toolcraft 0.0.798 → 0.0.800

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/composition.json CHANGED
@@ -138,7 +138,7 @@
138
138
  },
139
139
  {
140
140
  "name": "toolcraft",
141
- "version": "0.0.798",
141
+ "version": "0.0.800",
142
142
  "license": "MIT"
143
143
  },
144
144
  {
@@ -148,7 +148,7 @@
148
148
  },
149
149
  {
150
150
  "name": "toolcraft-schema",
151
- "version": "0.0.798",
151
+ "version": "0.0.800",
152
152
  "license": "MIT"
153
153
  },
154
154
  {
@@ -138,7 +138,7 @@
138
138
  },
139
139
  {
140
140
  "name": "toolcraft",
141
- "version": "0.0.798",
141
+ "version": "0.0.800",
142
142
  "license": "MIT"
143
143
  },
144
144
  {
@@ -148,7 +148,7 @@
148
148
  },
149
149
  {
150
150
  "name": "toolcraft-schema",
151
- "version": "0.0.798",
151
+ "version": "0.0.800",
152
152
  "license": "MIT"
153
153
  },
154
154
  {
@@ -9,7 +9,7 @@ export * from "./fs/webdav/index.js";
9
9
  export * from "./bridge/index.js";
10
10
  export * from "./python/index.js";
11
11
  export { parseXml, parseXmlSteps, parseXmlSourceSteps, normalizeXmlChunks, parseXmlStream, XmlLimitError } from "./xml.js";
12
- export type { XmlName, XmlAttribute, XmlContent, XmlElement, XmlLimits, XmlParserFrame, XmlFrameRequest, XmlNamespaceScope, XmlNamespaceRequest, XmlAttributeRequest, XmlAttributeRecord } from "./xml.js";
12
+ export type { XmlName, XmlAttribute, XmlContent, XmlElement, XmlLimits, XmlParserFrame, XmlFrameRequest, XmlNamespaceScope, XmlNamespaceRequest, XmlSourceRead, XmlSourceSpan, XmlAttributeRequest, XmlAttributeRecord } from "./xml.js";
13
13
  export * from "./contracts/object.js";
14
14
  export { ObjectAuthority } from "./fs/object-authority.js";
15
15
  export * from "./fs/s3/index.js";
@@ -64,24 +64,48 @@ size cost. The default string and stream APIs retain their existing ancestry beh
64
64
 
65
65
  Also set `storeNamespaces: true` to resolve namespace scopes through host-backed
66
66
  immutable storage. `namespaceOperation: "get"` supplies a scope and prefix; fill
67
- `value` when present and set `complete: true`. For `"set"`, bind the supplied prefix
67
+ `value` when present and set `complete: true`. For `"has"`, fill `found` using
68
+ membership without reading the URI value. For `"set"`, bind the supplied prefix
68
69
  and value in a new scope, then fill `result` with its storage `reference` and the
69
70
  number of distinct bindings (`size`). Frames carry this scope handle so shadowing
70
71
  never changes ancestor scopes. In this mode event elements have an empty
71
72
  `namespaces` map; their names and attributes still carry resolved namespace URIs.
72
- The codec remains filesystem-independent, and individual names, attribute values
73
- and namespace URIs still need a token storage strategy.
73
+ The codec remains filesystem-independent. With `deferNamespaces: true`, resolved
74
+ URI metadata uses optional `namespaceReference` handles instead of complete strings.
75
+ Service `namespaceOperation: "reference"` by supplying the token `reference` when
76
+ present and setting `complete: true`. These handles belong to the host scope store;
77
+ retain or stream them before retiring that store. Built-in xml/xmlns names keep
78
+ their short namespace strings. Individual names still retain their token cost;
79
+ attribute values can use the fragment protocol below.
74
80
 
75
81
  Set `storeAttributes: true` to move current-tag attribute values, source order,
76
82
  and duplicate-name state behind `attributeOperation` requests (`has`, `append`,
77
83
  `read`, and `expanded`). Service the request's `found` or `result` field as described
78
- by `XmlAttributeRequest`, keeping records in caller-owned backing. With this flag,
84
+ by `XmlAttributeRequest`, keeping records in caller-owned backing. When namespace
85
+ storage is also enabled, `expanded` supplies the local name and a `namespace` scope
86
+ and prefix. Check the pair of resolved URI and local name without joining them in
87
+ memory; an empty prefix means no attribute namespace, regardless of the default
88
+ element namespace. With this flag,
79
89
  element `attributes` arrays are empty. Supply `onAttribute(attribute, element)` to
80
90
  receive resolved attributes in source order, after the open event and before any
81
91
  content or close event. The parser yields between attributes so a host can await
82
92
  storage or a consumer. Namespace declarations may follow prefixed attributes;
83
93
  resolution and duplicate-expanded-name validation still happen after the complete
84
- start tag. Individual attribute names and values retain their token cost.
94
+ start tag. With `fragmentAttributes: true`, ordinary values are validated through source spans; all attribute
95
+ values are stored as spans on attribute records and replayed in at most 512 UTF-16
96
+ units. Hosts must preserve the optional `source` field in append/read requests.
97
+ The third `onAttribute` argument is `true` for continuations of the same logical
98
+ attribute; await each fragment before requesting the next step. Empty attributes
99
+ still emit once. Entity diagnostics are emitted during validation, not replay.
100
+ This mode requires `storeAttributes: true`. Combining it with `storeNamespaces`
101
+ also validates namespace declarations in bounded fragments. A namespace `set`
102
+ request's `value` may then be a generator: strings are decoded value fragments,
103
+ numbers are work checkpoints, and objects are source-read requests. Service each
104
+ step before advancing, consume the complete value before filling `result`, and
105
+ close the producer if storage fails. Reserved URI checks preserve their diagnostic
106
+ order without joining fragments. Expanded-name events return complete URI strings
107
+ unless deferred namespace metadata is enabled; individual names still retain
108
+ their token cost.
85
109
 
86
110
  Pass `undefined` as the source length to parse incrementally. Such read requests
87
111
  have `streaming: true` and request at most 512 UTF-16 units. Return any nonempty
@@ -100,5 +124,12 @@ decoding replay bounded source windows before publishing a logical node, retaini
100
124
  its diagnostic and limit ordering. Numeric entity references are decoded without
101
125
  materializing the token. Set `compactDeclaration: true` to retain normalized declaration fields without
102
126
  copying long whitespace spans; validation always reads bounded windows. The
103
- buffered API keeps the original declaration spelling. Names, attribute values
104
- and namespace scopes still require their own memory budget.
127
+ buffered API keeps the original declaration spelling.
128
+
129
+ Combine `fragmentContent` with `deferContentNames: true` to validate long
130
+ processing-instruction targets through source windows. Their events have an empty
131
+ `target` and a `targetSource` span; replay that span before retiring the source.
132
+ All body continuations carry the same target span. Short targets remain strings.
133
+ The stored document engine transfers these spans to its own token store and
134
+ streams targets during output. Element and attribute names still require their
135
+ own memory budget.
@@ -1,9 +1,14 @@
1
+ import type { XmlNamespaceScope } from "./frames.js";
1
2
  export interface XmlAttributeRecord {
2
3
  name: string;
3
4
  prefix: string;
4
5
  localName: string;
5
6
  value: string;
6
7
  next: number;
8
+ source?: {
9
+ start: number;
10
+ end: number;
11
+ };
7
12
  }
8
13
  export interface XmlAttributeState {
9
14
  reference: number;
@@ -16,6 +21,10 @@ export type XmlAttributeRequest = {
16
21
  state: XmlAttributeState;
17
22
  name: string;
18
23
  found?: boolean;
24
+ namespace?: {
25
+ scope: XmlNamespaceScope;
26
+ prefix: string;
27
+ };
19
28
  } | {
20
29
  attributeOperation: 'append';
21
30
  state: XmlAttributeState;
@@ -41,7 +50,13 @@ export declare class XmlAttributes {
41
50
  get size(): number;
42
51
  get first(): number;
43
52
  has(name: string): Generator<XmlAttributeRequest, boolean, void>;
44
- append(name: string, prefix: string, localName: string, value: string): Generator<XmlAttributeRequest, void, void>;
53
+ append(name: string, prefix: string, localName: string, value: string, source?: {
54
+ start: number;
55
+ end: number;
56
+ }): Generator<XmlAttributeRequest, void, void>;
45
57
  read(reference: number, values: boolean): Generator<XmlAttributeRequest, XmlAttributeRecord, void>;
46
- expanded(name: string): Generator<XmlAttributeRequest, boolean, void>;
58
+ expanded(name: string, namespace?: {
59
+ scope: XmlNamespaceScope;
60
+ prefix: string;
61
+ }): Generator<XmlAttributeRequest, boolean, void>;
47
62
  }
@@ -23,14 +23,14 @@ export class XmlAttributes {
23
23
  throw new TypeError('Incomplete XML attribute lookup');
24
24
  return request.found;
25
25
  }
26
- *append(name, prefix, localName, value) {
26
+ *append(name, prefix, localName, value, source) {
27
27
  if (!this.external) {
28
28
  this.names.set(name, value);
29
29
  this.records.push({ name, prefix, localName });
30
30
  this.state = { reference: 0, first: 1, length: this.records.length, size: this.names.size };
31
31
  return;
32
32
  }
33
- const request = { attributeOperation: 'append', state: this.state, attribute: { name, prefix, localName, value } };
33
+ const request = { attributeOperation: 'append', state: this.state, attribute: { name, prefix, localName, value, ...(source ? { source } : {}) } };
34
34
  yield request;
35
35
  if (!request.result)
36
36
  throw new TypeError('Incomplete XML attribute append');
@@ -47,13 +47,13 @@ export class XmlAttributes {
47
47
  throw new TypeError('Incomplete XML attribute read');
48
48
  return request.result;
49
49
  }
50
- *expanded(name) {
50
+ *expanded(name, namespace) {
51
51
  if (!this.external) {
52
52
  const found = this.expandedNames.has(name);
53
53
  this.expandedNames.add(name);
54
54
  return found;
55
55
  }
56
- const request = { attributeOperation: 'expanded', state: this.state, name };
56
+ const request = { attributeOperation: 'expanded', state: this.state, name, ...(namespace ? { namespace } : {}) };
57
57
  yield request;
58
58
  if (request.found === undefined)
59
59
  throw new TypeError('Incomplete XML expanded attribute lookup');
@@ -1,4 +1,5 @@
1
- import type { XmlContent, XmlElement } from './index.js';
1
+ import type { XmlSourceStep } from './source.js';
2
+ import type { XmlContent, XmlElement, XmlName } from './index.js';
2
3
  export interface XmlParserFrame {
3
4
  element: XmlElement;
4
5
  content: XmlContent[] | undefined;
@@ -33,6 +34,17 @@ export interface XmlNamespaceScope {
33
34
  readonly size: number;
34
35
  }
35
36
  export type XmlNamespaceRequest = {
37
+ readonly namespaceOperation: 'reference';
38
+ readonly scope: XmlNamespaceScope;
39
+ readonly prefix: string;
40
+ reference?: number;
41
+ complete?: true;
42
+ } | {
43
+ readonly namespaceOperation: 'has';
44
+ readonly scope: XmlNamespaceScope;
45
+ readonly prefix: string;
46
+ found?: boolean;
47
+ } | {
36
48
  readonly namespaceOperation: 'get';
37
49
  readonly scope: XmlNamespaceScope;
38
50
  readonly prefix: string;
@@ -42,8 +54,10 @@ export type XmlNamespaceRequest = {
42
54
  readonly namespaceOperation: 'set';
43
55
  readonly scope: XmlNamespaceScope;
44
56
  readonly prefix: string;
45
- readonly value: string;
57
+ readonly value: string | Generator<XmlSourceStep | string, void, void>;
46
58
  result?: XmlNamespaceScope;
47
59
  };
48
60
  export declare function namespaceValue(scope: Map<string, string> | XmlNamespaceScope, prefix: string): Generator<XmlNamespaceRequest, string | undefined, void>;
49
- export declare function bindNamespace(scope: Map<string, string> | XmlNamespaceScope, prefix: string, value: string): Generator<XmlNamespaceRequest, Map<string, string> | XmlNamespaceScope, void>;
61
+ export declare function namespaceMetadata(scope: Map<string, string> | XmlNamespaceScope, prefix: string, deferred?: boolean): Generator<XmlNamespaceRequest, Pick<XmlName, 'namespace' | 'namespaceReference'>, void>;
62
+ export declare function hasNamespace(scope: Map<string, string> | XmlNamespaceScope, prefix: string): Generator<XmlNamespaceRequest, boolean, void>;
63
+ export declare function bindNamespace(scope: Map<string, string> | XmlNamespaceScope, prefix: string, value: string | Generator<XmlSourceStep | string, void, void>): Generator<XmlNamespaceRequest, Map<string, string> | XmlNamespaceScope, void>;
@@ -52,8 +52,30 @@ export function* namespaceValue(scope, prefix) {
52
52
  throw new TypeError('Incomplete XML namespace read');
53
53
  return request.value;
54
54
  }
55
+ export function* namespaceMetadata(scope, prefix, deferred = false) {
56
+ if (!deferred || scope instanceof Map)
57
+ return { namespace: (yield* namespaceValue(scope, prefix)) ?? '' };
58
+ if (prefix === 'xml')
59
+ return { namespace: 'http://www.w3.org/XML/1998/namespace' };
60
+ const request = { namespaceOperation: 'reference', scope, prefix };
61
+ yield request;
62
+ if (!request.complete)
63
+ throw new TypeError('Incomplete XML namespace reference lookup');
64
+ return { namespace: '', ...(request.reference === undefined ? {} : { namespaceReference: request.reference }) };
65
+ }
66
+ export function* hasNamespace(scope, prefix) {
67
+ if (scope instanceof Map)
68
+ return scope.has(prefix);
69
+ const request = { namespaceOperation: 'has', scope, prefix };
70
+ yield request;
71
+ if (request.found === undefined)
72
+ throw new TypeError('Incomplete XML namespace membership lookup');
73
+ return request.found;
74
+ }
55
75
  export function* bindNamespace(scope, prefix, value) {
56
76
  if (scope instanceof Map) {
77
+ if (typeof value !== 'string')
78
+ throw new TypeError('Streamed XML namespace values require external scopes');
57
79
  scope.set(prefix, value);
58
80
  return scope;
59
81
  }
@@ -3,13 +3,14 @@ export type { XmlAttributeRequest, XmlAttributeRecord, XmlAttributeState } from
3
3
  import { type XmlNamespaceRequest, type XmlFrameRequest } from "./frames.js";
4
4
  export type { XmlParserFrame, XmlFrameRequest, XmlNamespaceScope, XmlNamespaceRequest } from "./frames.js";
5
5
  export { normalizeXmlChunks } from "./characters.js";
6
- import { type XmlSourceStep } from "./source.js";
7
- export type { XmlSourceRead, XmlSourceStep } from "./source.js";
6
+ import { type XmlSourceSpan, type XmlSourceStep } from "./source.js";
7
+ export type { XmlSourceRead, XmlSourceSpan, XmlSourceStep } from "./source.js";
8
8
  import type { XmlStreamEvent } from "./stream.js";
9
9
  export { XmlLimitError } from "./errors.js";
10
10
  export interface XmlName {
11
11
  readonly name: string;
12
12
  readonly namespace: string;
13
+ readonly namespaceReference?: number;
13
14
  readonly localName: string;
14
15
  }
15
16
  export interface XmlAttribute extends XmlName {
@@ -21,6 +22,8 @@ export type XmlContent = XmlElement | {
21
22
  } | {
22
23
  readonly kind: "processing-instruction";
23
24
  readonly target: string;
25
+ readonly targetSource?: XmlSourceSpan;
26
+ readonly targetReference?: number;
24
27
  readonly text: string;
25
28
  };
26
29
  export interface XmlElement extends XmlName {
@@ -65,9 +68,14 @@ export interface XmlSourceLimits extends XmlStepLimits {
65
68
  readonly storeFrames?: boolean;
66
69
  /** Resolve namespace scopes through host-serviced requests. Requires retainTree: false. */
67
70
  readonly storeNamespaces?: boolean;
71
+ /** Carry host namespace token references in event metadata. Requires stored namespaces. */
72
+ readonly deferNamespaces?: boolean;
73
+ /** Emit long processing-instruction targets as source spans. Requires fragmentContent. */
74
+ readonly deferContentNames?: boolean;
68
75
  /** Store attribute collections through host requests and emit resolved attributes individually. */
69
76
  readonly storeAttributes?: boolean;
70
- readonly onAttribute?: (attribute: XmlAttribute, element: XmlElement) => void;
77
+ readonly fragmentAttributes?: boolean;
78
+ readonly onAttribute?: (attribute: XmlAttribute, element: XmlElement, continuation?: boolean) => void;
71
79
  }
72
80
  export type XmlParseStep = XmlSourceStep | XmlFrameRequest | XmlNamespaceRequest | XmlAttributeRequest;
73
81
  /** Parse a validated, BOM-free, line-normalized UTF-16 source via bounded read requests. */
@@ -1,5 +1,5 @@
1
1
  import { XmlAttributes } from "./attributes.js";
2
- import { XmlFrames, bindNamespace, namespaceValue } from "./frames.js";
2
+ import { XmlFrames, bindNamespace, namespaceValue, namespaceMetadata, hasNamespace } from "./frames.js";
3
3
  import { validCharacter } from "./characters.js";
4
4
  export { normalizeXmlChunks } from "./characters.js";
5
5
  import { XmlSource } from "./source.js";
@@ -68,6 +68,40 @@ function qualifiedNameSync(name, cache) {
68
68
  }
69
69
  return parts;
70
70
  }
71
+ /** Validate a name span without retaining its spelling. */
72
+ function* validateNameSpan(source, start, end) {
73
+ let first = true, colon = false, work = 0;
74
+ for (let offset = start; offset < end;) {
75
+ let point = yield* source.charCodeAt(offset);
76
+ if (point >= 0xd800 && point <= 0xdbff && offset + 1 < end) {
77
+ const low = yield* source.charCodeAt(offset + 1);
78
+ if (low >= 0xdc00 && low <= 0xdfff)
79
+ point = 0x10000 + (point - 0xd800) * 1024 + low - 0xdc00;
80
+ }
81
+ if (point === 58) {
82
+ if (colon || first)
83
+ invalid("invalid qualified name");
84
+ colon = true;
85
+ first = true;
86
+ }
87
+ else {
88
+ if (!(first ? nameStart(point) : namePart(point)))
89
+ invalid("invalid qualified name");
90
+ first = false;
91
+ }
92
+ const size = point > 0xffff ? 2 : 1;
93
+ offset += size;
94
+ work += size;
95
+ if (work >= 512) {
96
+ yield 512;
97
+ work -= 512;
98
+ }
99
+ }
100
+ if (first)
101
+ invalid("invalid qualified name");
102
+ if (work)
103
+ yield work;
104
+ }
71
105
  function* entity(source, start, end, recover) {
72
106
  // Inspect only a bounded prefix, even for arbitrarily long numeric references.
73
107
  const prefix = typeof source === "string" ? source.slice(start, Math.min(end, start + 4))
@@ -102,7 +136,7 @@ function* entity(source, start, end, recover) {
102
136
  return String.fromCodePoint(point);
103
137
  }
104
138
  /** Decoded text fragments; source read/work requests remain host-serviced. */
105
- function* textParts(source, start, end, decode, recover) {
139
+ function* textParts(source, start, end, decode, recover, attribute = false) {
106
140
  let cursor = start;
107
141
  while (cursor < end) {
108
142
  let finish = Math.min(end, cursor + 512);
@@ -111,7 +145,13 @@ function* textParts(source, start, end, decode, recover) {
111
145
  if (last >= 0xd800 && last <= 0xdbff)
112
146
  finish--;
113
147
  }
114
- const raw = yield* source.slice(cursor, finish);
148
+ let raw = yield* source.slice(cursor, finish);
149
+ if (attribute) {
150
+ let normalized = "";
151
+ for (const character of raw)
152
+ normalized += "\t\n\r".includes(character) ? " " : character;
153
+ raw = normalized;
154
+ }
115
155
  const found = decode ? raw.indexOf("&") : -1;
116
156
  if (found < 0) {
117
157
  yield raw;
@@ -140,9 +180,9 @@ function* textParts(source, start, end, decode, recover) {
140
180
  cursor = semicolon + 1;
141
181
  }
142
182
  }
143
- function* textFragments(source, start, end, decode, recover) {
183
+ function* textFragments(source, start, end, decode, recover, attribute = false) {
144
184
  let pending = "";
145
- for (const part of textParts(source, start, end, decode, recover)) {
185
+ for (const part of textParts(source, start, end, decode, recover, attribute)) {
146
186
  if (typeof part !== "string") {
147
187
  yield part;
148
188
  continue;
@@ -254,10 +294,16 @@ function validateLimits(limits) {
254
294
  }
255
295
  export function* parseXmlSourceSteps(length, limits = {}) {
256
296
  validateLimits(limits);
297
+ if (limits.deferContentNames && !limits.fragmentContent)
298
+ throw new TypeError("Deferred XML content names require fragmentContent: true");
299
+ if (limits.fragmentAttributes && !limits.storeAttributes)
300
+ throw new TypeError("XML attribute fragments require storeAttributes: true");
257
301
  if (limits.storeFrames && limits.retainTree !== false)
258
302
  throw new TypeError("Stored XML frames require retainTree: false");
259
303
  if (limits.storeAttributes && limits.retainTree !== false)
260
304
  throw new TypeError("Stored XML attributes require retainTree: false");
305
+ if (limits.deferNamespaces && !limits.storeNamespaces)
306
+ throw new TypeError("Deferred XML namespaces require stored namespaces");
261
307
  if (limits.storeNamespaces && limits.retainTree !== false)
262
308
  throw new TypeError("Stored XML namespaces require retainTree: false");
263
309
  if (limits.fragmentContent && limits.retainTree !== false)
@@ -314,7 +360,7 @@ export function* parseXmlSourceSteps(length, limits = {}) {
314
360
  }
315
361
  return offset - start;
316
362
  };
317
- const scanName = function* () {
363
+ const scanName = function* (defer = false) {
318
364
  const start = offset;
319
365
  while (yield* source.has(offset)) {
320
366
  const c = (yield* source.charCodeAt(offset));
@@ -327,6 +373,10 @@ export function* parseXmlSourceSteps(length, limits = {}) {
327
373
  pendingWork -= 512;
328
374
  }
329
375
  }
376
+ if (defer && offset - start > 512) {
377
+ yield* validateNameSpan(source, start, offset);
378
+ return ["", "", "", { start, end: offset }];
379
+ }
330
380
  const name = (yield* source.slice(start, offset));
331
381
  pendingWork += offset - start;
332
382
  while (pendingWork >= 512) {
@@ -336,7 +386,7 @@ export function* parseXmlSourceSteps(length, limits = {}) {
336
386
  const [prefix, localName] = qualifiedNameSync(name, qualifiedNames);
337
387
  return [name, prefix, localName];
338
388
  };
339
- const rawContent = function* (kind, start, end, parent, target = "") {
389
+ const rawContent = function* (kind, start, end, parent, target = "", targetSource) {
340
390
  if (end - start > maxTextLength - textLength)
341
391
  throw new XmlLimitError("maxTextLength", "XML text limit exceeded");
342
392
  textLength += end - start;
@@ -350,7 +400,7 @@ export function* parseXmlSourceSteps(length, limits = {}) {
350
400
  finish--;
351
401
  }
352
402
  const text = yield* source.slice(cursor, finish);
353
- const content = kind === "processing-instruction" ? { kind, target, text } : { kind, text };
403
+ const content = kind === "processing-instruction" ? { kind, target, text, ...(targetSource ? { targetSource } : {}) } : { kind, text };
354
404
  if (kind === "cdata" && retainTree)
355
405
  parent.element.text += text;
356
406
  if (retain)
@@ -486,7 +536,7 @@ export function* parseXmlSourceSteps(length, limits = {}) {
486
536
  else if ((yield* source.startsWith("<?", offset))) {
487
537
  const start = offset;
488
538
  offset += 2;
489
- const [target] = yield* scanName();
539
+ const [target, , , targetSource] = yield* scanName(limits.deferContentNames);
490
540
  while (pendingWork >= 512) {
491
541
  yield 512;
492
542
  pendingWork -= 512;
@@ -519,7 +569,7 @@ export function* parseXmlSourceSteps(length, limits = {}) {
519
569
  yield 512;
520
570
  pendingWork -= 512;
521
571
  }
522
- yield* rawContent("processing-instruction", body, end, yield* stack.peek(), target);
572
+ yield* rawContent("processing-instruction", body, end, yield* stack.peek(), target, targetSource);
523
573
  }
524
574
  offset = end + 2;
525
575
  }
@@ -600,28 +650,56 @@ export function* parseXmlSourceSteps(length, limits = {}) {
600
650
  }
601
651
  if (end < 0)
602
652
  invalid("unterminated attribute");
603
- const raw = (yield* source.slice(offset, end));
604
- pendingWork += raw.length * 2;
605
- while (pendingWork >= 512) {
606
- yield 512;
607
- pendingWork -= 512;
653
+ const fragmented = limits.fragmentAttributes && (limits.storeNamespaces || attribute !== "xmlns" && attrPrefix !== "xmlns");
654
+ const valueStart = offset;
655
+ let value = "", size = 0, matchesXml = true, matchesXmlns = true;
656
+ if (fragmented) {
657
+ if ((yield* source.indexOf("<", offset, end)) >= 0)
658
+ invalid("less-than in attribute");
659
+ for (const part of textFragments(source, offset, end, true, limits.recover, true)) {
660
+ if (typeof part !== "string")
661
+ yield part;
662
+ else {
663
+ matchesXml = matchesXml && part === xmlNamespace.slice(size, size + part.length);
664
+ matchesXmlns = matchesXmlns && part === xmlnsNamespace.slice(size, size + part.length);
665
+ size += part.length;
666
+ }
667
+ }
668
+ matchesXml = matchesXml && size === xmlNamespace.length;
669
+ matchesXmlns = matchesXmlns && size === xmlnsNamespace.length;
670
+ if (size > maxTextLength - textLength)
671
+ throw new XmlLimitError("maxTextLength", "XML text limit exceeded");
672
+ textLength += size;
608
673
  }
609
- if (raw.indexOf("<") >= 0)
610
- invalid("less-than in attribute");
611
- const normalized = /[\t\n\r]/.test(raw) ? raw.replace(/[\t\n\r]/g, " ") : raw;
612
- const value = normalized.indexOf("&") < 0 ? (pendingWork += normalized.length, normalized) : yield* entities(normalized, limits.recover);
613
- while (pendingWork >= 512) {
614
- yield 512;
615
- pendingWork -= 512;
674
+ else {
675
+ const raw = (yield* source.slice(offset, end));
676
+ pendingWork += raw.length * 2;
677
+ while (pendingWork >= 512) {
678
+ yield 512;
679
+ pendingWork -= 512;
680
+ }
681
+ if (raw.indexOf("<") >= 0)
682
+ invalid("less-than in attribute");
683
+ let normalized = raw;
684
+ if (raw.includes("\t") || raw.includes("\n") || raw.includes("\r")) {
685
+ normalized = "";
686
+ for (const character of raw)
687
+ normalized += "\t\n\r".includes(character) ? " " : character;
688
+ }
689
+ value = normalized.indexOf("&") < 0 ? (pendingWork += normalized.length, normalized) : yield* entities(normalized, limits.recover);
690
+ while (pendingWork >= 512) {
691
+ yield 512;
692
+ pendingWork -= 512;
693
+ }
694
+ admitText(value);
616
695
  }
617
- admitText(value);
618
- yield* attributes.append(attribute, attrPrefix, attrLocal, value);
696
+ yield* attributes.append(attribute, attrPrefix, attrLocal, value, limits.fragmentAttributes ? { start: offset, end } : undefined);
619
697
  offset = end + 1;
620
698
  if (attribute === "xmlns" || attribute.startsWith("xmlns:")) {
621
699
  const nsPrefix = attribute === "xmlns" ? "" : attribute.slice(6);
622
- if (nsPrefix === "xmlns" || value === xmlnsNamespace
623
- || (nsPrefix === "xml") !== (value === xmlNamespace)
624
- || (nsPrefix !== "" && value === ""))
700
+ if (nsPrefix === "xmlns" || (fragmented ? matchesXmlns : value === xmlnsNamespace)
701
+ || (nsPrefix === "xml") !== (fragmented ? matchesXml : value === xmlNamespace)
702
+ || (nsPrefix !== "" && (fragmented ? size === 0 : value === "")))
625
703
  invalid("invalid namespace binding");
626
704
  if (!ownsNamespaces && namespaces instanceof Map) {
627
705
  const copy = new Map();
@@ -632,7 +710,8 @@ export function* parseXmlSourceSteps(length, limits = {}) {
632
710
  namespaces = copy;
633
711
  ownsNamespaces = true;
634
712
  }
635
- namespaces = yield* bindNamespace(namespaces, nsPrefix, value);
713
+ namespaces = yield* bindNamespace(namespaces, nsPrefix, fragmented
714
+ ? textFragments(source, valueStart, end, true, limits.recover ? () => { } : undefined, true) : value);
636
715
  if (namespaces.size > maxNamespaces)
637
716
  throw new XmlLimitError("maxNamespaces", "XML namespace scope limit exceeded");
638
717
  }
@@ -645,11 +724,12 @@ export function* parseXmlSourceSteps(length, limits = {}) {
645
724
  if (attribute === "xmlns" || attribute.startsWith("xmlns:"))
646
725
  continue;
647
726
  pendingWork += attribute.length;
648
- if (attrPrefix && (yield* namespaceValue(namespaces, attrPrefix)) === undefined)
727
+ if (attrPrefix && !(yield* hasNamespace(namespaces, attrPrefix)))
649
728
  invalid("unbound attribute prefix");
650
729
  if (attributes.length > 1) {
651
- const key = `${attrPrefix ? (yield* namespaceValue(namespaces, attrPrefix)) : ""}\0${attrLocal}`;
652
- if (yield* attributes.expanded(key))
730
+ const backed = limits.storeAttributes && !(namespaces instanceof Map) ? namespaces : undefined;
731
+ const key = backed ? attrLocal : `${attrPrefix ? (yield* namespaceValue(namespaces, attrPrefix)) : ""}\0${attrLocal}`;
732
+ if (yield* attributes.expanded(key, backed ? { scope: backed, prefix: attrPrefix } : undefined))
653
733
  invalid("duplicate expanded attribute");
654
734
  }
655
735
  }
@@ -659,14 +739,14 @@ export function* parseXmlSourceSteps(length, limits = {}) {
659
739
  yield 512;
660
740
  pendingWork -= 512;
661
741
  }
662
- if (prefix === "xmlns" || (prefix && (yield* namespaceValue(namespaces, prefix)) === undefined))
742
+ if (prefix === "xmlns" || (prefix && !(yield* hasNamespace(namespaces, prefix))))
663
743
  invalid("unbound element prefix");
664
744
  if (++nodes > maxNodes)
665
745
  throw new XmlLimitError("maxNodes", "XML resource limit exceeded");
666
746
  if (stack.length + 1 > maxDepth)
667
747
  throw new XmlLimitError("maxDepth", "XML resource limit exceeded");
668
- const namespace = (yield* namespaceValue(namespaces, prefix)) ?? "";
669
- limits.onElement?.({ name, namespace, localName }, (yield* stack.peek())?.element, stack.length + 1);
748
+ const namespace = yield* namespaceMetadata(namespaces, prefix, limits.deferNamespaces);
749
+ limits.onElement?.({ name, ...namespace, localName }, (yield* stack.peek())?.element, stack.length + 1);
670
750
  admitContent();
671
751
  const retainedAttributes = [];
672
752
  if (retainContent) {
@@ -686,7 +766,7 @@ export function* parseXmlSourceSteps(length, limits = {}) {
686
766
  pendingWork -= 512;
687
767
  }
688
768
  const content = retain ? [] : undefined;
689
- const element = { kind: "element", name, namespace, localName, children: [], text: "", content: content ?? emptyContent, attributes: retainContent ? retainedAttributes : emptyAttributes, namespaces: retainContent && namespaces instanceof Map ? namespaces : emptyNamespaces, ...(root === undefined && retainContent ? { prolog, epilog, ...(declaration === undefined ? {} : { declaration }) } : {}) };
769
+ const element = { kind: "element", name, ...namespace, localName, children: [], text: "", content: content ?? emptyContent, attributes: retainContent ? retainedAttributes : emptyAttributes, namespaces: retainContent && namespaces instanceof Map ? namespaces : emptyNamespaces, ...(root === undefined && retainContent ? { prolog, epilog, ...(declaration === undefined ? {} : { declaration }) } : {}) };
690
770
  const parent = (yield* stack.peek());
691
771
  if (parent) {
692
772
  if (retainTree) {
@@ -716,10 +796,24 @@ export function* parseXmlSourceSteps(length, limits = {}) {
716
796
  for (let cursor = attributes.first; cursor;) {
717
797
  const attribute = yield* attributes.read(cursor, true);
718
798
  cursor = attribute.next;
719
- const namespace = attribute.name === "xmlns" || attribute.prefix === "xmlns" ? xmlnsNamespace
720
- : attribute.prefix ? (yield* namespaceValue(namespaces, attribute.prefix)) : "";
721
- limits.onAttribute({ name: attribute.name, localName: attribute.localName, namespace, value: attribute.value }, element);
722
- yield 1;
799
+ const namespace = attribute.name === "xmlns" || attribute.prefix === "xmlns" ? { namespace: xmlnsNamespace }
800
+ : attribute.prefix ? yield* namespaceMetadata(namespaces, attribute.prefix, limits.deferNamespaces) : { namespace: "" };
801
+ let continuation = false;
802
+ if (attribute.source) {
803
+ for (const part of textFragments(source, attribute.source.start, attribute.source.end, true, limits.recover ? () => { } : undefined, true)) {
804
+ if (typeof part !== "string") {
805
+ yield part;
806
+ continue;
807
+ }
808
+ limits.onAttribute({ name: attribute.name, localName: attribute.localName, ...namespace, value: part }, element, continuation);
809
+ continuation = true;
810
+ yield 1;
811
+ }
812
+ }
813
+ if (!continuation) {
814
+ limits.onAttribute({ name: attribute.name, localName: attribute.localName, ...namespace, value: attribute.value }, element);
815
+ yield 1;
816
+ }
723
817
  }
724
818
  }
725
819
  if (!empty)
@@ -1,3 +1,7 @@
1
+ export interface XmlSourceSpan {
2
+ readonly start: number;
3
+ readonly end: number;
4
+ }
1
5
  /** Fill value before advancing. Streaming reads may be short; mark complete at EOF. */
2
6
  export interface XmlSourceRead {
3
7
  readonly offset: number;
@@ -8,7 +8,7 @@
8
8
  },
9
9
  {
10
10
  "name": "toolcraft-schema",
11
- "version": "0.0.798",
11
+ "version": "0.0.800",
12
12
  "license": "MIT"
13
13
  }
14
14
  ]
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "toolcraft-schema",
3
- "version": "0.0.798",
3
+ "version": "0.0.800",
4
4
  "license": "MIT",
5
5
  "type": "module",
6
6
  "main": "dist/index.js",
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "toolcraft",
3
- "version": "0.0.798",
3
+ "version": "0.0.800",
4
4
  "license": "MIT",
5
5
  "type": "module",
6
6
  "main": "dist/index.js",
@@ -177,7 +177,7 @@
177
177
  "saxes"
178
178
  ],
179
179
  "optionalDependencies": {
180
- "toolcraft-schema": "0.0.798",
180
+ "toolcraft-schema": "0.0.800",
181
181
  "toolcraft-design": "*",
182
182
  "@poe-code/frontmatter": "*",
183
183
  "@poe-code/agent-mcp-config": "*",