Spaces:
Paused
Paused
Yu Chen commited on
Commit ·
4567d57
1
Parent(s): 70cd4ae
update the csv corpus to generate / rewrite article
Browse files
public/middle_school_english_vocabulary_overview.csv
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
src/app/api/article-rewrite/route.ts
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import { openai } from '@ai-sdk/openai';
|
| 2 |
+
import { generateText } from 'ai';
|
| 3 |
+
import { langfuse } from '@/lib/langfuse';
|
| 4 |
+
import { AI_CONFIG } from '@/config/ai';
|
| 5 |
+
|
| 6 |
+
export async function POST(req: Request) {
|
| 7 |
+
try {
|
| 8 |
+
const {
|
| 9 |
+
grade,
|
| 10 |
+
unit,
|
| 11 |
+
textbookVocab,
|
| 12 |
+
additionalVocab,
|
| 13 |
+
grammar,
|
| 14 |
+
topic,
|
| 15 |
+
inputArticle,
|
| 16 |
+
} = await req.json();
|
| 17 |
+
|
| 18 |
+
const langfusePrompt = await langfuse.prompt.get('article/rewrite');
|
| 19 |
+
|
| 20 |
+
const prompt = langfusePrompt.prompt
|
| 21 |
+
.replace(/\{\{grade_values\}\}/g, Array.isArray(grade) ? grade.join(', ') : grade)
|
| 22 |
+
.replace(/\{\{unit_values\}\}/g, Array.isArray(unit) ? unit.join(', ') : unit)
|
| 23 |
+
.replace(/\{\{textbook_vocab_values\}\}/g, textbookVocab || '')
|
| 24 |
+
.replace(/\{\{additional_vocab_values\}\}/g, additionalVocab || '')
|
| 25 |
+
.replace(/\{\{grammar_values\}\}/g, Array.isArray(grammar) ? grammar.join(', ') : grammar)
|
| 26 |
+
.replace(/\{\{topic_values\}\}/g, Array.isArray(topic) ? topic.join(', ') : topic)
|
| 27 |
+
.replace(/\{\{input_article_value\}\}/g, inputArticle || '');
|
| 28 |
+
|
| 29 |
+
const result = await generateText({
|
| 30 |
+
model: openai(AI_CONFIG.model),
|
| 31 |
+
prompt,
|
| 32 |
+
});
|
| 33 |
+
|
| 34 |
+
return Response.json({
|
| 35 |
+
article: result.text,
|
| 36 |
+
createdAt: new Date().toISOString(),
|
| 37 |
+
});
|
| 38 |
+
} catch (error) {
|
| 39 |
+
console.error('Error rewriting article:', error);
|
| 40 |
+
return Response.json({ error: 'Failed to rewrite article' }, { status: 500 });
|
| 41 |
+
}
|
| 42 |
+
}
|
| 43 |
+
|
| 44 |
+
|
src/app/api/vocab/route.ts
ADDED
|
@@ -0,0 +1,154 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import fs from 'fs';
|
| 2 |
+
import path from 'path';
|
| 3 |
+
|
| 4 |
+
type VocabRequest = {
|
| 5 |
+
publisher: string[];
|
| 6 |
+
grade: string[];
|
| 7 |
+
unit: string[];
|
| 8 |
+
};
|
| 9 |
+
|
| 10 |
+
// Map UI labels to CSV values
|
| 11 |
+
function mapGradeToCsv(uiGrade: string): string {
|
| 12 |
+
const map: Record<string, string> = {
|
| 13 |
+
'七年級上學期': '七上',
|
| 14 |
+
'七年級下學期': '七下',
|
| 15 |
+
'八年級上學期': '八上',
|
| 16 |
+
'八年級下學期': '八下',
|
| 17 |
+
'九年級上學期': '九上',
|
| 18 |
+
'九年級下學期': '九下',
|
| 19 |
+
};
|
| 20 |
+
return map[uiGrade] ?? uiGrade;
|
| 21 |
+
}
|
| 22 |
+
|
| 23 |
+
function mapUnitToCsv(uiUnit: string): string {
|
| 24 |
+
const normalized = uiUnit.replace(/\s+/g, '');
|
| 25 |
+
const map: Record<string, string> = {
|
| 26 |
+
'第一課': 'lesson 1',
|
| 27 |
+
'第二課': 'lesson 2',
|
| 28 |
+
'第三課': 'lesson 3',
|
| 29 |
+
'第四課': 'lesson 4',
|
| 30 |
+
'第五課': 'lesson 5',
|
| 31 |
+
'第六課': 'lesson 6',
|
| 32 |
+
};
|
| 33 |
+
return map[normalized] ?? uiUnit;
|
| 34 |
+
}
|
| 35 |
+
|
| 36 |
+
function parseCSVLine(line: string): string[] {
|
| 37 |
+
const result: string[] = [];
|
| 38 |
+
let current = '';
|
| 39 |
+
let inQuotes = false;
|
| 40 |
+
let i = 0;
|
| 41 |
+
|
| 42 |
+
while (i < line.length) {
|
| 43 |
+
const char = line[i];
|
| 44 |
+
|
| 45 |
+
if (char === '"') {
|
| 46 |
+
if (inQuotes && line[i + 1] === '"') {
|
| 47 |
+
current += '"';
|
| 48 |
+
i += 2;
|
| 49 |
+
continue;
|
| 50 |
+
} else {
|
| 51 |
+
inQuotes = !inQuotes;
|
| 52 |
+
}
|
| 53 |
+
} else if (char === ',' && !inQuotes) {
|
| 54 |
+
result.push(current.trim());
|
| 55 |
+
current = '';
|
| 56 |
+
} else {
|
| 57 |
+
current += char;
|
| 58 |
+
}
|
| 59 |
+
|
| 60 |
+
i++;
|
| 61 |
+
}
|
| 62 |
+
|
| 63 |
+
result.push(current.trim());
|
| 64 |
+
return result;
|
| 65 |
+
}
|
| 66 |
+
|
| 67 |
+
function parseCSVEntry(lines: string[], startIndex: number): { values: string[]; nextIndex: number } {
|
| 68 |
+
let currentLine = startIndex;
|
| 69 |
+
let fullLine = '';
|
| 70 |
+
let openQuotes = 0;
|
| 71 |
+
|
| 72 |
+
while (currentLine < lines.length) {
|
| 73 |
+
const line = lines[currentLine];
|
| 74 |
+
fullLine += (fullLine ? '\n' : '') + line;
|
| 75 |
+
|
| 76 |
+
for (const ch of line) {
|
| 77 |
+
if (ch === '"') openQuotes++;
|
| 78 |
+
}
|
| 79 |
+
|
| 80 |
+
if (openQuotes % 2 === 0) {
|
| 81 |
+
break;
|
| 82 |
+
}
|
| 83 |
+
|
| 84 |
+
currentLine++;
|
| 85 |
+
}
|
| 86 |
+
|
| 87 |
+
const values = parseCSVLine(fullLine);
|
| 88 |
+
return { values, nextIndex: currentLine + 1 };
|
| 89 |
+
}
|
| 90 |
+
|
| 91 |
+
export async function POST(req: Request) {
|
| 92 |
+
try {
|
| 93 |
+
const { publisher, grade, unit } = (await req.json()) as VocabRequest;
|
| 94 |
+
|
| 95 |
+
const filePath = path.join(process.cwd(), 'public', 'middle_school_english_vocabulary_overview.csv');
|
| 96 |
+
const csvContent = fs.readFileSync(filePath, 'utf-8');
|
| 97 |
+
|
| 98 |
+
const lines = csvContent.split('\n');
|
| 99 |
+
if (lines.length === 0) {
|
| 100 |
+
return Response.json({ vocab: [], text: '' });
|
| 101 |
+
}
|
| 102 |
+
|
| 103 |
+
// Header indices
|
| 104 |
+
const header = parseCSVLine(lines[0]);
|
| 105 |
+
const idxPublisher = header.indexOf('出版社');
|
| 106 |
+
const idxGrade = header.indexOf('年級');
|
| 107 |
+
const idxUnit = header.indexOf('單元');
|
| 108 |
+
const idxWord = header.indexOf('單字');
|
| 109 |
+
|
| 110 |
+
if (idxPublisher === -1 || idxGrade === -1 || idxUnit === -1 || idxWord === -1) {
|
| 111 |
+
return Response.json({ vocab: [], text: '' });
|
| 112 |
+
}
|
| 113 |
+
|
| 114 |
+
const gradeSet = new Set(grade.map(mapGradeToCsv));
|
| 115 |
+
const unitSet = new Set(unit.map(mapUnitToCsv));
|
| 116 |
+
const publisherSet = new Set(publisher);
|
| 117 |
+
|
| 118 |
+
const vocabSet = new Set<string>();
|
| 119 |
+
|
| 120 |
+
let i = 1;
|
| 121 |
+
while (i < lines.length) {
|
| 122 |
+
if (!lines[i] || !lines[i].trim()) {
|
| 123 |
+
i++;
|
| 124 |
+
continue;
|
| 125 |
+
}
|
| 126 |
+
const entry = parseCSVEntry(lines, i);
|
| 127 |
+
const values = entry.values;
|
| 128 |
+
|
| 129 |
+
const recPublisher = values[idxPublisher];
|
| 130 |
+
const recGrade = values[idxGrade];
|
| 131 |
+
const recUnit = values[idxUnit];
|
| 132 |
+
const recWord = values[idxWord];
|
| 133 |
+
|
| 134 |
+
const matchPublisher = publisherSet.size === 0 || publisherSet.has(recPublisher);
|
| 135 |
+
const matchGrade = gradeSet.size === 0 || gradeSet.has(recGrade);
|
| 136 |
+
const matchUnit = unitSet.size === 0 || unitSet.has(recUnit);
|
| 137 |
+
|
| 138 |
+
if (matchPublisher && matchGrade && matchUnit && recWord) {
|
| 139 |
+
vocabSet.add(recWord);
|
| 140 |
+
}
|
| 141 |
+
|
| 142 |
+
i = entry.nextIndex;
|
| 143 |
+
}
|
| 144 |
+
|
| 145 |
+
const vocab = Array.from(vocabSet);
|
| 146 |
+
const text = vocab.join(', ');
|
| 147 |
+
return Response.json({ vocab, text });
|
| 148 |
+
} catch (error) {
|
| 149 |
+
console.error('Error filtering vocab:', error);
|
| 150 |
+
return Response.json({ vocab: [], text: '' }, { status: 500 });
|
| 151 |
+
}
|
| 152 |
+
}
|
| 153 |
+
|
| 154 |
+
|
src/components/article/ArticleGenerationModal.tsx
CHANGED
|
@@ -92,13 +92,37 @@ export function ArticleGenerationModal({
|
|
| 92 |
|
| 93 |
// Update textbook vocabulary when selections change
|
| 94 |
useEffect(() => {
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
|
| 99 |
-
|
| 100 |
-
|
| 101 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 102 |
}, [formData.grade, formData.unit, formData.publisher]);
|
| 103 |
|
| 104 |
const handleCheckboxChange = (field: keyof ArticleGenerationParams, value: string) => {
|
|
|
|
| 92 |
|
| 93 |
// Update textbook vocabulary when selections change
|
| 94 |
useEffect(() => {
|
| 95 |
+
let cancelled = false;
|
| 96 |
+
const fetchVocab = async () => {
|
| 97 |
+
if (formData.grade.length > 0 && formData.unit.length > 0 && formData.publisher.length > 0) {
|
| 98 |
+
try {
|
| 99 |
+
const res = await fetch('/api/vocab', {
|
| 100 |
+
method: 'POST',
|
| 101 |
+
headers: { 'Content-Type': 'application/json' },
|
| 102 |
+
body: JSON.stringify({
|
| 103 |
+
publisher: formData.publisher,
|
| 104 |
+
grade: formData.grade,
|
| 105 |
+
unit: formData.unit,
|
| 106 |
+
}),
|
| 107 |
+
});
|
| 108 |
+
if (!res.ok) throw new Error('Failed to fetch vocab');
|
| 109 |
+
const data: { vocab: string[]; text: string } = await res.json();
|
| 110 |
+
if (!cancelled) {
|
| 111 |
+
setFormData(prev => ({ ...prev, textbookVocab: data.text }));
|
| 112 |
+
}
|
| 113 |
+
} catch (e) {
|
| 114 |
+
if (!cancelled) {
|
| 115 |
+
setFormData(prev => ({ ...prev, textbookVocab: '' }));
|
| 116 |
+
}
|
| 117 |
+
}
|
| 118 |
+
} else {
|
| 119 |
+
if (!cancelled) {
|
| 120 |
+
setFormData(prev => ({ ...prev, textbookVocab: '' }));
|
| 121 |
+
}
|
| 122 |
+
}
|
| 123 |
+
};
|
| 124 |
+
fetchVocab();
|
| 125 |
+
return () => { cancelled = true; };
|
| 126 |
}, [formData.grade, formData.unit, formData.publisher]);
|
| 127 |
|
| 128 |
const handleCheckboxChange = (field: keyof ArticleGenerationParams, value: string) => {
|