nix-super/src/libexpr/primops/fetchTree.cc

538 lines
19 KiB
C++
Raw Normal View History

#include "primops.hh"
#include "eval-inline.hh"
#include "eval-settings.hh"
#include "store-api.hh"
#include "fetchers.hh"
2020-04-08 15:12:22 +03:00
#include "filetransfer.hh"
#include "registry.hh"
#include "url.hh"
#include <ctime>
#include <iomanip>
#include <regex>
namespace nix {
void emitTreeAttrs(
EvalState & state,
const fetchers::Tree & tree,
const fetchers::Input & input,
Value & v,
bool emptyRevFallback,
bool forceDirty)
{
assert(input.isLocked());
auto attrs = state.buildBindings(10);
Use `std::set<StringContextElem>` not `PathSet` for string contexts Motivation `PathSet` is not correct because string contexts have other forms (`Built` and `DrvDeep`) that are not rendered as plain store paths. Instead of wrongly using `PathSet`, or "stringly typed" using `StringSet`, use `std::std<StringContextElem>`. ----- In support of this change, `NixStringContext` is now defined as `std::std<StringContextElem>` not `std:vector<StringContextElem>`. The old definition was just used by a `getContext` method which was only used by the eval cache. It can be deleted altogether since the types are now unified and the preexisting `copyContext` function already suffices. Summarizing the previous paragraph: Old: - `value/context.hh`: `NixStringContext = std::vector<StringContextElem>` - `value.hh`: `NixStringContext Value::getContext(...)` - `value.hh`: `copyContext(...)` New: - `value/context.hh`: `NixStringContext = std::set<StringContextElem>` - `value.hh`: `copyContext(...)` ---- The string representation of string context elements no longer contains the store dir. The diff of `src/libexpr/tests/value/context.cc` should make clear what the new representation is, so we recommend reviewing that file first. This was done for two reasons: Less API churn: `Value::mkString` and friends did not take a `Store` before. But if `NixStringContextElem::{parse, to_string}` *do* take a store (as they did before), then we cannot have the `Value` functions use them (in order to work with the fully-structured `NixStringContext`) without adding that argument. That would have been a lot of churn of threading the store, and this diff is already large enough, so the easier and less invasive thing to do was simply make the element `parse` and `to_string` functions not take the `Store` reference, and the easiest way to do that was to simply drop the store dir. Space usage: Dropping the `/nix/store/` (or similar) from the internal representation will safe space in the heap of the Nix programming being interpreted. If the heap contains many strings with non-trivial contexts, the saving could add up to something significant. ---- The eval cache version is bumped. The eval cache serialization uses `NixStringContextElem::{parse, to_string}`, and since those functions are changed per the above, that means the on-disk representation is also changed. This is simply done by changing the name of the used for the eval cache from `eval-cache-v4` to eval-cache-v5`. ---- To avoid some duplication `EvalCache::mkPathString` is added to abstract over the simple case of turning a store path to a string with just that string in the context. Context This PR picks up where #7543 left off. That one introduced the fully structured `NixStringContextElem` data type, but kept `PathSet context` as an awkward middle ground between internal `char[][]` interpreter heap string contexts and `NixStringContext` fully parsed string contexts. The infelicity of `PathSet context` was specifically called out during Nix team group review, but it was agreeing that fixing it could be left as future work. This is that future work. A possible follow-up step would be to get rid of the `char[][]` evaluator heap representation, too, but it is not yet clear how to do that. To use `NixStringContextElem` there we would need to get the STL containers to GC pointers in the GC build, and I am not sure how to do that. ---- PR #7543 effectively is writing the inverse of a `mkPathString`, `mkOutputString`, and one more such function for the `DrvDeep` case. I would like that PR to have property tests ensuring it is actually the inverse as expected. This PR sets things up nicely so that reworking that PR to be in that more elegant and better tested way is possible. Co-authored-by: Théophane Hufschmitt <7226587+thufschmitt@users.noreply.github.com>
2023-01-29 03:31:10 +02:00
state.mkStorePathString(tree.storePath, attrs.alloc(state.sOutPath));
// FIXME: support arbitrary input attributes.
auto narHash = input.getNarHash();
assert(narHash);
attrs.alloc("narHash").mkString(narHash->to_string(SRI, true));
Merge legacy `fetchGit`-builtin with the generic `fetchTree`-function The original idea was to implement a git-fetcher in Nix's core that supports content hashes[1]. In #3549[2] it has been suggested to actually use `fetchTree` for this since it's a fairly generic wrapper over the new fetcher-API[3] and already supports content-hashes. This patch implements a new git-fetcher based on `fetchTree` by incorporating the following changes: * Removed the original `fetchGit`-implementation and replaced it with an alias on the `fetchTree` implementation. * Ensured that the `git`-fetcher from `libfetchers` always computes a content-hash and returns an "empty" revision on dirty trees (the latter one is needed to retain backwards-compatibility). * The hash-mismatch error in the fetcher-API exits with code 102 as it usually happens whenever a hash-mismatch is detected by Nix. * Removed the `flakes`-feature-flag: I didn't see a reason why this API is so tightly coupled to the flakes-API and at least `fetchGit` should remain usable without any feature-flags. * It's only possible to specify a `narHash` for a `git`-tree if either a `ref` or a `rev` is given[4]. * It's now possible to specify an URL without a protocol. If it's missing, `file://` is automatically added as it was the case in the original `fetchGit`-implementation. [1] https://github.com/NixOS/nix/pull/3216 [2] https://github.com/NixOS/nix/pull/3549#issuecomment-625194383 [3] https://github.com/NixOS/nix/pull/3459 [4] https://github.com/NixOS/nix/pull/3216#issuecomment-553956703
2020-04-29 23:39:58 +03:00
if (input.getType() == "git")
attrs.alloc("submodules").mkBool(
fetchers::maybeGetBoolAttr(input.attrs, "submodules").value_or(false));
Merge legacy `fetchGit`-builtin with the generic `fetchTree`-function The original idea was to implement a git-fetcher in Nix's core that supports content hashes[1]. In #3549[2] it has been suggested to actually use `fetchTree` for this since it's a fairly generic wrapper over the new fetcher-API[3] and already supports content-hashes. This patch implements a new git-fetcher based on `fetchTree` by incorporating the following changes: * Removed the original `fetchGit`-implementation and replaced it with an alias on the `fetchTree` implementation. * Ensured that the `git`-fetcher from `libfetchers` always computes a content-hash and returns an "empty" revision on dirty trees (the latter one is needed to retain backwards-compatibility). * The hash-mismatch error in the fetcher-API exits with code 102 as it usually happens whenever a hash-mismatch is detected by Nix. * Removed the `flakes`-feature-flag: I didn't see a reason why this API is so tightly coupled to the flakes-API and at least `fetchGit` should remain usable without any feature-flags. * It's only possible to specify a `narHash` for a `git`-tree if either a `ref` or a `rev` is given[4]. * It's now possible to specify an URL without a protocol. If it's missing, `file://` is automatically added as it was the case in the original `fetchGit`-implementation. [1] https://github.com/NixOS/nix/pull/3216 [2] https://github.com/NixOS/nix/pull/3549#issuecomment-625194383 [3] https://github.com/NixOS/nix/pull/3459 [4] https://github.com/NixOS/nix/pull/3216#issuecomment-553956703
2020-04-29 23:39:58 +03:00
if (!forceDirty) {
if (auto rev = input.getRev()) {
attrs.alloc("rev").mkString(rev->gitRev());
attrs.alloc("shortRev").mkString(rev->gitShortRev());
} else if (emptyRevFallback) {
// Backwards compat for `builtins.fetchGit`: dirty repos return an empty sha1 as rev
auto emptyHash = Hash(htSHA1);
attrs.alloc("rev").mkString(emptyHash.gitRev());
attrs.alloc("shortRev").mkString(emptyHash.gitShortRev());
}
if (auto revCount = input.getRevCount())
attrs.alloc("revCount").mkInt(*revCount);
else if (emptyRevFallback)
attrs.alloc("revCount").mkInt(0);
}
if (auto dirtyRev = fetchers::maybeGetStrAttr(input.attrs, "dirtyRev")) {
attrs.alloc("dirtyRev").mkString(*dirtyRev);
attrs.alloc("dirtyShortRev").mkString(*fetchers::maybeGetStrAttr(input.attrs, "dirtyShortRev"));
}
if (auto lastModified = input.getLastModified()) {
attrs.alloc("lastModified").mkInt(*lastModified);
attrs.alloc("lastModifiedDate").mkString(
fmt("%s", std::put_time(std::gmtime(&*lastModified), "%Y%m%d%H%M%S")));
}
v.mkAttrs(attrs);
}
2021-10-06 18:29:47 +03:00
std::string fixURI(std::string uri, EvalState & state, const std::string & defaultScheme = "file")
{
Merge legacy `fetchGit`-builtin with the generic `fetchTree`-function The original idea was to implement a git-fetcher in Nix's core that supports content hashes[1]. In #3549[2] it has been suggested to actually use `fetchTree` for this since it's a fairly generic wrapper over the new fetcher-API[3] and already supports content-hashes. This patch implements a new git-fetcher based on `fetchTree` by incorporating the following changes: * Removed the original `fetchGit`-implementation and replaced it with an alias on the `fetchTree` implementation. * Ensured that the `git`-fetcher from `libfetchers` always computes a content-hash and returns an "empty" revision on dirty trees (the latter one is needed to retain backwards-compatibility). * The hash-mismatch error in the fetcher-API exits with code 102 as it usually happens whenever a hash-mismatch is detected by Nix. * Removed the `flakes`-feature-flag: I didn't see a reason why this API is so tightly coupled to the flakes-API and at least `fetchGit` should remain usable without any feature-flags. * It's only possible to specify a `narHash` for a `git`-tree if either a `ref` or a `rev` is given[4]. * It's now possible to specify an URL without a protocol. If it's missing, `file://` is automatically added as it was the case in the original `fetchGit`-implementation. [1] https://github.com/NixOS/nix/pull/3216 [2] https://github.com/NixOS/nix/pull/3549#issuecomment-625194383 [3] https://github.com/NixOS/nix/pull/3459 [4] https://github.com/NixOS/nix/pull/3216#issuecomment-553956703
2020-04-29 23:39:58 +03:00
state.checkURI(uri);
if (uri.find("://") == std::string::npos) {
const auto p = ParsedURL {
.scheme = defaultScheme,
.authority = "",
.path = uri
};
return p.to_string();
} else {
return uri;
}
}
std::string fixURIForGit(std::string uri, EvalState & state)
{
/* Detects scp-style uris (e.g. git@github.com:NixOS/nix) and fixes
* them by removing the `:` and assuming a scheme of `ssh://`
* */
static std::regex scp_uri("([^/]*)@(.*):(.*)");
if (uri[0] != '/' && std::regex_match(uri, scp_uri))
return fixURI(std::regex_replace(uri, scp_uri, "$1@$2/$3"), state, "ssh");
else
return fixURI(uri, state);
Merge legacy `fetchGit`-builtin with the generic `fetchTree`-function The original idea was to implement a git-fetcher in Nix's core that supports content hashes[1]. In #3549[2] it has been suggested to actually use `fetchTree` for this since it's a fairly generic wrapper over the new fetcher-API[3] and already supports content-hashes. This patch implements a new git-fetcher based on `fetchTree` by incorporating the following changes: * Removed the original `fetchGit`-implementation and replaced it with an alias on the `fetchTree` implementation. * Ensured that the `git`-fetcher from `libfetchers` always computes a content-hash and returns an "empty" revision on dirty trees (the latter one is needed to retain backwards-compatibility). * The hash-mismatch error in the fetcher-API exits with code 102 as it usually happens whenever a hash-mismatch is detected by Nix. * Removed the `flakes`-feature-flag: I didn't see a reason why this API is so tightly coupled to the flakes-API and at least `fetchGit` should remain usable without any feature-flags. * It's only possible to specify a `narHash` for a `git`-tree if either a `ref` or a `rev` is given[4]. * It's now possible to specify an URL without a protocol. If it's missing, `file://` is automatically added as it was the case in the original `fetchGit`-implementation. [1] https://github.com/NixOS/nix/pull/3216 [2] https://github.com/NixOS/nix/pull/3549#issuecomment-625194383 [3] https://github.com/NixOS/nix/pull/3459 [4] https://github.com/NixOS/nix/pull/3216#issuecomment-553956703
2020-04-29 23:39:58 +03:00
}
struct FetchTreeParams {
bool emptyRevFallback = false;
bool allowNameArgument = false;
};
Merge legacy `fetchGit`-builtin with the generic `fetchTree`-function The original idea was to implement a git-fetcher in Nix's core that supports content hashes[1]. In #3549[2] it has been suggested to actually use `fetchTree` for this since it's a fairly generic wrapper over the new fetcher-API[3] and already supports content-hashes. This patch implements a new git-fetcher based on `fetchTree` by incorporating the following changes: * Removed the original `fetchGit`-implementation and replaced it with an alias on the `fetchTree` implementation. * Ensured that the `git`-fetcher from `libfetchers` always computes a content-hash and returns an "empty" revision on dirty trees (the latter one is needed to retain backwards-compatibility). * The hash-mismatch error in the fetcher-API exits with code 102 as it usually happens whenever a hash-mismatch is detected by Nix. * Removed the `flakes`-feature-flag: I didn't see a reason why this API is so tightly coupled to the flakes-API and at least `fetchGit` should remain usable without any feature-flags. * It's only possible to specify a `narHash` for a `git`-tree if either a `ref` or a `rev` is given[4]. * It's now possible to specify an URL without a protocol. If it's missing, `file://` is automatically added as it was the case in the original `fetchGit`-implementation. [1] https://github.com/NixOS/nix/pull/3216 [2] https://github.com/NixOS/nix/pull/3549#issuecomment-625194383 [3] https://github.com/NixOS/nix/pull/3459 [4] https://github.com/NixOS/nix/pull/3216#issuecomment-553956703
2020-04-29 23:39:58 +03:00
static void fetchTree(
2021-10-06 18:29:47 +03:00
EvalState & state,
const PosIdx pos,
2021-10-06 18:29:47 +03:00
Value * * args,
Value & v,
std::optional<std::string> type,
const FetchTreeParams & params = FetchTreeParams{}
Merge legacy `fetchGit`-builtin with the generic `fetchTree`-function The original idea was to implement a git-fetcher in Nix's core that supports content hashes[1]. In #3549[2] it has been suggested to actually use `fetchTree` for this since it's a fairly generic wrapper over the new fetcher-API[3] and already supports content-hashes. This patch implements a new git-fetcher based on `fetchTree` by incorporating the following changes: * Removed the original `fetchGit`-implementation and replaced it with an alias on the `fetchTree` implementation. * Ensured that the `git`-fetcher from `libfetchers` always computes a content-hash and returns an "empty" revision on dirty trees (the latter one is needed to retain backwards-compatibility). * The hash-mismatch error in the fetcher-API exits with code 102 as it usually happens whenever a hash-mismatch is detected by Nix. * Removed the `flakes`-feature-flag: I didn't see a reason why this API is so tightly coupled to the flakes-API and at least `fetchGit` should remain usable without any feature-flags. * It's only possible to specify a `narHash` for a `git`-tree if either a `ref` or a `rev` is given[4]. * It's now possible to specify an URL without a protocol. If it's missing, `file://` is automatically added as it was the case in the original `fetchGit`-implementation. [1] https://github.com/NixOS/nix/pull/3216 [2] https://github.com/NixOS/nix/pull/3549#issuecomment-625194383 [3] https://github.com/NixOS/nix/pull/3459 [4] https://github.com/NixOS/nix/pull/3216#issuecomment-553956703
2020-04-29 23:39:58 +03:00
) {
fetchers::Input input;
Use `std::set<StringContextElem>` not `PathSet` for string contexts Motivation `PathSet` is not correct because string contexts have other forms (`Built` and `DrvDeep`) that are not rendered as plain store paths. Instead of wrongly using `PathSet`, or "stringly typed" using `StringSet`, use `std::std<StringContextElem>`. ----- In support of this change, `NixStringContext` is now defined as `std::std<StringContextElem>` not `std:vector<StringContextElem>`. The old definition was just used by a `getContext` method which was only used by the eval cache. It can be deleted altogether since the types are now unified and the preexisting `copyContext` function already suffices. Summarizing the previous paragraph: Old: - `value/context.hh`: `NixStringContext = std::vector<StringContextElem>` - `value.hh`: `NixStringContext Value::getContext(...)` - `value.hh`: `copyContext(...)` New: - `value/context.hh`: `NixStringContext = std::set<StringContextElem>` - `value.hh`: `copyContext(...)` ---- The string representation of string context elements no longer contains the store dir. The diff of `src/libexpr/tests/value/context.cc` should make clear what the new representation is, so we recommend reviewing that file first. This was done for two reasons: Less API churn: `Value::mkString` and friends did not take a `Store` before. But if `NixStringContextElem::{parse, to_string}` *do* take a store (as they did before), then we cannot have the `Value` functions use them (in order to work with the fully-structured `NixStringContext`) without adding that argument. That would have been a lot of churn of threading the store, and this diff is already large enough, so the easier and less invasive thing to do was simply make the element `parse` and `to_string` functions not take the `Store` reference, and the easiest way to do that was to simply drop the store dir. Space usage: Dropping the `/nix/store/` (or similar) from the internal representation will safe space in the heap of the Nix programming being interpreted. If the heap contains many strings with non-trivial contexts, the saving could add up to something significant. ---- The eval cache version is bumped. The eval cache serialization uses `NixStringContextElem::{parse, to_string}`, and since those functions are changed per the above, that means the on-disk representation is also changed. This is simply done by changing the name of the used for the eval cache from `eval-cache-v4` to eval-cache-v5`. ---- To avoid some duplication `EvalCache::mkPathString` is added to abstract over the simple case of turning a store path to a string with just that string in the context. Context This PR picks up where #7543 left off. That one introduced the fully structured `NixStringContextElem` data type, but kept `PathSet context` as an awkward middle ground between internal `char[][]` interpreter heap string contexts and `NixStringContext` fully parsed string contexts. The infelicity of `PathSet context` was specifically called out during Nix team group review, but it was agreeing that fixing it could be left as future work. This is that future work. A possible follow-up step would be to get rid of the `char[][]` evaluator heap representation, too, but it is not yet clear how to do that. To use `NixStringContextElem` there we would need to get the STL containers to GC pointers in the GC build, and I am not sure how to do that. ---- PR #7543 effectively is writing the inverse of a `mkPathString`, `mkOutputString`, and one more such function for the `DrvDeep` case. I would like that PR to have property tests ensuring it is actually the inverse as expected. This PR sets things up nicely so that reworking that PR to be in that more elegant and better tested way is possible. Co-authored-by: Théophane Hufschmitt <7226587+thufschmitt@users.noreply.github.com>
2023-01-29 03:31:10 +02:00
NixStringContext context;
state.forceValue(*args[0], pos);
if (args[0]->type() == nAttrs) {
2022-04-28 13:54:14 +03:00
state.forceAttrs(*args[0], pos, "while evaluating the argument passed to builtins.fetchTree");
fetchers::Attrs attrs;
if (auto aType = args[0]->attrs->get(state.sType)) {
if (type)
2022-05-25 13:32:22 +03:00
state.debugThrowLastTrace(EvalError({
.msg = hintfmt("unexpected attribute 'type'"),
.errPos = state.positions[pos]
2022-05-25 13:32:22 +03:00
}));
type = state.forceStringNoCtx(*aType->value, aType->pos, "while evaluating the `type` attribute passed to builtins.fetchTree");
} else if (!type)
2022-05-25 13:32:22 +03:00
state.debugThrowLastTrace(EvalError({
.msg = hintfmt("attribute 'type' is missing in call to 'fetchTree'"),
.errPos = state.positions[pos]
2022-05-25 13:32:22 +03:00
}));
attrs.emplace("type", type.value());
for (auto & attr : *args[0]->attrs) {
if (attr.name == state.sType) continue;
state.forceValue(*attr.value, attr.pos);
2021-10-06 18:29:47 +03:00
if (attr.value->type() == nPath || attr.value->type() == nString) {
2022-11-29 01:25:36 +02:00
auto s = state.coerceToString(attr.pos, *attr.value, context, "", false, false).toOwned();
attrs.emplace(state.symbols[attr.name],
state.symbols[attr.name] == "url"
2021-10-06 18:29:47 +03:00
? type == "git"
? fixURIForGit(s, state)
: fixURI(s, state)
: s);
}
else if (attr.value->type() == nBool)
attrs.emplace(state.symbols[attr.name], Explicit<bool>{attr.value->boolean});
2020-12-29 03:40:04 +02:00
else if (attr.value->type() == nInt)
attrs.emplace(state.symbols[attr.name], uint64_t(attr.value->integer));
else
2022-05-25 13:32:22 +03:00
state.debugThrowLastTrace(TypeError("fetchTree argument '%s' is %s while a string, Boolean or integer is expected",
state.symbols[attr.name], showType(*attr.value)));
}
if (!params.allowNameArgument)
if (auto nameIter = attrs.find("name"); nameIter != attrs.end())
2022-05-25 13:32:22 +03:00
state.debugThrowLastTrace(EvalError({
.msg = hintfmt("attribute 'name' isnt supported in call to 'fetchTree'"),
.errPos = state.positions[pos]
2022-05-25 13:32:22 +03:00
}));
input = fetchers::Input::fromAttrs(std::move(attrs));
Merge legacy `fetchGit`-builtin with the generic `fetchTree`-function The original idea was to implement a git-fetcher in Nix's core that supports content hashes[1]. In #3549[2] it has been suggested to actually use `fetchTree` for this since it's a fairly generic wrapper over the new fetcher-API[3] and already supports content-hashes. This patch implements a new git-fetcher based on `fetchTree` by incorporating the following changes: * Removed the original `fetchGit`-implementation and replaced it with an alias on the `fetchTree` implementation. * Ensured that the `git`-fetcher from `libfetchers` always computes a content-hash and returns an "empty" revision on dirty trees (the latter one is needed to retain backwards-compatibility). * The hash-mismatch error in the fetcher-API exits with code 102 as it usually happens whenever a hash-mismatch is detected by Nix. * Removed the `flakes`-feature-flag: I didn't see a reason why this API is so tightly coupled to the flakes-API and at least `fetchGit` should remain usable without any feature-flags. * It's only possible to specify a `narHash` for a `git`-tree if either a `ref` or a `rev` is given[4]. * It's now possible to specify an URL without a protocol. If it's missing, `file://` is automatically added as it was the case in the original `fetchGit`-implementation. [1] https://github.com/NixOS/nix/pull/3216 [2] https://github.com/NixOS/nix/pull/3549#issuecomment-625194383 [3] https://github.com/NixOS/nix/pull/3459 [4] https://github.com/NixOS/nix/pull/3216#issuecomment-553956703
2020-04-29 23:39:58 +03:00
} else {
2022-11-29 01:25:36 +02:00
auto url = state.coerceToString(pos, *args[0], context,
"while evaluating the first argument passed to the fetcher",
false, false).toOwned();
Merge legacy `fetchGit`-builtin with the generic `fetchTree`-function The original idea was to implement a git-fetcher in Nix's core that supports content hashes[1]. In #3549[2] it has been suggested to actually use `fetchTree` for this since it's a fairly generic wrapper over the new fetcher-API[3] and already supports content-hashes. This patch implements a new git-fetcher based on `fetchTree` by incorporating the following changes: * Removed the original `fetchGit`-implementation and replaced it with an alias on the `fetchTree` implementation. * Ensured that the `git`-fetcher from `libfetchers` always computes a content-hash and returns an "empty" revision on dirty trees (the latter one is needed to retain backwards-compatibility). * The hash-mismatch error in the fetcher-API exits with code 102 as it usually happens whenever a hash-mismatch is detected by Nix. * Removed the `flakes`-feature-flag: I didn't see a reason why this API is so tightly coupled to the flakes-API and at least `fetchGit` should remain usable without any feature-flags. * It's only possible to specify a `narHash` for a `git`-tree if either a `ref` or a `rev` is given[4]. * It's now possible to specify an URL without a protocol. If it's missing, `file://` is automatically added as it was the case in the original `fetchGit`-implementation. [1] https://github.com/NixOS/nix/pull/3216 [2] https://github.com/NixOS/nix/pull/3549#issuecomment-625194383 [3] https://github.com/NixOS/nix/pull/3459 [4] https://github.com/NixOS/nix/pull/3216#issuecomment-553956703
2020-04-29 23:39:58 +03:00
if (type == "git") {
fetchers::Attrs attrs;
attrs.emplace("type", "git");
attrs.emplace("url", fixURIForGit(url, state));
Merge legacy `fetchGit`-builtin with the generic `fetchTree`-function The original idea was to implement a git-fetcher in Nix's core that supports content hashes[1]. In #3549[2] it has been suggested to actually use `fetchTree` for this since it's a fairly generic wrapper over the new fetcher-API[3] and already supports content-hashes. This patch implements a new git-fetcher based on `fetchTree` by incorporating the following changes: * Removed the original `fetchGit`-implementation and replaced it with an alias on the `fetchTree` implementation. * Ensured that the `git`-fetcher from `libfetchers` always computes a content-hash and returns an "empty" revision on dirty trees (the latter one is needed to retain backwards-compatibility). * The hash-mismatch error in the fetcher-API exits with code 102 as it usually happens whenever a hash-mismatch is detected by Nix. * Removed the `flakes`-feature-flag: I didn't see a reason why this API is so tightly coupled to the flakes-API and at least `fetchGit` should remain usable without any feature-flags. * It's only possible to specify a `narHash` for a `git`-tree if either a `ref` or a `rev` is given[4]. * It's now possible to specify an URL without a protocol. If it's missing, `file://` is automatically added as it was the case in the original `fetchGit`-implementation. [1] https://github.com/NixOS/nix/pull/3216 [2] https://github.com/NixOS/nix/pull/3549#issuecomment-625194383 [3] https://github.com/NixOS/nix/pull/3459 [4] https://github.com/NixOS/nix/pull/3216#issuecomment-553956703
2020-04-29 23:39:58 +03:00
input = fetchers::Input::fromAttrs(std::move(attrs));
} else {
input = fetchers::Input::fromURL(fixURI(url, state));
Merge legacy `fetchGit`-builtin with the generic `fetchTree`-function The original idea was to implement a git-fetcher in Nix's core that supports content hashes[1]. In #3549[2] it has been suggested to actually use `fetchTree` for this since it's a fairly generic wrapper over the new fetcher-API[3] and already supports content-hashes. This patch implements a new git-fetcher based on `fetchTree` by incorporating the following changes: * Removed the original `fetchGit`-implementation and replaced it with an alias on the `fetchTree` implementation. * Ensured that the `git`-fetcher from `libfetchers` always computes a content-hash and returns an "empty" revision on dirty trees (the latter one is needed to retain backwards-compatibility). * The hash-mismatch error in the fetcher-API exits with code 102 as it usually happens whenever a hash-mismatch is detected by Nix. * Removed the `flakes`-feature-flag: I didn't see a reason why this API is so tightly coupled to the flakes-API and at least `fetchGit` should remain usable without any feature-flags. * It's only possible to specify a `narHash` for a `git`-tree if either a `ref` or a `rev` is given[4]. * It's now possible to specify an URL without a protocol. If it's missing, `file://` is automatically added as it was the case in the original `fetchGit`-implementation. [1] https://github.com/NixOS/nix/pull/3216 [2] https://github.com/NixOS/nix/pull/3549#issuecomment-625194383 [3] https://github.com/NixOS/nix/pull/3459 [4] https://github.com/NixOS/nix/pull/3216#issuecomment-553956703
2020-04-29 23:39:58 +03:00
}
}
if (!evalSettings.pureEval && !input.isDirect())
input = lookupInRegistries(state.store, input).first;
if (evalSettings.pureEval && !input.isLocked())
2022-05-25 13:32:22 +03:00
state.debugThrowLastTrace(EvalError("in pure evaluation mode, 'fetchTree' requires a locked input, at %s", state.positions[pos]));
auto [tree, input2] = input.fetch(state.store);
state.allowPath(tree.storePath);
emitTreeAttrs(state, tree, input2, v, params.emptyRevFallback, false);
}
static void prim_fetchTree(EvalState & state, const PosIdx pos, Value * * args, Value & v)
Merge legacy `fetchGit`-builtin with the generic `fetchTree`-function The original idea was to implement a git-fetcher in Nix's core that supports content hashes[1]. In #3549[2] it has been suggested to actually use `fetchTree` for this since it's a fairly generic wrapper over the new fetcher-API[3] and already supports content-hashes. This patch implements a new git-fetcher based on `fetchTree` by incorporating the following changes: * Removed the original `fetchGit`-implementation and replaced it with an alias on the `fetchTree` implementation. * Ensured that the `git`-fetcher from `libfetchers` always computes a content-hash and returns an "empty" revision on dirty trees (the latter one is needed to retain backwards-compatibility). * The hash-mismatch error in the fetcher-API exits with code 102 as it usually happens whenever a hash-mismatch is detected by Nix. * Removed the `flakes`-feature-flag: I didn't see a reason why this API is so tightly coupled to the flakes-API and at least `fetchGit` should remain usable without any feature-flags. * It's only possible to specify a `narHash` for a `git`-tree if either a `ref` or a `rev` is given[4]. * It's now possible to specify an URL without a protocol. If it's missing, `file://` is automatically added as it was the case in the original `fetchGit`-implementation. [1] https://github.com/NixOS/nix/pull/3216 [2] https://github.com/NixOS/nix/pull/3549#issuecomment-625194383 [3] https://github.com/NixOS/nix/pull/3459 [4] https://github.com/NixOS/nix/pull/3216#issuecomment-553956703
2020-04-29 23:39:58 +03:00
{
fetchTree(state, pos, args, v, std::nullopt, FetchTreeParams { .allowNameArgument = false });
Merge legacy `fetchGit`-builtin with the generic `fetchTree`-function The original idea was to implement a git-fetcher in Nix's core that supports content hashes[1]. In #3549[2] it has been suggested to actually use `fetchTree` for this since it's a fairly generic wrapper over the new fetcher-API[3] and already supports content-hashes. This patch implements a new git-fetcher based on `fetchTree` by incorporating the following changes: * Removed the original `fetchGit`-implementation and replaced it with an alias on the `fetchTree` implementation. * Ensured that the `git`-fetcher from `libfetchers` always computes a content-hash and returns an "empty" revision on dirty trees (the latter one is needed to retain backwards-compatibility). * The hash-mismatch error in the fetcher-API exits with code 102 as it usually happens whenever a hash-mismatch is detected by Nix. * Removed the `flakes`-feature-flag: I didn't see a reason why this API is so tightly coupled to the flakes-API and at least `fetchGit` should remain usable without any feature-flags. * It's only possible to specify a `narHash` for a `git`-tree if either a `ref` or a `rev` is given[4]. * It's now possible to specify an URL without a protocol. If it's missing, `file://` is automatically added as it was the case in the original `fetchGit`-implementation. [1] https://github.com/NixOS/nix/pull/3216 [2] https://github.com/NixOS/nix/pull/3549#issuecomment-625194383 [3] https://github.com/NixOS/nix/pull/3459 [4] https://github.com/NixOS/nix/pull/3216#issuecomment-553956703
2020-04-29 23:39:58 +03:00
}
static RegisterPrimOp primop_fetchTree({
.name = "fetchTree",
.args = {"input"},
.doc = R"(
Fetch a source tree or a plain file using one of the supported backends.
*input* can be an attribute set representation of [flake reference](@docroot@/command-ref/new-cli/nix3-flake.md#flake-references) or a URL.
The input should be "locked", that is, it should contain a commit hash or content hash unless impure evaluation (`--impure`) is allowed.
Here are some examples of how to use `fetchTree`:
- Fetch a GitHub repository:
```nix
builtins.fetchTree {
type = "github";
owner = "NixOS";
repo = "nixpkgs";
rev = "ae2e6b3958682513d28f7d633734571fb18285dd";
}
```
This evaluates to attribute set:
```
{
lastModified = 1686503798;
lastModifiedDate = "20230611171638";
narHash = "sha256-rA9RqKP9OlBrgGCPvfd5HVAXDOy8k2SmPtB/ijShNXc=";
outPath = "/nix/store/l5m6qlvfs9sdw14ja3qbzpglcjlb6j1x-source";
rev = "ae2e6b3958682513d28f7d633734571fb18285dd";
shortRev = "ae2e6b3";
}
```
- Fetch a single file from a URL:
```nix
builtins.fetchTree "https://example.com/"
```
> **Note**
>
> This function requires the [`flakes` experimental feature](@docroot@/contributing/experimental-features.md#xp-feature-flakes) to be enabled.
)",
.fun = prim_fetchTree,
.experimentalFeature = Xp::Flakes,
});
static void fetch(EvalState & state, const PosIdx pos, Value * * args, Value & v,
const std::string & who, bool unpack, std::string name)
{
std::optional<std::string> url;
std::optional<Hash> expectedHash;
state.forceValue(*args[0], pos);
if (args[0]->type() == nAttrs) {
for (auto & attr : *args[0]->attrs) {
std::string_view n(state.symbols[attr.name]);
if (n == "url")
url = state.forceStringNoCtx(*attr.value, attr.pos, "while evaluating the url we should fetch");
else if (n == "sha256")
expectedHash = newHashAllowEmpty(state.forceStringNoCtx(*attr.value, attr.pos, "while evaluating the sha256 of the content we should fetch"), htSHA256);
else if (n == "name")
name = state.forceStringNoCtx(*attr.value, attr.pos, "while evaluating the name of the content we should fetch");
else
2022-05-25 13:32:22 +03:00
state.debugThrowLastTrace(EvalError({
.msg = hintfmt("unsupported argument '%s' to '%s'", n, who),
.errPos = state.positions[attr.pos]
2022-05-25 13:32:22 +03:00
}));
}
if (!url)
2022-05-25 13:32:22 +03:00
state.debugThrowLastTrace(EvalError({
Improve error formatting Changes: * The divider lines are gone. These were in practice a bit confusing, in particular with --show-trace or --keep-going, since then there were multiple lines, suggesting a start/end which wasn't the case. * Instead, multi-line error messages are now indented to align with the prefix (e.g. "error: "). * The 'description' field is gone since we weren't really using it. * 'hint' is renamed to 'msg' since it really wasn't a hint. * The error is now printed *before* the location info. * The 'name' field is no longer printed since most of the time it wasn't very useful since it was just the name of the exception (like EvalError). Ideally in the future this would be a unique, easily googleable error ID (like rustc). * "trace:" is now just "…". This assumes error contexts start with something like "while doing X". Example before: error: --- AssertionError ---------------------------------------------------------------------------------------- nix at: (7:7) in file: /home/eelco/Dev/nixpkgs/pkgs/applications/misc/hello/default.nix 6| 7| x = assert false; 1; | ^ 8| assertion 'false' failed ----------------------------------------------------- show-trace ----------------------------------------------------- trace: while evaluating the attribute 'x' of the derivation 'hello-2.10' at: (192:11) in file: /home/eelco/Dev/nixpkgs/pkgs/stdenv/generic/make-derivation.nix 191| // (lib.optionalAttrs (!(attrs ? name) && attrs ? pname && attrs ? version)) { 192| name = "${attrs.pname}-${attrs.version}"; | ^ 193| } // (lib.optionalAttrs (stdenv.hostPlatform != stdenv.buildPlatform && !dontAddHostSuffix && (attrs ? name || (attrs ? pname && attrs ? version)))) { Example after: error: assertion 'false' failed at: (7:7) in file: /home/eelco/Dev/nixpkgs/pkgs/applications/misc/hello/default.nix 6| 7| x = assert false; 1; | ^ 8| … while evaluating the attribute 'x' of the derivation 'hello-2.10' at: (192:11) in file: /home/eelco/Dev/nixpkgs/pkgs/stdenv/generic/make-derivation.nix 191| // (lib.optionalAttrs (!(attrs ? name) && attrs ? pname && attrs ? version)) { 192| name = "${attrs.pname}-${attrs.version}"; | ^ 193| } // (lib.optionalAttrs (stdenv.hostPlatform != stdenv.buildPlatform && !dontAddHostSuffix && (attrs ? name || (attrs ? pname && attrs ? version)))) {
2021-01-21 01:27:36 +02:00
.msg = hintfmt("'url' argument required"),
.errPos = state.positions[pos]
2022-05-25 13:32:22 +03:00
}));
} else
2022-04-28 13:54:14 +03:00
url = state.forceStringNoCtx(*args[0], pos, "while evaluating the url we should fetch");
if (who == "fetchTarball")
url = evalSettings.resolvePseudoUrl(*url);
state.checkURI(*url);
if (name == "")
name = baseNameOf(*url);
if (evalSettings.pureEval && !expectedHash)
2022-05-25 13:32:22 +03:00
state.debugThrowLastTrace(EvalError("in pure evaluation mode, '%s' requires a 'sha256' argument", who));
// early exit if pinned and already in the store
if (expectedHash && expectedHash->type == htSHA256) {
auto expectedPath = state.store->makeFixedOutputPath(
name,
FixedOutputInfo {
.method = unpack ? FileIngestionMethod::Recursive : FileIngestionMethod::Flat,
.hash = *expectedHash,
.references = {}
});
if (state.store->isValidPath(expectedPath)) {
state.allowAndSetStorePathString(expectedPath, v);
return;
}
}
// TODO: fetching may fail, yet the path may be substitutable.
// https://github.com/NixOS/nix/issues/4313
auto storePath =
unpack
? fetchers::downloadTarball(state.store, *url, name, (bool) expectedHash).tree.storePath
2020-09-29 14:05:19 +03:00
: fetchers::downloadFile(state.store, *url, name, (bool) expectedHash).storePath;
if (expectedHash) {
auto hash = unpack
? state.store->queryPathInfo(storePath)->narHash
: hashFile(htSHA256, state.store->toRealPath(storePath));
if (hash != *expectedHash)
2022-05-25 13:32:22 +03:00
state.debugThrowLastTrace(EvalError((unsigned int) 102, "hash mismatch in file downloaded from '%s':\n specified: %s\n got: %s",
*url, expectedHash->to_string(Base32, true), hash.to_string(Base32, true)));
}
state.allowAndSetStorePathString(storePath, v);
}
static void prim_fetchurl(EvalState & state, const PosIdx pos, Value * * args, Value & v)
{
fetch(state, pos, args, v, "fetchurl", false, "");
}
static RegisterPrimOp primop_fetchurl({
.name = "__fetchurl",
.args = {"url"},
.doc = R"(
Download the specified URL and return the path of the downloaded file.
Not available in [restricted evaluation mode](@docroot@/command-ref/conf-file.md#conf-restrict-eval).
)",
.fun = prim_fetchurl,
});
static void prim_fetchTarball(EvalState & state, const PosIdx pos, Value * * args, Value & v)
{
fetch(state, pos, args, v, "fetchTarball", true, "source");
}
static RegisterPrimOp primop_fetchTarball({
.name = "fetchTarball",
.args = {"args"},
.doc = R"(
Download the specified URL, unpack it and return the path of the
unpacked tree. The file must be a tape archive (`.tar`) compressed
with `gzip`, `bzip2` or `xz`. The top-level path component of the
files in the tarball is removed, so it is best if the tarball
contains a single directory at top level. The typical use of the
function is to obtain external Nix expression dependencies, such as
a particular version of Nixpkgs, e.g.
```nix
with import (fetchTarball https://github.com/NixOS/nixpkgs/archive/nixos-14.12.tar.gz) {};
stdenv.mkDerivation { }
```
2021-10-26 15:21:24 +03:00
The fetched tarball is cached for a certain amount of time (1
hour by default) in `~/.cache/nix/tarballs/`. You can change the
cache timeout either on the command line with `--tarball-ttl`
*number-of-seconds* or in the Nix configuration file by adding
the line `tarball-ttl = ` *number-of-seconds*.
2021-10-26 15:21:24 +03:00
Note that when obtaining the hash with `nix-prefetch-url` the
option `--unpack` is required.
This function can also verify the contents against a hash. In that
case, the function takes a set instead of a URL. The set requires
the attribute `url` and the attribute `sha256`, e.g.
```nix
with import (fetchTarball {
url = "https://github.com/NixOS/nixpkgs/archive/nixos-14.12.tar.gz";
sha256 = "1jppksrfvbk5ypiqdz4cddxdl8z6zyzdb2srq8fcffr327ld5jj2";
}) {};
stdenv.mkDerivation { }
```
Not available in [restricted evaluation mode](@docroot@/command-ref/conf-file.md#conf-restrict-eval).
)",
.fun = prim_fetchTarball,
});
static void prim_fetchGit(EvalState & state, const PosIdx pos, Value * * args, Value & v)
Merge legacy `fetchGit`-builtin with the generic `fetchTree`-function The original idea was to implement a git-fetcher in Nix's core that supports content hashes[1]. In #3549[2] it has been suggested to actually use `fetchTree` for this since it's a fairly generic wrapper over the new fetcher-API[3] and already supports content-hashes. This patch implements a new git-fetcher based on `fetchTree` by incorporating the following changes: * Removed the original `fetchGit`-implementation and replaced it with an alias on the `fetchTree` implementation. * Ensured that the `git`-fetcher from `libfetchers` always computes a content-hash and returns an "empty" revision on dirty trees (the latter one is needed to retain backwards-compatibility). * The hash-mismatch error in the fetcher-API exits with code 102 as it usually happens whenever a hash-mismatch is detected by Nix. * Removed the `flakes`-feature-flag: I didn't see a reason why this API is so tightly coupled to the flakes-API and at least `fetchGit` should remain usable without any feature-flags. * It's only possible to specify a `narHash` for a `git`-tree if either a `ref` or a `rev` is given[4]. * It's now possible to specify an URL without a protocol. If it's missing, `file://` is automatically added as it was the case in the original `fetchGit`-implementation. [1] https://github.com/NixOS/nix/pull/3216 [2] https://github.com/NixOS/nix/pull/3549#issuecomment-625194383 [3] https://github.com/NixOS/nix/pull/3459 [4] https://github.com/NixOS/nix/pull/3216#issuecomment-553956703
2020-04-29 23:39:58 +03:00
{
fetchTree(state, pos, args, v, "git", FetchTreeParams { .emptyRevFallback = true, .allowNameArgument = true });
Merge legacy `fetchGit`-builtin with the generic `fetchTree`-function The original idea was to implement a git-fetcher in Nix's core that supports content hashes[1]. In #3549[2] it has been suggested to actually use `fetchTree` for this since it's a fairly generic wrapper over the new fetcher-API[3] and already supports content-hashes. This patch implements a new git-fetcher based on `fetchTree` by incorporating the following changes: * Removed the original `fetchGit`-implementation and replaced it with an alias on the `fetchTree` implementation. * Ensured that the `git`-fetcher from `libfetchers` always computes a content-hash and returns an "empty" revision on dirty trees (the latter one is needed to retain backwards-compatibility). * The hash-mismatch error in the fetcher-API exits with code 102 as it usually happens whenever a hash-mismatch is detected by Nix. * Removed the `flakes`-feature-flag: I didn't see a reason why this API is so tightly coupled to the flakes-API and at least `fetchGit` should remain usable without any feature-flags. * It's only possible to specify a `narHash` for a `git`-tree if either a `ref` or a `rev` is given[4]. * It's now possible to specify an URL without a protocol. If it's missing, `file://` is automatically added as it was the case in the original `fetchGit`-implementation. [1] https://github.com/NixOS/nix/pull/3216 [2] https://github.com/NixOS/nix/pull/3549#issuecomment-625194383 [3] https://github.com/NixOS/nix/pull/3459 [4] https://github.com/NixOS/nix/pull/3216#issuecomment-553956703
2020-04-29 23:39:58 +03:00
}
static RegisterPrimOp primop_fetchGit({
.name = "fetchGit",
.args = {"args"},
.doc = R"(
Fetch a path from git. *args* can be a URL, in which case the HEAD
of the repo at that URL is fetched. Otherwise, it can be an
attribute with the following attributes (all except `url` optional):
- `url`
The URL of the repo.
- `name` (default: *basename of the URL*)
The name of the directory the repo should be exported to in the store.
- `rev` (default: *the tip of `ref`*)
The [Git revision] to fetch.
This is typically a commit hash.
[Git revision]: https://git-scm.com/docs/git-rev-parse#_specifying_revisions
2022-06-26 15:00:00 +03:00
- `ref` (default: `HEAD`)
The [Git reference] under which to look for the requested revision.
This is often a branch or tag name.
[Git reference]: https://git-scm.com/book/en/v2/Git-Internals-Git-References
By default, the `ref` value is prefixed with `refs/heads/`.
As of 2.3.0, Nix will not prefix `refs/heads/` if `ref` starts with `refs/`.
- `submodules` (default: `false`)
A Boolean parameter that specifies whether submodules should be checked out.
- `shallow` (default: `false`)
A Boolean parameter that specifies whether fetching a shallow clone is allowed.
- `allRefs`
Whether to fetch all references of the repository.
With this argument being true, it's possible to load a `rev` from *any* `ref`
(by default only `rev`s from the specified `ref` are supported).
Here are some examples of how to use `fetchGit`.
- To fetch a private repository over SSH:
```nix
builtins.fetchGit {
url = "git@github.com:my-secret/repository.git";
ref = "master";
rev = "adab8b916a45068c044658c4158d81878f9ed1c3";
}
```
- To fetch an arbitrary reference:
```nix
builtins.fetchGit {
url = "https://github.com/NixOS/nix.git";
ref = "refs/heads/0.5-release";
}
```
- If the revision you're looking for is in the default branch of
the git repository you don't strictly need to specify the branch
name in the `ref` attribute.
However, if the revision you're looking for is in a future
branch for the non-default branch you will need to specify the
the `ref` attribute as well.
```nix
builtins.fetchGit {
url = "https://github.com/nixos/nix.git";
rev = "841fcbd04755c7a2865c51c1e2d3b045976b7452";
ref = "1.11-maintenance";
}
```
> **Note**
>
> It is nice to always specify the branch which a revision
> belongs to. Without the branch being specified, the fetcher
> might fail if the default branch changes. Additionally, it can
> be confusing to try a commit from a non-default branch and see
> the fetch fail. If the branch is specified the fault is much
> more obvious.
- If the revision you're looking for is in the default branch of
the git repository you may omit the `ref` attribute.
```nix
builtins.fetchGit {
url = "https://github.com/nixos/nix.git";
rev = "841fcbd04755c7a2865c51c1e2d3b045976b7452";
}
```
- To fetch a specific tag:
```nix
builtins.fetchGit {
url = "https://github.com/nixos/nix.git";
ref = "refs/tags/1.9";
}
```
- To fetch the latest version of a remote branch:
```nix
builtins.fetchGit {
url = "ssh://git@github.com/nixos/nix.git";
ref = "master";
}
```
Nix will refetch the branch according to the [`tarball-ttl`](@docroot@/command-ref/conf-file.md#conf-tarball-ttl) setting.
This behavior is disabled in [pure evaluation mode](@docroot@/command-ref/conf-file.md#conf-pure-eval).
- To fetch the content of a checked-out work directory:
```nix
builtins.fetchGit ./work-dir
```
If the URL points to a local directory, and no `ref` or `rev` is
given, `fetchGit` will use the current content of the checked-out
files, even if they are not committed or added to Git's index. It will
only consider files added to the Git repository, as listed by `git ls-files`.
)",
.fun = prim_fetchGit,
});
}