swadhin / opengrammar public

refactor: simplify deduplicateIssues to fix complexity lint; fix useless spread in context-filter

Swadhin Biswas committed 6 months ago
ca3e50e
1 parent d39705a
commit ca3e50e
2 files changed
+52 -53
Showing 2 changed files+52-53
opengrammar/backend/src/analyzer.tsopengrammar/backend/src/rules/context-filter.ts
▼opengrammar/backend/src/analyzer.ts
+50-45
@@ -88,61 +88,26 @@ export class RuleBasedAnalyzer {
88
88
* When two issues share the same offset+length, keep the higher priority one.
89
89
* When one issue fully contains another, keep the more specific (shorter) one.
90
90
*/
91
- private static deduplicateIssues(issues: Issue[]): Issue[] {
91
+ private static getPriority(type: string): number {
92
92
const PRIORITY: Record<string, number> = { grammar: 4, spelling: 3, clarity: 2, style: 1 };
93
+ return PRIORITY[type] || 0;
94
+ }
93
95
94
- // Sort by offset, then by priority (highest first)
95
- const sorted = issues.sort((a, b) => {
96
+ private static deduplicateIssues(issues: Issue[]): Issue[] {
97
+ const sorted = [...issues].sort((a, b) => {
96
98
if (a.offset !== b.offset) return a.offset - b.offset;
97
- return (PRIORITY[b.type] || 0) - (PRIORITY[a.type] || 0);
99
+ return this.getPriority(b.type) - this.getPriority(a.type);
98
100
});
99
101
100
102
const result: Issue[] = [];
101
- const seenSpans = new Map<string, Issue>(); // key: "offset:length" or normalized original
103
+ const seenSpans = new Map<string, Issue>();
102
104
103
105
for (const issue of sorted) {
104
106
const spanKey = `${issue.offset}:${issue.length}`;
105
- const origKey = issue.original.toLowerCase().trim();
106
-
107
- // Case 1: Exact same span — keep higher priority
108
- if (seenSpans.has(spanKey)) {
109
- const existing = seenSpans.get(spanKey)!;
110
- if ((PRIORITY[issue.type] || 0) > (PRIORITY[existing.type] || 0)) {
111
- // Replace with higher-priority version
112
- const idx = result.indexOf(existing);
113
- if (idx >= 0) result[idx] = issue;
114
- seenSpans.set(spanKey, issue);
115
- }
116
- continue;
117
- }
118
107
119
- // Case 2: Same original text, same suggestion — skip duplicate
120
- const dedupKey = `${origKey}→${(issue.suggestion || '').toLowerCase().trim()}`;
121
- let isDuplicate = false;
122
- for (const seen of result) {
123
- const seenDedupKey = `${seen.original.toLowerCase().trim()}→${(seen.suggestion || '').toLowerCase().trim()}`;
124
- if (dedupKey === seenDedupKey && Math.abs(issue.offset - seen.offset) < 3) {
125
- isDuplicate = true;
126
- break;
127
- }
128
- }
129
- if (isDuplicate) continue;
130
-
131
- // Case 3: Different rules flagging the same original text at the same position
132
- let overlapping = false;
133
- for (const seen of result) {
134
- if (seen.original.toLowerCase().trim() === origKey &&
135
- Math.abs(issue.offset - seen.offset) <= issue.original.length) {
136
- overlapping = true;
137
- // Keep the one with higher priority
138
- if ((PRIORITY[issue.type] || 0) > (PRIORITY[seen.type] || 0)) {
139
- const idx = result.indexOf(seen);
140
- if (idx >= 0) result[idx] = issue;
141
- }
142
- break;
143
- }
144
- }
145
- if (overlapping) continue;
108
+ if (this.handleExactSpanMatch(issue, spanKey, seenSpans, result)) continue;
109
+ if (this.isIdenticalSuggestionDuplicate(issue, result)) continue;
110
+ if (this.handleOverlappingSameText(issue, result)) continue;
146
111
147
112
result.push(issue);
148
113
seenSpans.set(spanKey, issue);
@@ -151,6 +116,46 @@ export class RuleBasedAnalyzer {
151
116
return result;
152
117
}
153
118
119
+ private static handleExactSpanMatch(issue: Issue, spanKey: string, seenSpans: Map<string, Issue>, result: Issue[]): boolean {
120
+ if (!seenSpans.has(spanKey)) return false;
121
+
122
+ const existing = seenSpans.get(spanKey)!;
123
+ if (this.getPriority(issue.type) > this.getPriority(existing.type)) {
124
+ const idx = result.indexOf(existing);
125
+ if (idx >= 0) result[idx] = issue;
126
+ seenSpans.set(spanKey, issue);
127
+ }
128
+ return true;
129
+ }
130
+
131
+ private static isIdenticalSuggestionDuplicate(issue: Issue, result: Issue[]): boolean {
132
+ const origKey = issue.original.toLowerCase().trim();
133
+ const dedupKey = `${origKey}→${(issue.suggestion || '').toLowerCase().trim()}`;
134
+
135
+ for (const seen of result) {
136
+ const seenDedupKey = `${seen.original.toLowerCase().trim()}→${(seen.suggestion || '').toLowerCase().trim()}`;
137
+ if (dedupKey === seenDedupKey && Math.abs(issue.offset - seen.offset) < 3) {
138
+ return true;
139
+ }
140
+ }
141
+ return false;
142
+ }
143
+
144
+ private static handleOverlappingSameText(issue: Issue, result: Issue[]): boolean {
145
+ const origKey = issue.original.toLowerCase().trim();
146
+
147
+ for (const seen of result) {
148
+ if (seen.original.toLowerCase().trim() === origKey && Math.abs(issue.offset - seen.offset) <= issue.original.length) {
149
+ if (this.getPriority(issue.type) > this.getPriority(seen.type)) {
150
+ const idx = result.indexOf(seen);
151
+ if (idx >= 0) result[idx] = issue;
152
+ }
153
+ return true;
154
+ }
155
+ }
156
+ return false;
157
+ }
158
+
154
159
155
160
156
161
private static checkCustomRules(text: string): Issue[] {
157
162
▼opengrammar/backend/src/rules/context-filter.ts
+2-8
@@ -102,10 +102,7 @@ const CONTEXT_MODULES: Record<WritingContext, Set<string>> = {
102
102
'clarity', 'spelling-advanced', 'grammar-advanced',
103
103
]),
104
104
105
- document: new Set([
106
- // Full check — all modules
107
- ...Object.keys(MODULE_PREFIXES),
108
- ]),
105
+ document: new Set(Object.keys(MODULE_PREFIXES)),
109
106
110
107
academic: new Set([
111
108
'grammar', 'spelling', 'confused-words', 'punctuation',
@@ -116,10 +113,7 @@ const CONTEXT_MODULES: Record<WritingContext, Set<string>> = {
116
113
'formatting-idioms', 'spelling-advanced', 'grammar-advanced',
117
114
]),
118
115
119
- general: new Set([
120
- // Everything enabled
121
- ...Object.keys(MODULE_PREFIXES),
122
- ]),
116
+ general: new Set(Object.keys(MODULE_PREFIXES)),
123
117
};
124
118
125
119
/**
126
120