bitfab 0.65.5 → 0.65.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a9c94aca223ea9794e110a16dae0e3e4f23ef45a2cf345c6bbaf396fc6b10b5f
-  data.tar.gz: 9c99a4266ec0109d155caeea5e1bb7792fb38c5e9d510e3552a2df60d192160a
+  metadata.gz: 84d81613e344f01a011f28ac437476d22e73867eb7ab2df04579075730b1c73b
+  data.tar.gz: d3bb4d9f79d587dca9ee2ef2e1ee36a41fbeb59da7cb69960bf9fdbe75e8d663
 SHA512:
-  metadata.gz: fa01c2d028e06cc338df64527d011df63e0a01dcec25559b4816523c951aed2b55ab0034075c7c09babec22d927bb4e5f0654456e82fd76e5d41815d0851fa30
-  data.tar.gz: 6f865c897e2ac1975c1b2a64cfbdf5c933f24356db7294e97a8ce11712cbf1937b942701c5237cd62dd79a68ae422dfaf46ed6eba3545e6c8522bdf0070831af
+  metadata.gz: 18ceec3ef05433309efff835c6bcf050789439e9d0424f9919972a70dd5d5b358eb07641fbed220d36b52e92f503d719dee044ba18003fb1371d77d0dba2e8e3
+  data.tar.gz: a6743709c41b9b69597e43e5c1336af5038bf9e5c56d5e3b247051e12fb4527d9732d79bb98e3587f90aedfbc59f7e23ebeae6c0d02e1bfad712f37d6789dd00

data/lib/bitfab/assertions.rb

@@ -13,7 +13,11 @@ module Bitfab
     end
 
     # Each returned assertion carries "category_id" and, with the same value,
-    # the deprecated "category_assertion_id".
+    # the deprecated "category_assertion_id". It also carries "rewrite", the
+    # latest check of its text (the same hash a save returns), or nil when no
+    # check ever ran on it. Read a "pending" check from a save back here once
+    # it finishes, and save each of its "separateAssertions" as its own
+    # assertion.
     def get_assertions(trace_id)
       response = @http_client.get(assertions_path(trace_id))
       return response unless response.is_a?(Hash) && response["assertions"].is_a?(Array)
@@ -24,17 +28,41 @@ module Bitfab
     # Assertion hashes take "category_id": omit it to keep the assignment, or
     # pass nil to clear it. The deprecated "category_assertion_id" is still
     # accepted and sent as "category_id"; "category_id" wins when both are set.
-    def save_assertions(trace_id, assertions, source: "agent")
-      save_assertions_all([{"traceId" => trace_id, "assertions" => assertions}], source:)
+    #
+    # By default, after the save, Bitfab checks each assertion's text against its
+    # trace. When the text bundles two claims, adds a condition, points at a
+    # value instead of stating it, names an id from the recorded run, or cannot
+    # be measured, Bitfab rewrites it and saves the rewritten text in place of
+    # the text you sent. This method returns as soon as the text is saved: each
+    # checked assertion comes back with "status" "pending", and Bitfab rewrites
+    # it in place in the background, usually within a minute. Read the result
+    # later with {#get_assertions}, whose assertions carry the same "rewrite"
+    # hash. Pass skip_guided_rewrite: true to save the text exactly as sent,
+    # with no check.
+    #
+    # Each returned assertion carries a "rewrite" hash: "status" ("pending",
+    # "rewritten", "unchanged", "not_checked", or "check_failed"),
+    # "sentAssertion", "changes", "separateAssertions", "unresolved", and
+    # "error". A check not finished 15 minutes after the save reads as
+    # "check_failed", with the text as sent. Save each of
+    # "separateAssertions" as its own assertion, since the saved one keeps
+    # only one claim.
+    def save_assertions(trace_id, assertions, source: "agent", skip_guided_rewrite: false)
+      save_assertions_all([{"traceId" => trace_id, "assertions" => assertions}], source:, skip_guided_rewrite:)
     end
 
-    # Same assertion hash keys as {#save_assertions}, for several traces at once.
-    def save_assertions_all(updates, source: "agent")
+    # Same assertion hash keys, rewrite, and returned "rewrite" hash as
+    # {#save_assertions}, for several traces at once.
+    def save_assertions_all(updates, source: "agent", skip_guided_rewrite: false)
       return [] if updates.empty?
 
       saved = @http_client.request(
         "/api/sdk/traces/assertions",
-        {"updates" => updates.map { |update| with_new_category_key(update) }, "source" => source}
+        {
+          "updates" => updates.map { |update| with_new_category_key(update) },
+          "source" => source,
+          "guidedRewrite" => !skip_guided_rewrite
+        }
       )["assertions"]
       saved.is_a?(Array) ? saved.map { |assertion| with_both_category_keys(assertion) } : saved
     end

data/lib/bitfab/cloudReplay.py

@@ -154,6 +154,7 @@ CODE_CHANGE_PATH_ENV = "BITFAB_CODE_CHANGE_PATH"
 CODE_CHANGE_MAX_FILES = 60
 CODE_CHANGE_MAX_FILE_BYTES = 500_000
 CODE_CHANGE_MAX_TOTAL_BYTES = 2_000_000
+BRANCH_WALK_LIMIT = 1000
 CHECK_ARGUMENTS_SECONDS = 300
 CHECK_ARGUMENTS_FLAG = "--check-arguments"
 REPLAY_ARGUMENTS_FLAG = "--replay-arguments"
@@ -933,6 +934,98 @@ def current_branch(root):
         return None
 
 
+def quiet_git(root, *args):
+    try:
+        return git(root, *args)
+    except CommandError:
+        return ""
+
+
+def default_branch(root):
+    origin_head = quiet_git(root, "rev-parse", "--abbrev-ref", "origin/HEAD")
+    candidates = [
+        origin_head if origin_head and origin_head != "origin/HEAD" else None,
+        "origin/main",
+        "origin/master",
+        "main",
+        "master",
+    ]
+    return next(
+        (
+            ref
+            for ref in dict.fromkeys(filter(None, candidates))
+            if quiet_git(root, "rev-parse", "--verify", "-q", f"{ref}^{{commit}}")
+        ),
+        None,
+    )
+
+
+def current_branch_refs(root):
+    branch = current_branch(root)
+    if not branch:
+        return set()
+    return {f"refs/heads/{branch}"} | {
+        f"refs/remotes/{remote}/{branch}"
+        for remote in quiet_git(root, "remote").split()
+    }
+
+
+def own_branch_refs(root):
+    upstream = quiet_git(root, "rev-parse", "--symbolic-full-name", "@{upstream}")
+    own = current_branch_refs(root)
+    if upstream:
+        own.add(upstream)
+    return own
+
+
+def on_default_branch(root, default):
+    full_name = quiet_git(root, "rev-parse", "--symbolic-full-name", default)
+    return bool(full_name) and full_name in current_branch_refs(root)
+
+
+def branch_tips(root):
+    own = own_branch_refs(root)
+    listing = quiet_git(
+        root,
+        "for-each-ref",
+        "--format=%(objectname) %(refname)",
+        "refs/heads",
+        "refs/remotes",
+    )
+    tips = {}
+    for line in listing.splitlines():
+        sha, _, ref = line.partition(" ")
+        if ref and ref not in own and not ref.endswith("/HEAD"):
+            tips.setdefault(
+                sha, ref.removeprefix("refs/heads/").removeprefix("refs/remotes/")
+            )
+    return tips
+
+
+def branch_point(root):
+    head = quiet_git(root, "rev-parse", "HEAD")
+    if not head:
+        return None
+    default = default_branch(root)
+    if default and on_default_branch(root, default):
+        return None
+    default_base = quiet_git(root, "merge-base", "HEAD", default) if default else ""
+    walk = ["rev-list", "--first-parent", f"--max-count={BRANCH_WALK_LIMIT}", "HEAD"]
+    if default_base:
+        walk.append(f"^{default_base}")
+    tips = branch_tips(root)
+    for sha in quiet_git(root, *walk).split()[1:]:
+        if sha in tips:
+            return {"sha": sha, "ref": tips[sha]}
+    if default_base and default_base != head:
+        return {"sha": default_base, "ref": default}
+    return None
+
+
+def replay_artifact(path):
+    return path == ".bitfab" or path.startswith(".bitfab/")
+
+
 def sensitive(path):
     parts = Path(path).parts
     name = Path(path).name.lower()
@@ -1033,12 +1126,20 @@ def snapshot(
         }
         if preview:
             return summary
+        code_change_base = request.get("codeChangeBase") or {}
+        extra_parent = (
+            ["-p", code_change_base["sha"]]
+            if not any(not replay_artifact(path) for _, path in changes)
+            and SHA.fullmatch(str(code_change_base.get("sha", "")))
+            else []
+        )
         sha = git(
             root,
             "commit-tree",
             tree,
             "-p",
             head,
+            *extra_parent,
             input=f"{snapshot_subject(request['id'], on_push)}\n\n{json.dumps(request)}\n",
         )
         return {**summary, "sha": sha}
@@ -1360,6 +1461,9 @@ def submit(root, repo, workflow, parsed, path):
     branch = current_branch(root)
     if branch:
         request["branch"] = branch
+    code_change_base = branch_point(root)
+    if code_change_base:
+        request["codeChangeBase"] = code_change_base
     check_arguments(parsed["replay"])
     allowed = allowed_files(root, parsed)
     if "--cloud-preview" in parsed["switches"]:
@@ -2067,7 +2171,9 @@ def changed_paths(root, base, commit):
         yield status, before_path, path
 
 
-def snapshot_code_change(root, base, commit, name):
+def snapshot_code_change(
+    root, base, commit, name, against="uncommitted (vs HEAD)", subject_ref=None
+):
     files, total = [], 0
     for status, before_path, path in changed_paths(root, base, commit):
         if len(files) >= CODE_CHANGE_MAX_FILES:
@@ -2083,14 +2189,34 @@ def snapshot_code_change(root, base, commit, name):
         files.append({"path": path, "before": before, "after": after})
     if not files:
         return None
-    head = (name or "").strip() or git(root, "log", "-1", "--format=%s", base)
+    head = (name or "").strip() or git(
+        root, "log", "-1", "--format=%s", subject_ref or base
+    )
     word = "file" if len(files) == 1 else "files"
     return {
-        "description": f"{head or 'Working-tree change'} ({len(files)} {word} changed uncommitted (vs HEAD))",
+        "description": f"{head or 'Working-tree change'} ({len(files)} {word} changed {against})",
         "files": files,
     }
 
 
+def branch_code_change(root, base, commit, request):
+    code_change_base = request.get("codeChangeBase")
+    if not isinstance(code_change_base, dict):
+        return None
+    sha, ref = code_change_base.get("sha"), code_change_base.get("ref")
+    if not (
+        isinstance(sha, str)
+        and SHA.fullmatch(sha)
+        and isinstance(ref, str)
+        and BRANCH_NAME.fullmatch(ref)
+        and quiet_git(root, "rev-parse", "--verify", "-q", f"{commit}^2") == sha
+    ):
+        return None
+    return snapshot_code_change(
+        root, sha, commit, replay_name(request["args"]), f"vs {ref}", base
+    )
+
+
 def write_code_change(change):
     handle, path = tempfile.mkstemp(
         prefix="bitfab-code-change-",
@@ -2127,6 +2253,7 @@ def execute():
         None
         if explicit_change or base is None
         else snapshot_code_change(root, base, commit, replay_name(request["args"]))
+        or branch_code_change(root, base, commit, request)
     )
     run_environment = dict(environment)
     if change:

data/lib/bitfab/replay.rb

@@ -598,6 +598,7 @@ module Bitfab
     CC_MAX_FILES = 60
     CC_MAX_FILE_BYTES = 500_000
     CC_MAX_TOTAL_BYTES = 2_000_000
+    CC_MAX_BRANCH_WALK = 1000
 
     def resolve_skip_assertion_judging(skip_assertion_judging, judge_assertions)
       return skip_assertion_judging if judge_assertions.nil?
@@ -658,7 +659,9 @@ module Bitfab
     # in replay() (the one call guaranteed to run on every replay) so it works no
     # matter how the replay was launched. Precedence: a BITFAB_CODE_CHANGE_PATH
     # override, else the git diff of the working tree vs HEAD (or vs the
-    # merge-base with BITFAB_CODE_CHANGE_BASE). Best-effort: any failure yields nil.
+    # merge-base with BITFAB_CODE_CHANGE_BASE). A clean working tree falls back to
+    # the point where HEAD split from the default branch, so committed branch work
+    # still shows. Best-effort: any failure yields nil.
     # An explicit code_change_files always wins over this.
     def resolve_auto_code_change(label = nil)
       return nil if ENV["BITFAB_DISABLE_CODE_CHANGE_CAPTURE"]
@@ -711,6 +714,73 @@ module Bitfab
       cc_ref_exists?(root, "HEAD") ? ["HEAD", "uncommitted (vs HEAD)"] : nil
     end
 
+    def cc_list_changes(root, base)
+      tracked = git(root, ["diff", "--name-status", "--find-renames", "-z", base, "--", ":!.bitfab"]) || ""
+      untracked = git(root, ["ls-files", "--others", "--exclude-standard", "-z", "--", ":!.bitfab"]) || ""
+      entries = cc_parse_name_status_z(tracked)
+      untracked.split("\0").reject(&:empty?).each { |p| entries << ["A", p, p] }
+      entries
+    end
+
+    def cc_default_branch_candidates(root)
+      origin_head = (git(root, ["rev-parse", "--abbrev-ref", "origin/HEAD"]) || "").strip
+      origin_head = nil if origin_head.empty? || origin_head == "origin/HEAD"
+      [origin_head, "origin/main", "origin/master", "main", "master"].compact.uniq
+    end
+
+    def cc_default_branch(root)
+      cc_default_branch_candidates(root).find { |ref| cc_ref_exists?(root, ref) }
+    end
+
+    def cc_current_branch_refs(root)
+      branch = (git(root, ["symbolic-ref", "--short", "-q", "HEAD"]) || "").strip
+      return [] if branch.empty?
+
+      ["refs/heads/#{branch}"] + (git(root, ["remote"]) || "").split.map { |remote| "refs/remotes/#{remote}/#{branch}" }
+    end
+
+    def cc_own_branch_refs(root)
+      upstream = (git(root, ["rev-parse", "--symbolic-full-name", "@{upstream}"]) || "").strip
+      own = cc_current_branch_refs(root)
+      own << upstream unless upstream.empty?
+      own
+    end
+
+    def cc_on_default_branch?(root, default)
+      full_name = (git(root, ["rev-parse", "--symbolic-full-name", default]) || "").strip
+      !full_name.empty? && cc_current_branch_refs(root).include?(full_name)
+    end
+
+    def cc_branch_tips(root)
+      own = cc_own_branch_refs(root)
+      listing = git(root, ["for-each-ref", "--format=%(objectname) %(refname)", "refs/heads", "refs/remotes"]) || ""
+      listing.each_line.with_object({}) do |line, tips|
+        sha, ref = line.strip.split(" ", 2)
+        next if ref.nil? || own.include?(ref) || ref.end_with?("/HEAD") || tips.key?(sha)
+
+        tips[sha] = ref.sub(%r{\Arefs/(heads|remotes)/}, "")
+      end
+    end
+
+    def cc_resolve_branch_point(root)
+      head = (git(root, ["rev-parse", "HEAD"]) || "").strip
+      return nil if head.empty?
+
+      default = cc_default_branch(root)
+      return nil if default && cc_on_default_branch?(root, default)
+
+      default_base = default ? (git(root, ["merge-base", "HEAD", default]) || "").strip : ""
+      walk = ["rev-list", "--first-parent", "--max-count=#{CC_MAX_BRANCH_WALK}", "HEAD"]
+      walk << "^#{default_base}" unless default_base.empty?
+      tips = cc_branch_tips(root)
+      (git(root, walk) || "").split.drop(1).each do |sha|
+        return [sha, "vs #{tips[sha]}"] if tips.key?(sha)
+      end
+      return [default_base, "vs #{default}"] unless default_base.empty? || default_base == head
+
+      nil
+    end
+
     def cc_blob_bytes(root, base, path)
       out = git(root, ["cat-file", "-s", "#{base}:#{path}"])
       (out && !out.strip.empty?) ? out.strip.to_i : 0
@@ -732,15 +802,18 @@ module Bitfab
       resolved = cc_resolve_base(root)
       return nil if resolved.nil?
 
-      base, against = resolved
-
-      tracked = git(root, ["diff", "--name-status", "--find-renames", "-z", base, "--", ":!.bitfab"]) || ""
-      untracked = git(root, ["ls-files", "--others", "--exclude-standard", "-z", "--", ":!.bitfab"]) || ""
-
-      entries = cc_parse_name_status_z(tracked)
-      untracked.split("\0").reject(&:empty?).each { |p| entries << ["A", p, p] }
+      entries = cc_list_changes(root, resolved[0])
+      if entries.empty? && resolved[0] == "HEAD"
+        branch_point = cc_resolve_branch_point(root)
+        if branch_point
+          resolved = branch_point
+          entries = cc_list_changes(root, branch_point[0])
+        end
+      end
       return nil if entries.empty?
 
+      base, against = resolved
+
       files = []
       total = 0
       entries.each do |status, before_path, path|

data/lib/bitfab/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Bitfab
-  VERSION = "0.65.5"
+  VERSION = "0.65.7"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: bitfab
 version: !ruby/object:Gem::Version
-  version: 0.65.5
+  version: 0.65.7
 platform: ruby
 authors:
 - Harvest Team