Spaces:
Paused
Paused
Add scale=0.9 to PDF generation for better A4 page fitting
Browse files
app.py
CHANGED
|
@@ -1,10 +1,13 @@
|
|
| 1 |
import os
|
| 2 |
import uuid
|
| 3 |
import tempfile
|
|
|
|
|
|
|
| 4 |
from datetime import datetime
|
| 5 |
from pathlib import Path
|
| 6 |
from typing import Optional
|
| 7 |
|
|
|
|
| 8 |
from fastapi import FastAPI, HTTPException, File, UploadFile
|
| 9 |
from fastapi.responses import JSONResponse, RedirectResponse
|
| 10 |
from fastapi.middleware.cors import CORSMiddleware
|
|
@@ -17,6 +20,10 @@ class HTMLRequest(BaseModel):
|
|
| 17 |
html_content: str
|
| 18 |
|
| 19 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
class PDFResponse(BaseModel):
|
| 21 |
pdf_url: str
|
| 22 |
filename: str
|
|
@@ -24,6 +31,91 @@ class PDFResponse(BaseModel):
|
|
| 24 |
repository_url: str
|
| 25 |
|
| 26 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
async def html_to_pdf_api(html_content: str) -> tuple[str, str]:
|
| 28 |
"""
|
| 29 |
HTMLコンテンツをPDFに変換してHugging Faceデータセットリポジトリにアップロード
|
|
@@ -128,6 +220,7 @@ async def root():
|
|
| 128 |
"storage": f"Hugging Face Dataset Repository: {hf_repo_id}",
|
| 129 |
"endpoints": {
|
| 130 |
"convert": "/convert - HTMLをPDFに変換してHFリポジトリに保存",
|
|
|
|
| 131 |
"files": "/files - HFリポジトリ内のPDFファイル一覧",
|
| 132 |
"docs": "/docs - API仕様書",
|
| 133 |
"health": "/health - ヘルスチェック"
|
|
@@ -192,6 +285,43 @@ async def health_check():
|
|
| 192 |
return {"status": "healthy", "timestamp": datetime.now().isoformat()}
|
| 193 |
|
| 194 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 195 |
@app.get("/files")
|
| 196 |
async def list_files():
|
| 197 |
"""
|
|
|
|
| 1 |
import os
|
| 2 |
import uuid
|
| 3 |
import tempfile
|
| 4 |
+
import json
|
| 5 |
+
import re
|
| 6 |
from datetime import datetime
|
| 7 |
from pathlib import Path
|
| 8 |
from typing import Optional
|
| 9 |
|
| 10 |
+
import httpx
|
| 11 |
from fastapi import FastAPI, HTTPException, File, UploadFile
|
| 12 |
from fastapi.responses import JSONResponse, RedirectResponse
|
| 13 |
from fastapi.middleware.cors import CORSMiddleware
|
|
|
|
| 20 |
html_content: str
|
| 21 |
|
| 22 |
|
| 23 |
+
class YAMLRequest(BaseModel):
|
| 24 |
+
yaml_content: str
|
| 25 |
+
|
| 26 |
+
|
| 27 |
class PDFResponse(BaseModel):
|
| 28 |
pdf_url: str
|
| 29 |
filename: str
|
|
|
|
| 31 |
repository_url: str
|
| 32 |
|
| 33 |
|
| 34 |
+
async def call_gemini_api(yaml_content: str) -> str:
|
| 35 |
+
"""
|
| 36 |
+
Gemini APIを呼び出してYAMLコンテンツからHTMLを生成
|
| 37 |
+
"""
|
| 38 |
+
try:
|
| 39 |
+
# 環境変数から設定を取得
|
| 40 |
+
gemini_api_key = os.getenv("GEMINI_API_KEY")
|
| 41 |
+
model_id = os.getenv("MODEL", "gemini-2.5-flash")
|
| 42 |
+
system_instruction = os.getenv("SYSTEM", "YAMLデータを基にHTMLを生成してください。")
|
| 43 |
+
|
| 44 |
+
if not gemini_api_key:
|
| 45 |
+
raise HTTPException(status_code=500, detail="GEMINI_API_KEY環境変数が設定されていません")
|
| 46 |
+
|
| 47 |
+
# Gemini APIリクエストボディを構築
|
| 48 |
+
request_body = {
|
| 49 |
+
"contents": [
|
| 50 |
+
{
|
| 51 |
+
"role": "user",
|
| 52 |
+
"parts": [
|
| 53 |
+
{
|
| 54 |
+
"text": system_instruction
|
| 55 |
+
},
|
| 56 |
+
]
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"role": "model",
|
| 60 |
+
"parts": [
|
| 61 |
+
{
|
| 62 |
+
"text": "はい、システムインストラクションとして認識しました。\nYAMLデータを基にHTMLを生成いたします。"
|
| 63 |
+
},
|
| 64 |
+
]
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"role": "user",
|
| 68 |
+
"parts": [
|
| 69 |
+
{
|
| 70 |
+
"text": yaml_content
|
| 71 |
+
},
|
| 72 |
+
]
|
| 73 |
+
},
|
| 74 |
+
],
|
| 75 |
+
"generationConfig": {
|
| 76 |
+
"temperature": 0.75,
|
| 77 |
+
"responseMimeType": "text/plain",
|
| 78 |
+
},
|
| 79 |
+
}
|
| 80 |
+
|
| 81 |
+
# Gemini APIを呼び出し
|
| 82 |
+
url = f"https://generativelanguage.googleapis.com/v1beta/models/{model_id}:generateContent?key={gemini_api_key}"
|
| 83 |
+
|
| 84 |
+
async with httpx.AsyncClient(timeout=30.0) as client:
|
| 85 |
+
response = await client.post(url, json=request_body)
|
| 86 |
+
response.raise_for_status()
|
| 87 |
+
|
| 88 |
+
# レスポンスからHTMLを抽出
|
| 89 |
+
result = response.json()
|
| 90 |
+
|
| 91 |
+
# レスポンス構造を解析してHTMLを抽出
|
| 92 |
+
if "candidates" in result and len(result["candidates"]) > 0:
|
| 93 |
+
candidate = result["candidates"][0]
|
| 94 |
+
if "content" in candidate and "parts" in candidate["content"]:
|
| 95 |
+
parts = candidate["content"]["parts"]
|
| 96 |
+
for part in parts:
|
| 97 |
+
if "text" in part:
|
| 98 |
+
text = part["text"]
|
| 99 |
+
# HTMLタグを含む部分を抽出
|
| 100 |
+
html_match = re.search(r'<html.*?>.*?</html>', text, re.DOTALL | re.IGNORECASE)
|
| 101 |
+
if html_match:
|
| 102 |
+
return html_match.group(0)
|
| 103 |
+
# HTML全体がない場合はbodyタグを探す
|
| 104 |
+
body_match = re.search(r'<body.*?>.*?</body>', text, re.DOTALL | re.IGNORECASE)
|
| 105 |
+
if body_match:
|
| 106 |
+
return f"<html><head><meta charset='utf-8'></head>{body_match.group(0)}</html>"
|
| 107 |
+
# それでもない場合は全体をHTMLとして扱う
|
| 108 |
+
if "<" in text and ">" in text:
|
| 109 |
+
return text
|
| 110 |
+
|
| 111 |
+
raise HTTPException(status_code=500, detail="Gemini APIレスポンスからHTMLを抽出できませんでした")
|
| 112 |
+
|
| 113 |
+
except httpx.HTTPError as e:
|
| 114 |
+
raise HTTPException(status_code=500, detail=f"Gemini API呼び出しエラー: {str(e)}")
|
| 115 |
+
except Exception as e:
|
| 116 |
+
raise HTTPException(status_code=500, detail=f"Gemini API処理中にエラーが発生しました: {str(e)}")
|
| 117 |
+
|
| 118 |
+
|
| 119 |
async def html_to_pdf_api(html_content: str) -> tuple[str, str]:
|
| 120 |
"""
|
| 121 |
HTMLコンテンツをPDFに変換してHugging Faceデータセットリポジトリにアップロード
|
|
|
|
| 220 |
"storage": f"Hugging Face Dataset Repository: {hf_repo_id}",
|
| 221 |
"endpoints": {
|
| 222 |
"convert": "/convert - HTMLをPDFに変換してHFリポジトリに保存",
|
| 223 |
+
"convert-yaml": "/convert-yaml - YAMLをGemini APIでHTMLに変換してPDF化",
|
| 224 |
"files": "/files - HFリポジトリ内のPDFファイル一覧",
|
| 225 |
"docs": "/docs - API仕様書",
|
| 226 |
"health": "/health - ヘルスチェック"
|
|
|
|
| 285 |
return {"status": "healthy", "timestamp": datetime.now().isoformat()}
|
| 286 |
|
| 287 |
|
| 288 |
+
@app.post("/convert-yaml", response_model=PDFResponse)
|
| 289 |
+
async def convert_yaml_to_pdf(request: YAMLRequest):
|
| 290 |
+
"""
|
| 291 |
+
YAMLコンテンツをGemini APIでHTMLに変換し、PDFを生成してHugging Faceデータセットリポジトリに保存するエンドポイント
|
| 292 |
+
|
| 293 |
+
- **yaml_content**: 変換するYAMLコンテンツ
|
| 294 |
+
|
| 295 |
+
Returns:
|
| 296 |
+
- **pdf_url**: 生成されたPDFのダウンロードURL(Hugging Face上)
|
| 297 |
+
- **filename**: PDFファイル名
|
| 298 |
+
- **message**: 処理結果メッセージ
|
| 299 |
+
- **repository_url**: リポジトリURL
|
| 300 |
+
"""
|
| 301 |
+
if not request.yaml_content or not request.yaml_content.strip():
|
| 302 |
+
raise HTTPException(status_code=400, detail="YAMLコンテンツが空です")
|
| 303 |
+
|
| 304 |
+
try:
|
| 305 |
+
# YAMLからGemini APIでHTMLを生成
|
| 306 |
+
html_content = await call_gemini_api(request.yaml_content)
|
| 307 |
+
|
| 308 |
+
# 生成されたHTMLをPDFに変換
|
| 309 |
+
filename, pdf_url = await html_to_pdf_api(html_content)
|
| 310 |
+
hf_repo_id = os.getenv("HF_DATASET_REPO_ID")
|
| 311 |
+
repository_url = f"https://huggingface.co/datasets/{hf_repo_id}"
|
| 312 |
+
|
| 313 |
+
return PDFResponse(
|
| 314 |
+
pdf_url=pdf_url,
|
| 315 |
+
filename=filename,
|
| 316 |
+
message=f"✅ YAMLからPDFが正常に生成され、Hugging Faceリポジトリに保存されました: {filename}",
|
| 317 |
+
repository_url=repository_url
|
| 318 |
+
)
|
| 319 |
+
except HTTPException:
|
| 320 |
+
raise
|
| 321 |
+
except Exception as e:
|
| 322 |
+
raise HTTPException(status_code=500, detail=f"処理中にエラーが発生しました: {str(e)}")
|
| 323 |
+
|
| 324 |
+
|
| 325 |
@app.get("/files")
|
| 326 |
async def list_files():
|
| 327 |
"""
|