From 06d89a7e5e5ef56e8186d0dd02b5b7ef1951fe6e Mon Sep 17 00:00:00 2001
From: Armin Weigl <tb46305@gmail.com>
Date: Sat, 20 Mar 2021 15:47:06 +0100
Subject: [PATCH] Fix 500 on invalid file or directory name

This fixes server errors occuring when displaying a commit or tree, which
contains an non decodeable file or directory name
---
 gitsrht/blueprints/api/porcelain.py |  2 +-
 gitsrht/git.py                      | 23 ++++++++++++-----------
 gitsrht/templates/utils.html        | 20 ++++++++++----------
 3 files changed, 23 insertions(+), 22 deletions(-)

diff --git a/gitsrht/blueprints/api/porcelain.py b/gitsrht/blueprints/api/porcelain.py
index 7be92c1..ad45c1a 100644
--- a/gitsrht/blueprints/api/porcelain.py
+++ b/gitsrht/blueprints/api/porcelain.py
@@ -48,7 +48,7 @@ def tree_to_dict(t):
         "short_id": t.short_id,
         "entries": [
             {
-                "name": e.name,
+                "name": e.raw_name.decode("utf-8", "replace"),
                 "id": str(e.id),
                 "type": (e.type_str if hasattr(e, "type_str") else e.type),
                 "mode": e.filemode,
diff --git a/gitsrht/git.py b/gitsrht/git.py
index 669913d..1a4f03e 100644
--- a/gitsrht/git.py
+++ b/gitsrht/git.py
@@ -95,7 +95,7 @@ class AnnotatedTreeEntry:
         self.commit = None
         if entry is not None:
             self.id = entry.id.hex
-            self.name = entry.name
+            self.name = entry.raw_name.decode("utf-8", "replace")
             self.type = (entry.type_str
                     if hasattr(entry, "type_str") else entry.type)
             self.filemode = entry.filemode
@@ -143,28 +143,29 @@ def annotate_tree(repo, tree, commit):
 def _diffstat_name(delta, anchor):
     if delta.status == pygit2.GIT_DELTA_DELETED:
         return Markup(escape(delta.old_file.path))
-    if delta.old_file.path == delta.new_file.path:
+    if delta.old_file.raw_path == delta.new_file.raw_path:
         return Markup(
-                f"<a href='#{escape(anchor)}{escape(delta.old_file.path)}'>" +
-                f"{escape(delta.old_file.path)}" +
+                f"<a href='#{escape(anchor)}{escape(delta.old_file.raw_path.decode('utf-8', 'replace'))}'>" +
+                f"{escape(delta.old_file.raw_path.decode('utf-8', 'replace'))}" +
                 f"</a>")
     # Based on git/diff.c
     pfx_length = 0
-    old_path = delta.old_file.path
-    new_path = delta.new_file.path
+    old_path = delta.old_file.raw_path
+    new_path = delta.new_file.raw_path
     for i in range(max(len(old_path), len(new_path))):
         if i >= len(old_path) or i >= len(new_path):
             break
         if old_path[i] != new_path[i]:
             break
-        if old_path[i] == '/':
+        if old_path[i] == b'/'[0]:
             pfx_length = i + 1
     # TODO: detect common suffix
     if pfx_length != 0:
-        return (f"{delta.old_file.path[:pfx_length]}{{" +
-            f"{delta.old_file.path[pfx_length:]} =&gt; {delta.new_file.path[pfx_length:]}" +
-            f"}}")
-    return f"{delta.old_file.path} => {delta.new_file.path}"
+        return (f"{delta.old_file.raw_path[:pfx_length].decode('utf-8', 'replace')}{{" +
+            f"{delta.old_file.raw_path[pfx_length:].decode('utf-8', 'replace')} =&gt; " +
+            f"{delta.new_file.raw_path[pfx_length:].decode('utf-8', 'replace')}}}")
+    return (f"{delta.old_file.raw_path.decode('utf-8', 'replace')} => " +
+            f"{delta.new_file.raw_path.decode('utf-8', 'replace')}")
 
 def _diffstat_line(delta, patch, anchor):
     name = _diffstat_name(delta, anchor)
diff --git a/gitsrht/templates/utils.html b/gitsrht/templates/utils.html
index 3ca36c7..1e064eb 100644
--- a/gitsrht/templates/utils.html
+++ b/gitsrht/templates/utils.html
@@ -164,24 +164,24 @@ endif %}{% endfor %}
       owner=repo.owner.canonical_name,
       repo=repo.name,
       ref=parent.id.hex,
-      path=patch.delta.old_file.path)}}"
-   id="{{anchor}}{{patch.delta.old_file.path}}"
+      path=patch.delta.old_file.raw_path.decode('utf-8', 'replace'))}}"
+   id="{{anchor}}{{patch.delta.old_file.raw_path.decode('utf-8', 'replace')}}"
   {% if target_blank %}
   target="_blank"
   {% endif %}
- >{{patch.delta.old_file.path}}</a>{#
+ >{{patch.delta.old_file.raw_path.decode('utf-8', 'replace')}}</a>{#
  #}{% endif %} =&gt; {#
  #}<a
    href="{{url_for("repo.tree",
       owner=repo.owner.canonical_name,
       repo=repo.name,
       ref=commit.id.hex,
-      path=patch.delta.new_file.path)}}"
-   id="{{anchor}}{{patch.delta.new_file.path}}"
+      path=patch.delta.new_file.raw_path.decode('utf-8', 'replace'))}}"
+   id="{{anchor}}{{patch.delta.new_file.raw_path.decode('utf-8', 'replace')}}"
   {% if target_blank %}
   target="_blank"
   {% endif %}
- >{{patch.delta.new_file.path}}</a>{#
+ >{{patch.delta.new_file.raw_path.decode('utf-8', 'replace')}}</a>{#
  #} <span class="pull-right"><span class="text-success">+{{patch.line_stats[1]}}</span>{#
  #} <span class="text-danger">-{{patch.line_stats[2]}}</span></span>{%
     if patch.delta.old_file.mode != patch.delta.new_file.mode %}{#
@@ -198,7 +198,7 @@ endif %}{% endfor %}
     owner=repo.owner.canonical_name,
     repo=repo.name,
     ref=parent.id.hex,
-    path=patch.delta.old_file.path)}}#L{{hunk.old_start}}"
+    path=patch.delta.old_file.raw_path.decode('utf-8', 'replace'))}}#L{{hunk.old_start}}"
   {% if target_blank %}
   target="_blank"
   {% endif %}
@@ -209,7 +209,7 @@ endif %}{% endfor %}
     owner=repo.owner.canonical_name,
     repo=repo.name,
     ref=commit.id.hex,
-    path=patch.delta.new_file.path)}}#L{{hunk.new_start}}"
+    path=patch.delta.new_file.raw_path.decode('utf-8', 'replace'))}}#L{{hunk.new_start}}"
   {% if target_blank %}
   target="_blank"
   {% endif %}
@@ -220,8 +220,8 @@ endif %}{% endfor %}
   "+":"text-success",
   "-":"text-danger",
   }).get(line.origin) or ""}}"><a
-    href="#{{anchor}}{{patch.delta.old_file.path}}-{{hunk_index}}-{{loop.index}}"
-    id="{{anchor}}{{patch.delta.old_file.path}}-{{hunk_index}}-{{loop.index}}"
+    href="#{{anchor}}{{patch.delta.old_file.raw_path.decode('utf-8', 'replace')}}-{{hunk_index}}-{{loop.index}}"
+    id="{{anchor}}{{patch.delta.old_file.raw_path.decode('utf-8', 'replace')}}-{{hunk_index}}-{{loop.index}}"
     aria-hidden="true"
     class="lineno"
     style="color: inherit"
-- 
2.38.4