v1.0.148: PDF parsing via PyPDF2 on VM (replaces broken PDFBox on Lucee)

This commit is contained in:
2026-06-23 12:18:48 +04:00
parent 6c288b2b0e
commit 2399b01a26
3 changed files with 64 additions and 34 deletions
+46
View File
@@ -57,6 +57,13 @@ class Handler(BaseHTTPRequestHandler):
self.send_header("Access-Control-Allow-Headers", "Content-Type")
self.end_headers()
def do_GET(self):
parsed = urlparse(self.path)
if parsed.path == "/process-v2":
self._handle_process_v2(parsed)
else:
self.send_error(404)
def do_POST(self):
if self.path == "/llm-ops":
self._handle_llm_ops()
@@ -64,6 +71,8 @@ class Handler(BaseHTTPRequestHandler):
self._handle_unzip_upload()
elif self.path == "/upload":
self._handle_upload()
elif self.path == "/parse-pdf":
self._handle_parse_pdf()
else:
self._handle_doc_convert()
@@ -345,6 +354,43 @@ class Handler(BaseHTTPRequestHandler):
else:
self._send_error_cors(502, f"assemble failed: HTTP {resp.status_code}")
def _handle_parse_pdf(self):
"""Распарсить PDF через PyPDF2 (получает base64 тело от Lucee)."""
import base64
from PyPDF2 import PdfReader
import io as io_module
length = int(self.headers.get("Content-Length", 0))
body = self.rfile.read(length).decode("utf-8", errors="ignore").strip()
raw = base64.b64decode(body)
try:
reader = PdfReader(io_module.BytesIO(raw))
elements = []
for page in reader.pages:
text = page.extract_text()
if text:
for line in text.split('\n'):
line = line.strip()
if line:
elements.append({"type": "paragraph", "text": line, "style": ""})
result = {
"ok": True,
"status": "parsed",
"element_count": len(elements),
"elements": elements,
"tables": 0,
"paragraphs": len(elements),
}
except Exception as e:
result = {"ok": False, "status": "error", "error": str(e), "element_count": 0}
self.send_response(200)
self.send_header("Content-Type", "application/json; charset=utf-8")
self.send_header("Access-Control-Allow-Origin", "*")
self.end_headers()
self.wfile.write(json.dumps(result, ensure_ascii=False).encode())
def _handle_unzip_upload(self):
"""Распаковать ZIP и загрузить каждый файл в Lucee. Только файлы из корня архива."""
import zipfile, io