hgitaly 1.3.1__tar.gz → 2.10.0__tar.gz

{hgitaly-1.3.1/hgitaly.egg-info → hgitaly-2.10.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: hgitaly
-Version: 1.3.1
+Version: 2.10.0
 Summary: Server-side implementation of Gitaly protocol for Mercurial
 Home-page: https://foss.heptapod.net/heptapod/hgitaly
 Author: Georges Racinet
@@ -10,6 +10,15 @@ Keywords: hg mercurial heptapod gitlab
 Requires-Python: >=3.8
 Description-Content-Type: text/markdown
 License-File: LICENSE
+Requires-Dist: heptapod>=5.0.0dev0
+Requires-Dist: protobuf~=4.21.0
+Requires-Dist: grpcio~=1.58.0
+Requires-Dist: grpcio-status~=1.58.0
+Requires-Dist: grpc-interceptor
+Requires-Dist: hg-loggingmod>=0.4.1
+Requires-Dist: psutil
+Requires-Dist: importlib_resources~=2.0.0
+Requires-Dist: spdx-lookup
 
 # HGitaly
 

{hgitaly-1.3.1 → hgitaly-2.10.0}/hgext3rd/hgitaly/__init__.py

@@ -6,6 +6,8 @@
 # SPDX-License-Identifier: GPL-2.0-or-later
 
 # flake8: noqa E402
+import os
+import time
 from mercurial.i18n import _
 from mercurial import (
     demandimport,
@@ -81,6 +83,13 @@ def serve(ui, **opts):
     By default the root of repositories is read from the
     `heptapod.repositories-root` configuration item if present.
     """
+    # local time does not make sense for a server-oriented process,
+    # so let's switch to UTC. Mercurial itself is transparent about that
+    # server-side commits will be recorded as been done UTC, which is fine
+    # (it does not change the recorded time, which is always UTC, just some
+    # ways of displaying it).
+    os.environ['TZ'] = 'UTC'
+    time.tzset()
     listen_urls = [pycompat.sysstr(u) for u in opts['listen']]
     if not listen_urls:
         # Any default in the option declaration would be added to

hgitaly-2.10.0/hgitaly/VERSION

@@ -0,0 +1 @@
+2.10.0

{hgitaly-1.3.1 → hgitaly-2.10.0}/hgitaly/diff.py

@@ -46,17 +46,19 @@ Status_Type_Map = dict(
     added=ChangedPaths.Status.ADDED,
     modified=ChangedPaths.Status.MODIFIED,
     removed=ChangedPaths.Status.DELETED,
+    renamed=ChangedPaths.Status.RENAMED,
     # Note: Mercurial includes TYPE_CHANGE
     # (symlink, regular file, submodule...etc) in MODIFIED status
 )
 """Mapping status object attributes to ChangedPaths enum."""
 
 COPIED = ChangedPaths.Status.COPIED
+RENAMED = ChangedPaths.Status.RENAMED
 DIFF_HUNKS_START_RX = re.compile(rb'^(--- )|^(Binary file)')
 """To match the header line right before hunks start getting dumped."""
 
 
-def changed_paths(repo, from_ctx, to_ctx, base_path):
+def changed_paths(repo, from_ctx, to_ctx, base_path, find_renames=False):
     if base_path is None:
         matcher = None
         path_trim_at = 0
@@ -74,26 +76,33 @@ def changed_paths(repo, from_ctx, to_ctx, base_path):
             patterns=[b'path:' + base_path])
         path_trim_at = len(base_path) + 1
 
-    copied = list(copy_changed_paths(
-        from_ctx,
-        to_ctx,
-        copies.pathcopies(from_ctx, to_ctx, match=matcher),
-        trim_at=path_trim_at))
-    copied_paths = set(cp.path for cp in copied)
+    copy_info = copies.pathcopies(from_ctx, to_ctx, match=matcher)
+    # copies do not distinguish actual copies from renames. The difference
+    # will be that a rename goes with deletion of the original.
 
     status = from_ctx.status(to_ctx, match=matcher)
-    for path in status_changed_paths(from_ctx, to_ctx, status,
-                                     trim_at=path_trim_at):
-        if path.path not in copied_paths:
-            yield path
-
-    yield from iter(copied)
-
-
-def status_changed_paths(from_ctx, to_ctx, status, trim_at=0):
+    # this will remove renames from copy_info, keeping only actual copies
+    yield from status_changed_paths(from_ctx, to_ctx, status, copy_info,
+                                    find_renames=find_renames,
+                                    trim_at=path_trim_at)
+    yield from copy_changed_paths(from_ctx,
+                                  to_ctx,
+                                  copy_info,
+                                  trim_at=path_trim_at)
+
+
+def status_changed_paths(from_ctx, to_ctx, status, copy_info,
+                         find_renames=False, trim_at=0):
+    rcopy_info = {v: k for k, v in copy_info.items()}
     """Return ChangedPaths from Mercurial status object"""
     for stype in ['added', 'modified', 'removed']:
         for path in status.__getattribute__(stype):
+            copied_from = copy_info.get(path)
+            if copied_from is not None:
+                if copied_from not in status.removed or find_renames:
+                    continue
+
+            old_path = b''
             if stype == 'added':
                 old_mode = OBJECT_MODE_DOES_NOT_EXIST
                 old_blob_id = NULL_BLOB_OID
@@ -102,8 +111,21 @@ def status_changed_paths(from_ctx, to_ctx, status, trim_at=0):
                 old_blob_id = ctx_blob_oid(from_ctx, path)
 
             if stype == 'removed':
-                new_mode = OBJECT_MODE_DOES_NOT_EXIST
-                new_blob_id = NULL_BLOB_OID
+                new_path = rcopy_info.get(path)
+                if new_path is None:
+                    new_mode = OBJECT_MODE_DOES_NOT_EXIST
+                    new_blob_id = NULL_BLOB_OID
+                else:
+                    del copy_info[new_path]
+                    if find_renames:
+                        stype = 'renamed'
+                        old_path = path
+                        path = new_path
+                        new_mode = git_perms(to_ctx.filectx(new_path))
+                        new_blob_id = ctx_blob_oid(to_ctx, new_path)
+                    else:
+                        new_mode = OBJECT_MODE_DOES_NOT_EXIST
+                        new_blob_id = NULL_BLOB_OID
             else:
                 new_mode = git_perms(to_ctx.filectx(path))
                 new_blob_id = ctx_blob_oid(to_ctx, path)
@@ -114,11 +136,13 @@ def status_changed_paths(from_ctx, to_ctx, status, trim_at=0):
                 new_mode=new_mode,
                 old_blob_id=old_blob_id,
                 new_blob_id=new_blob_id,
+                old_path=old_path,
                 status=Status_Type_Map[stype]
             )
 
 
-def copy_changed_paths(from_ctx, to_ctx, path_copies, trim_at=0):
+def copy_changed_paths(from_ctx, to_ctx, path_copies,
+                       trim_at=0, find_renames=False):
     """Return ChangedPaths for the given paths, relative to base_path.
 
     Given that Gitaly currently (gitaly@c54d613d0) does not pass
@@ -129,8 +153,11 @@ def copy_changed_paths(from_ctx, to_ctx, path_copies, trim_at=0):
     for target, source in path_copies.items():
         yield ChangedPaths(path=target[trim_at:],
                            status=COPIED,
-                           old_mode=git_perms(to_ctx.filectx(source)),
+                           old_mode=git_perms(from_ctx.filectx(source)),
+                           old_blob_id=ctx_blob_oid(from_ctx, source),
                            new_mode=git_perms(to_ctx.filectx(target)),
+                           new_blob_id=ctx_blob_oid(to_ctx, target),
+                           old_path=source[trim_at:],
                            )
 
 

{hgitaly-1.3.1 → hgitaly-2.10.0}/hgitaly/errors.py

@@ -121,3 +121,27 @@ def operation_error_treatment(context, error_message_class, logger,
         #   // The post-receive hook runs after references have been updated
         #   // and any failures of it are ignored.
         logger.error("Error in GitLab post-receive hook: %r", exc.message)
+
+
+class MercurialNotFound(LookupError):
+    """General class for all failed lookups.
+
+    Better for us that to use `hg.errors.LookupError` because the latter is
+    not systematically raised. Anyway, we need our own articulation point.
+    """
+
+
+class MercurialPathNotFound(MercurialNotFound):
+    """Express that some path could not be found.
+
+    This is generally in the context of some changeset that the caller is
+    aware of, but it could be in an entire revset, in the workdir…
+    """
+
+
+class MercurialChangesetNotFound(MercurialNotFound):
+    """Express that one or serveral changesets could not be found.
+
+    In case the caller did a call possibly involving several changesets,
+    the failing, perhaps more specific, expression can be set as argument.
+    """

hgitaly-2.10.0/hgitaly/file_content.py

@@ -0,0 +1,160 @@
+# Copyright 2024 Georges Racinet <georges.racinet@cloudcrane.io>
+#
+# This software may be used and distributed according to the terms of the
+# GNU General Public License version 2 or any later version.
+#
+# SPDX-License-Identifier: GPL-2.0-or-later
+import re
+
+from mercurial import (
+    error as hgerror,
+)
+
+from .errors import (
+    MercurialPathNotFound,
+    MercurialChangesetNotFound,
+)
+from .oid import extract_blob_oid
+from .stub.analysis_pb2 import (
+    CheckBlobsGeneratedRequest,
+)
+
+
+def frozen_bytes_set(it):
+    return frozenset(s.encode('ascii') for s in it)
+
+
+# compare go-enry/data/generated.go
+GENERATED_FILE_PATH_REGEXPS = [
+    re.compile(b'|'.join((
+        rb'(^Pods|/Pods)/',  # Cocoa pods
+        rb'(^|/)Carthage/Build/',  # Carthage build
+        rb'\.feature\.css$',  # Generated NET specflow feature file
+        rb'node_modules/',  # Node modules
+        # Go vendor
+        rb'vendor/([-0-9A-Za-z]+\.)+(com|edu|gov|in|me|net|org|fm|io)',
+        rb'(Gopkg|glide).lock$',  # Go lock
+        rb'(^|/)(\w+\.)?esy.lock$`',  # Esy lock
+        rb'npm-shrinkwrap.json$',
+        rb'package-lock.json$',  # NPM package lock
+        rb'(^|/)\.pnp\.(c|m)?js$',  # Yarn plugnplay
+        rb'Godeps/',
+        rb'composer.lock$',
+        rb'.\.zep\.(?:c|h|php)$',  # Generated by zephir
+        rb'Cargo.lock$',
+        rb'Pipfile.lock$',  # Pipenv lock
+        rb'__generated__/',  # GraphQL relay
+        rb'\.(js|css)\.map$',  # Source map
+    ))),
+    re.compile(b'|'.join((
+        rb'\.designer\.(cs|vb)$',  # NET designer file
+    )), re.IGNORECASE),
+]
+
+# compare go-enry/data/generated.go
+GENERATED_FILE_EXTENSIONS = frozen_bytes_set((
+    # Xcode
+    'nib',
+    'xcworkspacedata',
+    'xcuserstate',
+    # TODO complete the list
+))
+
+GENERATED_LINE_RX = re.compile(
+    b'|'.join((
+        # Source Map
+        rb'\A{"version":\d+',
+        rb'\A/\*\*Begin line maps\. \*\*/{',
+        # Generated NetDoc  TODO should be only for .xml files
+        rb'\A.*$.*<doc>.*(\r)?\n.*<assembly>(\r)?\n.*</doc>(\r)?\n.*\Z',
+        # PEG.js
+        rb'^(?:[^/]|/[^\*])*/\*(?:[^\*]|\*[^/])*Generated by PEG.js',
+        # Generated Go (actually our version is more general, but anything
+        # matching this regexp is looking for trouble!
+        rb'Code generated by',
+        # Protocol Buffers, general case
+        b"Generated by the protocol buffer compiler.  DO NOT EDIT!",
+        # Protocol Buffers, JavaScript (should be only line 6 and crossed with
+        # extension, but given the content, matching this is acceptable
+        # TODO unless we are looking at a code generator!
+        b"GENERATED CODE -- DO NOT EDIT!",
+        # Apache Thrift
+        b"Autogenerated by Thrift Compiler",
+        # JNI header (again more general than go-enry, because we don't want
+        # to assert it to be JNI, just that it is generated)
+        rb"/\* DO NOT EDIT THIS FILE - it is machine generated \*/",
+        # VCR Cassette TODO should be for .yml files only
+        rb"\A.*(\r)?\n.*recorded_with: VCR",
+        # Compiled Cython TODO should be .c files only
+        rb"\A.*Generated by Cython",
+        # (Fortran) module
+        rb"\A.*(PCBNEW-LibModule-V|GFORTRAN module version ')",
+        # Racc
+        rb"# This file is automatically generated by Racc",
+        # JFlex
+        rb"/\* The following code was generated by JFlex ",
+        # Grammar Kit
+        rb"// This is a generated file. Not intended for manual editing\.",
+        # Roxygen 2
+        rb"% Generated by roxygen2: do not edit by hand",
+        # Jison
+        rb"/\* (parser generated by jison |generated by jison-lex )",
+        # gRPC C++
+        rb"// Generated by the gRPC",
+        # Dart
+        rb"generated code\W{2,3}do not modify",
+        # Perl's PPPort
+        rb"Automatically created by Devel::PPPort",
+        # GIMP
+        rb'/\* GIMP [a-zA-Z0-9\- ]+ C\-Source image dump \(.+?\.c\) \*/',
+        (rb'/\*  GIMP header image file format '
+         rb'\([a-zA-Z0-9\- ]+\)\: .+?\.h  \*/'),
+        # Visual Studio 6
+        rb'# Microsoft Developer Studio Generated Build File',
+        # Haxe
+        rb"Generated by Haxe",
+        # Jooq
+        rb"This file is generated by jOOQ\.",
+    )),
+    re.MULTILINE
+)
+
+# TODO isMinifiedFile, hasSourceMapReference, isCompiledCoffeeScript,
+# isGeneratedPostScript,
+# isGeneratedUnity3DMeta (extension restrictions should really apply)
+# isGeneratedGameMakerStudio, isGeneratedHTML
+
+
+def is_blob_generated(repo, blob: CheckBlobsGeneratedRequest.Blob):
+    """"Detect if a blob is made of generated data."""
+    # For now the Rails application sends revisions in `commit_id:path`
+    # format, so close to HGitaly blob oids (!). Probably upstream does not
+    # want to use actual Git blob oids, but Gitaly does support them
+    colon_split = blob.revision.split(b':', 1)
+    if len(colon_split) == 2:
+        csid, rpath = colon_split
+    else:  # direct oid
+        csid, rpath = extract_blob_oid(repo, blob.revision.decode('utf-8'))
+
+    if csid is None or rpath is None:
+        raise ValueError("Invalid blob oid")
+
+    extension = rpath.rsplit(b'.', 1)[-1]
+    if extension in GENERATED_FILE_EXTENSIONS:
+        return True
+
+    if any(rx.search(rpath) is not None
+           for rx in GENERATED_FILE_PATH_REGEXPS):
+        return True
+
+    try:
+        changeset = repo[csid]
+    except hgerror.RepoLookupError:
+        raise MercurialChangesetNotFound(csid)
+
+    try:
+        filectx = changeset.filectx(rpath)
+    except KeyError:
+        raise MercurialPathNotFound(rpath)
+
+    return GENERATED_LINE_RX.search(filectx.data()) is not None

{hgitaly-1.3.1 → hgitaly-2.10.0}/hgitaly/gitlab_ref.py

@@ -28,7 +28,11 @@ from hgext3rd.heptapod.special_ref import (
     parse_special_ref,
     special_refs,
 )
-from hgext3rd.heptapod.keep_around import iter_keep_arounds
+from hgext3rd.heptapod.keep_around import (
+    iter_keep_arounds,
+    KEEP_AROUND_REF_PREFIX,
+    KEEP_AROUND_REF_PREFIX_LEN,
+)
 
 
 def gitlab_special_ref_target(repo, ref_path):
@@ -130,7 +134,13 @@ def has_keep_around(repo, sha):
 
 def keep_around_ref_path(sha):
     # TODO should move to py-heptapod
-    return b'refs/keep-around/' + sha
+    return KEEP_AROUND_REF_PREFIX + sha
+
+
+def parse_keep_around_ref_path(ref):
+    if not ref.startswith(KEEP_AROUND_REF_PREFIX):
+        return None
+    return ref[KEEP_AROUND_REF_PREFIX_LEN:]
 
 
 def iter_keep_arounds_as_refs(repo, deref=True, patterns=None):

{hgitaly-1.3.1 → hgitaly-2.10.0}/hgitaly/logging.py

@@ -4,7 +4,9 @@
 # GNU General Public License version 2 or any later version.
 #
 # SPDX-License-Identifier: GPL-2.0-or-later
+import attr
 import logging
+from typing import Any
 
 
 _missing = object()
@@ -28,14 +30,16 @@ def extract_correlation(context):
     return corr_id
 
 
+@attr.define
 class LoggerAdapter(logging.LoggerAdapter):
-
-    def __init__(self, logger, context):
-        self.grpc_context = context
-        self.logger = logger
+    logger: Any
+    grpc_context: Any
 
     def process(self, msg, kwargs):
+        extra = kwargs.get('extra')
+        if extra is None:
+            kwargs['extra'] = extra = {}
+
         # set on context by interceptor
-        kwargs['extra'] = {'correlation_id':
-                           extract_correlation(self.grpc_context)}
+        extra['correlation_id'] = extract_correlation(self.grpc_context)
         return msg, kwargs

{hgitaly-1.3.1 → hgitaly-2.10.0}/hgitaly/message.py

@@ -87,6 +87,10 @@ def message_to_string(message):
     return result
 
 
+def as_dict(message):
+    return {descr.name: value for descr, value in message.ListFields()}
+
+
 class Logging:
     """Wrapper of requests and responses for sensible logging.
 

{hgitaly-1.3.1 → hgitaly-2.10.0}/hgitaly/repository.py

@@ -39,34 +39,6 @@ AUTO_PUBLISH_MAPPING = {
 AUTO_PUBLISH_REVERSE_MAPPING = {v: k for k, v in AUTO_PUBLISH_MAPPING.items()}
 
 
-def set_gitlab_project_full_path(repo, full_path: bytes):
-    """Store information about the full path of GitLab Project.
-
-    In GitLab terminology, ``full_path`` is the URI path, while ``path``
-    its the last segment of ``full_path``.
-
-    In Git repositories, this is stored in config. We could as well use
-    the repo-local hgrcs, but it is simpler to use a dedicated file, and
-    it makes sense to consider it not part of ``store`` (``svfs``), same
-    as ``hgrc``.
-    """
-    with repo.wlock():
-        repo.vfs.write(GITLAB_PROJECT_FULL_PATH_FILENAME, full_path)
-
-
-def get_gitlab_project_full_path(repo):
-    """Get the full path of GitLab Project.
-
-    See also :func:`set_gitlab_project_full_path`.
-
-    :return: the full path, or ``None`` if not set.
-    """
-    try:
-        return repo.vfs.read(GITLAB_PROJECT_FULL_PATH_FILENAME)
-    except FileNotFoundError:
-        return None
-
-
 def unbundle(repo, bundle_path: str, rails_sync=False):
     """Call unbundle with proper options and conversions.