marc-ts 0.1.0 → 0.2.0

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/dist/marcxml.js CHANGED
@@ -1,5 +1,5 @@
1
1
  import { t as $ } from "./types-c4Mo9m9u.js";
2
- var M = /* @__PURE__ */ new Map([
2
+ var v = /* @__PURE__ */ new Map([
3
3
  ["amp", "&"],
4
4
  ["lt", "<"],
5
5
  ["gt", ">"],
@@ -7,16 +7,16 @@ var M = /* @__PURE__ */ new Map([
7
7
  ["apos", "'"]
8
8
  ]);
9
9
  function h(e) {
10
- return e.replace(/&(?:#x([0-9a-fA-F]+)|#([0-9]+)|([a-zA-Z]+));/g, (i, r, o, t) => {
11
- if (r !== void 0) {
12
- const n = parseInt(r, 16);
10
+ return e.replace(/&(?:#x([0-9a-fA-F]+)|#([0-9]+)|([a-zA-Z]+));/g, (i, s, c, t) => {
11
+ if (s !== void 0) {
12
+ const n = parseInt(s, 16);
13
13
  return n >= 0 && n <= 1114111 ? String.fromCodePoint(n) : "�";
14
14
  }
15
- if (o !== void 0) {
16
- const n = parseInt(o, 10);
15
+ if (c !== void 0) {
16
+ const n = parseInt(c, 10);
17
17
  return n >= 0 && n <= 1114111 ? String.fromCodePoint(n) : "�";
18
18
  }
19
- return M.get(t) ?? i;
19
+ return v.get(t) ?? i;
20
20
  });
21
21
  }
22
22
  function u(e) {
@@ -27,38 +27,38 @@ function f(e) {
27
27
  return i === -1 ? e : e.slice(i + 1);
28
28
  }
29
29
  function m(e) {
30
- const i = {}, r = /([a-zA-Z_:][^\s=]*)\s*=\s*(?:"([^"]*)"|'([^']*)')/g;
31
- let o;
32
- for (; (o = r.exec(e)) !== null; ) {
33
- const t = f(o[1]);
34
- i[t] = h(o[2] ?? o[3] ?? "");
30
+ const i = {}, s = /([a-zA-Z_:][^\s=]*)\s*=\s*(?:"([^"]*)"|'([^']*)')/g;
31
+ let c;
32
+ for (; (c = s.exec(e)) !== null; ) {
33
+ const t = f(c[1]);
34
+ i[t] = h(c[2] ?? c[3] ?? "");
35
35
  }
36
36
  return i;
37
37
  }
38
- function v(e) {
38
+ function b(e) {
39
39
  const i = [];
40
- let r = 0;
41
- for (; r < e.length; ) {
42
- const o = e.indexOf("<", r);
43
- if (o === -1) {
44
- e.slice(r).trim() && i.push({
40
+ let s = 0;
41
+ for (; s < e.length; ) {
42
+ const c = e.indexOf("<", s);
43
+ if (c === -1) {
44
+ e.slice(s).trim() && i.push({
45
45
  type: "text",
46
- text: h(e.slice(r))
46
+ text: h(e.slice(s))
47
47
  });
48
48
  break;
49
49
  }
50
- if (o > r) {
51
- const s = e.slice(r, o);
52
- s.trim() && i.push({
50
+ if (c > s) {
51
+ const o = e.slice(s, c);
52
+ o.trim() && i.push({
53
53
  type: "text",
54
- text: h(s)
54
+ text: h(o)
55
55
  });
56
56
  }
57
- const t = e.indexOf(">", o);
57
+ const t = e.indexOf(">", c);
58
58
  if (t === -1) break;
59
- const n = e.slice(o + 1, t);
59
+ const n = e.slice(c + 1, t);
60
60
  if (n.startsWith("!") || n.startsWith("?")) {
61
- r = t + 1;
61
+ s = t + 1;
62
62
  continue;
63
63
  }
64
64
  if (n.startsWith("/")) i.push({
@@ -66,60 +66,60 @@ function v(e) {
66
66
  name: f(n.slice(1).trim())
67
67
  });
68
68
  else if (n.endsWith("/")) {
69
- const s = n.slice(0, -1).trim(), c = s.search(/\s/), l = c === -1 ? s : s.slice(0, c), p = c === -1 ? "" : s.slice(c);
69
+ const o = n.slice(0, -1).trim(), r = o.search(/\s/), l = r === -1 ? o : o.slice(0, r), p = r === -1 ? "" : o.slice(r);
70
70
  i.push({
71
71
  type: "self-close",
72
72
  name: f(l),
73
73
  attrs: m(p)
74
74
  });
75
75
  } else {
76
- const s = n.search(/\s/), c = s === -1 ? n : n.slice(0, s), l = s === -1 ? "" : n.slice(s);
76
+ const o = n.search(/\s/), r = o === -1 ? n : n.slice(0, o), l = o === -1 ? "" : n.slice(o);
77
77
  i.push({
78
78
  type: "open",
79
- name: f(c),
79
+ name: f(r),
80
80
  attrs: m(l)
81
81
  });
82
82
  }
83
- r = t + 1;
83
+ s = t + 1;
84
84
  }
85
85
  return i;
86
86
  }
87
87
  function w(e, i) {
88
- let r = "";
89
- const o = [];
88
+ let s = "";
89
+ const c = [];
90
90
  let t = i;
91
91
  for (; t < e.length; ) {
92
92
  const n = e[t];
93
93
  if (n.type === "close" && n.name === "record") return {
94
94
  record: {
95
- leader: r,
96
- fields: o
95
+ leader: s,
96
+ fields: c
97
97
  },
98
98
  end: t + 1
99
99
  };
100
100
  if (n.type === "open" && n.name === "leader") {
101
- t++, t < e.length && e[t].type === "text" && (r = e[t].text.trim(), t++), t < e.length && e[t].type === "close" && t++;
101
+ t++, t < e.length && e[t].type === "text" && (s = e[t].text.trim(), t++), t < e.length && e[t].type === "close" && t++;
102
102
  continue;
103
103
  }
104
104
  if (n.type === "self-close" && n.name === "controlfield") {
105
- o.push({
105
+ c.push({
106
106
  tag: n.attrs?.tag ?? "",
107
107
  data: ""
108
108
  }), t++;
109
109
  continue;
110
110
  }
111
111
  if (n.type === "open" && n.name === "controlfield") {
112
- const s = n.attrs?.tag ?? "";
112
+ const o = n.attrs?.tag ?? "";
113
113
  t++;
114
- let c = "";
115
- t < e.length && e[t].type === "text" && (c = e[t].text ?? "", t++), t < e.length && e[t].type === "close" && t++, o.push({
116
- tag: s,
117
- data: c
114
+ let r = "";
115
+ t < e.length && e[t].type === "text" && (r = e[t].text ?? "", t++), t < e.length && e[t].type === "close" && t++, c.push({
116
+ tag: o,
117
+ data: r
118
118
  });
119
119
  continue;
120
120
  }
121
121
  if (n.type === "self-close" && n.name === "datafield") {
122
- o.push({
122
+ c.push({
123
123
  tag: n.attrs?.tag ?? "",
124
124
  indicator1: n.attrs?.ind1 ?? " ",
125
125
  indicator2: n.attrs?.ind2 ?? " ",
@@ -128,7 +128,7 @@ function w(e, i) {
128
128
  continue;
129
129
  }
130
130
  if (n.type === "open" && n.name === "datafield") {
131
- const s = n.attrs?.tag ?? "", c = n.attrs?.ind1 ?? " ", l = n.attrs?.ind2 ?? " ", p = [];
131
+ const o = n.attrs?.tag ?? "", r = n.attrs?.ind1 ?? " ", l = n.attrs?.ind2 ?? " ", p = [];
132
132
  for (t++; t < e.length; ) {
133
133
  const d = e[t];
134
134
  if (d.type === "close" && d.name === "datafield") {
@@ -147,9 +147,9 @@ function w(e, i) {
147
147
  }
148
148
  t++;
149
149
  }
150
- o.push({
151
- tag: s,
152
- indicator1: c,
150
+ c.push({
151
+ tag: o,
152
+ indicator1: r,
153
153
  indicator2: l,
154
154
  subfields: p
155
155
  });
@@ -159,62 +159,55 @@ function w(e, i) {
159
159
  }
160
160
  return {
161
161
  record: {
162
- leader: r,
163
- fields: o
162
+ leader: s,
163
+ fields: c
164
164
  },
165
165
  end: t
166
166
  };
167
167
  }
168
- function b(e) {
169
- const i = v(e), r = [];
170
- let o = 0;
171
- for (; o < i.length; ) {
172
- const t = i[o];
168
+ function M(e) {
169
+ const i = b(e), s = [];
170
+ let c = 0;
171
+ for (; c < i.length; ) {
172
+ const t = i[c];
173
173
  if (t.type === "open" && t.name === "record") {
174
- const { record: n, end: s } = w(i, o + 1);
175
- r.push(n), o = s;
174
+ const { record: n, end: o } = w(i, c + 1);
175
+ s.push(n), c = o;
176
176
  continue;
177
177
  }
178
- o++;
178
+ c++;
179
179
  }
180
- return r;
181
- }
182
- function E(e) {
183
- const i = b(e);
184
- if (i.length === 0) throw new Error("No MARC record found in MARCXML input");
185
- return i[0];
180
+ return s;
186
181
  }
187
182
  var A = `<?xml version="1.0" encoding="UTF-8"?>
188
183
  `, y = 'xmlns="http://www.loc.gov/MARC21/slim"', a = " ";
189
184
  function C(e) {
190
185
  const i = [`<record ${y}>`];
191
186
  i.push(`${a}<leader>${u(e.leader)}</leader>`);
192
- for (const r of e.fields) if ($(r)) i.push(`${a}<controlfield tag="${r.tag}">${u(r.data)}</controlfield>`);
187
+ for (const s of e.fields) if ($(s)) i.push(`${a}<controlfield tag="${s.tag}">${u(s.data)}</controlfield>`);
193
188
  else {
194
- const o = r.indicator1 === " " ? " " : r.indicator1, t = r.indicator2 === " " ? " " : r.indicator2;
195
- i.push(`${a}<datafield tag="${r.tag}" ind1="${o}" ind2="${t}">`);
196
- for (const n of r.subfields) i.push(`${a}${a}<subfield code="${n.code}">${u(n.value)}</subfield>`);
189
+ const c = s.indicator1 === " " ? " " : s.indicator1, t = s.indicator2 === " " ? " " : s.indicator2;
190
+ i.push(`${a}<datafield tag="${s.tag}" ind1="${c}" ind2="${t}">`);
191
+ for (const n of s.subfields) i.push(`${a}${a}<subfield code="${n.code}">${u(n.value)}</subfield>`);
197
192
  i.push(`${a}</datafield>`);
198
193
  }
199
194
  return i.push("</record>"), i.join(`
200
195
  `);
201
196
  }
202
- function I(e) {
197
+ function X(e) {
203
198
  const i = [A, `<collection ${y}>`];
204
- for (const r of e) {
205
- const o = C(r).split(`
199
+ for (const s of e) {
200
+ const c = C(s).split(`
206
201
  `).map((t) => a + t).join(`
207
202
  `);
208
- i.push(o);
203
+ i.push(c);
209
204
  }
210
205
  return i.push("</collection>"), i.join(`
211
206
  `);
212
207
  }
213
208
  export {
214
- b as parseMarcXml,
215
- E as parseMarcXmlRecord,
216
- I as serializeMarcXml,
217
- C as serializeMarcXmlRecord
209
+ M as parseMarcXml,
210
+ X as serializeMarcXml
218
211
  };
219
212
 
220
213
  //# sourceMappingURL=marcxml.js.map
@@ -1 +1 @@
1
- {"version":3,"file":"marcxml.js","names":[],"sources":["../src/marcxml.ts"],"sourcesContent":["/**\n * MARCXML parser and serializer.\n *\n * Supports the Library of Congress MARCXML schema:\n * http://www.loc.gov/MARC21/slim\n *\n * Parsing is done with a hand-rolled state machine — no XML library needed.\n * The MARCXML format is sufficiently regular (fixed element names, no arbitrary\n * nesting) that a full DOM parser is unnecessary.\n */\n\nimport type { MarcRecord, ControlField, DataField, Subfield } from './types';\nimport { isControlField } from './types';\n\n// ─── XML entity handling ─────────────────────────────────────────────────────\n\nconst ENTITY_MAP: ReadonlyMap<string, string> = new Map([\n ['amp', '&'],\n ['lt', '<'],\n ['gt', '>'],\n ['quot', '\"'],\n ['apos', \"'\"],\n]);\n\nfunction unescapeXml(text: string): string {\n return text.replace(/&(?:#x([0-9a-fA-F]+)|#([0-9]+)|([a-zA-Z]+));/g, (_, hex, dec, name) => {\n if (hex !== undefined) {\n const cp = parseInt(hex, 16);\n return cp >= 0 && cp <= 0x10ffff ? String.fromCodePoint(cp) : '�';\n }\n if (dec !== undefined) {\n const cp = parseInt(dec, 10);\n return cp >= 0 && cp <= 0x10ffff ? String.fromCodePoint(cp) : '�';\n }\n return ENTITY_MAP.get(name) ?? _;\n });\n}\n\nfunction escapeXml(text: string): string {\n return text\n // XML 1.0 forbids most C0 control characters in document text. There is no\n // valid XML 1.0 representation for them, so substitute the Unicode\n // replacement character to keep the output well-formed.\n .replace(/[\\x00-\\x08\\x0B\\x0C\\x0E-\\x1F]/g, '�')\n .replace(/&/g, '&amp;')\n .replace(/</g, '&lt;')\n .replace(/>/g, '&gt;')\n .replace(/\"/g, '&quot;')\n // Preserve literal CR through the XML round-trip: XML parsers normalize\n // bare \\r and \\r\\n to \\n, so we must encode CR as a numeric reference.\n .replace(/\\r/g, '&#13;');\n}\n\n// ─── Minimal tokeniser ────────────────────────────────────────────────────────\n\ninterface Token {\n type: 'open' | 'close' | 'self-close' | 'text';\n /** Local name (no namespace prefix) */\n name?: string;\n attrs?: Record<string, string>;\n text?: string;\n}\n\n/**\n * Strip namespace prefix from a tag name, e.g. \"marc:record\" → \"record\".\n */\nfunction localName(raw: string): string {\n const colon = raw.indexOf(':');\n return colon === -1 ? raw : raw.slice(colon + 1);\n}\n\n/**\n * Parse `key=\"value\"` pairs out of an attribute string.\n */\nfunction parseAttrs(attrStr: string): Record<string, string> {\n const attrs: Record<string, string> = {};\n const re = /([a-zA-Z_:][^\\s=]*)\\s*=\\s*(?:\"([^\"]*)\"|'([^']*)')/g;\n let m: RegExpExecArray | null;\n while ((m = re.exec(attrStr)) !== null) {\n const key = localName(m[1]!);\n attrs[key] = unescapeXml(m[2] ?? m[3] ?? '');\n }\n return attrs;\n}\n\n/**\n * Tokenise an XML string into a flat stream of open/close/text tokens.\n * Skips processing instructions, comments, and DOCTYPE declarations.\n * Sufficient for the well-constrained MARCXML format.\n */\nfunction tokenise(xml: string): Token[] {\n const tokens: Token[] = [];\n let i = 0;\n\n while (i < xml.length) {\n const ltPos = xml.indexOf('<', i);\n\n // Text node before next tag\n if (ltPos === -1) {\n const text = xml.slice(i).trim();\n if (text) tokens.push({ type: 'text', text: unescapeXml(xml.slice(i)) });\n break;\n }\n\n if (ltPos > i) {\n const raw = xml.slice(i, ltPos);\n const text = raw.trim();\n if (text) tokens.push({ type: 'text', text: unescapeXml(raw) });\n }\n\n const gtPos = xml.indexOf('>', ltPos);\n if (gtPos === -1) break;\n\n const tag = xml.slice(ltPos + 1, gtPos);\n\n // Skip comments, PIs, DOCTYPE\n if (tag.startsWith('!') || tag.startsWith('?')) {\n i = gtPos + 1;\n continue;\n }\n\n if (tag.startsWith('/')) {\n tokens.push({ type: 'close', name: localName(tag.slice(1).trim()) });\n } else if (tag.endsWith('/')) {\n const inner = tag.slice(0, -1).trim();\n const spaceIdx = inner.search(/\\s/);\n const name = spaceIdx === -1 ? inner : inner.slice(0, spaceIdx);\n const attrStr = spaceIdx === -1 ? '' : inner.slice(spaceIdx);\n tokens.push({ type: 'self-close', name: localName(name), attrs: parseAttrs(attrStr) });\n } else {\n const spaceIdx = tag.search(/\\s/);\n const name = spaceIdx === -1 ? tag : tag.slice(0, spaceIdx);\n const attrStr = spaceIdx === -1 ? '' : tag.slice(spaceIdx);\n tokens.push({ type: 'open', name: localName(name), attrs: parseAttrs(attrStr) });\n }\n\n i = gtPos + 1;\n }\n\n return tokens;\n}\n\n// ─── MARCXML parser ───────────────────────────────────────────────────────────\n\n/**\n * Parse one `<record>` element's worth of tokens into a MarcRecord.\n * Mutates `pos` via the returned index.\n */\nfunction parseRecordTokens(tokens: Token[], start: number): { record: MarcRecord; end: number } {\n let leader = '';\n const fields: (ControlField | DataField)[] = [];\n let i = start;\n\n while (i < tokens.length) {\n const tok = tokens[i]!;\n\n if (tok.type === 'close' && tok.name === 'record') {\n return { record: { leader, fields }, end: i + 1 };\n }\n\n if (tok.type === 'open' && tok.name === 'leader') {\n i++;\n if (i < tokens.length && tokens[i]!.type === 'text') {\n leader = tokens[i]!.text!.trim();\n i++;\n }\n // consume </leader>\n if (i < tokens.length && tokens[i]!.type === 'close') i++;\n continue;\n }\n\n if (tok.type === 'self-close' && tok.name === 'controlfield') {\n fields.push({ tag: tok.attrs?.['tag'] ?? '', data: '' });\n i++;\n continue;\n }\n\n if (tok.type === 'open' && tok.name === 'controlfield') {\n const tag = tok.attrs?.['tag'] ?? '';\n i++;\n let data = '';\n if (i < tokens.length && tokens[i]!.type === 'text') {\n data = tokens[i]!.text ?? '';\n i++;\n }\n // consume </controlfield>\n if (i < tokens.length && tokens[i]!.type === 'close') i++;\n fields.push({ tag, data });\n continue;\n }\n\n if (tok.type === 'self-close' && tok.name === 'datafield') {\n fields.push({\n tag: tok.attrs?.['tag'] ?? '',\n indicator1: tok.attrs?.['ind1'] ?? ' ',\n indicator2: tok.attrs?.['ind2'] ?? ' ',\n subfields: [],\n });\n i++;\n continue;\n }\n\n if (tok.type === 'open' && tok.name === 'datafield') {\n const tag = tok.attrs?.['tag'] ?? '';\n const indicator1 = tok.attrs?.['ind1'] ?? ' ';\n const indicator2 = tok.attrs?.['ind2'] ?? ' ';\n const subfields: Subfield[] = [];\n i++;\n\n while (i < tokens.length) {\n const stok = tokens[i]!;\n if (stok.type === 'close' && stok.name === 'datafield') {\n i++;\n break;\n }\n if (stok.type === 'open' && stok.name === 'subfield') {\n const code = stok.attrs?.['code'] ?? '';\n i++;\n let value = '';\n if (i < tokens.length && tokens[i]!.type === 'text') {\n value = tokens[i]!.text ?? '';\n i++;\n }\n // consume </subfield>\n if (i < tokens.length && tokens[i]!.type === 'close') i++;\n subfields.push({ code, value });\n continue;\n }\n i++;\n }\n\n fields.push({ tag, indicator1, indicator2, subfields });\n continue;\n }\n\n i++;\n }\n\n return { record: { leader, fields }, end: i };\n}\n\n/**\n * Parse a MARCXML string containing one `<collection>` or one bare `<record>`.\n * Returns all records found.\n */\nexport function parseMarcXml(xml: string): MarcRecord[] {\n const tokens = tokenise(xml);\n const records: MarcRecord[] = [];\n let i = 0;\n\n while (i < tokens.length) {\n const tok = tokens[i]!;\n if (tok.type === 'open' && tok.name === 'record') {\n const { record, end } = parseRecordTokens(tokens, i + 1);\n records.push(record);\n i = end;\n continue;\n }\n i++;\n }\n\n return records;\n}\n\n/**\n * Parse a MARCXML string expected to contain exactly one `<record>`.\n * Throws if no record is found.\n */\nexport function parseMarcXmlRecord(xml: string): MarcRecord {\n const records = parseMarcXml(xml);\n if (records.length === 0) throw new Error('No MARC record found in MARCXML input');\n return records[0]!;\n}\n\n// ─── MARCXML serializer ───────────────────────────────────────────────────────\n\nconst XML_HEADER = '<?xml version=\"1.0\" encoding=\"UTF-8\"?>\\n';\nconst COLLECTION_NS = 'xmlns=\"http://www.loc.gov/MARC21/slim\"';\nconst INDENT = ' ';\n\n/**\n * Serialize a single MarcRecord to a `<record>` XML element string (no collection wrapper).\n */\nexport function serializeMarcXmlRecord(record: MarcRecord): string {\n const lines: string[] = [`<record ${COLLECTION_NS}>`];\n lines.push(`${INDENT}<leader>${escapeXml(record.leader)}</leader>`);\n\n for (const field of record.fields) {\n if (isControlField(field)) {\n lines.push(`${INDENT}<controlfield tag=\"${field.tag}\">${escapeXml(field.data)}</controlfield>`);\n } else {\n const ind1 = field.indicator1 === ' ' ? ' ' : field.indicator1;\n const ind2 = field.indicator2 === ' ' ? ' ' : field.indicator2;\n lines.push(`${INDENT}<datafield tag=\"${field.tag}\" ind1=\"${ind1}\" ind2=\"${ind2}\">`);\n for (const sf of field.subfields) {\n lines.push(\n `${INDENT}${INDENT}<subfield code=\"${sf.code}\">${escapeXml(sf.value)}</subfield>`\n );\n }\n lines.push(`${INDENT}</datafield>`);\n }\n }\n\n lines.push('</record>');\n return lines.join('\\n');\n}\n\n/**\n * Serialize one or more MarcRecords into a MARCXML `<collection>` document.\n */\nexport function serializeMarcXml(records: MarcRecord[]): string {\n const parts: string[] = [\n XML_HEADER,\n `<collection ${COLLECTION_NS}>`,\n ];\n\n for (const record of records) {\n // Indent each record element by one level inside <collection>\n const recordXml = serializeMarcXmlRecord(record)\n .split('\\n')\n .map((line) => INDENT + line)\n .join('\\n');\n parts.push(recordXml);\n }\n\n parts.push('</collection>');\n return parts.join('\\n');\n}\n"],"mappings":";AAgBA,IAAM,IAA0C,oBAAI,IAAI;AAAA,EACtD,CAAC,OAAO,GAAG;AAAA,EACX,CAAC,MAAM,GAAG;AAAA,EACV,CAAC,MAAM,GAAG;AAAA,EACV,CAAC,QAAQ,GAAG;AAAA,EACZ,CAAC,QAAQ,GAAG;AACd,CAAC;AAED,SAAS,EAAY,GAAsB;AACzC,SAAO,EAAK,QAAQ,iDAAA,CAAkD,GAAG,GAAK,GAAK,MAAS;AAC1F,QAAI,MAAQ,QAAW;AACrB,YAAM,IAAK,SAAS,GAAK,EAAE;AAC3B,aAAO,KAAM,KAAK,KAAM,UAAW,OAAO,cAAc,CAAE,IAAI;AAAA,IAChE;AACA,QAAI,MAAQ,QAAW;AACrB,YAAM,IAAK,SAAS,GAAK,EAAE;AAC3B,aAAO,KAAM,KAAK,KAAM,UAAW,OAAO,cAAc,CAAE,IAAI;AAAA,IAChE;AACA,WAAO,EAAW,IAAI,CAAI,KAAK;AAAA,EACjC,CAAC;AACH;AAEA,SAAS,EAAU,GAAsB;AACvC,SAAO,EAIJ,QAAQ,iCAAiC,GAAG,EAC5C,QAAQ,MAAM,OAAO,EACrB,QAAQ,MAAM,MAAM,EACpB,QAAQ,MAAM,MAAM,EACpB,QAAQ,MAAM,QAAQ,EAGtB,QAAQ,OAAO,OAAO;AAC3B;AAeA,SAAS,EAAU,GAAqB;AACtC,QAAM,IAAQ,EAAI,QAAQ,GAAG;AAC7B,SAAO,MAAU,KAAK,IAAM,EAAI,MAAM,IAAQ,CAAC;AACjD;AAKA,SAAS,EAAW,GAAyC;AAC3D,QAAM,IAAgC,CAAC,GACjC,IAAK;AACX,MAAI;AACJ,UAAQ,IAAI,EAAG,KAAK,CAAO,OAAO,QAAM;AACtC,UAAM,IAAM,EAAU,EAAE,CAAA,CAAG;AAC3B,IAAA,EAAM,CAAA,IAAO,EAAY,EAAE,CAAA,KAAM,EAAE,CAAA,KAAM,EAAE;AAAA,EAC7C;AACA,SAAO;AACT;AAOA,SAAS,EAAS,GAAsB;AACtC,QAAM,IAAkB,CAAC;AACzB,MAAI,IAAI;AAER,SAAO,IAAI,EAAI,UAAQ;AACrB,UAAM,IAAQ,EAAI,QAAQ,KAAK,CAAC;AAGhC,QAAI,MAAU,IAAI;AAEhB,MADa,EAAI,MAAM,CAAC,EAAE,KACtB,KAAM,EAAO,KAAK;AAAA,QAAE,MAAM;AAAA,QAAQ,MAAM,EAAY,EAAI,MAAM,CAAC,CAAC;AAAA,MAAE,CAAC;AACvE;AAAA,IACF;AAEA,QAAI,IAAQ,GAAG;AACb,YAAM,IAAM,EAAI,MAAM,GAAG,CAAK;AAE9B,MADa,EAAI,KACb,KAAM,EAAO,KAAK;AAAA,QAAE,MAAM;AAAA,QAAQ,MAAM,EAAY,CAAG;AAAA,MAAE,CAAC;AAAA,IAChE;AAEA,UAAM,IAAQ,EAAI,QAAQ,KAAK,CAAK;AACpC,QAAI,MAAU,GAAI;AAElB,UAAM,IAAM,EAAI,MAAM,IAAQ,GAAG,CAAK;AAGtC,QAAI,EAAI,WAAW,GAAG,KAAK,EAAI,WAAW,GAAG,GAAG;AAC9C,MAAA,IAAI,IAAQ;AACZ;AAAA,IACF;AAEA,QAAI,EAAI,WAAW,GAAG,EACpB,CAAA,EAAO,KAAK;AAAA,MAAE,MAAM;AAAA,MAAS,MAAM,EAAU,EAAI,MAAM,CAAC,EAAE,KAAK,CAAC;AAAA,IAAE,CAAC;AAAA,aAC1D,EAAI,SAAS,GAAG,GAAG;AAC5B,YAAM,IAAQ,EAAI,MAAM,GAAG,EAAE,EAAE,KAAK,GAC9B,IAAW,EAAM,OAAO,IAAI,GAC5B,IAAO,MAAa,KAAK,IAAQ,EAAM,MAAM,GAAG,CAAQ,GACxD,IAAU,MAAa,KAAK,KAAK,EAAM,MAAM,CAAQ;AAC3D,MAAA,EAAO,KAAK;AAAA,QAAE,MAAM;AAAA,QAAc,MAAM,EAAU,CAAI;AAAA,QAAG,OAAO,EAAW,CAAO;AAAA,MAAE,CAAC;AAAA,IACvF,OAAO;AACL,YAAM,IAAW,EAAI,OAAO,IAAI,GAC1B,IAAO,MAAa,KAAK,IAAM,EAAI,MAAM,GAAG,CAAQ,GACpD,IAAU,MAAa,KAAK,KAAK,EAAI,MAAM,CAAQ;AACzD,MAAA,EAAO,KAAK;AAAA,QAAE,MAAM;AAAA,QAAQ,MAAM,EAAU,CAAI;AAAA,QAAG,OAAO,EAAW,CAAO;AAAA,MAAE,CAAC;AAAA,IACjF;AAEA,IAAA,IAAI,IAAQ;AAAA,EACd;AAEA,SAAO;AACT;AAQA,SAAS,EAAkB,GAAiB,GAAoD;AAC9F,MAAI,IAAS;AACb,QAAM,IAAuC,CAAC;AAC9C,MAAI,IAAI;AAER,SAAO,IAAI,EAAO,UAAQ;AACxB,UAAM,IAAM,EAAO,CAAA;AAEnB,QAAI,EAAI,SAAS,WAAW,EAAI,SAAS,SACvC,QAAO;AAAA,MAAE,QAAQ;AAAA,QAAE,QAAA;AAAA,QAAQ,QAAA;AAAA,MAAO;AAAA,MAAG,KAAK,IAAI;AAAA,IAAE;AAGlD,QAAI,EAAI,SAAS,UAAU,EAAI,SAAS,UAAU;AAChD,MAAA,KACI,IAAI,EAAO,UAAU,EAAO,CAAA,EAAI,SAAS,WAC3C,IAAS,EAAO,CAAA,EAAI,KAAM,KAAK,GAC/B,MAGE,IAAI,EAAO,UAAU,EAAO,CAAA,EAAI,SAAS,WAAS;AACtD;AAAA,IACF;AAEA,QAAI,EAAI,SAAS,gBAAgB,EAAI,SAAS,gBAAgB;AAC5D,MAAA,EAAO,KAAK;AAAA,QAAE,KAAK,EAAI,OAAQ,OAAU;AAAA,QAAI,MAAM;AAAA,MAAG,CAAC,GACvD;AACA;AAAA,IACF;AAEA,QAAI,EAAI,SAAS,UAAU,EAAI,SAAS,gBAAgB;AACtD,YAAM,IAAM,EAAI,OAAQ,OAAU;AAClC,MAAA;AACA,UAAI,IAAO;AACX,MAAI,IAAI,EAAO,UAAU,EAAO,CAAA,EAAI,SAAS,WAC3C,IAAO,EAAO,CAAA,EAAI,QAAQ,IAC1B,MAGE,IAAI,EAAO,UAAU,EAAO,CAAA,EAAI,SAAS,WAAS,KACtD,EAAO,KAAK;AAAA,QAAE,KAAA;AAAA,QAAK,MAAA;AAAA,MAAK,CAAC;AACzB;AAAA,IACF;AAEA,QAAI,EAAI,SAAS,gBAAgB,EAAI,SAAS,aAAa;AACzD,MAAA,EAAO,KAAK;AAAA,QACV,KAAK,EAAI,OAAQ,OAAU;AAAA,QAC3B,YAAY,EAAI,OAAQ,QAAW;AAAA,QACnC,YAAY,EAAI,OAAQ,QAAW;AAAA,QACnC,WAAW,CAAC;AAAA,MACd,CAAC,GACD;AACA;AAAA,IACF;AAEA,QAAI,EAAI,SAAS,UAAU,EAAI,SAAS,aAAa;AACnD,YAAM,IAAM,EAAI,OAAQ,OAAU,IAC5B,IAAa,EAAI,OAAQ,QAAW,KACpC,IAAa,EAAI,OAAQ,QAAW,KACpC,IAAwB,CAAC;AAG/B,WAFA,KAEO,IAAI,EAAO,UAAQ;AACxB,cAAM,IAAO,EAAO,CAAA;AACpB,YAAI,EAAK,SAAS,WAAW,EAAK,SAAS,aAAa;AACtD,UAAA;AACA;AAAA,QACF;AACA,YAAI,EAAK,SAAS,UAAU,EAAK,SAAS,YAAY;AACpD,gBAAM,IAAO,EAAK,OAAQ,QAAW;AACrC,UAAA;AACA,cAAI,IAAQ;AACZ,UAAI,IAAI,EAAO,UAAU,EAAO,CAAA,EAAI,SAAS,WAC3C,IAAQ,EAAO,CAAA,EAAI,QAAQ,IAC3B,MAGE,IAAI,EAAO,UAAU,EAAO,CAAA,EAAI,SAAS,WAAS,KACtD,EAAU,KAAK;AAAA,YAAE,MAAA;AAAA,YAAM,OAAA;AAAA,UAAM,CAAC;AAC9B;AAAA,QACF;AACA,QAAA;AAAA,MACF;AAEA,MAAA,EAAO,KAAK;AAAA,QAAE,KAAA;AAAA,QAAK,YAAA;AAAA,QAAY,YAAA;AAAA,QAAY,WAAA;AAAA,MAAU,CAAC;AACtD;AAAA,IACF;AAEA,IAAA;AAAA,EACF;AAEA,SAAO;AAAA,IAAE,QAAQ;AAAA,MAAE,QAAA;AAAA,MAAQ,QAAA;AAAA,IAAO;AAAA,IAAG,KAAK;AAAA,EAAE;AAC9C;AAMA,SAAgB,EAAa,GAA2B;AACtD,QAAM,IAAS,EAAS,CAAG,GACrB,IAAwB,CAAC;AAC/B,MAAI,IAAI;AAER,SAAO,IAAI,EAAO,UAAQ;AACxB,UAAM,IAAM,EAAO,CAAA;AACnB,QAAI,EAAI,SAAS,UAAU,EAAI,SAAS,UAAU;AAChD,YAAM,EAAE,QAAA,GAAQ,KAAA,EAAA,IAAQ,EAAkB,GAAQ,IAAI,CAAC;AACvD,MAAA,EAAQ,KAAK,CAAM,GACnB,IAAI;AACJ;AAAA,IACF;AACA,IAAA;AAAA,EACF;AAEA,SAAO;AACT;AAMA,SAAgB,EAAmB,GAAyB;AAC1D,QAAM,IAAU,EAAa,CAAG;AAChC,MAAI,EAAQ,WAAW,EAAG,OAAM,IAAI,MAAM,uCAAuC;AACjF,SAAO,EAAQ,CAAA;AACjB;AAIA,IAAM,IAAa;AAAA,GACb,IAAgB,0CAChB,IAAS;AAKf,SAAgB,EAAuB,GAA4B;AACjE,QAAM,IAAkB,CAAC,WAAW,CAAA,GAAgB;AACpD,EAAA,EAAM,KAAK,GAAG,CAAA,WAAiB,EAAU,EAAO,MAAM,CAAA,WAAY;AAElE,aAAW,KAAS,EAAO,OACzB,KAAI,EAAe,CAAK,EACtB,CAAA,EAAM,KAAK,GAAG,CAAA,sBAA4B,EAAM,GAAA,KAAQ,EAAU,EAAM,IAAI,CAAA,iBAAkB;AAAA,OACzF;AACL,UAAM,IAAO,EAAM,eAAe,MAAM,MAAM,EAAM,YAC9C,IAAO,EAAM,eAAe,MAAM,MAAM,EAAM;AACpD,IAAA,EAAM,KAAK,GAAG,CAAA,mBAAyB,EAAM,GAAA,WAAc,CAAA,WAAe,CAAA,IAAQ;AAClF,eAAW,KAAM,EAAM,UACrB,CAAA,EAAM,KACJ,GAAG,CAAA,GAAS,CAAA,mBAAyB,EAAG,IAAA,KAAS,EAAU,EAAG,KAAK,CAAA,aACrE;AAEF,IAAA,EAAM,KAAK,GAAG,CAAA,cAAoB;AAAA,EACpC;AAGF,SAAA,EAAM,KAAK,WAAW,GACf,EAAM,KAAK;AAAA,CAAI;AACxB;AAKA,SAAgB,EAAiB,GAA+B;AAC9D,QAAM,IAAkB,CACtB,GACA,eAAe,CAAA,GACjB;AAEA,aAAW,KAAU,GAAS;AAE5B,UAAM,IAAY,EAAuB,CAAM,EAC5C,MAAM;AAAA,CAAI,EACV,IAAA,CAAK,MAAS,IAAS,CAAI,EAC3B,KAAK;AAAA,CAAI;AACZ,IAAA,EAAM,KAAK,CAAS;AAAA,EACtB;AAEA,SAAA,EAAM,KAAK,eAAe,GACnB,EAAM,KAAK;AAAA,CAAI;AACxB"}
1
+ {"version":3,"file":"marcxml.js","names":[],"sources":["../src/marcxml.ts"],"sourcesContent":["/**\n * MARCXML parser and serializer.\n *\n * Supports the Library of Congress MARCXML schema:\n * http://www.loc.gov/MARC21/slim\n *\n * Parsing is done with a hand-rolled state machine — no XML library needed.\n * The MARCXML format is sufficiently regular (fixed element names, no arbitrary\n * nesting) that a full DOM parser is unnecessary.\n */\n\nimport type { MarcRecord, ControlField, DataField, Subfield } from './types';\nimport { isControlField } from './types';\n\n// ─── XML entity handling ─────────────────────────────────────────────────────\n\nconst ENTITY_MAP: ReadonlyMap<string, string> = new Map([\n ['amp', '&'],\n ['lt', '<'],\n ['gt', '>'],\n ['quot', '\"'],\n ['apos', \"'\"],\n]);\n\nfunction unescapeXml(text: string): string {\n return text.replace(/&(?:#x([0-9a-fA-F]+)|#([0-9]+)|([a-zA-Z]+));/g, (_, hex, dec, name) => {\n if (hex !== undefined) {\n const cp = parseInt(hex, 16);\n return cp >= 0 && cp <= 0x10ffff ? String.fromCodePoint(cp) : '�';\n }\n if (dec !== undefined) {\n const cp = parseInt(dec, 10);\n return cp >= 0 && cp <= 0x10ffff ? String.fromCodePoint(cp) : '�';\n }\n return ENTITY_MAP.get(name) ?? _;\n });\n}\n\nfunction escapeXml(text: string): string {\n return text\n // XML 1.0 forbids most C0 control characters in document text. There is no\n // valid XML 1.0 representation for them, so substitute the Unicode\n // replacement character to keep the output well-formed.\n .replace(/[\\x00-\\x08\\x0B\\x0C\\x0E-\\x1F]/g, '�')\n .replace(/&/g, '&amp;')\n .replace(/</g, '&lt;')\n .replace(/>/g, '&gt;')\n .replace(/\"/g, '&quot;')\n // Preserve literal CR through the XML round-trip: XML parsers normalize\n // bare \\r and \\r\\n to \\n, so we must encode CR as a numeric reference.\n .replace(/\\r/g, '&#13;');\n}\n\n// ─── Minimal tokeniser ────────────────────────────────────────────────────────\n\ninterface Token {\n type: 'open' | 'close' | 'self-close' | 'text';\n /** Local name (no namespace prefix) */\n name?: string;\n attrs?: Record<string, string>;\n text?: string;\n}\n\n/**\n * Strip namespace prefix from a tag name, e.g. \"marc:record\" → \"record\".\n */\nfunction localName(raw: string): string {\n const colon = raw.indexOf(':');\n return colon === -1 ? raw : raw.slice(colon + 1);\n}\n\n/**\n * Parse `key=\"value\"` pairs out of an attribute string.\n */\nfunction parseAttrs(attrStr: string): Record<string, string> {\n const attrs: Record<string, string> = {};\n const re = /([a-zA-Z_:][^\\s=]*)\\s*=\\s*(?:\"([^\"]*)\"|'([^']*)')/g;\n let m: RegExpExecArray | null;\n while ((m = re.exec(attrStr)) !== null) {\n const key = localName(m[1]!);\n attrs[key] = unescapeXml(m[2] ?? m[3] ?? '');\n }\n return attrs;\n}\n\n/**\n * Tokenise an XML string into a flat stream of open/close/text tokens.\n * Skips processing instructions, comments, and DOCTYPE declarations.\n * Sufficient for the well-constrained MARCXML format.\n */\nfunction tokenise(xml: string): Token[] {\n const tokens: Token[] = [];\n let i = 0;\n\n while (i < xml.length) {\n const ltPos = xml.indexOf('<', i);\n\n // Text node before next tag\n if (ltPos === -1) {\n const text = xml.slice(i).trim();\n if (text) tokens.push({ type: 'text', text: unescapeXml(xml.slice(i)) });\n break;\n }\n\n if (ltPos > i) {\n const raw = xml.slice(i, ltPos);\n const text = raw.trim();\n if (text) tokens.push({ type: 'text', text: unescapeXml(raw) });\n }\n\n const gtPos = xml.indexOf('>', ltPos);\n if (gtPos === -1) break;\n\n const tag = xml.slice(ltPos + 1, gtPos);\n\n // Skip comments, PIs, DOCTYPE\n if (tag.startsWith('!') || tag.startsWith('?')) {\n i = gtPos + 1;\n continue;\n }\n\n if (tag.startsWith('/')) {\n tokens.push({ type: 'close', name: localName(tag.slice(1).trim()) });\n } else if (tag.endsWith('/')) {\n const inner = tag.slice(0, -1).trim();\n const spaceIdx = inner.search(/\\s/);\n const name = spaceIdx === -1 ? inner : inner.slice(0, spaceIdx);\n const attrStr = spaceIdx === -1 ? '' : inner.slice(spaceIdx);\n tokens.push({ type: 'self-close', name: localName(name), attrs: parseAttrs(attrStr) });\n } else {\n const spaceIdx = tag.search(/\\s/);\n const name = spaceIdx === -1 ? tag : tag.slice(0, spaceIdx);\n const attrStr = spaceIdx === -1 ? '' : tag.slice(spaceIdx);\n tokens.push({ type: 'open', name: localName(name), attrs: parseAttrs(attrStr) });\n }\n\n i = gtPos + 1;\n }\n\n return tokens;\n}\n\n// ─── MARCXML parser ───────────────────────────────────────────────────────────\n\n/**\n * Parse one `<record>` element's worth of tokens into a MarcRecord.\n * Mutates `pos` via the returned index.\n */\nfunction parseRecordTokens(tokens: Token[], start: number): { record: MarcRecord; end: number } {\n let leader = '';\n const fields: (ControlField | DataField)[] = [];\n let i = start;\n\n while (i < tokens.length) {\n const tok = tokens[i]!;\n\n if (tok.type === 'close' && tok.name === 'record') {\n return { record: { leader, fields }, end: i + 1 };\n }\n\n if (tok.type === 'open' && tok.name === 'leader') {\n i++;\n if (i < tokens.length && tokens[i]!.type === 'text') {\n leader = tokens[i]!.text!.trim();\n i++;\n }\n // consume </leader>\n if (i < tokens.length && tokens[i]!.type === 'close') i++;\n continue;\n }\n\n if (tok.type === 'self-close' && tok.name === 'controlfield') {\n fields.push({ tag: tok.attrs?.['tag'] ?? '', data: '' });\n i++;\n continue;\n }\n\n if (tok.type === 'open' && tok.name === 'controlfield') {\n const tag = tok.attrs?.['tag'] ?? '';\n i++;\n let data = '';\n if (i < tokens.length && tokens[i]!.type === 'text') {\n data = tokens[i]!.text ?? '';\n i++;\n }\n // consume </controlfield>\n if (i < tokens.length && tokens[i]!.type === 'close') i++;\n fields.push({ tag, data });\n continue;\n }\n\n if (tok.type === 'self-close' && tok.name === 'datafield') {\n fields.push({\n tag: tok.attrs?.['tag'] ?? '',\n indicator1: tok.attrs?.['ind1'] ?? ' ',\n indicator2: tok.attrs?.['ind2'] ?? ' ',\n subfields: [],\n });\n i++;\n continue;\n }\n\n if (tok.type === 'open' && tok.name === 'datafield') {\n const tag = tok.attrs?.['tag'] ?? '';\n const indicator1 = tok.attrs?.['ind1'] ?? ' ';\n const indicator2 = tok.attrs?.['ind2'] ?? ' ';\n const subfields: Subfield[] = [];\n i++;\n\n while (i < tokens.length) {\n const stok = tokens[i]!;\n if (stok.type === 'close' && stok.name === 'datafield') {\n i++;\n break;\n }\n if (stok.type === 'open' && stok.name === 'subfield') {\n const code = stok.attrs?.['code'] ?? '';\n i++;\n let value = '';\n if (i < tokens.length && tokens[i]!.type === 'text') {\n value = tokens[i]!.text ?? '';\n i++;\n }\n // consume </subfield>\n if (i < tokens.length && tokens[i]!.type === 'close') i++;\n subfields.push({ code, value });\n continue;\n }\n i++;\n }\n\n fields.push({ tag, indicator1, indicator2, subfields });\n continue;\n }\n\n i++;\n }\n\n return { record: { leader, fields }, end: i };\n}\n\n/**\n * Parse a MARCXML string containing one `<collection>` or one bare `<record>`.\n * Returns all records found.\n */\nexport function parseMarcXml(xml: string): MarcRecord[] {\n const tokens = tokenise(xml);\n const records: MarcRecord[] = [];\n let i = 0;\n\n while (i < tokens.length) {\n const tok = tokens[i]!;\n if (tok.type === 'open' && tok.name === 'record') {\n const { record, end } = parseRecordTokens(tokens, i + 1);\n records.push(record);\n i = end;\n continue;\n }\n i++;\n }\n\n return records;\n}\n\n// ─── MARCXML serializer ───────────────────────────────────────────────────────\n\nconst XML_HEADER = '<?xml version=\"1.0\" encoding=\"UTF-8\"?>\\n';\nconst COLLECTION_NS = 'xmlns=\"http://www.loc.gov/MARC21/slim\"';\nconst INDENT = ' ';\n\nfunction serializeMarcXmlRecord(record: MarcRecord): string {\n const lines: string[] = [`<record ${COLLECTION_NS}>`];\n lines.push(`${INDENT}<leader>${escapeXml(record.leader)}</leader>`);\n\n for (const field of record.fields) {\n if (isControlField(field)) {\n lines.push(`${INDENT}<controlfield tag=\"${field.tag}\">${escapeXml(field.data)}</controlfield>`);\n } else {\n const ind1 = field.indicator1 === ' ' ? ' ' : field.indicator1;\n const ind2 = field.indicator2 === ' ' ? ' ' : field.indicator2;\n lines.push(`${INDENT}<datafield tag=\"${field.tag}\" ind1=\"${ind1}\" ind2=\"${ind2}\">`);\n for (const sf of field.subfields) {\n lines.push(\n `${INDENT}${INDENT}<subfield code=\"${sf.code}\">${escapeXml(sf.value)}</subfield>`\n );\n }\n lines.push(`${INDENT}</datafield>`);\n }\n }\n\n lines.push('</record>');\n return lines.join('\\n');\n}\n\n/**\n * Serialize one or more MarcRecords into a MARCXML `<collection>` document.\n */\nexport function serializeMarcXml(records: MarcRecord[]): string {\n const parts: string[] = [\n XML_HEADER,\n `<collection ${COLLECTION_NS}>`,\n ];\n\n for (const record of records) {\n // Indent each record element by one level inside <collection>\n const recordXml = serializeMarcXmlRecord(record)\n .split('\\n')\n .map((line) => INDENT + line)\n .join('\\n');\n parts.push(recordXml);\n }\n\n parts.push('</collection>');\n return parts.join('\\n');\n}\n"],"mappings":";AAgBA,IAAM,IAA0C,oBAAI,IAAI;AAAA,EACtD,CAAC,OAAO,GAAG;AAAA,EACX,CAAC,MAAM,GAAG;AAAA,EACV,CAAC,MAAM,GAAG;AAAA,EACV,CAAC,QAAQ,GAAG;AAAA,EACZ,CAAC,QAAQ,GAAG;AACd,CAAC;AAED,SAAS,EAAY,GAAsB;AACzC,SAAO,EAAK,QAAQ,iDAAA,CAAkD,GAAG,GAAK,GAAK,MAAS;AAC1F,QAAI,MAAQ,QAAW;AACrB,YAAM,IAAK,SAAS,GAAK,EAAE;AAC3B,aAAO,KAAM,KAAK,KAAM,UAAW,OAAO,cAAc,CAAE,IAAI;AAAA,IAChE;AACA,QAAI,MAAQ,QAAW;AACrB,YAAM,IAAK,SAAS,GAAK,EAAE;AAC3B,aAAO,KAAM,KAAK,KAAM,UAAW,OAAO,cAAc,CAAE,IAAI;AAAA,IAChE;AACA,WAAO,EAAW,IAAI,CAAI,KAAK;AAAA,EACjC,CAAC;AACH;AAEA,SAAS,EAAU,GAAsB;AACvC,SAAO,EAIJ,QAAQ,iCAAiC,GAAG,EAC5C,QAAQ,MAAM,OAAO,EACrB,QAAQ,MAAM,MAAM,EACpB,QAAQ,MAAM,MAAM,EACpB,QAAQ,MAAM,QAAQ,EAGtB,QAAQ,OAAO,OAAO;AAC3B;AAeA,SAAS,EAAU,GAAqB;AACtC,QAAM,IAAQ,EAAI,QAAQ,GAAG;AAC7B,SAAO,MAAU,KAAK,IAAM,EAAI,MAAM,IAAQ,CAAC;AACjD;AAKA,SAAS,EAAW,GAAyC;AAC3D,QAAM,IAAgC,CAAC,GACjC,IAAK;AACX,MAAI;AACJ,UAAQ,IAAI,EAAG,KAAK,CAAO,OAAO,QAAM;AACtC,UAAM,IAAM,EAAU,EAAE,CAAA,CAAG;AAC3B,IAAA,EAAM,CAAA,IAAO,EAAY,EAAE,CAAA,KAAM,EAAE,CAAA,KAAM,EAAE;AAAA,EAC7C;AACA,SAAO;AACT;AAOA,SAAS,EAAS,GAAsB;AACtC,QAAM,IAAkB,CAAC;AACzB,MAAI,IAAI;AAER,SAAO,IAAI,EAAI,UAAQ;AACrB,UAAM,IAAQ,EAAI,QAAQ,KAAK,CAAC;AAGhC,QAAI,MAAU,IAAI;AAEhB,MADa,EAAI,MAAM,CAAC,EAAE,KACtB,KAAM,EAAO,KAAK;AAAA,QAAE,MAAM;AAAA,QAAQ,MAAM,EAAY,EAAI,MAAM,CAAC,CAAC;AAAA,MAAE,CAAC;AACvE;AAAA,IACF;AAEA,QAAI,IAAQ,GAAG;AACb,YAAM,IAAM,EAAI,MAAM,GAAG,CAAK;AAE9B,MADa,EAAI,KACb,KAAM,EAAO,KAAK;AAAA,QAAE,MAAM;AAAA,QAAQ,MAAM,EAAY,CAAG;AAAA,MAAE,CAAC;AAAA,IAChE;AAEA,UAAM,IAAQ,EAAI,QAAQ,KAAK,CAAK;AACpC,QAAI,MAAU,GAAI;AAElB,UAAM,IAAM,EAAI,MAAM,IAAQ,GAAG,CAAK;AAGtC,QAAI,EAAI,WAAW,GAAG,KAAK,EAAI,WAAW,GAAG,GAAG;AAC9C,MAAA,IAAI,IAAQ;AACZ;AAAA,IACF;AAEA,QAAI,EAAI,WAAW,GAAG,EACpB,CAAA,EAAO,KAAK;AAAA,MAAE,MAAM;AAAA,MAAS,MAAM,EAAU,EAAI,MAAM,CAAC,EAAE,KAAK,CAAC;AAAA,IAAE,CAAC;AAAA,aAC1D,EAAI,SAAS,GAAG,GAAG;AAC5B,YAAM,IAAQ,EAAI,MAAM,GAAG,EAAE,EAAE,KAAK,GAC9B,IAAW,EAAM,OAAO,IAAI,GAC5B,IAAO,MAAa,KAAK,IAAQ,EAAM,MAAM,GAAG,CAAQ,GACxD,IAAU,MAAa,KAAK,KAAK,EAAM,MAAM,CAAQ;AAC3D,MAAA,EAAO,KAAK;AAAA,QAAE,MAAM;AAAA,QAAc,MAAM,EAAU,CAAI;AAAA,QAAG,OAAO,EAAW,CAAO;AAAA,MAAE,CAAC;AAAA,IACvF,OAAO;AACL,YAAM,IAAW,EAAI,OAAO,IAAI,GAC1B,IAAO,MAAa,KAAK,IAAM,EAAI,MAAM,GAAG,CAAQ,GACpD,IAAU,MAAa,KAAK,KAAK,EAAI,MAAM,CAAQ;AACzD,MAAA,EAAO,KAAK;AAAA,QAAE,MAAM;AAAA,QAAQ,MAAM,EAAU,CAAI;AAAA,QAAG,OAAO,EAAW,CAAO;AAAA,MAAE,CAAC;AAAA,IACjF;AAEA,IAAA,IAAI,IAAQ;AAAA,EACd;AAEA,SAAO;AACT;AAQA,SAAS,EAAkB,GAAiB,GAAoD;AAC9F,MAAI,IAAS;AACb,QAAM,IAAuC,CAAC;AAC9C,MAAI,IAAI;AAER,SAAO,IAAI,EAAO,UAAQ;AACxB,UAAM,IAAM,EAAO,CAAA;AAEnB,QAAI,EAAI,SAAS,WAAW,EAAI,SAAS,SACvC,QAAO;AAAA,MAAE,QAAQ;AAAA,QAAE,QAAA;AAAA,QAAQ,QAAA;AAAA,MAAO;AAAA,MAAG,KAAK,IAAI;AAAA,IAAE;AAGlD,QAAI,EAAI,SAAS,UAAU,EAAI,SAAS,UAAU;AAChD,MAAA,KACI,IAAI,EAAO,UAAU,EAAO,CAAA,EAAI,SAAS,WAC3C,IAAS,EAAO,CAAA,EAAI,KAAM,KAAK,GAC/B,MAGE,IAAI,EAAO,UAAU,EAAO,CAAA,EAAI,SAAS,WAAS;AACtD;AAAA,IACF;AAEA,QAAI,EAAI,SAAS,gBAAgB,EAAI,SAAS,gBAAgB;AAC5D,MAAA,EAAO,KAAK;AAAA,QAAE,KAAK,EAAI,OAAQ,OAAU;AAAA,QAAI,MAAM;AAAA,MAAG,CAAC,GACvD;AACA;AAAA,IACF;AAEA,QAAI,EAAI,SAAS,UAAU,EAAI,SAAS,gBAAgB;AACtD,YAAM,IAAM,EAAI,OAAQ,OAAU;AAClC,MAAA;AACA,UAAI,IAAO;AACX,MAAI,IAAI,EAAO,UAAU,EAAO,CAAA,EAAI,SAAS,WAC3C,IAAO,EAAO,CAAA,EAAI,QAAQ,IAC1B,MAGE,IAAI,EAAO,UAAU,EAAO,CAAA,EAAI,SAAS,WAAS,KACtD,EAAO,KAAK;AAAA,QAAE,KAAA;AAAA,QAAK,MAAA;AAAA,MAAK,CAAC;AACzB;AAAA,IACF;AAEA,QAAI,EAAI,SAAS,gBAAgB,EAAI,SAAS,aAAa;AACzD,MAAA,EAAO,KAAK;AAAA,QACV,KAAK,EAAI,OAAQ,OAAU;AAAA,QAC3B,YAAY,EAAI,OAAQ,QAAW;AAAA,QACnC,YAAY,EAAI,OAAQ,QAAW;AAAA,QACnC,WAAW,CAAC;AAAA,MACd,CAAC,GACD;AACA;AAAA,IACF;AAEA,QAAI,EAAI,SAAS,UAAU,EAAI,SAAS,aAAa;AACnD,YAAM,IAAM,EAAI,OAAQ,OAAU,IAC5B,IAAa,EAAI,OAAQ,QAAW,KACpC,IAAa,EAAI,OAAQ,QAAW,KACpC,IAAwB,CAAC;AAG/B,WAFA,KAEO,IAAI,EAAO,UAAQ;AACxB,cAAM,IAAO,EAAO,CAAA;AACpB,YAAI,EAAK,SAAS,WAAW,EAAK,SAAS,aAAa;AACtD,UAAA;AACA;AAAA,QACF;AACA,YAAI,EAAK,SAAS,UAAU,EAAK,SAAS,YAAY;AACpD,gBAAM,IAAO,EAAK,OAAQ,QAAW;AACrC,UAAA;AACA,cAAI,IAAQ;AACZ,UAAI,IAAI,EAAO,UAAU,EAAO,CAAA,EAAI,SAAS,WAC3C,IAAQ,EAAO,CAAA,EAAI,QAAQ,IAC3B,MAGE,IAAI,EAAO,UAAU,EAAO,CAAA,EAAI,SAAS,WAAS,KACtD,EAAU,KAAK;AAAA,YAAE,MAAA;AAAA,YAAM,OAAA;AAAA,UAAM,CAAC;AAC9B;AAAA,QACF;AACA,QAAA;AAAA,MACF;AAEA,MAAA,EAAO,KAAK;AAAA,QAAE,KAAA;AAAA,QAAK,YAAA;AAAA,QAAY,YAAA;AAAA,QAAY,WAAA;AAAA,MAAU,CAAC;AACtD;AAAA,IACF;AAEA,IAAA;AAAA,EACF;AAEA,SAAO;AAAA,IAAE,QAAQ;AAAA,MAAE,QAAA;AAAA,MAAQ,QAAA;AAAA,IAAO;AAAA,IAAG,KAAK;AAAA,EAAE;AAC9C;AAMA,SAAgB,EAAa,GAA2B;AACtD,QAAM,IAAS,EAAS,CAAG,GACrB,IAAwB,CAAC;AAC/B,MAAI,IAAI;AAER,SAAO,IAAI,EAAO,UAAQ;AACxB,UAAM,IAAM,EAAO,CAAA;AACnB,QAAI,EAAI,SAAS,UAAU,EAAI,SAAS,UAAU;AAChD,YAAM,EAAE,QAAA,GAAQ,KAAA,EAAA,IAAQ,EAAkB,GAAQ,IAAI,CAAC;AACvD,MAAA,EAAQ,KAAK,CAAM,GACnB,IAAI;AACJ;AAAA,IACF;AACA,IAAA;AAAA,EACF;AAEA,SAAO;AACT;AAIA,IAAM,IAAa;AAAA,GACb,IAAgB,0CAChB,IAAS;AAEf,SAAS,EAAuB,GAA4B;AAC1D,QAAM,IAAkB,CAAC,WAAW,CAAA,GAAgB;AACpD,EAAA,EAAM,KAAK,GAAG,CAAA,WAAiB,EAAU,EAAO,MAAM,CAAA,WAAY;AAElE,aAAW,KAAS,EAAO,OACzB,KAAI,EAAe,CAAK,EACtB,CAAA,EAAM,KAAK,GAAG,CAAA,sBAA4B,EAAM,GAAA,KAAQ,EAAU,EAAM,IAAI,CAAA,iBAAkB;AAAA,OACzF;AACL,UAAM,IAAO,EAAM,eAAe,MAAM,MAAM,EAAM,YAC9C,IAAO,EAAM,eAAe,MAAM,MAAM,EAAM;AACpD,IAAA,EAAM,KAAK,GAAG,CAAA,mBAAyB,EAAM,GAAA,WAAc,CAAA,WAAe,CAAA,IAAQ;AAClF,eAAW,KAAM,EAAM,UACrB,CAAA,EAAM,KACJ,GAAG,CAAA,GAAS,CAAA,mBAAyB,EAAG,IAAA,KAAS,EAAU,EAAG,KAAK,CAAA,aACrE;AAEF,IAAA,EAAM,KAAK,GAAG,CAAA,cAAoB;AAAA,EACpC;AAGF,SAAA,EAAM,KAAK,WAAW,GACf,EAAM,KAAK;AAAA,CAAI;AACxB;AAKA,SAAgB,EAAiB,GAA+B;AAC9D,QAAM,IAAkB,CACtB,GACA,eAAe,CAAA,GACjB;AAEA,aAAW,KAAU,GAAS;AAE5B,UAAM,IAAY,EAAuB,CAAM,EAC5C,MAAM;AAAA,CAAI,EACV,IAAA,CAAK,MAAS,IAAS,CAAI,EAC3B,KAAK;AAAA,CAAI;AACZ,IAAA,EAAM,KAAK,CAAS;AAAA,EACtB;AAEA,SAAA,EAAM,KAAK,eAAe,GACnB,EAAM,KAAK;AAAA,CAAI;AACxB"}
package/dist/parser.d.ts CHANGED
@@ -1,43 +1,14 @@
1
- import { MarcRecord, ParseOptions, ParseResult } from './types';
1
+ import { MarcRecord, ParseOptions } from './types';
2
2
  /**
3
- * Parse a MARC21 record from ISO2709 binary format.
3
+ * Parse a concatenated ISO2709 binary stream into an array of MARC records.
4
4
  *
5
- * @param buffer - The binary data to parse (use Uint8Array for browser compatibility)
6
- * @param options - Parsing options (strict mode, max warnings)
7
- * @returns Parse result containing the record and any warnings
5
+ * Records in the stream are separated by 0x1D (RECORD_TERMINATOR). Each slice
6
+ * is parsed with {@link parseMarcRecord}; slices that produce a null record
7
+ * (e.g. due to encoding errors in lenient mode) are silently skipped.
8
8
  *
9
- * @example
10
- * ```typescript
11
- * const buffer = new Uint8Array([...]); // MARC21 binary data
12
- * const result = parseMarcRecord(buffer);
13
- *
14
- * if (result.record) {
15
- * console.log('Parsed successfully');
16
- * }
17
- *
18
- * if (result.warnings.length > 0) {
19
- * console.warn('Warnings:', result.warnings);
20
- * }
21
- * ```
22
- */
23
- export declare function parseMarcRecord(buffer: Uint8Array, options?: ParseOptions): ParseResult;
24
- /**
25
- * Convenience wrapper for strict parsing.
26
- * Throws an error if parsing fails.
27
- *
28
- * @param buffer - The binary data to parse
29
- * @returns The parsed MARC record
30
- * @throws Error if parsing fails
31
- *
32
- * @example
33
- * ```typescript
34
- * try {
35
- * const record = parseMarcRecordStrict(buffer);
36
- * console.log('Title:', title(record));
37
- * } catch (error) {
38
- * console.error('Parsing failed:', error);
39
- * }
40
- * ```
9
+ * @param buffer - Binary data containing one or more concatenated MARC records
10
+ * @param options - Parsing options forwarded to the per-record parser
11
+ * @returns Array of successfully parsed MARC records
41
12
  */
42
- export declare function parseMarcRecordStrict(buffer: Uint8Array): MarcRecord;
13
+ export declare function parseMarcBinary(buffer: Uint8Array, options?: ParseOptions): MarcRecord[];
43
14
  //# sourceMappingURL=parser.d.ts.map
@@ -1,4 +1,4 @@
1
- import { MarcRecord, MarcWarning } from './types';
1
+ import { MarcRecord } from './types';
2
2
  /**
3
3
  * Options for serializing MARC records.
4
4
  */
@@ -13,45 +13,14 @@ export interface SerializeOptions {
13
13
  readonly encoding?: 'utf8' | 'marc8';
14
14
  }
15
15
  /**
16
- * Serialize a MARC record to ISO2709 binary format.
16
+ * Serialize an array of MARC records to a concatenated ISO2709 binary stream.
17
17
  *
18
- * @param record - The MARC record to serialize
19
- * @returns Binary representation of the record (Uint8Array for browser compatibility)
18
+ * Each record is individually serialized (with its own 0x1D terminator) and
19
+ * the results are concatenated into a single Uint8Array.
20
20
  *
21
- * @example
22
- * ```typescript
23
- * const record: MarcRecord = {
24
- * leader: '00000nam 2200000 4500',
25
- * fields: [
26
- * { tag: '001', data: 'ocm12345678' },
27
- * {
28
- * tag: '245',
29
- * indicator1: '1',
30
- * indicator2: '0',
31
- * subfields: [{ code: 'a', value: 'Title' }],
32
- * },
33
- * ],
34
- * };
35
- *
36
- * const buffer = serializeMarcRecord(record);
37
- * // Can now be written to file or transmitted
38
- * ```
39
- */
40
- export declare function serializeMarcRecord(record: MarcRecord, options?: SerializeOptions): Uint8Array;
41
- /**
42
- * Result of {@link serializeMarcRecordWithWarnings}: the serialized bytes
43
- * along with any warnings generated by the encoder (e.g. lossy MARC-8
44
- * substitutions).
45
- */
46
- export interface SerializeResult {
47
- readonly bytes: Uint8Array;
48
- readonly warnings: readonly MarcWarning[];
49
- }
50
- /**
51
- * Serialize a MARC record and surface any warnings generated by the encoder.
52
- * Use this when you need programmatic visibility into lossy encodings — for
53
- * example, MARC-8 output of records containing characters with no MARC-8
54
- * equivalent.
21
+ * @param records - MARC records to serialize
22
+ * @param options - Encoding options forwarded to the per-record serializer
23
+ * @returns Concatenated binary representation of all records
55
24
  */
56
- export declare function serializeMarcRecordWithWarnings(record: MarcRecord, options?: SerializeOptions): SerializeResult;
25
+ export declare function serializeMarcBinary(records: MarcRecord[], options?: SerializeOptions): Uint8Array;
57
26
  //# sourceMappingURL=serializer.d.ts.map
@@ -1,5 +1,4 @@
1
- import { MarcWarning, MarcWarningType, ParseOptions, ParseResult } from './types';
2
- export type { MarcWarning, MarcWarningType, ParseOptions, ParseResult };
1
+ import { MarcWarning, MarcWarningType } from './types';
3
2
  /**
4
3
  * Create a warning object.
5
4
  *
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "marc-ts",
3
- "version": "0.1.0",
3
+ "version": "0.2.0",
4
4
  "description": "TypeScript MARC21 library for Node.js and browsers",
5
5
  "type": "module",
6
6
  "main": "./dist/index.cjs",