@gmickel/gno 1.27.1 → 1.29.0
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.
- package/README.md +46 -20
- package/assets/skill/SKILL.md +59 -1
- package/browser-extension/artifacts/gno-browser-clipper-v1.29.0.zip +0 -0
- package/browser-extension/artifacts/gno-browser-clipper-v1.29.0.zip.sha256 +1 -0
- package/browser-extension/dist/chunk-b2zm0jjd.js +50 -0
- package/browser-extension/dist/manifest.json +1 -1
- package/browser-extension/dist/preview.html +1 -1
- package/browser-extension/dist/service-worker.js +6 -6
- package/package.json +2 -2
- package/spec/cli.md +82 -5
- package/spec/db/schema.sql +31 -0
- package/spec/mcp.md +71 -0
- package/spec/output-schemas/ask.schema.json +156 -1
- package/spec/output-schemas/browser-clip-preview.schema.json +30 -0
- package/spec/output-schemas/collection-egress-check.schema.json +91 -0
- package/spec/output-schemas/collection-egress-policy-set.schema.json +56 -0
- package/spec/output-schemas/collection-egress-policy.schema.json +41 -0
- package/spec/output-schemas/context-capsule-v1.schema.json +183 -3
- package/spec/output-schemas/context-capsule-verification.schema.json +1 -1
- package/spec/output-schemas/egress-audit-management.schema.json +88 -0
- package/spec/output-schemas/get.schema.json +94 -0
- package/spec/output-schemas/mcp-http-error.schema.json +113 -2
- package/spec/output-schemas/mcp-job-status.schema.json +28 -0
- package/spec/output-schemas/multi-get.schema.json +128 -0
- package/spec/output-schemas/publish-artifact.schema.json +32 -0
- package/spec/output-schemas/record-import.schema.json +223 -0
- package/spec/output-schemas/retrieval-trace-common.schema.json +31 -0
- package/spec/output-schemas/retrieval-trace-export.schema.json +3 -0
- package/spec/output-schemas/retrieval-trace-qrels.schema.json +3 -0
- package/spec/output-schemas/retrieval-trace-summary.schema.json +4 -0
- package/spec/output-schemas/search-result.schema.json +94 -0
- package/spec/output-schemas/search-results.schema.json +155 -0
- package/spec/project-profile.schema.json +66 -0
- package/src/app/context-format.ts +1 -1
- package/src/app/context-runtime-contract.ts +16 -7
- package/src/cli/commands/ask.ts +7 -0
- package/src/cli/commands/collection/index.ts +5 -0
- package/src/cli/commands/collection/policy.ts +126 -0
- package/src/cli/commands/context-build.ts +5 -0
- package/src/cli/commands/daemon.ts +45 -3
- package/src/cli/commands/egress-audit.ts +68 -0
- package/src/cli/commands/embed.ts +2 -0
- package/src/cli/commands/get.ts +12 -2
- package/src/cli/commands/index-cmd.ts +13 -0
- package/src/cli/commands/multi-get.ts +9 -2
- package/src/cli/commands/query.ts +12 -0
- package/src/cli/commands/replay.ts +13 -3
- package/src/cli/commands/shared.ts +28 -0
- package/src/cli/commands/update.ts +5 -0
- package/src/cli/commands/vsearch.ts +3 -1
- package/src/cli/program.ts +127 -0
- package/src/config/index.ts +9 -0
- package/src/config/project-profile.ts +2 -0
- package/src/config/types.ts +66 -0
- package/src/converters/adapters/browser-export/adapter.ts +199 -0
- package/src/converters/adapters/browser-export/formats.ts +358 -0
- package/src/converters/adapters/email/adapter.ts +429 -0
- package/src/converters/adapters/email/html.ts +162 -0
- package/src/converters/adapters/email/mime.ts +454 -0
- package/src/converters/adapters/email/parameters.ts +77 -0
- package/src/converters/adapters/ical/adapter.ts +475 -0
- package/src/converters/adapters/ical/recurrence.ts +186 -0
- package/src/converters/adapters/jsonl/adapter.ts +238 -0
- package/src/converters/adapters/jsonl/config.ts +105 -0
- package/src/converters/adapters/shared/html-text.ts +165 -0
- package/src/converters/adapters/shared/record-utils.ts +79 -0
- package/src/converters/adapters/shared/utf8-lines.ts +141 -0
- package/src/converters/adapters/transcript/adapter.ts +295 -0
- package/src/converters/adapters/transcript/json.ts +184 -0
- package/src/converters/adapters/transcript/model.ts +171 -0
- package/src/converters/adapters/transcript/text.ts +58 -0
- package/src/converters/adapters/transcript/timed.ts +152 -0
- package/src/converters/index.ts +11 -1
- package/src/converters/mime.ts +8 -0
- package/src/converters/pipeline.ts +15 -1
- package/src/converters/registry.ts +37 -1
- package/src/converters/types.ts +136 -0
- package/src/core/browser-clip.ts +16 -1
- package/src/core/collection-egress-policy-projection.ts +28 -0
- package/src/core/collection-egress-policy-service.ts +443 -0
- package/src/core/collection-egress-policy-validation.ts +242 -0
- package/src/core/config-mutation.ts +31 -19
- package/src/core/context-capsule-schema.ts +198 -1
- package/src/core/context-capsule-validation.ts +4 -4
- package/src/core/context-capsule-verification.ts +5 -1
- package/src/core/context-capsule.ts +177 -112
- package/src/core/context-evidence.ts +93 -15
- package/src/core/destination-classifier.ts +402 -0
- package/src/core/document-capabilities.ts +12 -0
- package/src/core/egress-audit.ts +239 -0
- package/src/core/egress-authorization.ts +50 -0
- package/src/core/egress-enforcement.ts +264 -0
- package/src/core/egress-policy.ts +440 -0
- package/src/core/egress-provenance.ts +336 -0
- package/src/core/job-manager.ts +37 -0
- package/src/core/network-boundary-inventory.ts +261 -0
- package/src/core/project-profile-apply-state.ts +5 -0
- package/src/core/project-profile.ts +4 -0
- package/src/core/record-metadata.ts +49 -0
- package/src/core/retrieval-qrels.ts +6 -0
- package/src/core/retrieval-replay.ts +18 -4
- package/src/core/retrieval-trace-export.ts +23 -4
- package/src/core/retrieval-trace-management-helpers.ts +1 -0
- package/src/core/retrieval-trace-management-types.ts +3 -0
- package/src/core/retrieval-trace-management.ts +9 -1
- package/src/core/retrieval-trace-session.ts +94 -0
- package/src/core/retrieval-trace.ts +22 -0
- package/src/ingestion/record-adapter-canonical.ts +433 -0
- package/src/ingestion/record-adapter.ts +448 -0
- package/src/ingestion/record-container.ts +704 -0
- package/src/ingestion/record-path.ts +20 -0
- package/src/ingestion/record-sync.ts +70 -0
- package/src/ingestion/sync.ts +243 -36
- package/src/ingestion/types.ts +71 -1
- package/src/ingestion/walker.ts +31 -11
- package/src/llm/errors.ts +10 -0
- package/src/llm/http-inference.ts +175 -0
- package/src/llm/http-policy.ts +537 -0
- package/src/llm/httpEmbedding.ts +47 -28
- package/src/llm/httpGeneration.ts +31 -18
- package/src/llm/httpRerank.ts +30 -18
- package/src/llm/index.ts +1 -0
- package/src/llm/nodeLlamaCpp/adapter.ts +55 -20
- package/src/llm/pinned-http-connection.ts +177 -0
- package/src/mcp/context.ts +53 -1
- package/src/mcp/http-egress.ts +220 -0
- package/src/mcp/http-security.ts +32 -21
- package/src/mcp/http-session.ts +13 -2
- package/src/mcp/http-transport.ts +96 -6
- package/src/mcp/sync-egress.ts +24 -0
- package/src/mcp/tools/add-collection.ts +4 -0
- package/src/mcp/tools/ask.ts +1 -1
- package/src/mcp/tools/context.ts +9 -2
- package/src/mcp/tools/egress.ts +247 -0
- package/src/mcp/tools/embed.ts +1 -0
- package/src/mcp/tools/get.ts +10 -2
- package/src/mcp/tools/index-cmd.ts +7 -0
- package/src/mcp/tools/index.ts +98 -0
- package/src/mcp/tools/multi-get.ts +9 -2
- package/src/mcp/tools/query.ts +10 -0
- package/src/mcp/tools/remove-collection.ts +4 -0
- package/src/mcp/tools/sync.ts +6 -0
- package/src/mcp/tools/trace.ts +7 -1
- package/src/mcp/tools/vsearch.ts +1 -0
- package/src/mcp/tools/workspace-write.ts +2 -0
- package/src/pipeline/egress-lineage.ts +124 -0
- package/src/pipeline/filters.ts +1 -1
- package/src/pipeline/graph-retrieval.ts +7 -4
- package/src/pipeline/hybrid.ts +21 -4
- package/src/pipeline/result-context.ts +12 -4
- package/src/pipeline/search.ts +23 -4
- package/src/pipeline/types.ts +6 -0
- package/src/pipeline/vsearch.ts +67 -28
- package/src/publish/artifact.ts +31 -3
- package/src/publish/export-service.ts +25 -0
- package/src/sdk/client.ts +119 -2
- package/src/sdk/documents.ts +13 -5
- package/src/sdk/embed.ts +1 -0
- package/src/sdk/types.ts +33 -0
- package/src/serve/browse-tree.ts +4 -2
- package/src/serve/clipper-capture.ts +5 -1
- package/src/serve/closed-json.ts +61 -0
- package/src/serve/config-sync.ts +4 -1
- package/src/serve/context.ts +1 -0
- package/src/serve/doc-events.ts +110 -39
- package/src/serve/public/components/CollectionEgressCheckPanel.tsx +411 -0
- package/src/serve/public/components/CollectionEgressPolicyEditor.tsx +78 -0
- package/src/serve/public/components/CollectionModelDialog.tsx +93 -9
- package/src/serve/public/components/EgressAuditPanel.tsx +278 -0
- package/src/serve/public/globals.built.css +1 -1
- package/src/serve/public/pages/Collections.tsx +3 -0
- package/src/serve/resident-request.ts +80 -3
- package/src/serve/resident-runtime.ts +63 -1
- package/src/serve/routes/api.ts +237 -65
- package/src/serve/routes/clipper.ts +64 -1
- package/src/serve/routes/mcp.ts +6 -0
- package/src/serve/routes/traces.ts +27 -4
- package/src/serve/server.ts +118 -7
- package/src/store/migrations/022-record-export-lineage.ts +42 -0
- package/src/store/migrations/023-collection-egress-policy.ts +61 -0
- package/src/store/migrations/024-egress-derived-lineage.ts +121 -0
- package/src/store/migrations/025-collection-egress-policy-revision.ts +40 -0
- package/src/store/migrations/index.ts +8 -0
- package/src/store/retrieval-trace-codec.ts +15 -0
- package/src/store/sqlite/adapter.ts +284 -14
- package/src/store/sqlite/change-journal-store.ts +41 -2
- package/src/store/sqlite/egress-audit-store.ts +485 -0
- package/src/store/sqlite/retrieval-trace-management-store.ts +47 -6
- package/src/store/sqlite/retrieval-trace-retention.ts +2 -0
- package/src/store/sqlite/retrieval-trace-rows.ts +8 -1
- package/src/store/sqlite/retrieval-trace-store.ts +104 -5
- package/src/store/types.ts +166 -1
- package/browser-extension/artifacts/gno-browser-clipper-v1.27.1.zip +0 -0
- package/browser-extension/artifacts/gno-browser-clipper-v1.27.1.zip.sha256 +0 -1
- package/browser-extension/dist/chunk-vn5f663b.js +0 -50
|
@@ -0,0 +1,199 @@
|
|
|
1
|
+
import type {
|
|
2
|
+
RecordAdapter,
|
|
3
|
+
RecordAdapterEvent,
|
|
4
|
+
RecordAdapterInput,
|
|
5
|
+
} from "../../types";
|
|
6
|
+
|
|
7
|
+
import {
|
|
8
|
+
type BrowserExportRecord,
|
|
9
|
+
parseBrowserJson,
|
|
10
|
+
parseNetscapeBookmarks,
|
|
11
|
+
} from "./formats";
|
|
12
|
+
|
|
13
|
+
const LIVE_PROFILE_MARKERS = [
|
|
14
|
+
"/library/safari/",
|
|
15
|
+
"/chrome/user data/",
|
|
16
|
+
"/google/chrome/",
|
|
17
|
+
"/chromium/user data/",
|
|
18
|
+
"/.config/google-chrome/",
|
|
19
|
+
"/.config/chromium/",
|
|
20
|
+
"/appdata/local/google/chrome/user data/",
|
|
21
|
+
"/firefox/profiles/",
|
|
22
|
+
"/.mozilla/firefox/",
|
|
23
|
+
"/mozilla/firefox/",
|
|
24
|
+
"/appdata/roaming/mozilla/firefox/profiles/",
|
|
25
|
+
];
|
|
26
|
+
|
|
27
|
+
const safeMarkdown = (value: string): string =>
|
|
28
|
+
value
|
|
29
|
+
.normalize("NFC")
|
|
30
|
+
.replaceAll("&", "&")
|
|
31
|
+
.replaceAll("<", "<")
|
|
32
|
+
.replaceAll(">", ">")
|
|
33
|
+
.replace(/([\\`*_[\]#])/g, "\\$1")
|
|
34
|
+
.trim();
|
|
35
|
+
|
|
36
|
+
const liveProfileReason = (input: RecordAdapterInput): string | undefined => {
|
|
37
|
+
const normalized = input.sourcePath.replaceAll("\\", "/").toLowerCase();
|
|
38
|
+
if (LIVE_PROFILE_MARKERS.some((marker) => normalized.includes(marker))) {
|
|
39
|
+
return "live-profile";
|
|
40
|
+
}
|
|
41
|
+
const basename = normalized.split("/").at(-1);
|
|
42
|
+
if (
|
|
43
|
+
basename === "cookies" ||
|
|
44
|
+
basename === "history" ||
|
|
45
|
+
basename?.endsWith(".sqlite") ||
|
|
46
|
+
basename?.endsWith(".db")
|
|
47
|
+
) {
|
|
48
|
+
return "live-database";
|
|
49
|
+
}
|
|
50
|
+
return undefined;
|
|
51
|
+
};
|
|
52
|
+
|
|
53
|
+
const normalizeUrl = (value: string): string | undefined => {
|
|
54
|
+
try {
|
|
55
|
+
const url = new URL(value.trim());
|
|
56
|
+
if (url.protocol !== "http:" && url.protocol !== "https:") {
|
|
57
|
+
return undefined;
|
|
58
|
+
}
|
|
59
|
+
if (url.username || url.password) return undefined;
|
|
60
|
+
url.hash = "";
|
|
61
|
+
return url.toString();
|
|
62
|
+
} catch {
|
|
63
|
+
return undefined;
|
|
64
|
+
}
|
|
65
|
+
};
|
|
66
|
+
|
|
67
|
+
const readSource = async (input: RecordAdapterInput): Promise<string> => {
|
|
68
|
+
const decoder = new TextDecoder("utf-8", { fatal: true });
|
|
69
|
+
let source = "";
|
|
70
|
+
for await (const chunk of input.open()) {
|
|
71
|
+
source += decoder.decode(chunk, { stream: true });
|
|
72
|
+
}
|
|
73
|
+
return source + decoder.decode();
|
|
74
|
+
};
|
|
75
|
+
|
|
76
|
+
const recordEvent = (
|
|
77
|
+
record: BrowserExportRecord
|
|
78
|
+
): RecordAdapterEvent | undefined => {
|
|
79
|
+
const url = normalizeUrl(record.url);
|
|
80
|
+
if (!url) return undefined;
|
|
81
|
+
const title = safeMarkdown(record.title || url);
|
|
82
|
+
const folder = record.folder ? safeMarkdown(record.folder) : undefined;
|
|
83
|
+
const tags = record.tags?.map(safeMarkdown).filter(Boolean);
|
|
84
|
+
const lines = [`# ${title}`, "", url];
|
|
85
|
+
if (folder) lines.push("", `Folder: ${folder}`);
|
|
86
|
+
if (tags && tags.length > 0) lines.push(`Tags: ${tags.join(", ")}`);
|
|
87
|
+
for (const [name, value] of Object.entries(record.dates ?? {}).sort(
|
|
88
|
+
([left], [right]) => left.localeCompare(right)
|
|
89
|
+
)) {
|
|
90
|
+
lines.push(`${safeMarkdown(name)}: ${safeMarkdown(value)}`);
|
|
91
|
+
}
|
|
92
|
+
const identity = record.externalId
|
|
93
|
+
? `${record.kind}:id:${record.externalId}`
|
|
94
|
+
: `${record.kind}:url:${url}`;
|
|
95
|
+
return {
|
|
96
|
+
type: "record",
|
|
97
|
+
record: {
|
|
98
|
+
stableId: `browser:${identity}`,
|
|
99
|
+
sourceLocator: record.sourceLocator,
|
|
100
|
+
markdown: lines.join("\n"),
|
|
101
|
+
title,
|
|
102
|
+
metadata: {
|
|
103
|
+
categories: ["browser-export", record.kind, ...(tags ?? [])],
|
|
104
|
+
dateFields: record.dates,
|
|
105
|
+
},
|
|
106
|
+
anchors: [{ kind: "record", value: record.sourceLocator }],
|
|
107
|
+
},
|
|
108
|
+
};
|
|
109
|
+
};
|
|
110
|
+
|
|
111
|
+
async function* parseBrowserExport(
|
|
112
|
+
input: RecordAdapterInput
|
|
113
|
+
): AsyncGenerator<RecordAdapterEvent> {
|
|
114
|
+
const denied = liveProfileReason(input);
|
|
115
|
+
if (denied) {
|
|
116
|
+
yield {
|
|
117
|
+
type: "failure",
|
|
118
|
+
failure: {
|
|
119
|
+
code: "ADAPTER_FAILURE",
|
|
120
|
+
message: "Live browser profile access is not allowed.",
|
|
121
|
+
retryable: false,
|
|
122
|
+
sourceLocator: denied,
|
|
123
|
+
},
|
|
124
|
+
};
|
|
125
|
+
yield { type: "snapshot", state: "partial" };
|
|
126
|
+
return;
|
|
127
|
+
}
|
|
128
|
+
|
|
129
|
+
let source: string;
|
|
130
|
+
try {
|
|
131
|
+
source = await readSource(input);
|
|
132
|
+
} catch {
|
|
133
|
+
yield {
|
|
134
|
+
type: "failure",
|
|
135
|
+
failure: {
|
|
136
|
+
code: "MALFORMED_RECORD",
|
|
137
|
+
message: "Browser export could not be decoded.",
|
|
138
|
+
retryable: true,
|
|
139
|
+
sourceLocator: "export",
|
|
140
|
+
},
|
|
141
|
+
};
|
|
142
|
+
yield { type: "snapshot", state: "partial" };
|
|
143
|
+
return;
|
|
144
|
+
}
|
|
145
|
+
const trimmed = source.trimStart();
|
|
146
|
+
const parsed = trimmed.startsWith("<")
|
|
147
|
+
? parseNetscapeBookmarks(
|
|
148
|
+
source,
|
|
149
|
+
input.limits.maxRecords,
|
|
150
|
+
input.limits.maxFailures
|
|
151
|
+
)
|
|
152
|
+
: parseBrowserJson(
|
|
153
|
+
source,
|
|
154
|
+
input.limits.maxRecords,
|
|
155
|
+
input.limits.maxFailures
|
|
156
|
+
);
|
|
157
|
+
let failures = parsed.failures.length;
|
|
158
|
+
for (const candidate of parsed.records) {
|
|
159
|
+
const event = recordEvent(candidate);
|
|
160
|
+
if (event) yield event;
|
|
161
|
+
else {
|
|
162
|
+
failures += 1;
|
|
163
|
+
yield {
|
|
164
|
+
type: "failure",
|
|
165
|
+
failure: {
|
|
166
|
+
code: "MALFORMED_RECORD",
|
|
167
|
+
message: "Browser export contained an invalid or unsafe URL.",
|
|
168
|
+
retryable: false,
|
|
169
|
+
sourceLocator: candidate.sourceLocator,
|
|
170
|
+
},
|
|
171
|
+
};
|
|
172
|
+
}
|
|
173
|
+
}
|
|
174
|
+
for (const locator of parsed.failures) {
|
|
175
|
+
yield {
|
|
176
|
+
type: "failure",
|
|
177
|
+
failure: {
|
|
178
|
+
code: "MALFORMED_RECORD",
|
|
179
|
+
message: "Browser export record could not be parsed.",
|
|
180
|
+
retryable: false,
|
|
181
|
+
sourceLocator: locator,
|
|
182
|
+
},
|
|
183
|
+
};
|
|
184
|
+
}
|
|
185
|
+
yield {
|
|
186
|
+
type: "snapshot",
|
|
187
|
+
state: failures === 0 ? "complete" : "partial",
|
|
188
|
+
};
|
|
189
|
+
}
|
|
190
|
+
|
|
191
|
+
export const browserExportAdapter: RecordAdapter = {
|
|
192
|
+
id: "adapter/browser-export",
|
|
193
|
+
version: "1.0.0",
|
|
194
|
+
canHandle: (mime, ext) =>
|
|
195
|
+
mime === "application/x-gno-browser-export+json" ||
|
|
196
|
+
mime === "text/x-gno-browser-bookmarks+html" ||
|
|
197
|
+
ext === ".browser-export",
|
|
198
|
+
records: parseBrowserExport,
|
|
199
|
+
};
|
|
@@ -0,0 +1,358 @@
|
|
|
1
|
+
import {
|
|
2
|
+
decodeHtmlEntitiesOnce,
|
|
3
|
+
htmlFragmentToText,
|
|
4
|
+
} from "../shared/html-text";
|
|
5
|
+
|
|
6
|
+
export type BrowserExportKind = "bookmark" | "history" | "reading-list";
|
|
7
|
+
|
|
8
|
+
export interface BrowserExportRecord {
|
|
9
|
+
kind: BrowserExportKind;
|
|
10
|
+
url: string;
|
|
11
|
+
title?: string;
|
|
12
|
+
folder?: string;
|
|
13
|
+
tags?: string[];
|
|
14
|
+
dates?: Record<string, string>;
|
|
15
|
+
externalId?: string;
|
|
16
|
+
sourceLocator: string;
|
|
17
|
+
}
|
|
18
|
+
|
|
19
|
+
export interface BrowserExportParseResult {
|
|
20
|
+
records: BrowserExportRecord[];
|
|
21
|
+
failures: string[];
|
|
22
|
+
}
|
|
23
|
+
|
|
24
|
+
const HTML_TOKEN_PATTERN =
|
|
25
|
+
/<H3\b[^>]*>([\s\S]*?)<\/H3\s*>|<A\b([^>]*)>([\s\S]*?)<\/A\s*>|<DL\b[^>]*>|<\/DL\s*>/gi;
|
|
26
|
+
const ATTRIBUTE_PATTERN = /([\w-]+)\s*=\s*(?:"([^"]*)"|'([^']*)'|([^\s>]+))/g;
|
|
27
|
+
const MAX_JSON_DEPTH = 32;
|
|
28
|
+
const WEBKIT_TO_UNIX_EPOCH_MS = 11_644_473_600_000;
|
|
29
|
+
const SENSITIVE_JSON_KEYS = new Set([
|
|
30
|
+
"cookies",
|
|
31
|
+
"passwords",
|
|
32
|
+
"logins",
|
|
33
|
+
"sessions",
|
|
34
|
+
"creditcards",
|
|
35
|
+
]);
|
|
36
|
+
const SUPPORTED_JSON_KEYS = new Set([
|
|
37
|
+
"roots",
|
|
38
|
+
"children",
|
|
39
|
+
"items",
|
|
40
|
+
"bookmarks",
|
|
41
|
+
"history",
|
|
42
|
+
"readingList",
|
|
43
|
+
"reading_list",
|
|
44
|
+
]);
|
|
45
|
+
const BOOKMARK_EXPORT_DOCUMENT_PATTERN =
|
|
46
|
+
/<!DOCTYPE\s+(?:NETSCAPE-Bookmark-file-1\b[^>]*|html\s*)>/i;
|
|
47
|
+
|
|
48
|
+
const plainText = (value: string): string =>
|
|
49
|
+
htmlFragmentToText(value).replace(/\s+/g, " ").trim();
|
|
50
|
+
|
|
51
|
+
const attributes = (value: string): Map<string, string> => {
|
|
52
|
+
const result = new Map<string, string>();
|
|
53
|
+
for (const match of value.matchAll(ATTRIBUTE_PATTERN)) {
|
|
54
|
+
const key = match[1]?.toLowerCase();
|
|
55
|
+
const item = match[2] ?? match[3] ?? match[4];
|
|
56
|
+
if (key && item !== undefined)
|
|
57
|
+
result.set(key, decodeHtmlEntitiesOnce(item));
|
|
58
|
+
}
|
|
59
|
+
return result;
|
|
60
|
+
};
|
|
61
|
+
|
|
62
|
+
const hasSensitiveJsonKey = (value: Record<string, unknown>): boolean =>
|
|
63
|
+
Object.keys(value).some((key) => SENSITIVE_JSON_KEYS.has(key.toLowerCase()));
|
|
64
|
+
|
|
65
|
+
const tagCount = (source: string, pattern: RegExp): number =>
|
|
66
|
+
source.match(pattern)?.length ?? 0;
|
|
67
|
+
|
|
68
|
+
const hasCompleteNetscapeStructure = (source: string): boolean => {
|
|
69
|
+
if (!BOOKMARK_EXPORT_DOCUMENT_PATTERN.test(source)) return false;
|
|
70
|
+
const openingDl = tagCount(source, /<DL\b[^>]*>/gi);
|
|
71
|
+
const closingDl = tagCount(source, /<\/DL\s*>/gi);
|
|
72
|
+
const openingAnchors = tagCount(source, /<A\b[^>]*>/gi);
|
|
73
|
+
const closingAnchors = tagCount(source, /<\/A\s*>/gi);
|
|
74
|
+
const openingHeadings = tagCount(source, /<H3\b[^>]*>/gi);
|
|
75
|
+
const closingHeadings = tagCount(source, /<\/H3\s*>/gi);
|
|
76
|
+
return (
|
|
77
|
+
openingDl > 0 &&
|
|
78
|
+
openingDl === closingDl &&
|
|
79
|
+
openingAnchors === closingAnchors &&
|
|
80
|
+
openingHeadings === closingHeadings
|
|
81
|
+
);
|
|
82
|
+
};
|
|
83
|
+
|
|
84
|
+
const exportDate = (value: string | undefined): string | undefined => {
|
|
85
|
+
if (!value) return undefined;
|
|
86
|
+
const epoch = Number(value);
|
|
87
|
+
if (Number.isFinite(epoch) && epoch >= 0) {
|
|
88
|
+
let milliseconds: number;
|
|
89
|
+
if (epoch >= 10_000_000_000_000_000) {
|
|
90
|
+
milliseconds = epoch / 1_000 - WEBKIT_TO_UNIX_EPOCH_MS;
|
|
91
|
+
} else {
|
|
92
|
+
milliseconds = epoch > 10_000_000_000 ? epoch : epoch * 1_000;
|
|
93
|
+
}
|
|
94
|
+
const date = new Date(milliseconds);
|
|
95
|
+
if (!Number.isNaN(date.getTime())) return date.toISOString();
|
|
96
|
+
}
|
|
97
|
+
if (/^\d{4}-\d{2}-\d{2}$/.test(value)) return value;
|
|
98
|
+
if (!/(?:Z|[+-]\d{2}:?\d{2})$/i.test(value)) return undefined;
|
|
99
|
+
const date = new Date(value);
|
|
100
|
+
return Number.isNaN(date.getTime()) ? undefined : date.toISOString();
|
|
101
|
+
};
|
|
102
|
+
|
|
103
|
+
export function parseNetscapeBookmarks(
|
|
104
|
+
source: string,
|
|
105
|
+
maxRecords: number,
|
|
106
|
+
maxFailures = 1_000
|
|
107
|
+
): BrowserExportParseResult {
|
|
108
|
+
const records: BrowserExportRecord[] = [];
|
|
109
|
+
const failures: string[] = [];
|
|
110
|
+
const folders: string[] = [];
|
|
111
|
+
let pendingFolder: string | undefined;
|
|
112
|
+
let anchorNumber = 0;
|
|
113
|
+
let failureCapReached = false;
|
|
114
|
+
const addFailure = (locator: string): boolean => {
|
|
115
|
+
if (failureCapReached) return true;
|
|
116
|
+
if (failures.length < maxFailures) failures.push(locator);
|
|
117
|
+
if (failures.length < maxFailures) return false;
|
|
118
|
+
failures.push("failure-limit");
|
|
119
|
+
failureCapReached = true;
|
|
120
|
+
return failureCapReached;
|
|
121
|
+
};
|
|
122
|
+
if (!hasCompleteNetscapeStructure(source)) {
|
|
123
|
+
addFailure("unsupported-html");
|
|
124
|
+
return { records, failures };
|
|
125
|
+
}
|
|
126
|
+
let dlDepth = 0;
|
|
127
|
+
for (const token of source.matchAll(HTML_TOKEN_PATTERN)) {
|
|
128
|
+
const raw = token[0].toLowerCase();
|
|
129
|
+
if (token[1] !== undefined) {
|
|
130
|
+
pendingFolder = plainText(token[1]);
|
|
131
|
+
continue;
|
|
132
|
+
}
|
|
133
|
+
if (raw.startsWith("<dl")) {
|
|
134
|
+
dlDepth += 1;
|
|
135
|
+
if (pendingFolder) folders.push(pendingFolder);
|
|
136
|
+
pendingFolder = undefined;
|
|
137
|
+
continue;
|
|
138
|
+
}
|
|
139
|
+
if (raw.startsWith("</dl")) {
|
|
140
|
+
if (dlDepth === 0) {
|
|
141
|
+
addFailure("html-structure");
|
|
142
|
+
break;
|
|
143
|
+
}
|
|
144
|
+
dlDepth -= 1;
|
|
145
|
+
folders.pop();
|
|
146
|
+
continue;
|
|
147
|
+
}
|
|
148
|
+
if (token[2] === undefined || token[3] === undefined) continue;
|
|
149
|
+
anchorNumber += 1;
|
|
150
|
+
if (records.length >= maxRecords) {
|
|
151
|
+
addFailure(`bookmark:${anchorNumber}/record-limit`);
|
|
152
|
+
break;
|
|
153
|
+
}
|
|
154
|
+
const attrs = attributes(token[2]);
|
|
155
|
+
const url = attrs.get("href");
|
|
156
|
+
if (!url) {
|
|
157
|
+
if (addFailure(`bookmark:${anchorNumber}`)) break;
|
|
158
|
+
continue;
|
|
159
|
+
}
|
|
160
|
+
const rawAdded = attrs.get("add_date");
|
|
161
|
+
const rawModified = attrs.get("last_modified");
|
|
162
|
+
const added = exportDate(rawAdded);
|
|
163
|
+
const modified = exportDate(rawModified);
|
|
164
|
+
if ((rawAdded && !added) || (rawModified && !modified)) {
|
|
165
|
+
if (addFailure(`bookmark:${anchorNumber}/date`)) break;
|
|
166
|
+
}
|
|
167
|
+
records.push({
|
|
168
|
+
kind: "bookmark",
|
|
169
|
+
url,
|
|
170
|
+
title: plainText(token[3]),
|
|
171
|
+
folder: folders.join("/"),
|
|
172
|
+
tags: attrs
|
|
173
|
+
.get("tags")
|
|
174
|
+
?.split(",")
|
|
175
|
+
.map((tag) => tag.trim())
|
|
176
|
+
.filter(Boolean),
|
|
177
|
+
dates:
|
|
178
|
+
added || modified
|
|
179
|
+
? {
|
|
180
|
+
...(added ? { added } : {}),
|
|
181
|
+
...(modified ? { modified } : {}),
|
|
182
|
+
}
|
|
183
|
+
: undefined,
|
|
184
|
+
sourceLocator: `bookmark:${anchorNumber}`,
|
|
185
|
+
});
|
|
186
|
+
}
|
|
187
|
+
if (dlDepth !== 0 && !failureCapReached) addFailure("html-structure");
|
|
188
|
+
return { records, failures };
|
|
189
|
+
}
|
|
190
|
+
|
|
191
|
+
const stringValue = (
|
|
192
|
+
value: Record<string, unknown>,
|
|
193
|
+
keys: string[]
|
|
194
|
+
): string | undefined => {
|
|
195
|
+
for (const key of keys) {
|
|
196
|
+
if (typeof value[key] === "string") return value[key] as string;
|
|
197
|
+
if (typeof value[key] === "number") return String(value[key]);
|
|
198
|
+
}
|
|
199
|
+
return undefined;
|
|
200
|
+
};
|
|
201
|
+
|
|
202
|
+
const kindValue = (value: Record<string, unknown>): BrowserExportKind => {
|
|
203
|
+
const raw = stringValue(value, ["kind", "type"])?.toLowerCase();
|
|
204
|
+
if (raw?.includes("history") || raw === "visit") return "history";
|
|
205
|
+
if (raw?.includes("reading") || raw === "read") return "reading-list";
|
|
206
|
+
return "bookmark";
|
|
207
|
+
};
|
|
208
|
+
|
|
209
|
+
const itemRecord = (
|
|
210
|
+
value: Record<string, unknown>,
|
|
211
|
+
locator: string,
|
|
212
|
+
folders: string[]
|
|
213
|
+
): BrowserExportRecord | undefined => {
|
|
214
|
+
const url = stringValue(value, ["url", "href", "uri"]);
|
|
215
|
+
if (!url) return undefined;
|
|
216
|
+
const tags = Array.isArray(value.tags)
|
|
217
|
+
? value.tags.filter((tag): tag is string => typeof tag === "string")
|
|
218
|
+
: typeof value.tags === "string"
|
|
219
|
+
? value.tags.split(",").map((tag) => tag.trim())
|
|
220
|
+
: undefined;
|
|
221
|
+
const rawAdded = stringValue(value, ["date_added", "dateAdded", "added"]);
|
|
222
|
+
const rawVisited = stringValue(value, [
|
|
223
|
+
"last_visit_time",
|
|
224
|
+
"lastVisitTime",
|
|
225
|
+
"visited_at",
|
|
226
|
+
]);
|
|
227
|
+
const rawRead = stringValue(value, ["read_at", "readAt"]);
|
|
228
|
+
const added = exportDate(rawAdded);
|
|
229
|
+
const visited = exportDate(rawVisited);
|
|
230
|
+
const read = exportDate(rawRead);
|
|
231
|
+
if ((rawAdded && !added) || (rawVisited && !visited) || (rawRead && !read)) {
|
|
232
|
+
return undefined;
|
|
233
|
+
}
|
|
234
|
+
const dateEntries: Array<[string, string]> = [];
|
|
235
|
+
if (added) dateEntries.push(["added", added]);
|
|
236
|
+
if (visited) dateEntries.push(["visited", visited]);
|
|
237
|
+
if (read) dateEntries.push(["read", read]);
|
|
238
|
+
return {
|
|
239
|
+
kind: kindValue(value),
|
|
240
|
+
url,
|
|
241
|
+
title: stringValue(value, ["name", "title"]),
|
|
242
|
+
folder: stringValue(value, ["folder", "path"]) ?? folders.join("/"),
|
|
243
|
+
tags,
|
|
244
|
+
dates: dateEntries.length > 0 ? Object.fromEntries(dateEntries) : undefined,
|
|
245
|
+
externalId: stringValue(value, ["id", "guid", "uuid"]),
|
|
246
|
+
sourceLocator: locator,
|
|
247
|
+
};
|
|
248
|
+
};
|
|
249
|
+
|
|
250
|
+
export function parseBrowserJson(
|
|
251
|
+
source: string,
|
|
252
|
+
maxRecords: number,
|
|
253
|
+
maxFailures = 1_000
|
|
254
|
+
): BrowserExportParseResult {
|
|
255
|
+
let parsed: unknown;
|
|
256
|
+
try {
|
|
257
|
+
parsed = JSON.parse(source);
|
|
258
|
+
} catch {
|
|
259
|
+
return { records: [], failures: ["json"] };
|
|
260
|
+
}
|
|
261
|
+
if (!Array.isArray(parsed) && (!parsed || typeof parsed !== "object")) {
|
|
262
|
+
return { records: [], failures: ["unsupported-json"] };
|
|
263
|
+
}
|
|
264
|
+
if (!Array.isArray(parsed)) {
|
|
265
|
+
const rootKeys = Object.keys(parsed as Record<string, unknown>);
|
|
266
|
+
if (hasSensitiveJsonKey(parsed as Record<string, unknown>)) {
|
|
267
|
+
return { records: [], failures: ["sensitive-json"] };
|
|
268
|
+
}
|
|
269
|
+
if (!rootKeys.some((key) => SUPPORTED_JSON_KEYS.has(key))) {
|
|
270
|
+
return { records: [], failures: ["unsupported-json"] };
|
|
271
|
+
}
|
|
272
|
+
}
|
|
273
|
+
const records: BrowserExportRecord[] = [];
|
|
274
|
+
const failures: string[] = [];
|
|
275
|
+
let capReached = false;
|
|
276
|
+
let failureCapReached = false;
|
|
277
|
+
const addFailure = (locator: string): boolean => {
|
|
278
|
+
if (failureCapReached) return true;
|
|
279
|
+
if (failures.length < maxFailures) failures.push(locator);
|
|
280
|
+
if (failures.length < maxFailures) return false;
|
|
281
|
+
failures.push("failure-limit");
|
|
282
|
+
failureCapReached = true;
|
|
283
|
+
capReached = true;
|
|
284
|
+
return true;
|
|
285
|
+
};
|
|
286
|
+
const stopAtCap = (locator: string): boolean => {
|
|
287
|
+
if (records.length < maxRecords) return false;
|
|
288
|
+
if (!capReached) addFailure(`${locator}/record-limit`);
|
|
289
|
+
capReached = true;
|
|
290
|
+
return true;
|
|
291
|
+
};
|
|
292
|
+
const visit = (
|
|
293
|
+
value: unknown,
|
|
294
|
+
locator: string,
|
|
295
|
+
folders: string[],
|
|
296
|
+
depth: number
|
|
297
|
+
): void => {
|
|
298
|
+
if (capReached) return;
|
|
299
|
+
if (depth > MAX_JSON_DEPTH) {
|
|
300
|
+
addFailure(locator);
|
|
301
|
+
return;
|
|
302
|
+
}
|
|
303
|
+
if (Array.isArray(value)) {
|
|
304
|
+
for (const [index, child] of value.entries()) {
|
|
305
|
+
if (stopAtCap(`${locator}/${index}`)) break;
|
|
306
|
+
visit(child, `${locator}/${index}`, folders, depth + 1);
|
|
307
|
+
}
|
|
308
|
+
return;
|
|
309
|
+
}
|
|
310
|
+
if (!value || typeof value !== "object") return;
|
|
311
|
+
const object = value as Record<string, unknown>;
|
|
312
|
+
if (hasSensitiveJsonKey(object)) {
|
|
313
|
+
addFailure(`${locator}/sensitive`);
|
|
314
|
+
return;
|
|
315
|
+
}
|
|
316
|
+
const record = itemRecord(object, locator, folders);
|
|
317
|
+
if (record) {
|
|
318
|
+
if (stopAtCap(locator)) return;
|
|
319
|
+
records.push(record);
|
|
320
|
+
}
|
|
321
|
+
const folderName =
|
|
322
|
+
!record && typeof object.name === "string" ? object.name : undefined;
|
|
323
|
+
const nextFolders = folderName ? [...folders, folderName] : folders;
|
|
324
|
+
const supportedChildren = [...SUPPORTED_JSON_KEYS].filter((key) =>
|
|
325
|
+
Object.hasOwn(object, key)
|
|
326
|
+
);
|
|
327
|
+
if (!record && supportedChildren.length === 0) {
|
|
328
|
+
addFailure(locator);
|
|
329
|
+
return;
|
|
330
|
+
}
|
|
331
|
+
for (const key of supportedChildren) {
|
|
332
|
+
const child = object[key];
|
|
333
|
+
const validRoots =
|
|
334
|
+
key === "roots" &&
|
|
335
|
+
child !== null &&
|
|
336
|
+
typeof child === "object" &&
|
|
337
|
+
!Array.isArray(child);
|
|
338
|
+
const validItems = key !== "roots" && Array.isArray(child);
|
|
339
|
+
if (!(validRoots || validItems)) {
|
|
340
|
+
if (addFailure(`${locator}/${key}`)) break;
|
|
341
|
+
continue;
|
|
342
|
+
}
|
|
343
|
+
if (validRoots) {
|
|
344
|
+
for (const [rootName, root] of Object.entries(
|
|
345
|
+
child as Record<string, unknown>
|
|
346
|
+
)) {
|
|
347
|
+
if (capReached) break;
|
|
348
|
+
visit(root, `${locator}/roots/${rootName}`, nextFolders, depth + 1);
|
|
349
|
+
}
|
|
350
|
+
} else {
|
|
351
|
+
visit(child, `${locator}/${key}`, nextFolders, depth + 1);
|
|
352
|
+
}
|
|
353
|
+
if (capReached) break;
|
|
354
|
+
}
|
|
355
|
+
};
|
|
356
|
+
visit(parsed, "json", [], 0);
|
|
357
|
+
return { records, failures };
|
|
358
|
+
}
|