From 109ec9a7d86bcadb16541873d95068f737586692 Mon Sep 17 00:00:00 2001
From: Frederick Yin <fkfd@macaw.me>
Date: Fri, 10 Jul 2020 20:30:54 +0800
Subject: Enhance blob view

Changes:
- set maximum size of displayable blobs
- raw blobs are now of type `bytes`
- do not try to decode handle binary blobs
---
 git-gmi/const.py   |  3 ++-
 git-gmi/gateway.py |  3 ++-
 git-gmi/git.py     | 25 +++++++++++++++++++------
 3 files changed, 23 insertions(+), 8 deletions(-)

diff --git a/git-gmi/const.py b/git-gmi/const.py
index 44818c6..f1f757f 100644
--- a/git-gmi/const.py
+++ b/git-gmi/const.py
@@ -2,4 +2,5 @@ STATUS_SUCCESS = "20"
 STATUS_NOT_FOUND = "51 NOT FOUND"
 STATUS_TEMPORARY_FAILURE = "40 TEMPORARY FAILURE"
 META_GEMINI = "text/gemini"
-META_PLAINTEXT = "text/plain"
\ No newline at end of file
+META_PLAINTEXT = "text/plain"
+MAX_DISPLAYED_BLOB_SIZE = 500 * 1024  # 500KB
diff --git a/git-gmi/gateway.py b/git-gmi/gateway.py
index 29f0268..b0c9b0a 100644
--- a/git-gmi/gateway.py
+++ b/git-gmi/gateway.py
@@ -2,6 +2,7 @@ from git import *
 from const import *
 from config import *
 from os import environ, listdir
+import sys
 
 # be careful when using print(); stdout is passed to the client.
 # this cgi uses \n as newline.
@@ -56,7 +57,7 @@ def handle_cgi_request(path: str, query: str):
         except FileNotFoundError:  # is file
             try:
                 if query == "raw":
-                    print(repo.view_raw_blob(branch, location))
+                    sys.stdout.buffer.write(repo.view_raw_blob(branch, location))
                 else:
                     print(repo.view_blob(branch, location))
             except FileNotFoundError:
diff --git a/git-gmi/git.py b/git-gmi/git.py
index 3de904e..f249ec5 100644
--- a/git-gmi/git.py
+++ b/git-gmi/git.py
@@ -263,16 +263,29 @@ class GitGmiRepo:
             f"{STATUS_SUCCESS} {META_GEMINI}\r\n"
             + self.generate_header()
             + f"## {self.name}/{'/'.join(location)} | {convert_filesize(blob.size)}\n\n"
-            f"=> {blob.name}?raw view raw\n\n"
-            f"```\n"
         )
-        response += blob.data.decode("utf-8") + "\n```"
+
+        if blob.is_binary:
+            response += (
+                "This file seems to be binary. Open link below to download.\n"
+                f"=> {blob.name}?raw download"
+            )
+        elif blob.size < MAX_DISPLAYED_BLOB_SIZE:
+            response += (
+                f"=> {blob.name}?raw view raw\n\n"
+                "```\n" + blob.data.decode("utf-8") + "\n```"
+            )
+        else:
+            response += (
+                "This file is too large to be displayed. Open link below to download.\n"
+                f"=> {blob.name}?raw download\n\n"
+            )
         return response
 
-    def view_raw_blob(self, branch: str, location=[]) -> str:
+    def view_raw_blob(self, branch: str, location=[]) -> bytes:
         blob = self.get_blob(branch, location)
         # if mimetypes can't make out the type, set it to plaintext
         guessed_mimetype = mimetypes.guess_type(blob.name)[0] or META_PLAINTEXT
-        response = f"{STATUS_SUCCESS} {guessed_mimetype}\r\n"
-        response += blob.data.decode("utf-8")
+        response = bytes(f"{STATUS_SUCCESS} {guessed_mimetype}\r\n", encoding="utf-8")
+        response += blob.data
         return response
-- 
cgit v1.2.3