texgen (WIP)

2 files changed, 94 insertions, 9 deletions

diff --git a/texgen.py b/texgen.py
new file mode 100644
index 0000000..851df83
--- /dev/null
+++ b/texgen.py
@@ -0,0 +1,65 @@
+import asyncio
+import contextlib
+import hashlib
+import os
+import shutil
+import typing
+import subprocess
+
+
[email protected]
+def temp_dir(path: str):
+    os.mkdir(path)
+    try:
+        yield path
+    finally:
+        # shutil.rmtree(path)
+        pass
+
+
+class TexGenerationError(Exception):
+    pass
+
+
+class TexGenerator:
+
+    def __init__(self, storage_path: str, temp_path: str, task_timeout):
+        self._storage_path = os.path.abspath(storage_path)
+        self._temp_path = os.path.abspath(temp_path)
+        self._task_timeout = task_timeout
+        for p in (self._temp_path, self._storage_path):
+            try:
+                os.mkdir(p)
+            except FileExistsError:
+                pass
+
+    async def xelatex(self, tex_source: str) -> str:
+        tex_hash = hashlib.sha256(tex_source.encode('utf-8')).hexdigest()
+        cache_file_path = os.path.join(self._storage_path, f'{tex_hash}.pdf')
+        if os.path.exists(cache_file_path):
+            return cache_file_path
+        with temp_dir(os.path.join(self._temp_path, os.urandom(24).hex())) as workdir:
+            job_name = 'texput'
+            with subprocess.Popen(
+                    [
+                        'xelatex',
+                        '-interaction=nonstopmode',
+                        '-halt-on-error',
+                        # this seems not working, as the output file name is always set to default `texput.pdf`
+                        f'-jobname={job_name}',
+                    ],
+                    cwd=workdir,
+                    stdin=subprocess.PIPE,
+                    stdout=subprocess.PIPE,
+                    stderr=subprocess.STDOUT,
+                    text=True,
+                    shell=True,
+            ) as proc:
+                stdout, stderr = proc.communicate(input=tex_source, timeout=self._task_timeout)
+                print('STDOUT', stdout)
+                print('STDERR', stderr)
+                if proc.returncode != 0:
+                    raise TexGenerationError(f'xelatex process exited with non-zero code {proc.returncode}')
+
+        os.rename(os.path.join(workdir, f'{job_name}.pdf'), cache_file_path)
+        return cache_file_path
diff --git a/web.py b/web.py
index 7aadac4..007e93d 100644
--- a/web.py
+++ b/web.py
@@ -2,26 +2,46 @@ import re
 
 import aiohttp
 from fastapi import FastAPI, Response
+from fastapi.responses import FileResponse
 
 import main
+import texgen
 
 app = FastAPI()
 
 utaten_pattern = re.compile(r'[a-z0-9]+')
+tex_generator = texgen.TexGenerator('pdf_cache', 'temp', 20)
 
 
[email protected]("/utaten/{item_id}/pdf")
-async def get_utaten_lyric_pdf(item_id: str):
-    raise NotImplementedError
+class TexSourceGenerationError(Exception):
+    pass
 
 
[email protected]("/utaten/{item_id}.tex")
-async def get_utaten_lyric_pdf(item_id: str, resp: Response):
+async def _get_utaten_tex_source(item_id: str) -> str:
     async with aiohttp.ClientSession() as ses:
         async with ses.get(f'https://utaten.com/lyric/{item_id}/') as r:
             if not r.ok:
-                resp.status_code = 503
-                return
+                raise TexSourceGenerationError('HTTP request failed when reading page source')
             html = await r.text()
-            tex = main.html_to_tex(html)
-            return Response(content=tex, media_type='application/x-tex')
+            return main.html_to_tex(html)
+
+
[email protected]("/utaten/{item_id}.pdf")
+async def get_utaten_lyric_pdf(item_id: str):
+    try:
+        print('_get_utaten_tex_source')
+        tex = await _get_utaten_tex_source(item_id)
+        print('xelatex')
+        pdf_path = await tex_generator.xelatex(tex)
+        return FileResponse(pdf_path, media_type='application/pdf')
+    except texgen.TexGenerationError as e:
+        return Response(content=f'Failed to generate tex file: {e}', status_code=502)
+
+
[email protected]("/utaten/{item_id}.tex")
+async def get_utaten_lyric_tex(item_id: str):
+    try:
+        tex = await _get_utaten_tex_source(item_id)
+        return Response(content=tex, media_type='application/x-tex')
+    except TexSourceGenerationError as e:
+        return Response(content=str(e), status_code=503)