intentumdiff-python 0.0.2b2__py3-none-win_amd64.whl → 0.0.3b1__py3-none-win_amd64.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- intentumdiff/__init__.py +7 -0
- intentumdiff/_differ_gate.py +0 -136
- intentumdiff/_differ_presentation.py +1 -72
- intentumdiff/_differ_runtime.py +0 -17
- intentumdiff/analysis/cross_file.py +21 -21
- intentumdiff/analysis/diagnostics.py +1 -164
- intentumdiff/analysis/guardrail_reports.py +24 -28
- intentumdiff/analysis/guardrails.py +2 -67
- intentumdiff/analysis/schema_resolver.py +23 -87
- intentumdiff/analysis/user_schemas.py +2 -17
- intentumdiff/cli/_commands.py +70 -71
- intentumdiff/cli/_parser.py +40 -55
- intentumdiff/cli/_shared.py +55 -4
- intentumdiff/core/commit_differ.py +65 -99
- intentumdiff/core/indexer.py +0 -1
- intentumdiff/core/models.py +226 -69
- intentumdiff/differ.py +125 -432
- intentumdiff/github_app.py +5 -8
- intentumdiff/intentumdiff_rust_core/intentumdiff_rust_core.dll +0 -0
- intentumdiff/live_server.py +17 -17
- intentumdiff/lsp/client.py +42 -42
- intentumdiff/lsp/servers.py +10 -11
- intentumdiff/lsp_server/_handlers.py +1 -1
- intentumdiff/lsp_server/server.py +0 -1
- intentumdiff/plugins/exceptions.py +24 -0
- intentumdiff/plugins/hub.py +38 -41
- intentumdiff/plugins/registry.py +10 -10
- intentumdiff/plugins/registry.schema.json +156 -15
- intentumdiff/plugins/registry_schema.py +47 -140
- intentumdiff/rust_core.py +112 -775
- intentumdiff/serve/_app.py +23 -25
- intentumdiff/sources/git_source.py +0 -183
- intentumdiff/vcs/git_cli.py +0 -47
- intentumdiff/wasm/abap_parser.wasm +0 -0
- intentumdiff/wasm/adf_parser.wasm +0 -0
- intentumdiff/wasm/asciidoc_parser.wasm +0 -0
- intentumdiff/wasm/asm_parser.wasm +0 -0
- intentumdiff/wasm/assemblyscript_parser.wasm +0 -0
- intentumdiff/wasm/astro_parser.wasm +0 -0
- intentumdiff/wasm/bash_parser.wasm +0 -0
- intentumdiff/wasm/cpp_parser.wasm +0 -0
- intentumdiff/wasm/csharp_parser.wasm +0 -0
- intentumdiff/wasm/databricks_parser.wasm +0 -0
- intentumdiff/wasm/dax_parser.wasm +0 -0
- intentumdiff/wasm/dockerfile_parser.wasm +0 -0
- intentumdiff/wasm/freebasic_parser.wasm +0 -0
- intentumdiff/wasm/generic_parser.wasm +0 -0
- intentumdiff/wasm/gitignore_parser.wasm +0 -0
- intentumdiff/wasm/graphql_parser.wasm +0 -0
- intentumdiff/wasm/groovy_parser.wasm +0 -0
- intentumdiff/wasm/js_ts_parser.wasm +0 -0
- intentumdiff/wasm/json_parser.wasm +0 -0
- intentumdiff/wasm/kotlin_parser.wasm +0 -0
- intentumdiff/wasm/latex_parser.wasm +0 -0
- intentumdiff/wasm/make_parser.wasm +0 -0
- intentumdiff/wasm/mdx_parser.wasm +0 -0
- intentumdiff/wasm/ocaml_parser.wasm +0 -0
- intentumdiff/wasm/odin_parser.wasm +0 -0
- intentumdiff/wasm/perl_parser.wasm +0 -0
- intentumdiff/wasm/php_parser.wasm +0 -0
- intentumdiff/wasm/plsql_parser.wasm +0 -0
- intentumdiff/wasm/postscript_parser.wasm +0 -0
- intentumdiff/wasm/powershell_parser.wasm +0 -0
- intentumdiff/wasm/python_parser.wasm +0 -0
- intentumdiff/wasm/qsharp_parser.wasm +0 -0
- intentumdiff/wasm/reasonml_parser.wasm +0 -0
- intentumdiff/wasm/rust_parser.wasm +0 -0
- intentumdiff/wasm/sas_parser.wasm +0 -0
- intentumdiff/wasm/sql_parser.wasm +0 -0
- intentumdiff/wasm/svelte_parser.wasm +0 -0
- intentumdiff/wasm/swift_parser.wasm +0 -0
- intentumdiff/wasm/tsql_parser.wasm +0 -0
- intentumdiff/wasm/vbnet_parser.wasm +0 -0
- intentumdiff/wasm/vue_parser.wasm +0 -0
- intentumdiff/wasm/wasm_provenance.json +219 -219
- intentumdiff/wasm/wat_parser.wasm +0 -0
- intentumdiff/wasm/yaml_parser.wasm +0 -0
- intentumdiff/watcher.py +15 -17
- {intentumdiff_python-0.0.2b2.dist-info → intentumdiff_python-0.0.3b1.dist-info}/METADATA +38 -6
- intentumdiff_python-0.0.3b1.dist-info/RECORD +164 -0
- {intentumdiff_python-0.0.2b2.dist-info → intentumdiff_python-0.0.3b1.dist-info}/sboms/intentumdiff-rust-core.cyclonedx.json +2 -2
- intentumdiff/analysis/invariances.py +0 -208
- intentumdiff/analysis/keyed_profiles.py +0 -25
- intentumdiff/analysis/resource_profiles.py +0 -14
- intentumdiff/analysis/text_review.py +0 -47
- intentumdiff/invariances/__init__.py +0 -2
- intentumdiff/invariances/rules.schema.json +0 -283
- intentumdiff/invariances/rules.yaml +0 -349
- intentumdiff_python-0.0.2b2.dist-info/RECORD +0 -171
- {intentumdiff_python-0.0.2b2.dist-info → intentumdiff_python-0.0.3b1.dist-info}/WHEEL +0 -0
- {intentumdiff_python-0.0.2b2.dist-info → intentumdiff_python-0.0.3b1.dist-info}/entry_points.txt +0 -0
- {intentumdiff_python-0.0.2b2.dist-info → intentumdiff_python-0.0.3b1.dist-info}/licenses/LICENSE +0 -0
intentumdiff/__init__.py
CHANGED
|
@@ -33,6 +33,8 @@ from intentumdiff.core.models import (
|
|
|
33
33
|
GuardrailCheckResult,
|
|
34
34
|
GuardrailSeverity,
|
|
35
35
|
GuardrailViolation,
|
|
36
|
+
IncompleteReport,
|
|
37
|
+
IncompleteReview,
|
|
36
38
|
NodePosition,
|
|
37
39
|
RefactoringKind,
|
|
38
40
|
ReferenceKind,
|
|
@@ -54,6 +56,7 @@ from intentumdiff.sources.string_source import StringSource
|
|
|
54
56
|
from intentumdiff.cache.store import CacheStore
|
|
55
57
|
from intentumdiff.cache.sqlite_store import SqliteCacheStore
|
|
56
58
|
from intentumdiff.live_server import LiveServer
|
|
59
|
+
from intentumdiff.rust_core import CoreSizeLimitError
|
|
57
60
|
|
|
58
61
|
__all__ = [
|
|
59
62
|
# Differ
|
|
@@ -81,6 +84,8 @@ __all__ = [
|
|
|
81
84
|
"DetectionResult",
|
|
82
85
|
"FUEL_UNLIMITED",
|
|
83
86
|
"SemanticDiff",
|
|
87
|
+
"IncompleteReview",
|
|
88
|
+
"IncompleteReport",
|
|
84
89
|
"Change",
|
|
85
90
|
"ChangeGroup",
|
|
86
91
|
"ChangeGroupKind",
|
|
@@ -106,6 +111,8 @@ __all__ = [
|
|
|
106
111
|
"Indexer",
|
|
107
112
|
# Live server
|
|
108
113
|
"LiveServer",
|
|
114
|
+
# Errors
|
|
115
|
+
"CoreSizeLimitError",
|
|
109
116
|
# Testing
|
|
110
117
|
"PluginTestHarness",
|
|
111
118
|
]
|
intentumdiff/_differ_gate.py
CHANGED
|
@@ -51,142 +51,6 @@ def _rust_certified_languages() -> frozenset[str]:
|
|
|
51
51
|
}
|
|
52
52
|
|
|
53
53
|
|
|
54
|
-
# Languages whose per-stage draft FINALIZATION routes through the Rust core (issue #57 —
|
|
55
|
-
# the retirement path for python refinement/presentation). Grows per parity-verified
|
|
56
|
-
# language; INTENTUMDIFF_RUST_FINALIZE_LANGS (comma-separated) adds pilots.
|
|
57
|
-
RUST_FINALIZE_LANGUAGES: frozenset[str] = frozenset(
|
|
58
|
-
{
|
|
59
|
-
"go", "kotlin", "ruby", "swift", "java", "c", "cpp", "gitignore",
|
|
60
|
-
"csharp", "elixir", "perl", "scala", "xml", "haskell",
|
|
61
|
-
# Structured-language batch (issue #57): parity-verified clean under routing —
|
|
62
|
-
# no engine change, per-language contracts + cross-cutting infra all green.
|
|
63
|
-
"adf", "asciidoc", "assemblyscript", "clojure", "databricks",
|
|
64
|
-
"databricks-workflow", "dax", "groovy", "latex", "lua", "ocaml", "odin",
|
|
65
|
-
"php", "plsql", "po", "postscript", "qsharp", "reasonml",
|
|
66
|
-
"sas", "squirrel", "tsql", "vbnet", "wast", "wat",
|
|
67
|
-
"puppet", "hcl",
|
|
68
|
-
# Body-reference variable renames corroborated by an anchored callable's param rename
|
|
69
|
-
# (promote_corroborated_variable_renames) — the routed analogue of refactoring.py's
|
|
70
|
-
# inferred_rename_pairs, which the default path used for dart's single-letter renames.
|
|
71
|
-
"dart",
|
|
72
|
-
# Structured/markup batch (#57): parity-verified clean under routing (full-suite gated).
|
|
73
|
-
# generic was piloted too but breaks the generic-text/markdown/text fallback path
|
|
74
|
-
# (#35/#36/#44) — deferred to its own fix.
|
|
75
|
-
"svelte", "astro", "vue", "mdx", "scss",
|
|
76
|
-
# dockerfile: needs the resource-profile shell-command identity keying (dockerfile_key),
|
|
77
|
-
# now ported to Rust so an inserted RUN doesn't cross-pair positionally.
|
|
78
|
-
"dockerfile",
|
|
79
|
-
# asm (experimental, #57): statement-profile language — instructions key by mnemonic +
|
|
80
|
-
# first-operand identity (asm_key) so an operand-value edit is a MODIFICATION, not churn.
|
|
81
|
-
"asm",
|
|
82
|
-
# css (#57): unblocked by applying semantic invariances in the routed short-circuit —
|
|
83
|
-
# the color-canonicalization equivalence (red≡#ff0000, blue≡#0000ff) now fires under
|
|
84
|
-
# routing. (yaml/json graceful-xfail shape + python dedent still gap-blocked.)
|
|
85
|
-
"css",
|
|
86
|
-
# bash (#57): bash_key ported (commands key by NAME) + the cross-key MODIFICATION split —
|
|
87
|
-
# `:` -> `echo Hello` unpaires into DELETE+ADD with descendant scaffold folded.
|
|
88
|
-
"bash",
|
|
89
|
-
# delphi (#57): delphi_key ported (statements keyed by call/assign identity) + statement
|
|
90
|
-
# review containers anchor in suppress_parent_modifications — the WriteLn edit presents
|
|
91
|
-
# as ONE statement MODIFICATION with descendant mods folded.
|
|
92
|
-
"delphi",
|
|
93
|
-
# m / Power Query (#57): served by the dax parser (dax routed-green); the step-edit
|
|
94
|
-
# contract reformulated to anti-swallowing (routed pairs the predicate edit as ONE
|
|
95
|
-
# MODIFICATION carrying both sides — honest, not the default's DELETE+ADD split).
|
|
96
|
-
"m",
|
|
97
|
-
# powershell (#57/#66): unblocked by the entity-anchored matching port — reordered
|
|
98
|
-
# functions pair by identity and surface as MOVEs (31/31 force-routed).
|
|
99
|
-
"powershell",
|
|
100
|
-
# json/yaml (#57): keyed-data profile port — pairs/items key by ancestor key path; array
|
|
101
|
-
# scalars/items by CONTENT identity (not position labels), so an inserted element never
|
|
102
|
-
# cross-pairs later unchanged values; identical-content positional churn suppresses with
|
|
103
|
-
# evidence; yaml block↔flow representation equivalence (YAML 1.2 §3) is style, not change.
|
|
104
|
-
"json", "yaml",
|
|
105
|
-
# html (#57/#64): path-profile keying port — elements key by their element PATH with
|
|
106
|
-
# identity attributes (id/name/key/…) beating same-tag ordinals, and relocated
|
|
107
|
-
# identical-content elements keep their match (stable-same-label escape) — a text edit
|
|
108
|
-
# inside an id-bearing or restructured element survives sibling inserts.
|
|
109
|
-
"html",
|
|
110
|
-
# rust (#57/#60): unlocked by the entity-anchored matching port (augment_entity_matching)
|
|
111
|
-
# — match→if-let now surfaces as a statement-rewrite MODIFICATION instead of losing the
|
|
112
|
-
# relocated arm bodies to DELETE+ADD churn. Full rust suite 112/112 force-routed.
|
|
113
|
-
"rust",
|
|
114
|
-
# abap (#57): entity-anchored grouping ported (entity_child_content_groups) — a child
|
|
115
|
-
# modification under a same-identity form groups as "GREET changed", not a free leaf edit.
|
|
116
|
-
"abap",
|
|
117
|
-
# r (#57): unblocked by the any→all truthiness fix in the signature-coverage suppression —
|
|
118
|
-
# a NEW function sharing param labels with a rename is no longer erased.
|
|
119
|
-
"r",
|
|
120
|
-
# zig (#57): parity-verified clean under routing post-anchors (14/14 force-routed) —
|
|
121
|
-
# pure routing, no engine change.
|
|
122
|
-
"zig",
|
|
123
|
-
# sql (#57): query-profile port (sql_key + matching/changes augmentation) — clauses,
|
|
124
|
-
# relations, and fields pair by role + normalized identity within their statement, so an
|
|
125
|
-
# added JOIN doesn't shift FROM into a bogus MOVE and unchanged SELECT fields never churn.
|
|
126
|
-
# The dax half of the profile is deliberately NOT ported (dax routed green without it).
|
|
127
|
-
"sql",
|
|
128
|
-
# freebasic/terraform (#57): parity-verified clean under routing (dedicated tests +
|
|
129
|
-
# the cross-cutting suites in the full gate) — pure routing, no engine change.
|
|
130
|
-
"freebasic", "terraform",
|
|
131
|
-
# graphql (#57/#67): unblocked by porting surface_changed_in_place_entity — a matched
|
|
132
|
-
# named entity (type/query/fragment) whose body changed carries its label into a group.
|
|
133
|
-
"graphql",
|
|
134
|
-
# javascript/typescript/tsx (#57): three ports close the js-family gap — (1) the
|
|
135
|
-
# function-valued-declaration profile (const f = () => … pairs with function f(…) as
|
|
136
|
-
# one entity) in the anchors code, (2) the style rule relabelling suppression residue
|
|
137
|
-
# as an IGNORED_STYLE group (provenance "suppression"), and (3) moved-pair literal
|
|
138
|
-
# update recovery (the #37 leaf score, move-pair scoped): edit-script pruning of
|
|
139
|
-
# unmatched leaves under moved subtrees was swallowing md5→sha256 inside a hoisted
|
|
140
|
-
# calc_hash. Same-label DELETE+ADD container churn around reported renames suppresses
|
|
141
|
-
# under a rename-coverage truthiness guard (tighter than the python pass it ports).
|
|
142
|
-
"javascript", "typescript", "tsx",
|
|
143
|
-
# generic (#57): the text review and both markdown section passes were already
|
|
144
|
-
# Rust-first; the routed short-circuit now mirrors the stage-12 orchestration
|
|
145
|
-
# (token churn replaced wholesale by Rust line spans, markdown section passes for
|
|
146
|
-
# .md filenames, stage-12 style-only resolution for empty results).
|
|
147
|
-
"generic",
|
|
148
|
-
# python (#57): the flagship, last. The certified BATCH path (sources->final in
|
|
149
|
-
# Rust, semantic_contract rust_finalized_v1) still serves default configs FIRST;
|
|
150
|
-
# the tier behind it is the full-Rust 9-fin routing (the stage-11 hybrid and the
|
|
151
|
-
# transitional python stages were deleted in the stage-4b payoff).
|
|
152
|
-
"python",
|
|
153
|
-
# markdown (#44): the tree-sitter markdown parser takes .md off the generic
|
|
154
|
-
# fallback — sections/headings/fences are REAL nodes (the corpus "section"
|
|
155
|
-
# entity vocabulary and the reorder→MOVE promotion consume them natively).
|
|
156
|
-
"markdown",
|
|
157
|
-
# toml (#48): keyed config format — tables/pairs are labeled by header/key
|
|
158
|
-
# identity in the parser itself, so the routed finalize pairs value edits
|
|
159
|
-
# under stable keys natively.
|
|
160
|
-
"toml",
|
|
161
|
-
# ini (#48): keyed config — sections by header, settings by key, same identity
|
|
162
|
-
# model as toml; certified on the same template.
|
|
163
|
-
"ini",
|
|
164
|
-
# gomod (#48): keyed dependency manifest — specs by module path, versions as
|
|
165
|
-
# semantic children; a version bump is ONE modification under stable identity.
|
|
166
|
-
"gomod",
|
|
167
|
-
# make (#48): rules by target identity, assignments by variable name, recipe
|
|
168
|
-
# lines as semantic children — a recipe edit pairs under its target.
|
|
169
|
-
"make",
|
|
170
|
-
# proto (#48): messages/services/enums/rpcs by name, fields by identifier with
|
|
171
|
-
# type + field number as semantic children — the wire-compat review story.
|
|
172
|
-
"proto",
|
|
173
|
-
# cmake (#48): commands labeled identifier(first-arg) so target identity is
|
|
174
|
-
# part of the command; function/macro defs by their defined name.
|
|
175
|
-
"cmake",
|
|
176
|
-
}
|
|
177
|
-
)
|
|
178
|
-
|
|
179
|
-
|
|
180
|
-
def _rust_finalize_languages() -> frozenset[str]:
|
|
181
|
-
# The set is TOTAL (all 67 supported languages, issue #57 2026-07-15), so it now
|
|
182
|
-
# serves as the certified-language allowlist. The INTENTUMDIFF_RUST_FINALIZE_LANGS
|
|
183
|
-
# pilot escape hatch is retired with the campaign: it existed to force-route
|
|
184
|
-
# candidates during per-language certification, and every candidate is routed.
|
|
185
|
-
# New languages certify by joining RUST_FINALIZE_LANGUAGES in code, behind the
|
|
186
|
-
# same protocol (force-routed probe -> Rust pin -> full gate -> flip).
|
|
187
|
-
return RUST_FINALIZE_LANGUAGES
|
|
188
|
-
|
|
189
|
-
|
|
190
54
|
def _strict_rust_only_enabled() -> bool:
|
|
191
55
|
return os.getenv(STRICT_RUST_GATE_ENV, "").strip() == "1"
|
|
192
56
|
|
|
@@ -1,21 +1,11 @@
|
|
|
1
|
-
"""Tree
|
|
1
|
+
"""Tree and fallback-diff helpers for intentumdiff.differ."""
|
|
2
2
|
|
|
3
3
|
from __future__ import annotations
|
|
4
4
|
|
|
5
|
-
import hashlib
|
|
6
|
-
import json
|
|
7
5
|
import logging
|
|
8
6
|
from typing import Any
|
|
9
7
|
|
|
10
|
-
from intentumdiff.analysis.text_review import (
|
|
11
|
-
PresentationResult,
|
|
12
|
-
)
|
|
13
8
|
from intentumdiff.core.models import (
|
|
14
|
-
Change,
|
|
15
|
-
ChangeGroup,
|
|
16
|
-
ChangeGroupKind,
|
|
17
|
-
ChangeType,
|
|
18
|
-
NodePosition,
|
|
19
9
|
SemanticDiff,
|
|
20
10
|
SemanticNode,
|
|
21
11
|
)
|
|
@@ -44,37 +34,6 @@ def _validate_tree_ids(root: SemanticNode, context: str) -> None:
|
|
|
44
34
|
)
|
|
45
35
|
|
|
46
36
|
|
|
47
|
-
def _root_structural_hash(node: SemanticNode) -> str:
|
|
48
|
-
"""Return the structural hash of a SemanticNode root (already computed)."""
|
|
49
|
-
return node.structural_hash
|
|
50
|
-
|
|
51
|
-
|
|
52
|
-
def _node_to_dict(node: SemanticNode) -> dict[str, Any]:
|
|
53
|
-
return json.loads(node.model_dump_json())
|
|
54
|
-
|
|
55
|
-
|
|
56
|
-
def _compute_structural_hash_for_tree(cst_json: str) -> str:
|
|
57
|
-
"""
|
|
58
|
-
Compute the structural hash of the FILTERED (trivia-stripped) CST.
|
|
59
|
-
Used for the style-only shortcut before running the full diff algorithm.
|
|
60
|
-
"""
|
|
61
|
-
from intentumdiff.plugins.loader import _structural_hash_impl
|
|
62
|
-
|
|
63
|
-
return _structural_hash_impl(cst_json)
|
|
64
|
-
|
|
65
|
-
|
|
66
|
-
def _count_cst_nodes(cst_json: str) -> int:
|
|
67
|
-
"""Count total nodes in a CST JSON string (recursive depth-first)."""
|
|
68
|
-
|
|
69
|
-
def _count(node: Any) -> int:
|
|
70
|
-
return 1 + sum(_count(c) for c in node.get("children", ()))
|
|
71
|
-
|
|
72
|
-
try:
|
|
73
|
-
return _count(json.loads(cst_json))
|
|
74
|
-
except Exception:
|
|
75
|
-
return 0
|
|
76
|
-
|
|
77
|
-
|
|
78
37
|
def _count_semantic_nodes(root: SemanticNode) -> int:
|
|
79
38
|
return 1 + len(root.descendants())
|
|
80
39
|
|
|
@@ -84,31 +43,6 @@ def _empty_semantic_tree(language: str) -> SemanticNode:
|
|
|
84
43
|
return SemanticNode.model_validate(_c_abi_call("empty_semantic_tree", language))
|
|
85
44
|
|
|
86
45
|
|
|
87
|
-
def _markdown_presentation(presented: PresentationResult, *, old_source: str, new_source: str,
|
|
88
|
-
old_filename: str, new_filename: str, phase: str) -> PresentationResult:
|
|
89
|
-
"""DTO adapter for complete Rust reconciliation; core errors propagate."""
|
|
90
|
-
from intentumdiff.rust_core import _c_abi_call, _require_source_anchored_changes
|
|
91
|
-
result = _c_abi_call("reconcile_markdown", {
|
|
92
|
-
"changes": [change.model_dump(mode="json") for change in presented.changes],
|
|
93
|
-
"change_groups": [group.model_dump(mode="json") for group in presented.change_groups],
|
|
94
|
-
"ignored_style_changes": presented.ignored_style_changes,
|
|
95
|
-
}, old_source, new_source, old_filename, new_filename, phase)
|
|
96
|
-
return PresentationResult(
|
|
97
|
-
changes=[Change.model_validate(value) for value in
|
|
98
|
-
_require_source_anchored_changes(result["changes"], "Rust Markdown reconciliation")],
|
|
99
|
-
change_groups=[ChangeGroup.model_validate(value) for value in result["change_groups"]],
|
|
100
|
-
ignored_style_changes=result["ignored_style_changes"],
|
|
101
|
-
)
|
|
102
|
-
|
|
103
|
-
|
|
104
|
-
def _markdown_section_move_presentation(presented: PresentationResult, **kwargs: Any) -> PresentationResult:
|
|
105
|
-
return _markdown_presentation(presented, phase="moves", **kwargs)
|
|
106
|
-
|
|
107
|
-
|
|
108
|
-
def _markdown_section_heading_rename_presentation(presented: PresentationResult, **kwargs: Any) -> PresentationResult:
|
|
109
|
-
return _markdown_presentation(presented, phase="renames", **kwargs)
|
|
110
|
-
|
|
111
|
-
|
|
112
46
|
def _has_error_node(tree: SemanticNode) -> bool:
|
|
113
47
|
"""Ask Rust whether a semantic tree contains a parse error."""
|
|
114
48
|
from intentumdiff.rust_core import parse_errors_present
|
|
@@ -127,11 +61,6 @@ def _token_fallback_diff(
|
|
|
127
61
|
return result.model_copy(update={"metadata": {**details, **dict(result.metadata)}})
|
|
128
62
|
|
|
129
63
|
|
|
130
|
-
# ---------------------------------------------------------------------------
|
|
131
|
-
# Streaming helpers
|
|
132
|
-
# ---------------------------------------------------------------------------
|
|
133
|
-
|
|
134
|
-
|
|
135
64
|
def _enrich_literal_labels(root: SemanticNode, source: str) -> SemanticNode:
|
|
136
65
|
"""Delegate literal-label enrichment to Rust; never reconstruct semantics."""
|
|
137
66
|
from intentumdiff.rust_core import enrich_literal_labels
|
intentumdiff/_differ_runtime.py
CHANGED
|
@@ -72,8 +72,6 @@ class _PhaseProfiler:
|
|
|
72
72
|
}
|
|
73
73
|
|
|
74
74
|
|
|
75
|
-
_DEFAULT_PLUGIN_FUEL = 100_000_000
|
|
76
|
-
_EXPLICIT_FUEL_EXHAUSTION_TEST_CAP = 1_000_000
|
|
77
75
|
# Hotspot thresholds are calibrated to the measured worst legitimate parser plus ~30%
|
|
78
76
|
# headroom. Two effects moved the floor in 2026-07: literal-container text capture in the
|
|
79
77
|
# tree-sitter converters (issue #46 — a deliberate correctness cost), and whole-binary LTO
|
|
@@ -103,21 +101,6 @@ def _attach_content_type_metadata(
|
|
|
103
101
|
return diff.model_copy(update={"metadata": metadata})
|
|
104
102
|
|
|
105
103
|
|
|
106
|
-
def _fuel_budget(base: int, extra: int) -> int:
|
|
107
|
-
"""Return an adaptive budget without hiding explicit low-fuel tests.
|
|
108
|
-
|
|
109
|
-
Real editor/CLI fuel settings are a floor, not a guarantee that ordinary
|
|
110
|
-
source files will fail once they cross that exact number. Tiny budgets are
|
|
111
|
-
still respected so fuel-pressure tests can prove runaway plugins fail
|
|
112
|
-
explicitly instead of being silently promoted to a larger cap.
|
|
113
|
-
"""
|
|
114
|
-
if base == -1:
|
|
115
|
-
return base
|
|
116
|
-
if base < _EXPLICIT_FUEL_EXHAUSTION_TEST_CAP:
|
|
117
|
-
return base
|
|
118
|
-
return max(base, extra)
|
|
119
|
-
|
|
120
|
-
|
|
121
104
|
def _drain_plugin_telemetry(plugin_adapter: Any) -> list[dict[str, Any]]:
|
|
122
105
|
drain = getattr(plugin_adapter, "drain_telemetry", None)
|
|
123
106
|
if not callable(drain):
|
|
@@ -46,24 +46,24 @@ def detect_cross_file_changes(
|
|
|
46
46
|
|
|
47
47
|
result = try_rust_diff_symbol_tables(old_json, new_json)
|
|
48
48
|
|
|
49
|
-
|
|
50
|
-
|
|
51
|
-
|
|
52
|
-
|
|
53
|
-
|
|
54
|
-
|
|
55
|
-
|
|
56
|
-
|
|
57
|
-
|
|
58
|
-
|
|
59
|
-
|
|
60
|
-
|
|
61
|
-
|
|
62
|
-
|
|
63
|
-
|
|
64
|
-
|
|
65
|
-
|
|
66
|
-
|
|
67
|
-
|
|
68
|
-
)
|
|
69
|
-
|
|
49
|
+
return [cross_file_change_from_core(item) for item in result]
|
|
50
|
+
|
|
51
|
+
|
|
52
|
+
def cross_file_change_from_core(item: dict) -> CrossFileChange:
|
|
53
|
+
"""DTO mapping for one core cross-file change."""
|
|
54
|
+
return CrossFileChange(
|
|
55
|
+
change_type=ChangeType(item["change_type"]),
|
|
56
|
+
symbol_name=item["symbol_name"],
|
|
57
|
+
old_file=item["old_file"],
|
|
58
|
+
new_file=item["new_file"],
|
|
59
|
+
old_node_id=item.get("old_node_id"),
|
|
60
|
+
new_node_id=item.get("new_node_id"),
|
|
61
|
+
old_position=item.get("old_position"),
|
|
62
|
+
new_position=item.get("new_position"),
|
|
63
|
+
old_language=item.get("old_language", ""),
|
|
64
|
+
new_language=item.get("new_language", ""),
|
|
65
|
+
node_type=item.get("node_type", ""),
|
|
66
|
+
symbol_kind=item.get("symbol_kind", ""),
|
|
67
|
+
confidence=item.get("confidence", 1.0),
|
|
68
|
+
description=item.get("description", ""),
|
|
69
|
+
)
|
|
@@ -6,14 +6,10 @@ from collections.abc import Mapping, Sequence
|
|
|
6
6
|
from dataclasses import dataclass, field
|
|
7
7
|
from typing import Any
|
|
8
8
|
|
|
9
|
-
from intentumdiff.core.models import
|
|
9
|
+
from intentumdiff.core.models import ChangeGroup, SemanticNode
|
|
10
10
|
|
|
11
11
|
|
|
12
12
|
DIAGNOSTICS_VERSION = 2
|
|
13
|
-
_PAIR_SAMPLE_LIMIT = 8
|
|
14
|
-
_CANDIDATE_SAMPLE_LIMIT = 8
|
|
15
|
-
|
|
16
|
-
|
|
17
13
|
@dataclass
|
|
18
14
|
class DiagnosticsRecorder:
|
|
19
15
|
"""Collect JSON-safe diagnostics events when enabled."""
|
|
@@ -101,40 +97,6 @@ class DiagnosticsRecorder:
|
|
|
101
97
|
metadata=payload,
|
|
102
98
|
)
|
|
103
99
|
|
|
104
|
-
def record_candidate_summary(
|
|
105
|
-
self,
|
|
106
|
-
*,
|
|
107
|
-
stage: str,
|
|
108
|
-
rule_id: str,
|
|
109
|
-
action: str = "candidate_summary",
|
|
110
|
-
reason: str = "",
|
|
111
|
-
evaluated_count: int = 0,
|
|
112
|
-
accepted_count: int = 0,
|
|
113
|
-
rejected_count: int = 0,
|
|
114
|
-
rejection_reasons: Mapping[str, int] | None = None,
|
|
115
|
-
accepted_samples: Sequence[Mapping[str, Any]] = (),
|
|
116
|
-
rejected_samples: Sequence[Mapping[str, Any]] = (),
|
|
117
|
-
metadata: Mapping[str, Any] | None = None,
|
|
118
|
-
) -> None:
|
|
119
|
-
"""Record bounded accepted/rejected candidate evidence."""
|
|
120
|
-
|
|
121
|
-
payload = {
|
|
122
|
-
"evaluated_count": evaluated_count,
|
|
123
|
-
"accepted_count": accepted_count,
|
|
124
|
-
"rejected_count": rejected_count,
|
|
125
|
-
"rejection_reasons": dict(rejection_reasons or {}),
|
|
126
|
-
"accepted_samples": list(accepted_samples[:_CANDIDATE_SAMPLE_LIMIT]),
|
|
127
|
-
"rejected_samples": list(rejected_samples[:_CANDIDATE_SAMPLE_LIMIT]),
|
|
128
|
-
**dict(metadata or {}),
|
|
129
|
-
}
|
|
130
|
-
self.record(
|
|
131
|
-
stage=stage,
|
|
132
|
-
action=action,
|
|
133
|
-
rule_id=rule_id,
|
|
134
|
-
reason=reason,
|
|
135
|
-
metadata=payload,
|
|
136
|
-
)
|
|
137
|
-
|
|
138
100
|
def record_group(self, *, stage: str, group: ChangeGroup) -> None:
|
|
139
101
|
kind = group.kind.value if hasattr(group.kind, "value") else str(group.kind)
|
|
140
102
|
self.record(
|
|
@@ -158,39 +120,6 @@ class DiagnosticsRecorder:
|
|
|
158
120
|
},
|
|
159
121
|
)
|
|
160
122
|
|
|
161
|
-
def record_change(
|
|
162
|
-
self,
|
|
163
|
-
*,
|
|
164
|
-
stage: str,
|
|
165
|
-
change: Change,
|
|
166
|
-
action: str = "change",
|
|
167
|
-
rule_id: str = "",
|
|
168
|
-
reason: str = "",
|
|
169
|
-
) -> None:
|
|
170
|
-
change_type = (
|
|
171
|
-
change.change_type.value
|
|
172
|
-
if isinstance(change.change_type, ChangeType)
|
|
173
|
-
else str(change.change_type)
|
|
174
|
-
)
|
|
175
|
-
self.record(
|
|
176
|
-
stage=stage,
|
|
177
|
-
action=action,
|
|
178
|
-
rule_id=rule_id,
|
|
179
|
-
reason=reason,
|
|
180
|
-
old_nodes=[change.old_node],
|
|
181
|
-
new_nodes=[change.new_node],
|
|
182
|
-
confidence=change.confidence,
|
|
183
|
-
metadata={
|
|
184
|
-
"change_type": change_type,
|
|
185
|
-
"description": change.description,
|
|
186
|
-
"refactoring_kind": (
|
|
187
|
-
change.refactoring_kind.value
|
|
188
|
-
if change.refactoring_kind is not None
|
|
189
|
-
else None
|
|
190
|
-
),
|
|
191
|
-
},
|
|
192
|
-
)
|
|
193
|
-
|
|
194
123
|
def snapshot(self) -> dict[str, Any]:
|
|
195
124
|
return {
|
|
196
125
|
"version": DIAGNOSTICS_VERSION,
|
|
@@ -200,98 +129,6 @@ class DiagnosticsRecorder:
|
|
|
200
129
|
}
|
|
201
130
|
|
|
202
131
|
|
|
203
|
-
def record_matching_delta(
|
|
204
|
-
recorder: DiagnosticsRecorder,
|
|
205
|
-
*,
|
|
206
|
-
stage: str,
|
|
207
|
-
rule_id: str,
|
|
208
|
-
before: Sequence[Any],
|
|
209
|
-
after: Sequence[Any],
|
|
210
|
-
) -> None:
|
|
211
|
-
"""Record before/after matching counts and sampled pair changes."""
|
|
212
|
-
|
|
213
|
-
if not recorder.enabled:
|
|
214
|
-
return
|
|
215
|
-
|
|
216
|
-
before_pairs = {_pair_key(match): match for match in before}
|
|
217
|
-
after_pairs = {_pair_key(match): match for match in after}
|
|
218
|
-
added = [after_pairs[key] for key in after_pairs.keys() - before_pairs.keys()]
|
|
219
|
-
removed = [before_pairs[key] for key in before_pairs.keys() - after_pairs.keys()]
|
|
220
|
-
if not added and not removed and len(before_pairs) == len(after_pairs):
|
|
221
|
-
recorder.record(
|
|
222
|
-
stage=stage,
|
|
223
|
-
action="augment_matching",
|
|
224
|
-
rule_id=rule_id,
|
|
225
|
-
reason="no matching changes",
|
|
226
|
-
metadata={"before_pairs": len(before_pairs), "after_pairs": len(after_pairs)},
|
|
227
|
-
)
|
|
228
|
-
return
|
|
229
|
-
|
|
230
|
-
recorder.record(
|
|
231
|
-
stage=stage,
|
|
232
|
-
action="augment_matching",
|
|
233
|
-
rule_id=rule_id,
|
|
234
|
-
reason="matching augmentation applied",
|
|
235
|
-
metadata={
|
|
236
|
-
"before_pairs": len(before_pairs),
|
|
237
|
-
"after_pairs": len(after_pairs),
|
|
238
|
-
"added_pair_count": len(added),
|
|
239
|
-
"removed_pair_count": len(removed),
|
|
240
|
-
"added_pairs": [_pair_digest(match) for match in added[:_PAIR_SAMPLE_LIMIT]],
|
|
241
|
-
"removed_pairs": [
|
|
242
|
-
_pair_digest(match) for match in removed[:_PAIR_SAMPLE_LIMIT]
|
|
243
|
-
],
|
|
244
|
-
},
|
|
245
|
-
)
|
|
246
|
-
|
|
247
|
-
|
|
248
|
-
def _pair_key(match: Any) -> tuple[str, str]:
|
|
249
|
-
return (match.old_node.id, match.new_node.id)
|
|
250
|
-
|
|
251
|
-
|
|
252
|
-
def _pair_digest(match: Any) -> dict[str, Any]:
|
|
253
|
-
return {
|
|
254
|
-
"old_id": match.old_node.id,
|
|
255
|
-
"new_id": match.new_node.id,
|
|
256
|
-
"old_type": match.old_node.node_type,
|
|
257
|
-
"new_type": match.new_node.node_type,
|
|
258
|
-
"old_label": match.old_node.label,
|
|
259
|
-
"new_label": match.new_node.label,
|
|
260
|
-
}
|
|
261
|
-
|
|
262
|
-
|
|
263
|
-
def node_digest(node: SemanticNode | None) -> dict[str, Any]:
|
|
264
|
-
"""Return a small diagnostics-safe node reference."""
|
|
265
|
-
|
|
266
|
-
if node is None:
|
|
267
|
-
return {}
|
|
268
|
-
return {
|
|
269
|
-
"id": node.id,
|
|
270
|
-
"type": node.node_type,
|
|
271
|
-
"label": node.label,
|
|
272
|
-
}
|
|
273
|
-
|
|
274
|
-
|
|
275
|
-
def pair_digest(
|
|
276
|
-
old_node: SemanticNode | None,
|
|
277
|
-
new_node: SemanticNode | None,
|
|
278
|
-
*,
|
|
279
|
-
score: float | None = None,
|
|
280
|
-
reason: str = "",
|
|
281
|
-
) -> dict[str, Any]:
|
|
282
|
-
"""Return a small diagnostics-safe old/new candidate reference."""
|
|
283
|
-
|
|
284
|
-
payload: dict[str, Any] = {
|
|
285
|
-
"old": node_digest(old_node),
|
|
286
|
-
"new": node_digest(new_node),
|
|
287
|
-
}
|
|
288
|
-
if score is not None:
|
|
289
|
-
payload["score"] = round(score, 4)
|
|
290
|
-
if reason:
|
|
291
|
-
payload["reason"] = reason
|
|
292
|
-
return payload
|
|
293
|
-
|
|
294
|
-
|
|
295
132
|
def _json_safe(value: Any, *, depth: int = 0) -> Any:
|
|
296
133
|
if depth > 6:
|
|
297
134
|
return str(value)
|
|
@@ -22,25 +22,10 @@ def build_guardrail_check_result(
|
|
|
22
22
|
strict: bool,
|
|
23
23
|
metadata: Mapping[str, Any] | None = None,
|
|
24
24
|
) -> GuardrailCheckResult:
|
|
25
|
-
"""
|
|
25
|
+
"""Core's guardrail check verdict for a list of violations (``guardrail_check_result``)."""
|
|
26
26
|
|
|
27
|
-
items =
|
|
28
|
-
|
|
29
|
-
violation.severity == GuardrailSeverity.IMMUTABLE for violation in items
|
|
30
|
-
)
|
|
31
|
-
important_count = sum(
|
|
32
|
-
violation.severity == GuardrailSeverity.IMPORTANT for violation in items
|
|
33
|
-
)
|
|
34
|
-
return GuardrailCheckResult(
|
|
35
|
-
violations=items,
|
|
36
|
-
violation_count=len(items),
|
|
37
|
-
immutable_count=immutable_count,
|
|
38
|
-
important_count=important_count,
|
|
39
|
-
checked_files=checked_files,
|
|
40
|
-
strict=strict,
|
|
41
|
-
passed=not (strict and immutable_count),
|
|
42
|
-
metadata=metadata or {},
|
|
43
|
-
)
|
|
27
|
+
items = [violation.model_dump(mode="json") for violation in violations]
|
|
28
|
+
return _core_check_result({"violations": items}, strict, checked_files, metadata)
|
|
44
29
|
|
|
45
30
|
|
|
46
31
|
def guardrail_result_from_diffs(
|
|
@@ -50,20 +35,31 @@ def guardrail_result_from_diffs(
|
|
|
50
35
|
checked_files: int | None = None,
|
|
51
36
|
metadata: Mapping[str, Any] | None = None,
|
|
52
37
|
) -> GuardrailCheckResult:
|
|
53
|
-
"""
|
|
38
|
+
"""Core's verdict over the diffs' violations; core counts the diffs as checked files."""
|
|
54
39
|
|
|
55
40
|
items = [diffs] if isinstance(diffs, SemanticDiff) else list(diffs)
|
|
56
|
-
|
|
57
|
-
|
|
41
|
+
# Core reads only each diff's ``guardrail_violations``, so only that field is marshalled.
|
|
42
|
+
evidence = [
|
|
43
|
+
{"guardrail_violations": [v.model_dump(mode="json") for v in diff.guardrail_violations]}
|
|
58
44
|
for diff in items
|
|
59
|
-
for violation in diff.guardrail_violations
|
|
60
45
|
]
|
|
61
|
-
return
|
|
62
|
-
|
|
63
|
-
|
|
64
|
-
|
|
65
|
-
|
|
66
|
-
|
|
46
|
+
return _core_check_result({"diffs": evidence}, strict, checked_files, metadata)
|
|
47
|
+
|
|
48
|
+
|
|
49
|
+
def _core_check_result(
|
|
50
|
+
evidence: dict[str, Any],
|
|
51
|
+
strict: bool,
|
|
52
|
+
checked_files: int | None,
|
|
53
|
+
metadata: Mapping[str, Any] | None,
|
|
54
|
+
) -> GuardrailCheckResult:
|
|
55
|
+
# Core owns the counts and the gate (``passed = !strict || no violations``, #296); the DTO
|
|
56
|
+
# only validates the result and rejects counts that disagree with the violations.
|
|
57
|
+
from intentumdiff.rust_core import _c_abi_call # noqa: PLC0415
|
|
58
|
+
|
|
59
|
+
request = {**evidence, "strict": strict, "checked_files": checked_files,
|
|
60
|
+
"metadata": None if metadata is None else dict(metadata)}
|
|
61
|
+
result = _c_abi_call("guardrail_check_result", json.dumps(request))
|
|
62
|
+
return GuardrailCheckResult.model_validate(result)
|
|
67
63
|
|
|
68
64
|
|
|
69
65
|
def render_guardrail_json(result: GuardrailCheckResult, *, indent: int = 2) -> str:
|