lfs merge=lfs -text"; std::string content1 = "litfile filter=lfs diff=lfs merge=lfs -text";
auto rules1 = parseGitAttrFile(content1); auto rules1 = parseGitAttrFile(content1);
@ -251,23 +254,25 @@ protected:
Fetch fetch2; Fetch fetch2;
}; };
TEST_F(FetchAttributeTest, ExactMatch) { TEST_F(FetchAttributeTest, ExactMatch)
{
EXPECT_TRUE(fetch1.hasAttribute("litfile", "filter", "lfs")); EXPECT_TRUE(fetch1.hasAttribute("litfile", "filter", "lfs"));
EXPECT_FALSE(fetch1.hasAttribute("other", "filter", "lfs")); EXPECT_FALSE(fetch1.hasAttribute("other", "filter", "lfs"));
} }
TEST_F(FetchAttributeTest, WildcardMatch) { TEST_F(FetchAttributeTest, WildcardMatch)
{
EXPECT_TRUE(fetch2.hasAttribute("match.wildcard", "filter", "lfs")); EXPECT_TRUE(fetch2.hasAttribute("match.wildcard", "filter", "lfs"));
EXPECT_FALSE(fetch2.hasAttribute("nomatch.otherext", "filter", "lfs")); EXPECT_FALSE(fetch2.hasAttribute("nomatch.otherext", "filter", "lfs"));
EXPECT_FALSE(fetch2.hasAttribute("nomatch.wildcard.extra", "filter", "lfs")); EXPECT_FALSE(fetch2.hasAttribute("nomatch.wildcard.extra", "filter", "lfs"));
} }
TEST_F(FetchAttributeTest, EmptyPath) { TEST_F(FetchAttributeTest, EmptyPath)
{
EXPECT_FALSE(fetch1.hasAttribute("", "filter", "lfs")); EXPECT_FALSE(fetch1.hasAttribute("", "filter", "lfs"));
EXPECT_FALSE(fetch2.hasAttribute("", "filter", "lfs")); EXPECT_FALSE(fetch2.hasAttribute("", "filter", "lfs"));
} }
} // namespace lfs } // namespace lfs
} // namespace nix } // namespace nix

131
src/libfetchers/git-lfs-fetch.hh
View file

@ -16,25 +16,28 @@
#include "processes.hh" #include "processes.hh"
#include "url.hh" #include "url.hh"
namespace fs = std::filesystem; namespace fs = std::filesystem;
namespace nix { namespace nix {
namespace lfs { namespace lfs {
// see Fetch::rules // see Fetch::rules
struct AttrRule { struct AttrRule
{
std::string pattern; std::string pattern;
std::unordered_map<std::string, std::string> attributes; std::unordered_map<std::string, std::string> attributes;
git_pathspec * pathspec = nullptr; git_pathspec * pathspec = nullptr;
AttrRule() = default; AttrRule() = default;
explicit AttrRule(std::string pat) : pattern(std::move(pat)) { explicit AttrRule(std::string pat)
: pattern(std::move(pat))
{
initPathspec(); initPathspec();
} }
~AttrRule() { ~AttrRule()
{
if (pathspec) { if (pathspec) {
git_pathspec_free(pathspec); git_pathspec_free(pathspec);
} }
@ -50,7 +53,8 @@ struct AttrRule {
} }
} }
void initPathspec() { void initPathspec()
{
git_strarray patterns = {0}; git_strarray patterns = {0};
const char * pattern_str = pattern.c_str(); const char * pattern_str = pattern.c_str();
patterns.strings = const_cast<char **>(&pattern_str); patterns.strings = const_cast<char **>(&pattern_str);
@ -63,43 +67,44 @@ struct AttrRule {
}; };
// git-lfs metadata about a file // git-lfs metadata about a file
struct Md { struct Md
{
std::string path; // fs path relative to repo root, no ./ prefix std::string path; // fs path relative to repo root, no ./ prefix
std::string oid; // git-lfs managed object id. you give this to the lfs server std::string oid; // git-lfs managed object id. you give this to the lfs server
// for downloads // for downloads
size_t size; // in bytes size_t size; // in bytes
}; };
struct GitUrl { struct GitUrl
{
std::string protocol; std::string protocol;
std::string user; std::string user;
std::string host; std::string host;
std::string port; std::string port;
std::string path; std::string path;
std::string toHttp() const { std::string toHttp() const
{
if (protocol.empty() || host.empty()) { if (protocol.empty() || host.empty()) {
return ""; return "";
} }
std::string prefix = ((protocol == "ssh") ? "https" : protocol) + "://"; std::string prefix = ((protocol == "ssh") ? "https" : protocol) + "://";
return prefix + host + return prefix + host + (port.empty() ? "" : ":" + port) + "/" + path;
(port.empty() ? "" : ":" + port) + "/" + path;
} }
// [host, path] // [host, path]
std::pair<std::string, std::string> toSsh() const { std::pair<std::string, std::string> toSsh() const
{
if (host.empty()) { if (host.empty()) {
return {"", ""}; return {"", ""};
} }
std::string userPart = user.empty() ? "" : user + "@"; std::string userPart = user.empty() ? "" : user + "@";
return { return {userPart + host, path};
userPart + host,
path
};
} }
}; };
struct Fetch { struct Fetch
{
// only true after init() // only true after init()
bool ready = false; bool ready = false;
@ -121,15 +126,15 @@ struct Fetch {
std::vector<nlohmann::json> fetchUrls(const std::vector<Md> & metadatas) const; std::vector<nlohmann::json> fetchUrls(const std::vector<Md> & metadatas) const;
}; };
static size_t writeCallback(void * contents, size_t size, size_t nmemb, std::string * s)
static size_t writeCallback(void *contents, size_t size, size_t nmemb, {
std::string *s) {
size_t newLength = size * nmemb; size_t newLength = size * nmemb;
s->append((char *) contents, newLength); s->append((char *) contents, newLength);
return newLength; return newLength;
} }
struct SinkCallbackData { struct SinkCallbackData
{
Sink * sink; Sink * sink;
std::string_view sha256Expected; std::string_view sha256Expected;
HashSink hashSink; HashSink hashSink;
@ -138,10 +143,12 @@ struct SinkCallbackData {
: sink(sink) : sink(sink)
, sha256Expected(sha256) , sha256Expected(sha256)
, hashSink(HashAlgorithm::SHA256) , hashSink(HashAlgorithm::SHA256)
{} {
}
}; };
static size_t sinkWriteCallback(void *contents, size_t size, size_t nmemb, SinkCallbackData *data) { static size_t sinkWriteCallback(void * contents, size_t size, size_t nmemb, SinkCallbackData * data)
{
size_t totalSize = size * nmemb; size_t totalSize = size * nmemb;
data->hashSink({(char *) contents, totalSize}); data->hashSink({(char *) contents, totalSize});
(*data->sink)({(char *) contents, totalSize}); (*data->sink)({(char *) contents, totalSize});
@ -149,7 +156,9 @@ static size_t sinkWriteCallback(void *contents, size_t size, size_t nmemb, SinkC
} }
// if authHeader is "", downloadToSink assumes to auth is expected // if authHeader is "", downloadToSink assumes to auth is expected
void downloadToSink(const std::string &url, const std::string &authHeader, Sink &sink, std::string_view sha256Expected) { void downloadToSink(
const std::string & url, const std::string & authHeader, Sink & sink, std::string_view sha256Expected)
{
CURL * curl; CURL * curl;
CURLcode res; CURLcode res;
@ -177,15 +186,17 @@ void downloadToSink(const std::string &url, const std::string &authHeader, Sink
const auto sha256Actual = data.hashSink.finish().first.to_string(HashFormat::Base16, false); const auto sha256Actual = data.hashSink.finish().first.to_string(HashFormat::Base16, false);
if (sha256Actual != data.sha256Expected) { if (sha256Actual != data.sha256Expected) {
throw std::runtime_error("sha256 mismatch: while fetching " + url + ": expected " + std::string(data.sha256Expected) + " but got " + sha256Actual); throw std::runtime_error(
"sha256 mismatch: while fetching " + url + ": expected " + std::string(data.sha256Expected) + " but got "
+ sha256Actual);
} }
curl_slist_free_all(headers); curl_slist_free_all(headers);
curl_easy_cleanup(curl); curl_easy_cleanup(curl);
} }
std::string getLfsApiToken(const GitUrl & u)
std::string getLfsApiToken(const GitUrl& u) { {
const auto [maybeUserAndHost, path] = u.toSsh(); const auto [maybeUserAndHost, path] = u.toSsh();
auto [status, output] = runProgram(RunOptions{ auto [status, output] = runProgram(RunOptions{
.program = "ssh", .program = "ssh",
@ -193,7 +204,9 @@ std::string getLfsApiToken(const GitUrl& u) {
}); });
if (output.empty()) if (output.empty())
throw std::runtime_error("git-lfs-authenticate: no output (cmd: ssh " + maybeUserAndHost + " git-lfs-authenticate " + path + " download)"); throw std::runtime_error(
"git-lfs-authenticate: no output (cmd: ssh " + maybeUserAndHost + " git-lfs-authenticate " + path
+ " download)");
nlohmann::json query_resp = nlohmann::json::parse(output); nlohmann::json query_resp = nlohmann::json::parse(output);
if (!query_resp.contains("header")) if (!query_resp.contains("header"))
@ -206,7 +219,8 @@ std::string getLfsApiToken(const GitUrl& u) {
return res; return res;
} }
std::string getLfsEndpointUrl(git_repository *repo) { std::string getLfsEndpointUrl(git_repository * repo)
{
int err; int err;
git_remote * remote = NULL; git_remote * remote = NULL;
err = git_remote_lookup(&remote, repo, "origin"); err = git_remote_lookup(&remote, repo, "origin");
@ -222,7 +236,8 @@ std::string getLfsEndpointUrl(git_repository *repo) {
return std::string(url_c_str); return std::string(url_c_str);
} }
std::string git_attr_value_to_string(git_attr_value_t value) { std::string git_attr_value_to_string(git_attr_value_t value)
{
switch (value) { switch (value) {
case GIT_ATTR_VALUE_UNSPECIFIED: case GIT_ATTR_VALUE_UNSPECIFIED:
return "GIT_ATTR_VALUE_UNSPECIFIED"; return "GIT_ATTR_VALUE_UNSPECIFIED";
@ -237,8 +252,8 @@ std::string git_attr_value_to_string(git_attr_value_t value) {
} }
} }
std::optional<Md> parseLfsMetadata(const std::string & content, const std::string & filename)
std::optional<Md> parseLfsMetadata(const std::string &content, const std::string &filename) { {
// https://github.com/git-lfs/git-lfs/blob/2ef4108/docs/spec.md // https://github.com/git-lfs/git-lfs/blob/2ef4108/docs/spec.md
// //
// example git-lfs pointer file: // example git-lfs pointer file:
@ -293,19 +308,18 @@ std::optional<Md> parseLfsMetadata(const std::string &content, const std::string
return std::make_optional(Md{filename, oid, std::stoul(size)}); return std::make_optional(Md{filename, oid, std::stoul(size)});
} }
// there's already a ParseURL here
// there's already a ParseURL here https://github.com/b-camacho/nix/blob/ef6fa54e05cd4134ec41b0d64c1a16db46237f83/src/libutil/url.cc#L13 // https://github.com/b-camacho/nix/blob/ef6fa54e05cd4134ec41b0d64c1a16db46237f83/src/libutil/url.cc#L13 but that does
// but that does not handle git's custom scp-like syntax // not handle git's custom scp-like syntax
GitUrl parseGitUrl(const std::string& url) { GitUrl parseGitUrl(const std::string & url)
{
GitUrl result; GitUrl result;
// regular protocols // regular protocols
const std::regex r_url( const std::regex r_url(R"(^(ssh|git|https?|ftps?)://(?:([^@]+)@)?([^:/]+)(?::(\d+))?/(.*))");
R"(^(ssh|git|https?|ftps?)://(?:([^@]+)@)?([^:/]+)(?::(\d+))?/(.*))");
// "alternative scp-like syntax" https://git-scm.com/docs/git-fetch#_git_urls // "alternative scp-like syntax" https://git-scm.com/docs/git-fetch#_git_urls
const std::regex r_scp_like_url( const std::regex r_scp_like_url(R"(^(?:([^@]+)@)?([^:/]+):(/?.*))");
R"(^(?:([^@]+)@)?([^:/]+):(/?.*))");
std::smatch matches; std::smatch matches;
if (std::regex_match(url, matches, r_url)) { if (std::regex_match(url, matches, r_url)) {
@ -314,8 +328,7 @@ GitUrl parseGitUrl(const std::string& url) {
result.host = matches[3].str(); result.host = matches[3].str();
result.port = matches[4].str(); result.port = matches[4].str();
result.path = matches[5].str(); result.path = matches[5].str();
} } else if (std::regex_match(url, matches, r_scp_like_url)) {
else if (std::regex_match(url, matches, r_scp_like_url)) {
result.protocol = "ssh"; result.protocol = "ssh";
result.user = matches[1].str(); result.user = matches[1].str();
@ -355,7 +368,8 @@ std::vector<AttrRule> parseGitAttrFile(const std::string& content)
if (git_pathspec_new(&rule.pathspec, &patterns) != 0) { if (git_pathspec_new(&rule.pathspec, &patterns) != 0) {
auto error = git_error_last(); auto error = git_error_last();
std::stringstream ss; std::stringstream ss;
ss << "git_pathspec_new parsing '" << line << "': " << (error ? error->message : "unknown error") << std::endl; ss << "git_pathspec_new parsing '" << line << "': " << (error ? error->message : "unknown error")
<< std::endl;
warn(ss.str()); warn(ss.str());
continue; continue;
} }
@ -396,7 +410,8 @@ std::vector<AttrRule> parseGitAttrFile(const std::string& content)
return rules; return rules;
} }
void Fetch::init(git_repository* repo, const std::string& gitattributesContent) { void Fetch::init(git_repository * repo, const std::string & gitattributesContent)
{
const auto remoteUrl = lfs::getLfsEndpointUrl(repo); const auto remoteUrl = lfs::getLfsEndpointUrl(repo);
this->gitUrl = parseGitUrl(remoteUrl); this->gitUrl = parseGitUrl(remoteUrl);
@ -407,15 +422,13 @@ void Fetch::init(git_repository* repo, const std::string& gitattributesContent)
this->ready = true; this->ready = true;
} }
bool Fetch::hasAttribute(const std::string & path, const std::string & attrName, const std::string & attrValue) const bool Fetch::hasAttribute(const std::string & path, const std::string & attrName, const std::string & attrValue) const
{ {
for (auto it = rules.rbegin(); it != rules.rend(); ++it) { for (auto it = rules.rbegin(); it != rules.rend(); ++it) {
int match = git_pathspec_matches_path( int match = git_pathspec_matches_path(
it->pathspec, it->pathspec,
0, // no flags 0, // no flags
path.c_str() path.c_str());
);
if (match > 0) { if (match > 0) {
auto attr = it->attributes.find(attrName); auto attr = it->attributes.find(attrName);
@ -428,7 +441,8 @@ bool Fetch::hasAttribute(const std::string& path, const std::string& attrName, c
return false; return false;
} }
nlohmann::json mdToPayload(const std::vector<Md> &items) { nlohmann::json mdToPayload(const std::vector<Md> & items)
{
nlohmann::json jArray = nlohmann::json::array(); nlohmann::json jArray = nlohmann::json::array();
for (const auto & md : items) { for (const auto & md : items) {
jArray.push_back({{"oid", md.oid}, {"size", md.size}}); jArray.push_back({{"oid", md.oid}, {"size", md.size}});
@ -436,8 +450,8 @@ nlohmann::json mdToPayload(const std::vector<Md> &items) {
return jArray; return jArray;
} }
std::vector<nlohmann::json> Fetch::fetchUrls(const std::vector<Md> & metadatas) const
std::vector<nlohmann::json> Fetch::fetchUrls(const std::vector<Md> &metadatas) const { {
nlohmann::json oidList = mdToPayload(metadatas); nlohmann::json oidList = mdToPayload(metadatas);
nlohmann::json data = { nlohmann::json data = {
{"operation", "download"}, {"operation", "download"},
@ -461,8 +475,7 @@ std::vector<nlohmann::json> Fetch::fetchUrls(const std::vector<Md> &metadatas) c
headers = curl_slist_append(headers, authHeader.c_str()); headers = curl_slist_append(headers, authHeader.c_str());
} }
headers = headers = curl_slist_append(headers, "Content-Type: application/vnd.git-lfs+json");
curl_slist_append(headers, "Content-Type: application/vnd.git-lfs+json");
headers = curl_slist_append(headers, "Accept: application/vnd.git-lfs+json"); headers = curl_slist_append(headers, "Accept: application/vnd.git-lfs+json");
curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers); curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers);
curl_easy_setopt(curl, CURLOPT_POST, 1L); curl_easy_setopt(curl, CURLOPT_POST, 1L);
@ -487,8 +500,7 @@ std::vector<nlohmann::json> Fetch::fetchUrls(const std::vector<Md> &metadatas) c
try { try {
auto resp = nlohmann::json::parse(responseString); auto resp = nlohmann::json::parse(responseString);
if (resp.contains("objects")) { if (resp.contains("objects")) {
objects.insert(objects.end(), resp["objects"].begin(), objects.insert(objects.end(), resp["objects"].begin(), resp["objects"].end());
resp["objects"].end());
} else { } else {
throw std::runtime_error("response does not contain 'objects'"); throw std::runtime_error("response does not contain 'objects'");
} }
@ -498,11 +510,11 @@ std::vector<nlohmann::json> Fetch::fetchUrls(const std::vector<Md> &metadatas) c
std::stringstream ss; std::stringstream ss;
ss << "response did not parse as json: " << responseString; ss << "response did not parse as json: " << responseString;
throw std::runtime_error(ss.str()); throw std::runtime_error(ss.str());
} }
} }
void Fetch::fetch(const git_blob* pointerBlob, const std::string& pointerFilePath, Sink& sink) const { void Fetch::fetch(const git_blob * pointerBlob, const std::string & pointerFilePath, Sink & sink) const
{
constexpr size_t chunkSize = 128 * 1024; // 128 KiB constexpr size_t chunkSize = 128 * 1024; // 128 KiB
auto size = git_blob_rawsize(pointerBlob); auto size = git_blob_rawsize(pointerBlob);
@ -510,7 +522,8 @@ void Fetch::fetch(const git_blob* pointerBlob, const std::string& pointerFilePat
debug("Skip git-lfs, pointer file too large"); debug("Skip git-lfs, pointer file too large");
warn("Encountered a file that should have been a pointer, but wasn't: %s", pointerFilePath); warn("Encountered a file that should have been a pointer, but wasn't: %s", pointerFilePath);
for (size_t offset = 0; offset < size; offset += chunkSize) { for (size_t offset = 0; offset < size; offset += chunkSize) {
sink(std::string((const char *) git_blob_rawcontent(pointerBlob) + offset, std::min(chunkSize, size - offset))); sink(std::string(
(const char *) git_blob_rawcontent(pointerBlob) + offset, std::min(chunkSize, size - offset)));
} }
return; return;
} }
@ -521,7 +534,8 @@ void Fetch::fetch(const git_blob* pointerBlob, const std::string& pointerFilePat
debug("Skip git-lfs, invalid pointer file"); debug("Skip git-lfs, invalid pointer file");
warn("Encountered a file that should have been a pointer, but wasn't: %s", pointerFilePath); warn("Encountered a file that should have been a pointer, but wasn't: %s", pointerFilePath);
for (size_t offset = 0; offset < size; offset += chunkSize) { for (size_t offset = 0; offset < size; offset += chunkSize) {
sink(std::string((const char *) git_blob_rawcontent(pointerBlob) + offset, std::min(chunkSize, size - offset))); sink(std::string(
(const char *) git_blob_rawcontent(pointerBlob) + offset, std::min(chunkSize, size - offset)));
} }
return; return;
} }
@ -535,7 +549,8 @@ void Fetch::fetch(const git_blob* pointerBlob, const std::string& pointerFilePat
std::string oid = obj.at("oid"); std::string oid = obj.at("oid");
std::string ourl = obj.at("actions").at("download").at("href"); std::string ourl = obj.at("actions").at("download").at("href");
std::string authHeader = ""; std::string authHeader = "";
if (obj.at("actions").at("download").contains("header") && obj.at("actions").at("download").at("header").contains("Authorization")) { if (obj.at("actions").at("download").contains("header")
&& obj.at("actions").at("download").at("header").contains("Authorization")) {
authHeader = obj["actions"]["download"]["header"]["Authorization"]; authHeader = obj["actions"]["download"]["header"]["Authorization"];
} }
// oid is also the sha256 // oid is also the sha256