docrev 0.10.1 → 0.11.0

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.
Files changed (167) hide show
  1. package/.gitattributes +1 -1
  2. package/CHANGELOG.md +184 -173
  3. package/PLAN-tables-and-postprocess.md +850 -850
  4. package/README.md +431 -431
  5. package/bin/rev.js +11 -11
  6. package/bin/rev.ts +145 -145
  7. package/completions/rev.bash +127 -127
  8. package/completions/rev.ps1 +210 -210
  9. package/completions/rev.zsh +207 -207
  10. package/dist/lib/anchor-match.d.ts +4 -2
  11. package/dist/lib/anchor-match.d.ts.map +1 -1
  12. package/dist/lib/anchor-match.js +52 -24
  13. package/dist/lib/anchor-match.js.map +1 -1
  14. package/dist/lib/annotations.d.ts +0 -6
  15. package/dist/lib/annotations.d.ts.map +1 -1
  16. package/dist/lib/annotations.js +26 -8
  17. package/dist/lib/annotations.js.map +1 -1
  18. package/dist/lib/build.d.ts.map +1 -1
  19. package/dist/lib/build.js +13 -10
  20. package/dist/lib/build.js.map +1 -1
  21. package/dist/lib/commands/build.d.ts.map +1 -1
  22. package/dist/lib/commands/build.js +5 -0
  23. package/dist/lib/commands/build.js.map +1 -1
  24. package/dist/lib/commands/doi.d.ts.map +1 -1
  25. package/dist/lib/commands/doi.js +14 -2
  26. package/dist/lib/commands/doi.js.map +1 -1
  27. package/dist/lib/commands/sync.d.ts.map +1 -1
  28. package/dist/lib/commands/sync.js +98 -107
  29. package/dist/lib/commands/sync.js.map +1 -1
  30. package/dist/lib/commands/utilities.js +164 -164
  31. package/dist/lib/commands/word-tools.js +8 -8
  32. package/dist/lib/comment-realign.d.ts +12 -21
  33. package/dist/lib/comment-realign.d.ts.map +1 -1
  34. package/dist/lib/comment-realign.js +40 -348
  35. package/dist/lib/comment-realign.js.map +1 -1
  36. package/dist/lib/dependencies.d.ts +10 -0
  37. package/dist/lib/dependencies.d.ts.map +1 -1
  38. package/dist/lib/dependencies.js +41 -9
  39. package/dist/lib/dependencies.js.map +1 -1
  40. package/dist/lib/doi.d.ts.map +1 -1
  41. package/dist/lib/doi.js +68 -80
  42. package/dist/lib/doi.js.map +1 -1
  43. package/dist/lib/errors.d.ts.map +1 -1
  44. package/dist/lib/errors.js +3 -1
  45. package/dist/lib/errors.js.map +1 -1
  46. package/dist/lib/grammar.js +3 -3
  47. package/dist/lib/import.d.ts +7 -4
  48. package/dist/lib/import.d.ts.map +1 -1
  49. package/dist/lib/import.js +58 -62
  50. package/dist/lib/import.js.map +1 -1
  51. package/dist/lib/macro-filter.lua +201 -201
  52. package/dist/lib/ooxml.d.ts +207 -0
  53. package/dist/lib/ooxml.d.ts.map +1 -0
  54. package/dist/lib/ooxml.js +634 -0
  55. package/dist/lib/ooxml.js.map +1 -0
  56. package/dist/lib/pdf-comments.js +44 -44
  57. package/dist/lib/plugins.js +57 -57
  58. package/dist/lib/pptx-color-filter.lua +37 -37
  59. package/dist/lib/pptx-themes.js +115 -115
  60. package/dist/lib/rate-limiter.d.ts +8 -0
  61. package/dist/lib/rate-limiter.d.ts.map +1 -1
  62. package/dist/lib/rate-limiter.js +41 -4
  63. package/dist/lib/rate-limiter.js.map +1 -1
  64. package/dist/lib/sections.d.ts.map +1 -1
  65. package/dist/lib/sections.js +5 -2
  66. package/dist/lib/sections.js.map +1 -1
  67. package/dist/lib/spelling.js +2 -2
  68. package/dist/lib/templates.js +387 -387
  69. package/dist/lib/themes.js +51 -51
  70. package/dist/lib/types.d.ts +9 -1
  71. package/dist/lib/types.d.ts.map +1 -1
  72. package/dist/lib/word-extraction.d.ts.map +1 -1
  73. package/dist/lib/word-extraction.js +48 -307
  74. package/dist/lib/word-extraction.js.map +1 -1
  75. package/dist/lib/word.d.ts.map +1 -1
  76. package/dist/lib/word.js +80 -175
  77. package/dist/lib/word.js.map +1 -1
  78. package/dist/lib/wordcomments.d.ts.map +1 -1
  79. package/dist/lib/wordcomments.js +53 -92
  80. package/dist/lib/wordcomments.js.map +1 -1
  81. package/docs-src/build.py +113 -113
  82. package/docs-src/extra.css +208 -208
  83. package/docs-src/md-to-html.lua +6 -6
  84. package/docs-src/template.html +116 -116
  85. package/eslint.config.js +27 -27
  86. package/lib/anchor-match.ts +307 -276
  87. package/lib/annotations.ts +664 -644
  88. package/lib/build.ts +1770 -1766
  89. package/lib/citations.ts +160 -160
  90. package/lib/commands/build.ts +860 -855
  91. package/lib/commands/citations.ts +515 -515
  92. package/lib/commands/comments.ts +1050 -1050
  93. package/lib/commands/context.ts +176 -176
  94. package/lib/commands/core.ts +309 -309
  95. package/lib/commands/doi.ts +451 -435
  96. package/lib/commands/file-ops.ts +372 -372
  97. package/lib/commands/history.ts +320 -320
  98. package/lib/commands/index.ts +87 -87
  99. package/lib/commands/init.ts +259 -259
  100. package/lib/commands/merge-resolve.ts +378 -378
  101. package/lib/commands/preview.ts +178 -178
  102. package/lib/commands/project-info.ts +244 -244
  103. package/lib/commands/quality.ts +517 -517
  104. package/lib/commands/response.ts +454 -454
  105. package/lib/commands/section-boundaries.ts +82 -82
  106. package/lib/commands/sections.ts +451 -451
  107. package/lib/commands/sync.ts +689 -709
  108. package/lib/commands/text-ops.ts +449 -449
  109. package/lib/commands/utilities.ts +448 -448
  110. package/lib/commands/verify-anchors.ts +272 -272
  111. package/lib/commands/word-tools.ts +340 -340
  112. package/lib/comment-realign.ts +99 -517
  113. package/lib/config.ts +84 -84
  114. package/lib/crossref.ts +781 -781
  115. package/lib/csl.ts +191 -191
  116. package/lib/dependencies.ts +132 -98
  117. package/lib/diff-engine.ts +465 -465
  118. package/lib/doi-cache.ts +115 -115
  119. package/lib/doi.ts +879 -897
  120. package/lib/equations.ts +506 -506
  121. package/lib/errors.ts +350 -346
  122. package/lib/format.ts +541 -541
  123. package/lib/git.ts +326 -326
  124. package/lib/grammar.ts +303 -303
  125. package/lib/image-registry.ts +180 -180
  126. package/lib/import.ts +906 -911
  127. package/lib/journals.ts +543 -543
  128. package/lib/macro-filter.lua +201 -201
  129. package/lib/macros.ts +273 -273
  130. package/lib/merge.ts +633 -633
  131. package/lib/ooxml.ts +768 -0
  132. package/lib/orcid.ts +144 -144
  133. package/lib/pdf-comments.ts +263 -263
  134. package/lib/pdf-import.ts +524 -524
  135. package/lib/plugins.ts +362 -362
  136. package/lib/postprocess.ts +188 -188
  137. package/lib/pptx-color-filter.lua +37 -37
  138. package/lib/pptx-template.ts +469 -469
  139. package/lib/pptx-themes.ts +483 -483
  140. package/lib/protect-restore.ts +520 -520
  141. package/lib/rate-limiter.ts +127 -94
  142. package/lib/response.ts +197 -197
  143. package/lib/restore-references.ts +240 -240
  144. package/lib/review.ts +327 -327
  145. package/lib/schema.ts +488 -488
  146. package/lib/scientific-words.ts +73 -73
  147. package/lib/sections.ts +428 -425
  148. package/lib/slides.ts +756 -756
  149. package/lib/spelling.ts +334 -334
  150. package/lib/templates.ts +526 -526
  151. package/lib/themes.ts +742 -742
  152. package/lib/trackchanges.ts +247 -247
  153. package/lib/tui.ts +450 -450
  154. package/lib/types.ts +558 -550
  155. package/lib/undo.ts +250 -250
  156. package/lib/utils.ts +69 -69
  157. package/lib/variables.ts +179 -179
  158. package/lib/word-extraction.ts +525 -806
  159. package/lib/word.ts +526 -643
  160. package/lib/wordcomments.ts +789 -840
  161. package/mkdocs.yml +64 -64
  162. package/package.json +137 -137
  163. package/scripts/postbuild.js +47 -47
  164. package/skill/REFERENCE.md +539 -539
  165. package/skill/SKILL.md +295 -295
  166. package/tsconfig.json +26 -26
  167. package/types/index.d.ts +525 -525
package/lib/undo.ts CHANGED
@@ -1,250 +1,250 @@
1
- /**
2
- * In-session undo stack for comment/annotation operations
3
- *
4
- * Provides undo/redo functionality during interactive sessions
5
- */
6
-
7
- interface StackEntry {
8
- state: string | object;
9
- description: string;
10
- timestamp: number;
11
- }
12
-
13
- interface HistoryEntry {
14
- description: string;
15
- current: boolean;
16
- index: number;
17
- }
18
-
19
- export interface StackInfo {
20
- position: number;
21
- size: number;
22
- undoSteps: number;
23
- redoSteps: number;
24
- }
25
-
26
- interface UndoStack {
27
- push(state: string | object, description?: string): void;
28
- undo(): StackEntry | null;
29
- redo(): StackEntry | null;
30
- current(): StackEntry | null;
31
- canUndo(): boolean;
32
- canRedo(): boolean;
33
- info(): StackInfo;
34
- history(limit?: number): HistoryEntry[];
35
- getStack(): StackEntry[];
36
- }
37
-
38
- export interface DocumentChange {
39
- text: string;
40
- description: string;
41
- }
42
-
43
- export interface DocumentSession {
44
- getText(): string;
45
- applyChange(newText: string, description: string): void;
46
- undo(): DocumentChange | null;
47
- redo(): DocumentChange | null;
48
- canUndo(): boolean;
49
- canRedo(): boolean;
50
- info(): StackInfo;
51
- history(limit?: number): HistoryEntry[];
52
- }
53
-
54
- /**
55
- * Create an undo stack
56
- */
57
- export function createUndoStack(maxSize: number = 50): UndoStack {
58
- const stack: StackEntry[] = [];
59
- let position = -1;
60
-
61
- return {
62
- /**
63
- * Push a new state onto the stack
64
- */
65
- push(state: string | object, description: string = ''): void {
66
- // Remove any states after current position (for redo)
67
- if (position < stack.length - 1) {
68
- stack.splice(position + 1);
69
- }
70
-
71
- // Add new state
72
- stack.push({
73
- state: typeof state === 'string' ? state : JSON.parse(JSON.stringify(state)),
74
- description,
75
- timestamp: Date.now(),
76
- });
77
-
78
- // Enforce max size
79
- while (stack.length > maxSize) {
80
- stack.shift();
81
- }
82
-
83
- position = stack.length - 1;
84
- },
85
-
86
- /**
87
- * Undo to previous state
88
- */
89
- undo(): StackEntry | null {
90
- if (position <= 0) {
91
- return null;
92
- }
93
-
94
- position--;
95
- return stack[position] || null;
96
- },
97
-
98
- /**
99
- * Redo to next state
100
- */
101
- redo(): StackEntry | null {
102
- if (position >= stack.length - 1) {
103
- return null;
104
- }
105
-
106
- position++;
107
- return stack[position] || null;
108
- },
109
-
110
- /**
111
- * Get current state
112
- */
113
- current(): StackEntry | null {
114
- if (position < 0 || position >= stack.length) {
115
- return null;
116
- }
117
- return stack[position] || null;
118
- },
119
-
120
- /**
121
- * Check if undo is available
122
- */
123
- canUndo(): boolean {
124
- return position > 0;
125
- },
126
-
127
- /**
128
- * Check if redo is available
129
- */
130
- canRedo(): boolean {
131
- return position < stack.length - 1;
132
- },
133
-
134
- /**
135
- * Get stack info
136
- */
137
- info(): StackInfo {
138
- return {
139
- position,
140
- size: stack.length,
141
- undoSteps: position,
142
- redoSteps: stack.length - position - 1,
143
- };
144
- },
145
-
146
- /**
147
- * Get history of changes
148
- */
149
- history(limit: number = 10): HistoryEntry[] {
150
- const start = Math.max(0, position - Math.floor(limit / 2));
151
- const end = Math.min(stack.length, start + limit);
152
-
153
- return stack.slice(start, end).map((item, i) => ({
154
- description: item.description,
155
- current: start + i === position,
156
- index: start + i,
157
- }));
158
- },
159
-
160
- /**
161
- * Get the full stack (for debugging)
162
- */
163
- getStack(): StackEntry[] {
164
- return [...stack];
165
- },
166
- };
167
- }
168
-
169
- /**
170
- * Create a document session with undo support
171
- */
172
- export function createDocumentSession(initialText: string): DocumentSession {
173
- const undoStack = createUndoStack();
174
-
175
- // Save initial state
176
- undoStack.push(initialText, 'Initial state');
177
-
178
- return {
179
- /**
180
- * Get current text
181
- */
182
- getText(): string {
183
- const current = undoStack.current();
184
- return current ? current.state as string : initialText;
185
- },
186
-
187
- /**
188
- * Apply a change
189
- */
190
- applyChange(newText: string, description: string): void {
191
- undoStack.push(newText, description);
192
- },
193
-
194
- /**
195
- * Undo last change
196
- */
197
- undo(): DocumentChange | null {
198
- const result = undoStack.undo();
199
- if (result) {
200
- return {
201
- text: result.state as string,
202
- description: result.description,
203
- };
204
- }
205
- return null;
206
- },
207
-
208
- /**
209
- * Redo last undone change
210
- */
211
- redo(): DocumentChange | null {
212
- const result = undoStack.redo();
213
- if (result) {
214
- return {
215
- text: result.state as string,
216
- description: result.description,
217
- };
218
- }
219
- return null;
220
- },
221
-
222
- /**
223
- * Check if undo is available
224
- */
225
- canUndo(): boolean {
226
- return undoStack.canUndo();
227
- },
228
-
229
- /**
230
- * Check if redo is available
231
- */
232
- canRedo(): boolean {
233
- return undoStack.canRedo();
234
- },
235
-
236
- /**
237
- * Get stack info
238
- */
239
- info(): StackInfo {
240
- return undoStack.info();
241
- },
242
-
243
- /**
244
- * Get change history
245
- */
246
- history(limit: number = 10): HistoryEntry[] {
247
- return undoStack.history(limit);
248
- },
249
- };
250
- }
1
+ /**
2
+ * In-session undo stack for comment/annotation operations
3
+ *
4
+ * Provides undo/redo functionality during interactive sessions
5
+ */
6
+
7
+ interface StackEntry {
8
+ state: string | object;
9
+ description: string;
10
+ timestamp: number;
11
+ }
12
+
13
+ interface HistoryEntry {
14
+ description: string;
15
+ current: boolean;
16
+ index: number;
17
+ }
18
+
19
+ export interface StackInfo {
20
+ position: number;
21
+ size: number;
22
+ undoSteps: number;
23
+ redoSteps: number;
24
+ }
25
+
26
+ interface UndoStack {
27
+ push(state: string | object, description?: string): void;
28
+ undo(): StackEntry | null;
29
+ redo(): StackEntry | null;
30
+ current(): StackEntry | null;
31
+ canUndo(): boolean;
32
+ canRedo(): boolean;
33
+ info(): StackInfo;
34
+ history(limit?: number): HistoryEntry[];
35
+ getStack(): StackEntry[];
36
+ }
37
+
38
+ export interface DocumentChange {
39
+ text: string;
40
+ description: string;
41
+ }
42
+
43
+ export interface DocumentSession {
44
+ getText(): string;
45
+ applyChange(newText: string, description: string): void;
46
+ undo(): DocumentChange | null;
47
+ redo(): DocumentChange | null;
48
+ canUndo(): boolean;
49
+ canRedo(): boolean;
50
+ info(): StackInfo;
51
+ history(limit?: number): HistoryEntry[];
52
+ }
53
+
54
+ /**
55
+ * Create an undo stack
56
+ */
57
+ export function createUndoStack(maxSize: number = 50): UndoStack {
58
+ const stack: StackEntry[] = [];
59
+ let position = -1;
60
+
61
+ return {
62
+ /**
63
+ * Push a new state onto the stack
64
+ */
65
+ push(state: string | object, description: string = ''): void {
66
+ // Remove any states after current position (for redo)
67
+ if (position < stack.length - 1) {
68
+ stack.splice(position + 1);
69
+ }
70
+
71
+ // Add new state
72
+ stack.push({
73
+ state: typeof state === 'string' ? state : JSON.parse(JSON.stringify(state)),
74
+ description,
75
+ timestamp: Date.now(),
76
+ });
77
+
78
+ // Enforce max size
79
+ while (stack.length > maxSize) {
80
+ stack.shift();
81
+ }
82
+
83
+ position = stack.length - 1;
84
+ },
85
+
86
+ /**
87
+ * Undo to previous state
88
+ */
89
+ undo(): StackEntry | null {
90
+ if (position <= 0) {
91
+ return null;
92
+ }
93
+
94
+ position--;
95
+ return stack[position] || null;
96
+ },
97
+
98
+ /**
99
+ * Redo to next state
100
+ */
101
+ redo(): StackEntry | null {
102
+ if (position >= stack.length - 1) {
103
+ return null;
104
+ }
105
+
106
+ position++;
107
+ return stack[position] || null;
108
+ },
109
+
110
+ /**
111
+ * Get current state
112
+ */
113
+ current(): StackEntry | null {
114
+ if (position < 0 || position >= stack.length) {
115
+ return null;
116
+ }
117
+ return stack[position] || null;
118
+ },
119
+
120
+ /**
121
+ * Check if undo is available
122
+ */
123
+ canUndo(): boolean {
124
+ return position > 0;
125
+ },
126
+
127
+ /**
128
+ * Check if redo is available
129
+ */
130
+ canRedo(): boolean {
131
+ return position < stack.length - 1;
132
+ },
133
+
134
+ /**
135
+ * Get stack info
136
+ */
137
+ info(): StackInfo {
138
+ return {
139
+ position,
140
+ size: stack.length,
141
+ undoSteps: position,
142
+ redoSteps: stack.length - position - 1,
143
+ };
144
+ },
145
+
146
+ /**
147
+ * Get history of changes
148
+ */
149
+ history(limit: number = 10): HistoryEntry[] {
150
+ const start = Math.max(0, position - Math.floor(limit / 2));
151
+ const end = Math.min(stack.length, start + limit);
152
+
153
+ return stack.slice(start, end).map((item, i) => ({
154
+ description: item.description,
155
+ current: start + i === position,
156
+ index: start + i,
157
+ }));
158
+ },
159
+
160
+ /**
161
+ * Get the full stack (for debugging)
162
+ */
163
+ getStack(): StackEntry[] {
164
+ return [...stack];
165
+ },
166
+ };
167
+ }
168
+
169
+ /**
170
+ * Create a document session with undo support
171
+ */
172
+ export function createDocumentSession(initialText: string): DocumentSession {
173
+ const undoStack = createUndoStack();
174
+
175
+ // Save initial state
176
+ undoStack.push(initialText, 'Initial state');
177
+
178
+ return {
179
+ /**
180
+ * Get current text
181
+ */
182
+ getText(): string {
183
+ const current = undoStack.current();
184
+ return current ? current.state as string : initialText;
185
+ },
186
+
187
+ /**
188
+ * Apply a change
189
+ */
190
+ applyChange(newText: string, description: string): void {
191
+ undoStack.push(newText, description);
192
+ },
193
+
194
+ /**
195
+ * Undo last change
196
+ */
197
+ undo(): DocumentChange | null {
198
+ const result = undoStack.undo();
199
+ if (result) {
200
+ return {
201
+ text: result.state as string,
202
+ description: result.description,
203
+ };
204
+ }
205
+ return null;
206
+ },
207
+
208
+ /**
209
+ * Redo last undone change
210
+ */
211
+ redo(): DocumentChange | null {
212
+ const result = undoStack.redo();
213
+ if (result) {
214
+ return {
215
+ text: result.state as string,
216
+ description: result.description,
217
+ };
218
+ }
219
+ return null;
220
+ },
221
+
222
+ /**
223
+ * Check if undo is available
224
+ */
225
+ canUndo(): boolean {
226
+ return undoStack.canUndo();
227
+ },
228
+
229
+ /**
230
+ * Check if redo is available
231
+ */
232
+ canRedo(): boolean {
233
+ return undoStack.canRedo();
234
+ },
235
+
236
+ /**
237
+ * Get stack info
238
+ */
239
+ info(): StackInfo {
240
+ return undoStack.info();
241
+ },
242
+
243
+ /**
244
+ * Get change history
245
+ */
246
+ history(limit: number = 10): HistoryEntry[] {
247
+ return undoStack.history(limit);
248
+ },
249
+ };
250
+ }
package/lib/utils.ts CHANGED
@@ -1,69 +1,69 @@
1
- /**
2
- * Shared utility functions
3
- */
4
-
5
- /**
6
- * Count words in text (excluding markdown syntax)
7
- * @param text - Markdown text
8
- * @returns Word count
9
- */
10
- export function countWords(text: string): number {
11
- return text
12
- .replace(/^---[\s\S]*?---/m, '') // Remove YAML frontmatter
13
- .replace(/!\[.*?\]\(.*?\)/g, '') // Remove images
14
- .replace(/\[([^\]]+)\]\([^)]+\)/g, '$1') // Keep link text
15
- .replace(/#+\s*/g, '') // Remove headers
16
- .replace(/\*\*|__|[*_`]/g, '') // Remove formatting
17
- .replace(/```[\s\S]*?```/g, '') // Remove code blocks
18
- .replace(/\{[^}]+\}/g, '') // Remove CriticMarkup and attributes
19
- .replace(/@\w+:\w+/g, '') // Remove cross-references
20
- .replace(/@\w+/g, '') // Remove citations
21
- .replace(/\|[^|]+\|/g, ' ') // Remove table cells
22
- .replace(/[-=]{3,}/g, '') // Remove horizontal rules
23
- .replace(/\n+/g, ' ') // Newlines to spaces
24
- .replace(/\s+/g, ' ') // Collapse multiple spaces
25
- .trim()
26
- .split(/\s+/)
27
- .filter(w => w.length > 0).length;
28
- }
29
-
30
- /**
31
- * Normalize whitespace in text
32
- * @param text - Input text
33
- * @returns Normalized text
34
- */
35
- export function normalizeWhitespace(text: string): string {
36
- return text
37
- .replace(/\r\n/g, '\n')
38
- .replace(/\t/g, ' ')
39
- .replace(/ +/g, ' ')
40
- .trim();
41
- }
42
-
43
- /**
44
- * Escape XML special characters
45
- * @param str - Input string
46
- * @returns XML-safe string
47
- */
48
- export function escapeXml(str: string): string {
49
- return str
50
- .replace(/&/g, '&amp;')
51
- .replace(/</g, '&lt;')
52
- .replace(/>/g, '&gt;')
53
- .replace(/"/g, '&quot;')
54
- .replace(/'/g, '&apos;');
55
- }
56
-
57
- /**
58
- * Escape LaTeX special characters
59
- * @param text - Text to escape
60
- * @returns Escaped text
61
- */
62
- export function escapeLatex(text: string): string {
63
- return text
64
- .replace(/\\/g, '\\textbackslash{}')
65
- .replace(/([#$%&_{}])/g, '\\$1')
66
- .replace(/\^/g, '\\textasciicircum{}')
67
- .replace(/~/g, '\\textasciitilde{}')
68
- .replace(/\n/g, ' ');
69
- }
1
+ /**
2
+ * Shared utility functions
3
+ */
4
+
5
+ /**
6
+ * Count words in text (excluding markdown syntax)
7
+ * @param text - Markdown text
8
+ * @returns Word count
9
+ */
10
+ export function countWords(text: string): number {
11
+ return text
12
+ .replace(/^---[\s\S]*?---/m, '') // Remove YAML frontmatter
13
+ .replace(/!\[.*?\]\(.*?\)/g, '') // Remove images
14
+ .replace(/\[([^\]]+)\]\([^)]+\)/g, '$1') // Keep link text
15
+ .replace(/#+\s*/g, '') // Remove headers
16
+ .replace(/\*\*|__|[*_`]/g, '') // Remove formatting
17
+ .replace(/```[\s\S]*?```/g, '') // Remove code blocks
18
+ .replace(/\{[^}]+\}/g, '') // Remove CriticMarkup and attributes
19
+ .replace(/@\w+:\w+/g, '') // Remove cross-references
20
+ .replace(/@\w+/g, '') // Remove citations
21
+ .replace(/\|[^|]+\|/g, ' ') // Remove table cells
22
+ .replace(/[-=]{3,}/g, '') // Remove horizontal rules
23
+ .replace(/\n+/g, ' ') // Newlines to spaces
24
+ .replace(/\s+/g, ' ') // Collapse multiple spaces
25
+ .trim()
26
+ .split(/\s+/)
27
+ .filter(w => w.length > 0).length;
28
+ }
29
+
30
+ /**
31
+ * Normalize whitespace in text
32
+ * @param text - Input text
33
+ * @returns Normalized text
34
+ */
35
+ export function normalizeWhitespace(text: string): string {
36
+ return text
37
+ .replace(/\r\n/g, '\n')
38
+ .replace(/\t/g, ' ')
39
+ .replace(/ +/g, ' ')
40
+ .trim();
41
+ }
42
+
43
+ /**
44
+ * Escape XML special characters
45
+ * @param str - Input string
46
+ * @returns XML-safe string
47
+ */
48
+ export function escapeXml(str: string): string {
49
+ return str
50
+ .replace(/&/g, '&amp;')
51
+ .replace(/</g, '&lt;')
52
+ .replace(/>/g, '&gt;')
53
+ .replace(/"/g, '&quot;')
54
+ .replace(/'/g, '&apos;');
55
+ }
56
+
57
+ /**
58
+ * Escape LaTeX special characters
59
+ * @param text - Text to escape
60
+ * @returns Escaped text
61
+ */
62
+ export function escapeLatex(text: string): string {
63
+ return text
64
+ .replace(/\\/g, '\\textbackslash{}')
65
+ .replace(/([#$%&_{}])/g, '\\$1')
66
+ .replace(/\^/g, '\\textasciicircum{}')
67
+ .replace(/~/g, '\\textasciitilde{}')
68
+ .replace(/\n/g, ' ');
69
+ }