proskomma-core 0.9.10 → 0.9.13
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/.eslintrc.json +79 -0
- package/.gitattributes +2 -0
- package/docs/_config.yml +1 -0
- package/docs/index.md +3 -0
- package/docs/pk_edit_0_9.odt +0 -0
- package/docs/pk_edit_0_9.pdf +0 -0
- package/docs/proskomma_101_slideshow.odp +0 -0
- package/docs/proskomma_101_slideshow.pdf +0 -0
- package/docs/proskomma_edit_html.png +0 -0
- package/docs/proskomma_edit_html.svg +451 -0
- package/docs/proskomma_edit_reference.png +0 -0
- package/docs/proskomma_edit_reference.svg +349 -0
- package/docs/repo_logo.png +0 -0
- package/esm/package.json +3 -0
- package/esm/wrapper.mjs +5 -0
- package/index.cjs +6 -0
- package/package.json +18 -15
- package/src/graph/index.cjs +201 -0
- package/src/graph/lib/do_cv.cjs +249 -0
- package/src/graph/lib/dump.cjs +40 -0
- package/src/graph/lib/remake_blocks.cjs +57 -0
- package/src/graph/lib/sequence_chars.cjs +84 -0
- package/src/graph/lib/sort.cjs +39 -0
- package/src/graph/lib/tribos.cjs +97 -0
- package/src/graph/lib/tribos_expression.cjs +556 -0
- package/src/graph/lib/tribos_ohm_UNUSED.cjs +37 -0
- package/src/graph/lib/tribos_step.cjs +434 -0
- package/src/graph/mutations/add.cjs +91 -0
- package/src/graph/mutations/delete.cjs +61 -0
- package/src/graph/mutations/index.cjs +54 -0
- package/src/graph/mutations/rehash.cjs +17 -0
- package/src/graph/mutations/tags.cjs +120 -0
- package/src/graph/mutations/update.cjs +210 -0
- package/src/graph/mutations/versification.cjs +96 -0
- package/src/graph/queries/block.cjs +228 -0
- package/src/graph/queries/cIndex.cjs +87 -0
- package/src/graph/queries/cell.cjs +42 -0
- package/src/graph/queries/cv.cjs +19 -0
- package/src/graph/queries/cvBook.cjs +16 -0
- package/src/graph/queries/cvChapter.cjs +16 -0
- package/src/graph/queries/cvIndex.cjs +47 -0
- package/src/graph/queries/cvNavigation.cjs +116 -0
- package/src/graph/queries/cvVerseElement.cjs +84 -0
- package/src/graph/queries/cvVerses.cjs +14 -0
- package/src/graph/queries/doc_set.cjs +220 -0
- package/src/graph/queries/document.cjs +555 -0
- package/src/graph/queries/idParts.cjs +30 -0
- package/src/graph/queries/index.cjs +151 -0
- package/src/graph/queries/inputBlockSpec.cjs +17 -0
- package/src/graph/queries/inputItemObject.cjs +12 -0
- package/src/graph/queries/input_att_spec.cjs +14 -0
- package/src/graph/queries/input_key_matches.cjs +11 -0
- package/src/graph/queries/input_key_value.cjs +11 -0
- package/src/graph/queries/input_key_values.cjs +11 -0
- package/src/graph/queries/input_selector_spec.cjs +19 -0
- package/src/graph/queries/item.cjs +71 -0
- package/src/graph/queries/itemGroup.cjs +75 -0
- package/src/graph/queries/key_value.cjs +18 -0
- package/src/graph/queries/kv_entry.cjs +21 -0
- package/src/graph/queries/kv_sequence.cjs +151 -0
- package/src/graph/queries/node.cjs +44 -0
- package/src/graph/queries/orig.cjs +11 -0
- package/src/graph/queries/regex_index.cjs +18 -0
- package/src/graph/queries/row_equals_spec.cjs +11 -0
- package/src/graph/queries/row_match_spec.cjs +11 -0
- package/src/graph/queries/selector_spec.cjs +29 -0
- package/src/graph/queries/sequence.cjs +314 -0
- package/src/graph/queries/table_sequence.cjs +192 -0
- package/src/graph/queries/tree_sequence.cjs +58 -0
- package/src/graph/queries/verseNumber.cjs +44 -0
- package/src/graph/queries/verseRange.cjs +11 -0
- package/src/graph/queries/versification.cjs +25 -0
- package/src/index.cjs +583 -0
- package/src/model/doc_set.cjs +389 -0
- package/src/model/doc_set_helpers/block.cjs +232 -0
- package/src/model/doc_set_helpers/enum.cjs +87 -0
- package/src/model/doc_set_helpers/item.cjs +198 -0
- package/src/model/doc_set_helpers/rehash.cjs +53 -0
- package/src/model/doc_set_helpers/selectors.cjs +58 -0
- package/src/model/doc_set_helpers/serialize.cjs +22 -0
- package/src/model/doc_set_helpers/unsuccinctify.cjs +288 -0
- package/src/model/doc_set_helpers/update.cjs +290 -0
- package/src/model/document.cjs +369 -0
- package/src/model/document_helpers/blocks.cjs +135 -0
- package/src/model/document_helpers/chapter_verse.cjs +291 -0
- package/src/model/document_helpers/pre_enums.cjs +67 -0
- package/src/model/document_helpers/sequences.cjs +205 -0
- package/src/model/document_helpers/serialize_succinct.cjs +55 -0
- package/src/model/document_helpers/succinct_filter.cjs +102 -0
- package/src/model/versifications/eng_vrs.cjs +455 -0
- package/src/model/versifications/index.cjs +38 -0
- package/src/model/versifications/lxx_vrs.cjs +950 -0
- package/src/model/versifications/org_vrs.cjs +155 -0
- package/src/model/versifications/rsc_vrs.cjs +342 -0
- package/src/model/versifications/rso_vrs.cjs +541 -0
- package/src/model/versifications/vul_vrs.cjs +835 -0
- package/src/parser/index.cjs +506 -0
- package/src/parser/lexers/index.cjs +11 -0
- package/src/parser/lexers/lexingRegexes.cjs +45 -0
- package/src/parser/lexers/nodes.cjs +123 -0
- package/src/parser/lexers/object_for_fragment.cjs +123 -0
- package/src/parser/lexers/tsv.cjs +45 -0
- package/src/parser/lexers/usfm.cjs +17 -0
- package/src/parser/lexers/usx/index.cjs +7 -0
- package/src/parser/lexers/usx/usx_lexer.cjs +321 -0
- package/src/parser/lib/tokenize.cjs +37 -0
- package/src/parser/model/index.cjs +637 -0
- package/src/parser/parser_specs.cjs +781 -0
- package/src/pipelines/perf2x/index.cjs +7 -0
- package/src/pipelines/perf2x/perf2PkJsonPipeline.json +41 -0
- package/src/pipelines/perf2x/perf2usfmPipeline.json +65 -0
- package/src/transforms/calculateUsfmChapterPositions.cjs +106 -0
- package/src/transforms/index.cjs +9 -0
- package/src/transforms/perf2PkJson.cjs +534 -0
- package/src/transforms/perf2usfm.cjs +294 -0
- package/src/util/byteArray.cjs +251 -0
- package/src/util/canons.cjs +117 -0
- package/src/util/enums.cjs +41 -0
- package/src/util/generateId.cjs +6 -0
- package/src/util/graftDefs.cjs +15 -0
- package/src/util/index.cjs +29 -0
- package/src/util/inspectSuccinct.cjs +59 -0
- package/src/util/itemDefs.cjs +33 -0
- package/src/util/parserConstantDefs.cjs +30 -0
- package/src/util/scopeDefs.cjs +173 -0
- package/src/util/scriptlike/blocksSpec.cjs +192 -0
- package/src/util/scriptlike/dodson_lexicon.xsl +19 -0
- package/src/util/scriptlike/dodson_lexicon_to_entries.cjs +85 -0
- package/src/util/scriptlike/open_scriptures_hebrew_lexicon.xsl +28 -0
- package/src/util/scriptlike/open_scriptures_hebrew_lexicon_to_entries.cjs +85 -0
- package/src/util/succinct.cjs +159 -0
- package/src/util/tags.cjs +28 -0
- package/src/util/tokenDefs.cjs +27 -0
- package/src/util/versification.cjs +364 -0
- package/dist/index.js +0 -8
- package/dist/index.js.LICENSE.txt +0 -23
|
@@ -0,0 +1,637 @@
|
|
|
1
|
+
const utils = require('../../util/index.cjs');
|
|
2
|
+
const ByteArray = utils.ByteArray;
|
|
3
|
+
const {
|
|
4
|
+
pushSuccinctGraftBytes,
|
|
5
|
+
pushSuccinctScopeBytes,
|
|
6
|
+
pushSuccinctTokenBytes,
|
|
7
|
+
} = utils.succinct;
|
|
8
|
+
const { addTag } = utils.tags;
|
|
9
|
+
const { labelForScope } = utils.scopeDefs;
|
|
10
|
+
const { itemEnum } = utils.itemDefs;
|
|
11
|
+
const { scopeEnum } = utils.scopeDefs;
|
|
12
|
+
const { tokenCategory, tokenEnum } = utils.tokenDefs;
|
|
13
|
+
|
|
14
|
+
const Sequence = class {
|
|
15
|
+
constructor(sType) {
|
|
16
|
+
this.id = utils.generateId();
|
|
17
|
+
this.type = sType;
|
|
18
|
+
this.tags = new Set([]);
|
|
19
|
+
this.blocks = [];
|
|
20
|
+
this.activeScopes = [];
|
|
21
|
+
}
|
|
22
|
+
|
|
23
|
+
addTag(tag) {
|
|
24
|
+
addTag(this.tags, tag);
|
|
25
|
+
}
|
|
26
|
+
|
|
27
|
+
plainText() {
|
|
28
|
+
return this.blocks.map(b => b.plainText()).join('').trim();
|
|
29
|
+
}
|
|
30
|
+
|
|
31
|
+
addItem(i) {
|
|
32
|
+
this.lastBlock().addItem(i);
|
|
33
|
+
}
|
|
34
|
+
|
|
35
|
+
addBlockGraft(g) {
|
|
36
|
+
this.newBlock('hangingGraft');
|
|
37
|
+
this.lastBlock().bg.push(g);
|
|
38
|
+
}
|
|
39
|
+
|
|
40
|
+
lastBlock() {
|
|
41
|
+
if (this.blocks.length === 0) {
|
|
42
|
+
this.newBlock('orphanTokens');
|
|
43
|
+
}
|
|
44
|
+
return this.blocks[this.blocks.length - 1];
|
|
45
|
+
}
|
|
46
|
+
|
|
47
|
+
newBlock(label) {
|
|
48
|
+
if (this.blocks.length > 0 && ['orphanTokens', 'hangingGraft'].includes(this.blocks[this.blocks.length - 1].bs.payload)) {
|
|
49
|
+
this.lastBlock().bs = {
|
|
50
|
+
type: 'scope',
|
|
51
|
+
subType: 'start',
|
|
52
|
+
payload: label,
|
|
53
|
+
};
|
|
54
|
+
} else {
|
|
55
|
+
this.blocks.push(new Block(label));
|
|
56
|
+
}
|
|
57
|
+
}
|
|
58
|
+
|
|
59
|
+
trim() {
|
|
60
|
+
this.blocks.forEach(b => b.trim());
|
|
61
|
+
}
|
|
62
|
+
|
|
63
|
+
reorderSpanWithAtts() {
|
|
64
|
+
this.blocks.forEach(b => b.reorderSpanWithAtts());
|
|
65
|
+
}
|
|
66
|
+
|
|
67
|
+
makeNoteGrafts(parser) {
|
|
68
|
+
this.blocks.forEach(b => b.makeNoteGrafts(parser));
|
|
69
|
+
}
|
|
70
|
+
|
|
71
|
+
close(parser) {
|
|
72
|
+
for (const activeScope of this.activeScopes.filter(() => true).reverse()) {
|
|
73
|
+
this.closeActiveScope(parser, activeScope);
|
|
74
|
+
}
|
|
75
|
+
this.activeScopes = [];
|
|
76
|
+
}
|
|
77
|
+
|
|
78
|
+
closeActiveScope(parser, sc) {
|
|
79
|
+
this.addItem({
|
|
80
|
+
type: 'scope',
|
|
81
|
+
subType: 'end',
|
|
82
|
+
payload: sc.label,
|
|
83
|
+
});
|
|
84
|
+
|
|
85
|
+
if (sc.onEnd) {
|
|
86
|
+
sc.onEnd(parser, sc.label);
|
|
87
|
+
}
|
|
88
|
+
}
|
|
89
|
+
|
|
90
|
+
filterGrafts(options) {
|
|
91
|
+
return this.blocks.map(b => b.filterGrafts(options)).reduce((acc, current) => acc.concat(current), []);
|
|
92
|
+
}
|
|
93
|
+
|
|
94
|
+
filterScopes(options) {
|
|
95
|
+
this.blocks.forEach(b => b.filterScopes(options));
|
|
96
|
+
}
|
|
97
|
+
|
|
98
|
+
text() {
|
|
99
|
+
return this.blocks.map(b => b.text()).join('');
|
|
100
|
+
}
|
|
101
|
+
|
|
102
|
+
addTableScopes() {
|
|
103
|
+
let inTable = false;
|
|
104
|
+
|
|
105
|
+
for (const [blockNo, block] of this.blocks.entries()) {
|
|
106
|
+
if (!inTable && block.bs.payload === 'blockTag/tr') {
|
|
107
|
+
inTable = true;
|
|
108
|
+
this.blocks[blockNo].items.unshift({
|
|
109
|
+
type: 'scope',
|
|
110
|
+
subType: 'start',
|
|
111
|
+
payload: labelForScope('table', []),
|
|
112
|
+
});
|
|
113
|
+
} else if (inTable && block.bs.payload !== 'blockTag/tr') {
|
|
114
|
+
inTable = false;
|
|
115
|
+
this.blocks[(blockNo - 1)].items.push({
|
|
116
|
+
type: 'scope',
|
|
117
|
+
subType: 'end',
|
|
118
|
+
payload: labelForScope('table', []),
|
|
119
|
+
});
|
|
120
|
+
}
|
|
121
|
+
}
|
|
122
|
+
|
|
123
|
+
if (inTable) {
|
|
124
|
+
this.lastBlock().items.push({
|
|
125
|
+
type: 'scope',
|
|
126
|
+
subType: 'end',
|
|
127
|
+
payload: labelForScope('table', []),
|
|
128
|
+
});
|
|
129
|
+
}
|
|
130
|
+
}
|
|
131
|
+
|
|
132
|
+
graftifyIntroductionHeadings(parser) {
|
|
133
|
+
let blockEntries = [...this.blocks.entries()];
|
|
134
|
+
blockEntries.reverse();
|
|
135
|
+
const introHeadingTags = ['iot', 'is'].concat(parser.customTags.introHeading);
|
|
136
|
+
|
|
137
|
+
for (const [n, block] of blockEntries) {
|
|
138
|
+
const blockTag = block.bs.payload.split('/')[1].replace(/[0-9]/g, '');
|
|
139
|
+
|
|
140
|
+
if (introHeadingTags.includes(blockTag)) {
|
|
141
|
+
const headingSequence = new Sequence('heading');
|
|
142
|
+
parser.sequences.heading.push(headingSequence);
|
|
143
|
+
headingSequence.blocks.push(block);
|
|
144
|
+
const headingGraft = {
|
|
145
|
+
type: 'graft',
|
|
146
|
+
subType: 'heading',
|
|
147
|
+
payload: headingSequence.id,
|
|
148
|
+
};
|
|
149
|
+
|
|
150
|
+
if (this.blocks.length < n + 2) {
|
|
151
|
+
this.newBlock('blockTag/hangingGraft');
|
|
152
|
+
}
|
|
153
|
+
this.blocks[n + 1].bg.unshift(headingGraft);
|
|
154
|
+
this.blocks.splice(n, 1);
|
|
155
|
+
} else if (blockTag.startsWith('imt')) {
|
|
156
|
+
const titleType = (blockTag.startsWith('imte') ? 'introEndTitle' : 'introTitle');
|
|
157
|
+
let titleSequence;
|
|
158
|
+
|
|
159
|
+
if (parser.sequences[titleType]) {
|
|
160
|
+
titleSequence = parser.sequences[titleType];
|
|
161
|
+
} else {
|
|
162
|
+
const graftType = (blockTag.startsWith('imte') ? 'endTitle' : 'title');
|
|
163
|
+
titleSequence = new Sequence(graftType);
|
|
164
|
+
parser.sequences[titleType] = titleSequence;
|
|
165
|
+
const titleGraft = {
|
|
166
|
+
type: 'graft',
|
|
167
|
+
subType: graftType,
|
|
168
|
+
payload: titleSequence.id,
|
|
169
|
+
};
|
|
170
|
+
|
|
171
|
+
if (this.blocks.length < n + 2) {
|
|
172
|
+
this.newBlock('blockTag/hangingGraft');
|
|
173
|
+
}
|
|
174
|
+
this.blocks[n + 1].bg.unshift(titleGraft);
|
|
175
|
+
}
|
|
176
|
+
this.blocks.splice(n, 1);
|
|
177
|
+
titleSequence.blocks.unshift(block);
|
|
178
|
+
}
|
|
179
|
+
}
|
|
180
|
+
}
|
|
181
|
+
|
|
182
|
+
moveOrphanScopes() {
|
|
183
|
+
if (this.blocks.length > 1) {
|
|
184
|
+
this.moveOrphanStartScopes();
|
|
185
|
+
this.moveOrphanEndScopes();
|
|
186
|
+
}
|
|
187
|
+
}
|
|
188
|
+
|
|
189
|
+
moveOrphanStartScopes() {
|
|
190
|
+
for (const [blockNo, block] of this.blocks.entries()) {
|
|
191
|
+
if (blockNo >= this.blocks.length - 1) {
|
|
192
|
+
continue;
|
|
193
|
+
}
|
|
194
|
+
|
|
195
|
+
for (const item of [...block.items].reverse()) {
|
|
196
|
+
if (item.subType !== 'start' || item.payload.startsWith('altChapter')) {
|
|
197
|
+
break;
|
|
198
|
+
}
|
|
199
|
+
this.blocks[blockNo + 1].items.unshift(this.blocks[blockNo].items.pop());
|
|
200
|
+
}
|
|
201
|
+
}
|
|
202
|
+
}
|
|
203
|
+
|
|
204
|
+
moveOrphanStartScopes2() {
|
|
205
|
+
for (const [blockNo, block] of this.blocks.entries()) {
|
|
206
|
+
if (blockNo >= this.blocks.length - 1) {
|
|
207
|
+
continue;
|
|
208
|
+
}
|
|
209
|
+
|
|
210
|
+
for (const item of [...block.items].reverse()) {
|
|
211
|
+
if (item.subType !== 'start') {
|
|
212
|
+
break;
|
|
213
|
+
}
|
|
214
|
+
this.blocks[blockNo + 1].items.unshift(this.blocks[blockNo].items.pop());
|
|
215
|
+
}
|
|
216
|
+
}
|
|
217
|
+
}
|
|
218
|
+
|
|
219
|
+
moveOrphanEndScopes() {
|
|
220
|
+
for (const [blockNo, block] of this.blocks.entries()) {
|
|
221
|
+
if (blockNo === 0) {
|
|
222
|
+
continue;
|
|
223
|
+
}
|
|
224
|
+
|
|
225
|
+
for (const item of [...block.items]) {
|
|
226
|
+
if (item.subType !== 'end') {
|
|
227
|
+
break;
|
|
228
|
+
}
|
|
229
|
+
this.blocks[blockNo - 1].items.push(this.blocks[blockNo].items.shift());
|
|
230
|
+
}
|
|
231
|
+
}
|
|
232
|
+
}
|
|
233
|
+
|
|
234
|
+
removeEmptyBlocks(customCanBeEmpty) {
|
|
235
|
+
const canBeEmpty = ['blockTag/b', 'blockTag/ib'].concat(customCanBeEmpty);
|
|
236
|
+
const emptyBlocks = [];
|
|
237
|
+
let changed = false;
|
|
238
|
+
|
|
239
|
+
for (const blockRecord of this.blocks.entries()) {
|
|
240
|
+
if (blockRecord[1].tokens().length === 0 && !canBeEmpty.includes(blockRecord[1].bs.payload)) {
|
|
241
|
+
emptyBlocks.push(blockRecord);
|
|
242
|
+
}
|
|
243
|
+
}
|
|
244
|
+
|
|
245
|
+
for (const [n, block] of emptyBlocks.reverse()) {
|
|
246
|
+
if (n < this.blocks.length - 1) {
|
|
247
|
+
for (const bg of [...block.bg].reverse()) {
|
|
248
|
+
this.blocks[n + 1].bg.unshift(bg);
|
|
249
|
+
}
|
|
250
|
+
|
|
251
|
+
for (const i of block.items.reverse()) {
|
|
252
|
+
this.blocks[n + 1].items.unshift(i);
|
|
253
|
+
}
|
|
254
|
+
this.blocks.splice(n, 1);
|
|
255
|
+
changed = true;
|
|
256
|
+
} else if (block.bg.length === 0 && block.items.length === 0) {
|
|
257
|
+
this.blocks.splice(n, 1);
|
|
258
|
+
changed = true;
|
|
259
|
+
}
|
|
260
|
+
}
|
|
261
|
+
|
|
262
|
+
if (changed) {
|
|
263
|
+
this.removeEmptyBlocks(customCanBeEmpty);
|
|
264
|
+
}
|
|
265
|
+
}
|
|
266
|
+
|
|
267
|
+
removeGraftsToEmptySequences(emptySequences) {
|
|
268
|
+
this.blocks.forEach(b => b.removeGraftsToEmptySequences(emptySequences));
|
|
269
|
+
}
|
|
270
|
+
|
|
271
|
+
succinctifyBlocks(docSet) {
|
|
272
|
+
const ret = [];
|
|
273
|
+
let openScopes = [];
|
|
274
|
+
|
|
275
|
+
const updateOpenScopes = item => {
|
|
276
|
+
if (item.subType === 'start') {
|
|
277
|
+
const existingScopes = openScopes.filter(s => s.payload === item.payload);
|
|
278
|
+
|
|
279
|
+
if (existingScopes.length === 0) {
|
|
280
|
+
openScopes.push(item);
|
|
281
|
+
}
|
|
282
|
+
} else {
|
|
283
|
+
openScopes = openScopes.filter(s => s.payload !== item.payload);
|
|
284
|
+
}
|
|
285
|
+
};
|
|
286
|
+
|
|
287
|
+
let nextToken = 0;
|
|
288
|
+
|
|
289
|
+
for (const block of this.blocks) {
|
|
290
|
+
const contentBA = new ByteArray(block.length);
|
|
291
|
+
const blockGraftsBA = new ByteArray(1);
|
|
292
|
+
const openScopesBA = new ByteArray(1);
|
|
293
|
+
const includedScopesBA = new ByteArray(1);
|
|
294
|
+
const nextTokenBA = new ByteArray(1);
|
|
295
|
+
|
|
296
|
+
nextTokenBA.pushNByte(nextToken);
|
|
297
|
+
|
|
298
|
+
for (const bg of block.bg) {
|
|
299
|
+
this.pushSuccinctGraft(blockGraftsBA, docSet, bg);
|
|
300
|
+
}
|
|
301
|
+
|
|
302
|
+
for (const os of openScopes) {
|
|
303
|
+
this.pushSuccinctScope(openScopesBA, docSet, os);
|
|
304
|
+
}
|
|
305
|
+
|
|
306
|
+
const includedScopes = [];
|
|
307
|
+
|
|
308
|
+
for (const item of block.items) {
|
|
309
|
+
switch (item.type) {
|
|
310
|
+
case 'token':
|
|
311
|
+
this.pushSuccinctToken(contentBA, docSet, item);
|
|
312
|
+
|
|
313
|
+
if (item.subType === 'wordLike') {
|
|
314
|
+
nextToken++;
|
|
315
|
+
}
|
|
316
|
+
break;
|
|
317
|
+
case 'graft':
|
|
318
|
+
this.pushSuccinctGraft(contentBA, docSet, item);
|
|
319
|
+
break;
|
|
320
|
+
case 'scope':
|
|
321
|
+
this.pushSuccinctScope(contentBA, docSet, item);
|
|
322
|
+
updateOpenScopes(item);
|
|
323
|
+
|
|
324
|
+
if (item.subType === 'start') {
|
|
325
|
+
includedScopes.push(item);
|
|
326
|
+
}
|
|
327
|
+
break;
|
|
328
|
+
default:
|
|
329
|
+
throw new Error(`Item type ${item.type} is not handled in succinctifyBlocks`);
|
|
330
|
+
}
|
|
331
|
+
}
|
|
332
|
+
|
|
333
|
+
const blockScopeBA = new ByteArray(10);
|
|
334
|
+
this.pushSuccinctScope(blockScopeBA, docSet, block.bs);
|
|
335
|
+
|
|
336
|
+
for (const is of includedScopes) {
|
|
337
|
+
this.pushSuccinctScope(includedScopesBA, docSet, is);
|
|
338
|
+
}
|
|
339
|
+
contentBA.trim();
|
|
340
|
+
blockGraftsBA.trim();
|
|
341
|
+
blockScopeBA.trim();
|
|
342
|
+
openScopesBA.trim();
|
|
343
|
+
includedScopesBA.trim();
|
|
344
|
+
ret.push({
|
|
345
|
+
c: contentBA,
|
|
346
|
+
bs: blockScopeBA,
|
|
347
|
+
bg: blockGraftsBA,
|
|
348
|
+
os: openScopesBA,
|
|
349
|
+
is: includedScopesBA,
|
|
350
|
+
nt: nextTokenBA,
|
|
351
|
+
});
|
|
352
|
+
}
|
|
353
|
+
return ret;
|
|
354
|
+
}
|
|
355
|
+
|
|
356
|
+
pushSuccinctToken(bA, docSet, item) {
|
|
357
|
+
const charsEnumIndex = docSet.enumForCategoryValue(tokenCategory[item.subType], item.payload);
|
|
358
|
+
pushSuccinctTokenBytes(bA, tokenEnum[item.subType], charsEnumIndex);
|
|
359
|
+
}
|
|
360
|
+
|
|
361
|
+
pushSuccinctGraft(bA, docSet, item) {
|
|
362
|
+
const graftTypeEnumIndex = docSet.enumForCategoryValue('graftTypes', item.subType);
|
|
363
|
+
const seqEnumIndex = docSet.enumForCategoryValue('ids', item.payload);
|
|
364
|
+
pushSuccinctGraftBytes(bA, graftTypeEnumIndex, seqEnumIndex);
|
|
365
|
+
}
|
|
366
|
+
|
|
367
|
+
pushSuccinctScope(bA, docSet, item) {
|
|
368
|
+
const scopeBits = item.payload.split('/');
|
|
369
|
+
const scopeTypeByte = scopeEnum[scopeBits[0]];
|
|
370
|
+
const scopeBitBytes = scopeBits.slice(1).map(b => docSet.enumForCategoryValue('scopeBits', b));
|
|
371
|
+
pushSuccinctScopeBytes(bA, itemEnum[`${item.subType}Scope`], scopeTypeByte, scopeBitBytes);
|
|
372
|
+
}
|
|
373
|
+
};
|
|
374
|
+
const Block = class {
|
|
375
|
+
constructor(blockScope) {
|
|
376
|
+
this.id = utils.generateId();
|
|
377
|
+
this.items = [];
|
|
378
|
+
this.bg = [];
|
|
379
|
+
this.bs = {
|
|
380
|
+
type: 'scope',
|
|
381
|
+
subType: 'start',
|
|
382
|
+
payload: blockScope,
|
|
383
|
+
};
|
|
384
|
+
this.os = [];
|
|
385
|
+
}
|
|
386
|
+
|
|
387
|
+
addItem(i) {
|
|
388
|
+
this.items.push(i);
|
|
389
|
+
}
|
|
390
|
+
|
|
391
|
+
plainText() {
|
|
392
|
+
return this.items.filter(i => i.type === 'token').map(i => i.payload).join('');
|
|
393
|
+
}
|
|
394
|
+
|
|
395
|
+
trim() {
|
|
396
|
+
this.items = this.trimEnd(this.trimStart(this.items));
|
|
397
|
+
}
|
|
398
|
+
|
|
399
|
+
reorderSpanWithAtts() {
|
|
400
|
+
const swaStarts = [];
|
|
401
|
+
|
|
402
|
+
for (const [pos, item] of this.items.entries()) {
|
|
403
|
+
if (item.subType === 'start' && item.payload.startsWith('spanWithAtts')) {
|
|
404
|
+
swaStarts.push(pos + 1);
|
|
405
|
+
}
|
|
406
|
+
}
|
|
407
|
+
|
|
408
|
+
for (const swaStart of swaStarts) {
|
|
409
|
+
let pos = swaStart;
|
|
410
|
+
let tokens = [];
|
|
411
|
+
let scopes = [];
|
|
412
|
+
|
|
413
|
+
// eslint-disable-next-line no-constant-condition
|
|
414
|
+
while (true) {
|
|
415
|
+
if (pos >= this.items.length) {
|
|
416
|
+
break;
|
|
417
|
+
}
|
|
418
|
+
|
|
419
|
+
const item = this.items[pos];
|
|
420
|
+
|
|
421
|
+
if (item.type === 'token') {
|
|
422
|
+
tokens.push(item);
|
|
423
|
+
} else if (item.subType === 'start' && item.payload.startsWith('attribute/spanWithAtts')) {
|
|
424
|
+
scopes.push(item);
|
|
425
|
+
} else {
|
|
426
|
+
break;
|
|
427
|
+
}
|
|
428
|
+
pos++;
|
|
429
|
+
}
|
|
430
|
+
|
|
431
|
+
if (tokens.length !== 0 && scopes.length !== 0) {
|
|
432
|
+
let pos = swaStart;
|
|
433
|
+
|
|
434
|
+
for (const s of scopes) {
|
|
435
|
+
this.items[pos] = s;
|
|
436
|
+
pos++;
|
|
437
|
+
}
|
|
438
|
+
|
|
439
|
+
for (const t of tokens) {
|
|
440
|
+
this.items[pos] = t;
|
|
441
|
+
pos++;
|
|
442
|
+
}
|
|
443
|
+
}
|
|
444
|
+
}
|
|
445
|
+
}
|
|
446
|
+
|
|
447
|
+
inlineToEnd() {
|
|
448
|
+
let toAppend = null;
|
|
449
|
+
|
|
450
|
+
for (const [pos, item] of this.items.entries()) {
|
|
451
|
+
if (item.subType === 'end' && ['inline/f', 'inline/fe', 'inline/x'].includes(item.payload)) {
|
|
452
|
+
toAppend = item;
|
|
453
|
+
this.items.splice(pos, 1);
|
|
454
|
+
break;
|
|
455
|
+
}
|
|
456
|
+
}
|
|
457
|
+
|
|
458
|
+
if (toAppend) {
|
|
459
|
+
this.addItem(toAppend);
|
|
460
|
+
}
|
|
461
|
+
}
|
|
462
|
+
|
|
463
|
+
makeNoteGrafts(parser) {
|
|
464
|
+
const noteStarts = [];
|
|
465
|
+
|
|
466
|
+
for (const [pos, item] of this.items.entries()) {
|
|
467
|
+
if (item.subType === 'start' && (item.payload.startsWith('inline/f') || item.payload.startsWith('inline/x'))) {
|
|
468
|
+
noteStarts.push(pos);
|
|
469
|
+
}
|
|
470
|
+
}
|
|
471
|
+
|
|
472
|
+
for (const noteStart of noteStarts) {
|
|
473
|
+
const noteLabel = this.items[noteStart].payload;
|
|
474
|
+
const callerToken = this.items[noteStart + 1];
|
|
475
|
+
|
|
476
|
+
if (callerToken.type === 'token' && callerToken.payload.length === 1) {
|
|
477
|
+
const callerSequence = new Sequence('noteCaller');
|
|
478
|
+
callerSequence.newBlock(noteLabel);
|
|
479
|
+
callerSequence.addItem(callerToken);
|
|
480
|
+
parser.sequences.noteCaller.push(callerSequence);
|
|
481
|
+
this.items[noteStart + 1] = {
|
|
482
|
+
type: 'graft',
|
|
483
|
+
subType: 'noteCaller',
|
|
484
|
+
payload: callerSequence.id,
|
|
485
|
+
};
|
|
486
|
+
}
|
|
487
|
+
}
|
|
488
|
+
}
|
|
489
|
+
|
|
490
|
+
trimStart(items) {
|
|
491
|
+
if (items.length === 0) {
|
|
492
|
+
return items;
|
|
493
|
+
}
|
|
494
|
+
|
|
495
|
+
const firstItem = items[0];
|
|
496
|
+
|
|
497
|
+
if (['lineSpace', 'eol'].includes(firstItem.subType)) {
|
|
498
|
+
return this.trimStart(items.slice(1));
|
|
499
|
+
}
|
|
500
|
+
|
|
501
|
+
if (firstItem.type === 'token') {
|
|
502
|
+
return items;
|
|
503
|
+
}
|
|
504
|
+
return [firstItem, ...this.trimStart(items.slice(1))];
|
|
505
|
+
}
|
|
506
|
+
|
|
507
|
+
trimEnd(items) {
|
|
508
|
+
if (items.length === 0) {
|
|
509
|
+
return items;
|
|
510
|
+
}
|
|
511
|
+
|
|
512
|
+
const lastItem = items[items.length - 1];
|
|
513
|
+
|
|
514
|
+
if (['lineSpace', 'eol'].includes(lastItem.subType)) {
|
|
515
|
+
return this.trimEnd(items.slice(0, items.length - 1));
|
|
516
|
+
}
|
|
517
|
+
|
|
518
|
+
if (lastItem.type === 'token') {
|
|
519
|
+
return items;
|
|
520
|
+
}
|
|
521
|
+
return [...this.trimEnd(items.slice(0, items.length - 1)), lastItem];
|
|
522
|
+
}
|
|
523
|
+
|
|
524
|
+
filterGrafts(options) {
|
|
525
|
+
// Each graft should be removed or returned
|
|
526
|
+
const ret = [];
|
|
527
|
+
let toRemove = [];
|
|
528
|
+
|
|
529
|
+
for (const [pos, item] of this.grafts()) {
|
|
530
|
+
if (this.graftPassesOptions(item, options)) {
|
|
531
|
+
ret.push(item.payload);
|
|
532
|
+
} else {
|
|
533
|
+
toRemove.push(pos);
|
|
534
|
+
}
|
|
535
|
+
}
|
|
536
|
+
|
|
537
|
+
for (const [count, pos] of Array.from(toRemove.entries())) {
|
|
538
|
+
this.items.splice(pos - count, 1);
|
|
539
|
+
}
|
|
540
|
+
toRemove = [];
|
|
541
|
+
|
|
542
|
+
for (const [pos, item] of this.bg.entries()) {
|
|
543
|
+
if (this.graftPassesOptions(item, options)) {
|
|
544
|
+
ret.push(item.payload);
|
|
545
|
+
} else {
|
|
546
|
+
toRemove.push(pos);
|
|
547
|
+
}
|
|
548
|
+
}
|
|
549
|
+
|
|
550
|
+
for (const [count, pos] of Array.from(toRemove.entries())) {
|
|
551
|
+
this.bg.splice(pos - count, 1);
|
|
552
|
+
}
|
|
553
|
+
return ret;
|
|
554
|
+
}
|
|
555
|
+
|
|
556
|
+
filterScopes(options) {
|
|
557
|
+
const toRemove = [];
|
|
558
|
+
|
|
559
|
+
for (const [pos, item] of this.scopes()) {
|
|
560
|
+
if (!this.scopePassesOptions(item, options)) {
|
|
561
|
+
toRemove.push(pos);
|
|
562
|
+
}
|
|
563
|
+
}
|
|
564
|
+
|
|
565
|
+
for (const [count, pos] of Array.from(toRemove.entries())) {
|
|
566
|
+
this.items.splice(pos - count, 1);
|
|
567
|
+
}
|
|
568
|
+
}
|
|
569
|
+
|
|
570
|
+
graftPassesOptions(item, options) {
|
|
571
|
+
return (
|
|
572
|
+
(!('includeGrafts' in options) || options.includeGrafts.includes(item.subType)) &&
|
|
573
|
+
(!('excludeGrafts' in options) || !options.excludeGrafts.includes(item.subType))
|
|
574
|
+
);
|
|
575
|
+
}
|
|
576
|
+
|
|
577
|
+
scopePassesOptions(item, options) {
|
|
578
|
+
return (
|
|
579
|
+
(!('includeScopes' in options) || this.scopeMatchesOptionArray(item.payload, options.includeScopes)) &&
|
|
580
|
+
(!('excludeScopes' in options) || !this.scopeMatchesOptionArray(item.payload, options.excludeScopes))
|
|
581
|
+
);
|
|
582
|
+
}
|
|
583
|
+
|
|
584
|
+
scopeMatchesOptionArray(itemString, optionArray) {
|
|
585
|
+
for (const optionString of optionArray) {
|
|
586
|
+
if (itemString.startsWith(optionString)) {
|
|
587
|
+
return true;
|
|
588
|
+
}
|
|
589
|
+
}
|
|
590
|
+
return false;
|
|
591
|
+
}
|
|
592
|
+
|
|
593
|
+
removeGraftsToEmptySequences(emptySequences) {
|
|
594
|
+
const ret = [];
|
|
595
|
+
let toRemove = [];
|
|
596
|
+
|
|
597
|
+
for (const [pos, item] of this.grafts()) {
|
|
598
|
+
if (emptySequences.includes(item.payload)) {
|
|
599
|
+
toRemove.push(pos);
|
|
600
|
+
}
|
|
601
|
+
}
|
|
602
|
+
|
|
603
|
+
for (const [count, pos] of Array.from(toRemove.entries())) {
|
|
604
|
+
this.items.splice(pos - count, 1);
|
|
605
|
+
}
|
|
606
|
+
toRemove = [];
|
|
607
|
+
|
|
608
|
+
for (const [pos, item] of this.bg.entries()) {
|
|
609
|
+
if (emptySequences.includes(item.payload)) {
|
|
610
|
+
toRemove.push(pos);
|
|
611
|
+
}
|
|
612
|
+
}
|
|
613
|
+
|
|
614
|
+
for (const [count, pos] of Array.from(toRemove.entries())) {
|
|
615
|
+
this.bg.splice(pos - count, 1);
|
|
616
|
+
}
|
|
617
|
+
return ret;
|
|
618
|
+
}
|
|
619
|
+
|
|
620
|
+
grafts() {
|
|
621
|
+
return Array.from(this.items.entries()).filter(ip => ip[1].type === 'graft');
|
|
622
|
+
}
|
|
623
|
+
|
|
624
|
+
scopes() {
|
|
625
|
+
return Array.from(this.items.entries()).filter(ip => ip[1].type === 'scope');
|
|
626
|
+
}
|
|
627
|
+
|
|
628
|
+
tokens() {
|
|
629
|
+
return Array.from(this.items.entries()).filter(ip => !['scope', 'graft'].includes(ip[1].type));
|
|
630
|
+
}
|
|
631
|
+
|
|
632
|
+
text() {
|
|
633
|
+
return this.tokens().map(t => t[1].payload).join('');
|
|
634
|
+
}
|
|
635
|
+
};
|
|
636
|
+
|
|
637
|
+
module.exports = { Sequence, Block };
|