Update README

This patch updates README, converting it to markdown, adding more links
and references, and explicitly mention the Python 3 dependency.

.gitignore

@@ -1,3 +1,4 @@
 *.pyc
 __pycache__
-.*.swp
+.*
+!.gitignore

README

@@ -1,56 +0,0 @@
-
-git-arr - A git repository browser
-----------------------------------
-
-git-arr is a git repository browser that can generate static HTML instead of
-having to run dynamically.
-
-It is smaller, with less features and a different set of tradeoffs than
-other similar software, so if you're looking for a robust and featureful git
-browser, please look at gitweb or cgit instead.
-
-However, if you want to generate static HTML at the expense of features, then
-it's probably going to be useful.
-
-It's open source under the MIT licence, please see the LICENSE file for more
-information.
-
-
-Getting started
----------------
-
-You will need Python, and the bottle.py framework (the package is usually
-called python-bottle in most distributions).
-
-If pygments is available, it will be used for syntax highlighting, otherwise
-everything will work fine, just in black and white.
-
-
-First, create a configuration file for your repositories. You can start by
-copying sample.conf, which has the list of the available options.
-
-Then, to generate the output to "/var/www/git-arr/" directory, run:
-
-  $ ./git-arr --config config.conf generate --output /var/www/git-arr/
-
-That's it!
-
-The first time you generate, depending on the size of your repositories, it
-can take some time. Subsequent runs should take less time, as it is smart
-enough to only generate what has changed.
-
-
-You can also use git-arr dynamically, although it's not its intended mode of
-use, by running:
-
-  $ ./git-arr --config config.conf serve
-
-That can be useful when making changes to the software itself.
-
-
-Contact
--------
-
-If you want to report bugs, send patches, or have any questions or comments,
-just let me know at albertito@blitiri.com.ar.
-

README.md

@@ -0,0 +1,65 @@
+
+# git-arr - A git repository browser
+
+[git-arr] is a [git] repository browser that can generate static HTML.
+
+It is smaller, with less features and a different set of tradeoffs than
+other similar software, so if you're looking for a robust and featureful git
+browser, please look at [gitweb] or [cgit] instead.
+
+However, if you want to generate static HTML at the expense of features, then
+it's probably going to be useful.
+
+It's open source under the MIT licence, please see the `LICENSE` file for more
+information.
+
+[git-arr]: https://blitiri.com.ar/p/git-arr/
+[git]: https://git-scm.com/
+[gitweb]: https://git-scm.com/docs/gitweb
+[cgit]: https://git.zx2c4.com/cgit/about/
+
+
+## Getting started
+
+You will need [Python 3], and the [bottle.py] framework (the package is usually
+called `python3-bottle` in most distributions).
+
+If [pygments] is available, it will be used for syntax highlighting, otherwise
+everything will work fine, just in black and white.
+
+
+First, create a configuration file for your repositories. You can start by
+copying `sample.conf`, which has the list of the available options.
+
+Then, to generate the output to `/var/www/git-arr/` directory, run:
+
+```sh
+./git-arr --config config.conf generate --output /var/www/git-arr/
+```
+
+That's it!
+
+The first time you generate, depending on the size of your repositories, it
+can take some time. Subsequent runs should take less time, as it is smart
+enough to only generate what has changed.
+
+You can also use git-arr dynamically, although it's not its intended mode of
+use, by running:
+
+```
+./git-arr --config config.conf serve
+```
+
+That can be useful when making changes to the software itself.
+
+
+[Python 3]: https://www.python.org/
+[bottle.py]: https://bottlepy.org/
+[pygments]: https://pygments.org/
+
+
+## Contact
+
+If you want to report bugs, send patches, or have any questions or comments,
+just let me know at albertito@blitiri.com.ar.
+

TODO

@@ -1,13 +0,0 @@
-
-In no particular order.
-
-- Atom/RSS.
-- Nicer diff:
-  - Better stat section, with nicer handling of filenames. We should switch to
-    --patch-with-raw and parse from that.
-  - Nicer output, don't use pygments but do our own.
-  - Anchors in diff sections so we can link to them.
-- Short symlinks to commits, with configurable length.
-- Handle symlinks properly.
-- "X hours ago" via javascript (only if it's not too ugly).
-

git-arr

@@ -1,22 +1,17 @@
-#!/usr/bin/env python
+#!/usr/bin/env python3
 """
 git-arr: A git web html generator.
 """
 
-from __future__ import print_function
-
+import configparser
 import math
 import optparse
 import os
 import re
 import sys
+from typing import Union
 
-try:
-    import configparser
-except ImportError:
-    import ConfigParser as configparser
-
-import bottle
+import bottle  # type: ignore
 
 import git
 import utils
@@ -26,12 +21,13 @@ import utils
 # Note this assumes they live next to the executable, and that is not a good
 # assumption; but it's good enough for now.
 bottle.TEMPLATE_PATH.insert(
-        0, os.path.abspath(os.path.dirname(sys.argv[0])) + '/views/')
+    0, os.path.abspath(os.path.dirname(sys.argv[0])) + "/views/"
+)
 
 # The path to our static files.
 # Note this assumes they live next to the executable, and that is not a good
 # assumption; but it's good enough for now.
-static_path = os.path.abspath(os.path.dirname(sys.argv[0])) + '/static/'
+static_path = os.path.abspath(os.path.dirname(sys.argv[0])) + "/static/"
 
 
 # The list of repositories is a global variable for convenience. It will be
@@ -46,39 +42,39 @@ def load_config(path):
     as configured.
     """
     defaults = {
-        'tree': 'yes',
-        'rootdiff': 'yes',
-        'desc': '',
-        'recursive': 'no',
-        'prefix': '',
-        'commits_in_summary': '10',
-        'commits_per_page': '50',
-        'max_pages': '250',
-        'web_url': '',
-        'web_url_file': 'web_url',
-        'git_url': '',
-        'git_url_file': 'cloneurl',
-        'embed_markdown': 'yes',
-        'embed_images': 'no',
-        'ignore': '',
-        'generate_patch': 'yes',
+        "tree": "yes",
+        "rootdiff": "yes",
+        "desc": "",
+        "recursive": "no",
+        "prefix": "",
+        "commits_in_summary": "10",
+        "commits_per_page": "50",
+        "max_pages": "250",
+        "web_url": "",
+        "web_url_file": "web_url",
+        "git_url": "",
+        "git_url_file": "cloneurl",
+        "embed_markdown": "yes",
+        "embed_images": "no",
+        "ignore": "",
+        "generate_patch": "yes",
     }
 
-    config = configparser.SafeConfigParser(defaults)
+    config = configparser.ConfigParser(defaults)
     config.read(path)
 
     # Do a first pass for general sanity checking and recursive expansion.
     for s in config.sections():
-        if config.getboolean(s, 'recursive'):
-            root = config.get(s, 'path')
-            prefix = config.get(s, 'prefix')
+        if config.getboolean(s, "recursive"):
+            root = config.get(s, "path")
+            prefix = config.get(s, "prefix")
 
             for path in os.listdir(root):
-                fullpath = find_git_dir(root + '/' + path)
+                fullpath = find_git_dir(root + "/" + path)
                 if not fullpath:
                     continue
 
-                if os.path.exists(fullpath + '/disable_gitweb'):
+                if os.path.exists(fullpath + "/disable_gitweb"):
                     continue
 
                 section = prefix + path
@@ -86,58 +82,60 @@ def load_config(path):
                     continue
 
                 config.add_section(section)
-                for opt, value in config.items(s, raw = True):
+                for opt, value in config.items(s, raw=True):
                     config.set(section, opt, value)
 
-                config.set(section, 'path', fullpath)
-                config.set(section, 'recursive', 'no')
+                config.set(section, "path", fullpath)
+                config.set(section, "recursive", "no")
 
             # This recursive section is no longer useful.
             config.remove_section(s)
 
     for s in config.sections():
-        if config.get(s, 'ignore') and re.search(config.get(s, 'ignore'), s):
+        if config.get(s, "ignore") and re.search(config.get(s, "ignore"), s):
             continue
 
-        fullpath = find_git_dir(config.get(s, 'path'))
+        fullpath = find_git_dir(config.get(s, "path"))
         if not fullpath:
             raise ValueError(
-                    '%s: path %s is not a valid git repository' % (
-                        s, config.get(s, 'path')))
+                "%s: path %s is not a valid git repository"
+                % (s, config.get(s, "path"))
+            )
 
-        config.set(s, 'path', fullpath)
-        config.set(s, 'name', s)
+        config.set(s, "path", fullpath)
+        config.set(s, "name", s)
 
-        desc = config.get(s, 'desc')
-        if not desc and os.path.exists(fullpath + '/description'):
-            desc = open(fullpath + '/description').read().strip()
+        desc = config.get(s, "desc")
+        if not desc and os.path.exists(fullpath + "/description"):
+            desc = open(fullpath + "/description").read().strip()
 
-        r = git.Repo(fullpath, name = s)
+        r = git.Repo(fullpath, name=s)
         r.info.desc = desc
-        r.info.commits_in_summary = config.getint(s, 'commits_in_summary')
-        r.info.commits_per_page = config.getint(s, 'commits_per_page')
-        r.info.max_pages = config.getint(s, 'max_pages')
+        r.info.commits_in_summary = config.getint(s, "commits_in_summary")
+        r.info.commits_per_page = config.getint(s, "commits_per_page")
+        r.info.max_pages = config.getint(s, "max_pages")
         if r.info.max_pages <= 0:
-            r.info.max_pages = sys.maxint
-        r.info.generate_tree = config.getboolean(s, 'tree')
-        r.info.root_diff = config.getboolean(s, 'rootdiff')
-        r.info.generate_patch = config.getboolean(s, 'generate_patch')
+            r.info.max_pages = sys.maxsize
+        r.info.generate_tree = config.getboolean(s, "tree")
+        r.info.root_diff = config.getboolean(s, "rootdiff")
+        r.info.generate_patch = config.getboolean(s, "generate_patch")
 
-        r.info.web_url = config.get(s, 'web_url')
-        web_url_file = fullpath + '/' + config.get(s, 'web_url_file')
+        r.info.web_url = config.get(s, "web_url")
+        web_url_file = fullpath + "/" + config.get(s, "web_url_file")
         if not r.info.web_url and os.path.isfile(web_url_file):
             r.info.web_url = open(web_url_file).read()
 
-        r.info.git_url = config.get(s, 'git_url')
-        git_url_file = fullpath + '/' + config.get(s, 'git_url_file')
+        r.info.git_url = config.get(s, "git_url")
+        git_url_file = fullpath + "/" + config.get(s, "git_url_file")
         if not r.info.git_url and os.path.isfile(git_url_file):
             r.info.git_url = open(git_url_file).read()
 
-        r.info.embed_markdown = config.getboolean(s, 'embed_markdown')
-        r.info.embed_images = config.getboolean(s, 'embed_images')
+        r.info.embed_markdown = config.getboolean(s, "embed_markdown")
+        r.info.embed_images = config.getboolean(s, "embed_images")
 
         repos[r.name] = r
 
+
 def find_git_dir(path):
     """Returns the path to the git directory for the given repository.
 
@@ -147,25 +145,26 @@ def find_git_dir(path):
 
     An empty string is returned if the given path is not a valid repository.
     """
+
     def check(p):
         """A dirty check for whether this is a git dir or not."""
         # Note silent stderr because we expect this to fail and don't want the
         # noise; and also we strip the final \n from the output.
-        return git.run_git(p,
-                ['rev-parse', '--git-dir'],
-                silent_stderr = True).read()[:-1]
+        return git.run_git(
+            p, ["rev-parse", "--git-dir"], silent_stderr=True
+        ).read()[:-1]
 
-    for p in [ path, path + '/.git' ]:
+    for p in [path, path + "/.git"]:
         if check(p):
             return p
 
-    return ''
+    return ""
 
 
 def repo_filter(unused_conf):
     """Bottle route filter for repos."""
     # TODO: consider allowing /, which is tricky.
-    regexp = r'[\w\.~-]+'
+    regexp = r"[\w\.~-]+"
 
     def to_python(s):
         """Return the corresponding Python object."""
@@ -179,8 +178,9 @@ def repo_filter(unused_conf):
 
     return regexp, to_python, to_url
 
+
 app = bottle.Bottle()
-app.router.add_filter('repo', repo_filter)
+app.router.add_filter("repo", repo_filter)
 bottle.app.push(app)
 
 
@@ -191,18 +191,18 @@ def with_utils(f):
     templates.
     """
     utilities = {
-        'shorten': utils.shorten,
-        'can_colorize': utils.can_colorize,
-        'colorize_diff': utils.colorize_diff,
-        'colorize_blob': utils.colorize_blob,
-        'can_markdown': utils.can_markdown,
-        'markdown_blob': utils.markdown_blob,
-        'can_embed_image': utils.can_embed_image,
-        'embed_image_blob': utils.embed_image_blob,
-        'is_binary': utils.is_binary,
-        'hexdump': utils.hexdump,
-        'abort': bottle.abort,
-        'smstr': git.smstr,
+        "shorten": utils.shorten,
+        "can_colorize": utils.can_colorize,
+        "colorize_diff": utils.colorize_diff,
+        "colorize_blob": utils.colorize_blob,
+        "can_markdown": utils.can_markdown,
+        "markdown_blob": utils.markdown_blob,
+        "can_embed_image": utils.can_embed_image,
+        "embed_image_blob": utils.embed_image_blob,
+        "is_binary": utils.is_binary,
+        "hexdump": utils.hexdump,
+        "abort": bottle.abort,
+        "smstr": git.smstr,
     }
 
     def wrapped(*args, **kwargs):
@@ -216,89 +216,108 @@ def with_utils(f):
 
     return wrapped
 
-@bottle.route('/')
-@bottle.view('index')
+
+@bottle.route("/")
+@bottle.view("index")
 @with_utils
 def index():
-    return dict(repos = repos)
+    return dict(repos=repos)
 
-@bottle.route('/r/<repo:repo>/')
-@bottle.view('summary')
+
+@bottle.route("/r/<repo:repo>/")
+@bottle.view("summary")
 @with_utils
 def summary(repo):
-    return dict(repo = repo)
+    return dict(repo=repo)
 
-@bottle.route('/r/<repo:repo>/c/<cid:re:[0-9a-f]{5,40}>/')
-@bottle.view('commit')
+
+@bottle.route("/r/<repo:repo>/c/<cid:re:[0-9a-f]{5,40}>/")
+@bottle.view("commit")
 @with_utils
 def commit(repo, cid):
     c = repo.commit(cid)
     if not c:
-        bottle.abort(404, 'Commit not found')
+        bottle.abort(404, "Commit not found")
 
-    return dict(repo = repo, c=c)
+    return dict(repo=repo, c=c)
 
-@bottle.route('/r/<repo:repo>/c/<cid:re:[0-9a-f]{5,40}>.patch')
-@bottle.view('patch',
-        # Output is text/plain, don't do HTML escaping.
-        template_settings={"noescape": True})
+
+@bottle.route("/r/<repo:repo>/c/<cid:re:[0-9a-f]{5,40}>.patch")
+@bottle.view(
+    "patch",
+    # Output is text/plain, don't do HTML escaping.
+    template_settings={"noescape": True},
+)
 def patch(repo, cid):
     c = repo.commit(cid)
     if not c:
-        bottle.abort(404, 'Commit not found')
+        bottle.abort(404, "Commit not found")
 
-    bottle.response.content_type = 'text/plain; charset=utf8'
+    bottle.response.content_type = "text/plain; charset=utf8"
 
-    return dict(repo = repo, c=c)
+    return dict(repo=repo, c=c)
 
-@bottle.route('/r/<repo:repo>/b/<bname:path>/t/f=<fname:path>.html')
-@bottle.route('/r/<repo:repo>/b/<bname:path>/t/<dirname:path>/f=<fname:path>.html')
-@bottle.view('blob')
+
+@bottle.route("/r/<repo:repo>/b/<bname:path>/t/f=<fname:path>.html")
+@bottle.route(
+    "/r/<repo:repo>/b/<bname:path>/t/<dirname:path>/f=<fname:path>.html"
+)
+@bottle.view("blob")
 @with_utils
-def blob(repo, bname, fname, dirname = ''):
-    if dirname and not dirname.endswith('/'):
-        dirname = dirname + '/'
+def blob(repo, bname, fname, dirname=""):
+    if dirname and not dirname.endswith("/"):
+        dirname = dirname + "/"
 
     dirname = git.smstr.from_url(dirname)
     fname = git.smstr.from_url(fname)
     path = dirname.raw + fname.raw
 
+    # Handle backslash-escaped characters, which are not utf8.
+    # This matches the generated links from git.unquote().
+    path = path.encode("utf8").decode("unicode-escape").encode("latin1")
+
     content = repo.blob(path, bname)
     if content is None:
         bottle.abort(404, "File %r not found in branch %s" % (path, bname))
 
-    return dict(repo = repo, branch = bname, dirname = dirname, fname = fname,
-                blob = content)
+    return dict(
+        repo=repo, branch=bname, dirname=dirname, fname=fname, blob=content
+    )
 
-@bottle.route('/r/<repo:repo>/b/<bname:path>/t/')
-@bottle.route('/r/<repo:repo>/b/<bname:path>/t/<dirname:path>/')
-@bottle.view('tree')
+
+@bottle.route("/r/<repo:repo>/b/<bname:path>/t/")
+@bottle.route("/r/<repo:repo>/b/<bname:path>/t/<dirname:path>/")
+@bottle.view("tree")
 @with_utils
-def tree(repo, bname, dirname = ''):
-    if dirname and not dirname.endswith('/'):
-        dirname = dirname + '/'
+def tree(repo, bname, dirname=""):
+    if dirname and not dirname.endswith("/"):
+        dirname = dirname + "/"
 
     dirname = git.smstr.from_url(dirname)
 
-    return dict(repo = repo, branch = bname, tree = repo.tree(bname),
-                dirname = dirname)
+    return dict(
+        repo=repo, branch=bname, tree=repo.tree(bname), dirname=dirname
+    )
 
-@bottle.route('/r/<repo:repo>/b/<bname:path>/')
-@bottle.route('/r/<repo:repo>/b/<bname:path>/<offset:int>.html')
-@bottle.view('branch')
+
+@bottle.route("/r/<repo:repo>/b/<bname:path>/")
+@bottle.route("/r/<repo:repo>/b/<bname:path>/<offset:int>.html")
+@bottle.view("branch")
 @with_utils
-def branch(repo, bname, offset = 0):
-    return dict(repo = repo, branch = bname, offset = offset)
+def branch(repo, bname, offset=0):
+    return dict(repo=repo, branch=bname, offset=offset)
 
-@bottle.route('/static/<path:path>')
+
+@bottle.route("/static/<path:path>")
 def static(path):
-    return bottle.static_file(path, root = static_path)
+    return bottle.static_file(path, root=static_path)
 
 
 #
 # Static HTML generation
 #
 
+
 def is_404(e):
     """True if e is an HTTPError with status 404, False otherwise."""
     # We need this because older bottle.py versions put the status code in
@@ -309,17 +328,19 @@ def is_404(e):
     else:
         return e.status_code == 404
 
-def generate(output, only = None):
+
+def generate(output: str, only=None):
     """Generate static html to the output directory."""
-    def write_to(path, func_or_str, args = (), mtime = None):
-        path = output + '/' + path
+
+    def write_to(path: str, func_or_str, args=(), mtime=None):
+        path = output + "/" + path
         dirname = os.path.dirname(path)
 
         if not os.path.exists(dirname):
             os.makedirs(dirname)
 
         if mtime:
-            path_mtime = 0
+            path_mtime: Union[float, int] = 0
             if os.path.exists(path):
                 path_mtime = os.stat(path).st_mtime
 
@@ -339,7 +360,7 @@ def generate(output, only = None):
         else:
             # Otherwise, be lazy if we were given a function to run, or write
             # always if they gave us a string.
-            if isinstance(func_or_str, (str, unicode)):
+            if isinstance(func_or_str, str):
                 print(path)
                 s = func_or_str
             else:
@@ -348,71 +369,99 @@ def generate(output, only = None):
                 print(path)
                 s = func_or_str(*args)
 
-        open(path, 'w').write(s.encode('utf8', errors = 'xmlcharrefreplace'))
+        open(path, "w").write(s)
         if mtime:
             os.utime(path, (mtime, mtime))
 
     def link(from_path, to_path):
-        from_path = output + '/' + from_path
+        from_path = output + "/" + from_path
 
         if os.path.lexists(from_path):
             return
-        print(from_path, '->', to_path)
+        print(from_path, "->", to_path)
         os.symlink(to_path, from_path)
 
-    def write_tree(r, bn, mtime):
-        t = r.tree(bn)
+    def write_tree(r: git.Repo, bn: str, mtime):
+        t: git.Tree = r.tree(bn)
 
-        write_to('r/%s/b/%s/t/index.html' % (r.name, bn),
-                tree, (r, bn), mtime)
+        write_to("r/%s/b/%s/t/index.html" % (r.name, bn), tree, (r, bn), mtime)
 
-        for otype, oname, _ in t.ls('', recursive = True):
+        for otype, oname, _ in t.ls("", recursive=True):
             # FIXME: bottle cannot route paths with '\n' so those are sadly
             # expected to fail for now; we skip them.
-            if '\n' in oname.raw:
-                print('skipping file with \\n: %r' % (oname.raw))
+            if "\n" in oname.raw:
+                print("skipping file with \\n: %r" % (oname.raw))
                 continue
 
-            if otype == 'blob':
+            if otype == "blob":
                 dirname = git.smstr(os.path.dirname(oname.raw))
                 fname = git.smstr(os.path.basename(oname.raw))
                 write_to(
-                    'r/%s/b/%s/t/%s%sf=%s.html' %
-                        (str(r.name), str(bn),
-                         dirname.raw, '/' if dirname.raw else '', fname.raw),
-                    blob, (r, bn, fname.url, dirname.url), mtime)
+                    "r/%s/b/%s/t/%s%sf=%s.html"
+                    % (
+                        str(r.name),
+                        str(bn),
+                        dirname.raw,
+                        "/" if dirname.raw else "",
+                        fname.raw,
+                    ),
+                    blob,
+                    (r, bn, fname.url, dirname.url),
+                    mtime,
+                )
             else:
-                write_to('r/%s/b/%s/t/%s/index.html' %
-                            (str(r.name), str(bn), oname.raw),
-                        tree, (r, bn, oname.url), mtime)
+                write_to(
+                    "r/%s/b/%s/t/%s/index.html"
+                    % (str(r.name), str(bn), oname.raw),
+                    tree,
+                    (r, bn, oname.url),
+                    mtime,
+                )
 
     # Always generate the index, to keep the "last updated" time fresh.
-    write_to('index.html', index())
+    write_to("index.html", index())
 
     # We can't call static() because it relies on HTTP headers.
     read_f = lambda f: open(f).read()
-    write_to('static/git-arr.css', read_f, [static_path + '/git-arr.css'],
-            os.stat(static_path + '/git-arr.css').st_mtime)
-    write_to('static/git-arr.js', read_f, [static_path + '/git-arr.js'],
-            os.stat(static_path + '/git-arr.js').st_mtime)
-    write_to('static/syntax.css', read_f, [static_path + '/syntax.css'],
-            os.stat(static_path + '/syntax.css').st_mtime)
+    write_to(
+        "static/git-arr.css",
+        read_f,
+        [static_path + "/git-arr.css"],
+        os.stat(static_path + "/git-arr.css").st_mtime,
+    )
+    write_to(
+        "static/git-arr.js",
+        read_f,
+        [static_path + "/git-arr.js"],
+        os.stat(static_path + "/git-arr.js").st_mtime,
+    )
+    write_to(
+        "static/syntax.css",
+        read_f,
+        [static_path + "/syntax.css"],
+        os.stat(static_path + "/syntax.css").st_mtime,
+    )
 
-    rs = sorted(repos.values(), key = lambda r: r.name)
+    rs = sorted(list(repos.values()), key=lambda r: r.name)
     if only:
         rs = [r for r in rs if r.name in only]
 
     for r in rs:
-        write_to('r/%s/index.html' % r.name, summary(r))
+        write_to("r/%s/index.html" % r.name, summary(r))
         for bn in r.branch_names():
             commit_count = 0
-            commit_ids = r.commit_ids('refs/heads/' + bn,
-                    limit = r.info.commits_per_page * r.info.max_pages)
+            commit_ids = r.commit_ids(
+                "refs/heads/" + bn,
+                limit=r.info.commits_per_page * r.info.max_pages,
+            )
             for cid in commit_ids:
-                write_to('r/%s/c/%s/index.html' % (r.name, cid),
-                        commit, (r, cid))
+                write_to(
+                    "r/%s/c/%s/index.html" % (r.name, cid), commit, (r, cid)
+                )
                 if r.info.generate_patch:
-                    write_to('r/%s/c/%s.patch' % (r.name, cid), patch, (r, cid))
+                    write_to(
+                        "r/%s/c/%s.patch" % (r.name, cid), patch, (r, cid)
+                    )
                 commit_count += 1
 
             # To avoid regenerating files that have not changed, we will
@@ -421,65 +470,83 @@ def generate(output, only = None):
             # write.
             branch_mtime = r.commit(bn).committer_date.epoch
 
-            nr_pages = int(math.ceil(
-                            float(commit_count) / r.info.commits_per_page))
+            nr_pages = int(
+                math.ceil(float(commit_count) / r.info.commits_per_page)
+            )
             nr_pages = min(nr_pages, r.info.max_pages)
 
             for page in range(nr_pages):
-                write_to('r/%s/b/%s/%d.html' % (r.name, bn, page),
-                            branch, (r, bn, page), branch_mtime)
+                write_to(
+                    "r/%s/b/%s/%d.html" % (r.name, bn, page),
+                    branch,
+                    (r, bn, page),
+                    branch_mtime,
+                )
 
-            link(from_path = 'r/%s/b/%s/index.html' % (r.name, bn),
-                    to_path = '0.html')
+            link(
+                from_path="r/%s/b/%s/index.html" % (r.name, bn),
+                to_path="0.html",
+            )
 
             if r.info.generate_tree:
                 write_tree(r, bn, branch_mtime)
 
         for tag_name, obj_id in r.tags():
             try:
-                write_to('r/%s/c/%s/index.html' % (r.name, obj_id),
-                        commit, (r, obj_id))
+                write_to(
+                    "r/%s/c/%s/index.html" % (r.name, obj_id),
+                    commit,
+                    (r, obj_id),
+                )
             except bottle.HTTPError as e:
                 # Some repos can have tags pointing to non-commits. This
                 # happens in the Linux Kernel's v2.6.11, which points directly
                 # to a tree. Ignore them.
                 if is_404(e):
-                    print('404 in tag %s (%s)' % (tag_name, obj_id))
+                    print("404 in tag %s (%s)" % (tag_name, obj_id))
                 else:
                     raise
 
 
 def main():
-    parser = optparse.OptionParser('usage: %prog [options] serve|generate')
-    parser.add_option('-c', '--config', metavar = 'FILE',
-            help = 'configuration file')
-    parser.add_option('-o', '--output', metavar = 'DIR',
-            help = 'output directory (for generate)')
-    parser.add_option('', '--only', metavar = 'REPO', action = 'append',
-            default = [],
-            help = 'generate/serve only this repository')
+    parser = optparse.OptionParser("usage: %prog [options] serve|generate")
+    parser.add_option(
+        "-c", "--config", metavar="FILE", help="configuration file"
+    )
+    parser.add_option(
+        "-o", "--output", metavar="DIR", help="output directory (for generate)"
+    )
+    parser.add_option(
+        "",
+        "--only",
+        metavar="REPO",
+        action="append",
+        default=[],
+        help="generate/serve only this repository",
+    )
     opts, args = parser.parse_args()
 
     if not opts.config:
-        parser.error('--config is mandatory')
+        parser.error("--config is mandatory")
 
     try:
         load_config(opts.config)
     except (configparser.NoOptionError, ValueError) as e:
-        print('Error parsing config:', e)
+        print("Error parsing config:", e)
         return
 
     if not args:
-        parser.error('Must specify an action (serve|generate)')
+        parser.error("Must specify an action (serve|generate)")
 
-    if args[0] == 'serve':
-        bottle.run(host = 'localhost', port = 8008, reloader = True)
-    elif args[0] == 'generate':
+    if args[0] == "serve":
+        bottle.run(host="localhost", port=8008, reloader=True)
+    elif args[0] == "generate":
         if not opts.output:
-            parser.error('Must specify --output')
-        generate(output = opts.output, only = opts.only)
+            parser.error("Must specify --output")
+        generate(output=opts.output, only=opts.only)
     else:
-        parser.error('Unknown action %s' % args[0])
+        parser.error("Unknown action %s" % args[0])
 
-if __name__ == '__main__':
+
+if __name__ == "__main__":
     main()

git.py

@@ -12,101 +12,86 @@ import subprocess
 from collections import defaultdict
 import email.utils
 import datetime
-import urllib
-from cgi import escape
+import urllib.request, urllib.parse, urllib.error
+from html import escape
+from typing import Any, Dict, IO, Iterable, List, Optional, Tuple, Union
 
 
 # Path to the git binary.
 GIT_BIN = "git"
 
-class EncodeWrapper:
-    """File-like wrapper that returns data utf8 encoded."""
-    def __init__(self, fd, encoding = 'utf8', errors = 'replace'):
-        self.fd = fd
-        self.encoding = encoding
-        self.errors = errors
 
-    def __iter__(self):
-        for line in self.fd:
-            yield line.decode(self.encoding, errors = self.errors)
-
-    def read(self):
-        """Returns the whole content."""
-        s = self.fd.read()
-        return s.decode(self.encoding, errors = self.errors)
-
-    def readline(self):
-        """Returns a single line."""
-        s = self.fd.readline()
-        return s.decode(self.encoding, errors = self.errors)
-
-
-def run_git(repo_path, params, stdin = None, silent_stderr = False, raw = False):
+def run_git(
+    repo_path: str, params, stdin: bytes = None, silent_stderr=False, raw=False
+) -> Union[IO[str], IO[bytes]]:
     """Invokes git with the given parameters.
 
     This function invokes git with the given parameters, and returns a
     file-like object with the output (from a pipe).
     """
-    params = [GIT_BIN, '--git-dir=%s' % repo_path] + list(params)
+    params = [GIT_BIN, "--git-dir=%s" % repo_path] + list(params)
 
     stderr = None
     if silent_stderr:
         stderr = subprocess.PIPE
 
     if not stdin:
-        p = subprocess.Popen(params,
-                stdin = None, stdout = subprocess.PIPE, stderr = stderr)
+        p = subprocess.Popen(
+            params, stdin=None, stdout=subprocess.PIPE, stderr=stderr
+        )
     else:
-        p = subprocess.Popen(params,
-                stdin = subprocess.PIPE, stdout = subprocess.PIPE,
-                stderr = stderr)
+        p = subprocess.Popen(
+            params,
+            stdin=subprocess.PIPE,
+            stdout=subprocess.PIPE,
+            stderr=stderr,
+        )
+
+        assert p.stdin is not None
         p.stdin.write(stdin)
         p.stdin.close()
 
+    assert p.stdout is not None
+
     if raw:
         return p.stdout
 
-    # We need to wrap stdout if we want to decode it as utf8, subprocess
-    # doesn't support us telling it the encoding.
-    if sys.version_info.major == 3:
-        return io.TextIOWrapper(p.stdout, encoding = 'utf8',
-                errors = 'replace')
-    else:
-        return EncodeWrapper(p.stdout)
+    return io.TextIOWrapper(
+        p.stdout, encoding="utf8", errors="backslashreplace"
+    )
 
 
-class GitCommand (object):
+class GitCommand(object):
     """Convenient way of invoking git."""
-    def __init__(self, path, cmd, *args, **kwargs):
+
+    def __init__(self, path: str, cmd: str):
         self._override = True
         self._path = path
         self._cmd = cmd
-        self._args = list(args)
-        self._kwargs = {}
-        self._stdin_buf = None
+        self._args: List[str] = []
+        self._kwargs: Dict[str, str] = {}
+        self._stdin_buf: Optional[bytes] = None
         self._raw = False
         self._override = False
-        for k, v in kwargs:
-            self.__setattr__(k, v)
 
     def __setattr__(self, k, v):
-        if k == '_override' or self._override:
+        if k == "_override" or self._override:
             self.__dict__[k] = v
             return
-        k = k.replace('_', '-')
+        k = k.replace("_", "-")
         self._kwargs[k] = v
 
-    def arg(self, a):
+    def arg(self, a: str):
         """Adds an argument."""
         self._args.append(a)
 
-    def raw(self, b):
+    def raw(self, b: bool):
         """Request raw rather than utf8-encoded command output."""
         self._override = True
         self._raw = b
         self._override = False
 
-    def stdin(self, s):
+    def stdin(self, s: bytes):
         """Sets the contents we will send in stdin."""
         self._override = True
         self._stdin_buf = s
@@ -116,46 +101,37 @@ class GitCommand (object):
         """Runs the git command."""
         params = [self._cmd]
 
-        for k, v in self._kwargs.items():
-            dash = '--' if len(k) > 1 else '-'
+        for k, v in list(self._kwargs.items()):
+            dash = "--" if len(k) > 1 else "-"
             if v is None:
-                params.append('%s%s' % (dash, k))
+                params.append("%s%s" % (dash, k))
             else:
-                params.append('%s%s=%s' % (dash, k, str(v)))
+                params.append("%s%s=%s" % (dash, k, str(v)))
 
         params.extend(self._args)
 
-        return run_git(self._path, params, self._stdin_buf, raw = self._raw)
+        return run_git(self._path, params, self._stdin_buf, raw=self._raw)
 
 
-class SimpleNamespace (object):
+class SimpleNamespace(object):
     """An entirely flexible object, which provides a convenient namespace."""
+
     def __init__(self, **kwargs):
         self.__dict__.update(kwargs)
 
 
 class smstr:
-    """A "smart" string, containing many representations for ease of use.
+    """A "smart" string, containing many representations for ease of use."""
 
-    This is a string class that contains:
-        .raw     -> raw string, authoritative source.
-        .unicode -> unicode representation, may not be perfect if .raw is not
-                    proper utf8 but should be good enough to show.
-        .url     -> escaped for safe embedding in URLs, can be not quite
-                    readable.
-        .html    -> an HTML-embeddable representation.
-    """
-    def __init__(self, raw):
-        if not isinstance(raw, str):
-            raise TypeError("The raw string must be instance of 'str'")
-        self.raw = raw
-        self.unicode = raw.decode('utf8', errors = 'replace')
-        self.url = urllib.pathname2url(raw)
+    raw: str  # string, probably utf8-encoded, good enough to show.
+    url: str  # escaped for safe embedding in URLs (not human-readable).
+    html: str  # HTML-embeddable representation.
+
+    def __init__(self, s: str):
+        self.raw = s
+        self.url = urllib.request.pathname2url(s)
         self.html = self._to_html()
 
-    def __cmp__(self, other):
-        return cmp(self.raw, other.raw)
-
     # Note we don't define __repr__() or __str__() to prevent accidental
     # misuse. It does mean that some uses become more annoying, so it's a
     # tradeoff that may change in the future.
@@ -163,11 +139,11 @@ class smstr:
     @staticmethod
     def from_url(url):
         """Returns an smstr() instance from an url-encoded string."""
-        return smstr(urllib.url2pathname(url))
+        return smstr(urllib.request.url2pathname(url))
 
     def split(self, sep):
         """Like str.split()."""
-        return [ smstr(s) for s in self.raw.split(sep) ]
+        return [smstr(s) for s in self.raw.split(sep)]
 
     def __add__(self, other):
         if isinstance(other, smstr):
@@ -176,10 +152,10 @@ class smstr:
 
     def _to_html(self):
         """Returns an html representation of the unicode string."""
-        html = u''
-        for c in escape(self.unicode):
-            if c in '\t\r\n\r\f\a\b\v\0':
-                esc_c = c.encode('ascii').encode('string_escape')
+        html = ""
+        for c in escape(self.raw):
+            if c in "\t\r\n\r\f\a\b\v\0":
+                esc_c = c.encode("unicode-escape").decode("utf8")
                 html += '<span class="ctrlchr">%s</span>' % esc_c
             else:
                 html += c
@@ -187,17 +163,26 @@ class smstr:
         return html
 
 
-def unquote(s):
+def unquote(s: str):
     """Git can return quoted file names, unquote them. Always return a str."""
     if not (s[0] == '"' and s[-1] == '"'):
-        # Unquoted strings are always safe, no need to mess with them; just
-        # make sure we return str.
-        s = s.encode('ascii')
+        # Unquoted strings are always safe, no need to mess with them
         return s
 
-    # Get rid of the quotes, we never want them in the output, and convert to
-    # a raw string, un-escaping the backslashes.
-    s = s[1:-1].decode('string-escape')
+    # The string will be of the form `"<escaped>"`, where <escaped> is a
+    # backslash-escaped representation of the name of the file.
+    # Examples:  "with\ttwo\ttabs" , "\303\261aca-utf8", "\361aca-latin1"
+
+    # Get rid of the quotes, we never want them in the output.
+    s = s[1:-1]
+
+    # Un-escape the backslashes.
+    # latin1 is ok to use here because in Python it just maps the code points
+    # 0-255 to the bytes 0x-0xff, which is what we expect.
+    s = s.encode("latin1").decode("unicode-escape")
+
+    # Convert to utf8.
+    s = s.encode("latin1").decode("utf8", errors="backslashreplace")
 
     return s
 
@@ -205,18 +190,18 @@ def unquote(s):
 class Repo:
     """A git repository."""
 
-    def __init__(self, path, name = None, info = None):
+    def __init__(self, path: str, name=None, info=None):
         self.path = path
         self.name = name
-        self.info = info or SimpleNamespace()
+        self.info: Any = info or SimpleNamespace()
 
     def cmd(self, cmd):
         """Returns a GitCommand() on our path."""
         return GitCommand(self.path, cmd)
 
-    def for_each_ref(self, pattern = None, sort = None, count = None):
+    def for_each_ref(self, pattern=None, sort=None, count=None):
         """Returns a list of references."""
-        cmd = self.cmd('for-each-ref')
+        cmd = self.cmd("for-each-ref")
         if sort:
             cmd.sort = sort
         if count:
@@ -228,61 +213,57 @@ class Repo:
             obj_id, obj_type, ref = l.split()
             yield obj_id, obj_type, ref
 
-    def branches(self, sort = '-authordate'):
+    def branches(self, sort="-authordate"):
         """Get the (name, obj_id) of the branches."""
-        refs = self.for_each_ref(pattern = 'refs/heads/', sort = sort)
+        refs = self.for_each_ref(pattern="refs/heads/", sort=sort)
         for obj_id, _, ref in refs:
-            yield ref[len('refs/heads/'):], obj_id
+            yield ref[len("refs/heads/") :], obj_id
 
     def branch_names(self):
         """Get the names of the branches."""
-        return ( name for name, _ in self.branches() )
+        return (name for name, _ in self.branches())
 
-    def tags(self, sort = '-taggerdate'):
+    def tags(self, sort="-taggerdate"):
         """Get the (name, obj_id) of the tags."""
-        refs = self.for_each_ref(pattern = 'refs/tags/', sort = sort)
+        refs = self.for_each_ref(pattern="refs/tags/", sort=sort)
         for obj_id, _, ref in refs:
-            yield ref[len('refs/tags/'):], obj_id
+            yield ref[len("refs/tags/") :], obj_id
 
-    def tag_names(self):
-        """Get the names of the tags."""
-        return ( name for name, _ in self.tags() )
-
-    def commit_ids(self, ref, limit = None):
+    def commit_ids(self, ref, limit=None):
         """Generate commit ids."""
-        cmd = self.cmd('rev-list')
+        cmd = self.cmd("rev-list")
         if limit:
             cmd.max_count = limit
 
         cmd.arg(ref)
-        cmd.arg('--')
+        cmd.arg("--")
 
         for l in cmd.run():
-            yield l.rstrip('\n')
+            yield l.rstrip("\n")
 
     def commit(self, commit_id):
         """Return a single commit."""
-        cs = list(self.commits(commit_id, limit = 1))
+        cs = list(self.commits(commit_id, limit=1))
         if len(cs) != 1:
             return None
         return cs[0]
 
-    def commits(self, ref, limit = None, offset = 0):
+    def commits(self, ref, limit=None, offset=0):
         """Generate commit objects for the ref."""
-        cmd = self.cmd('rev-list')
+        cmd = self.cmd("rev-list")
         if limit:
             cmd.max_count = limit + offset
 
         cmd.header = None
 
         cmd.arg(ref)
-        cmd.arg('--')
+        cmd.arg("--")
 
-        info_buffer = ''
+        info_buffer = ""
         count = 0
         for l in cmd.run():
-            if '\0' in l:
-                pre, post = l.split('\0', 1)
+            if "\0" in l:
+                pre, post = l.split("\0", 1)
                 info_buffer += pre
 
                 count += 1
@@ -301,11 +282,11 @@ class Repo:
 
     def diff(self, ref):
         """Return a Diff object for the ref."""
-        cmd = self.cmd('diff-tree')
+        cmd = self.cmd("diff-tree")
         cmd.patch = None
         cmd.numstat = None
         cmd.find_renames = None
-        if (self.info.root_diff):
+        if self.info.root_diff:
             cmd.root = None
         # Note we intentionally do not use -z, as the filename is just for
         # reference, and it is safer to let git do the escaping.
@@ -316,13 +297,13 @@ class Repo:
 
     def refs(self):
         """Return a dict of obj_id -> ref."""
-        cmd = self.cmd('show-ref')
+        cmd = self.cmd("show-ref")
         cmd.dereference = None
 
         r = defaultdict(list)
         for l in cmd.run():
             l = l.strip()
-            obj_id, ref = l.split(' ', 1)
+            obj_id, ref = l.split(" ", 1)
             r[obj_id].append(ref)
 
         return r
@@ -333,39 +314,49 @@ class Repo:
 
     def blob(self, path, ref):
         """Returns a Blob instance for the given path."""
-        cmd = self.cmd('cat-file')
+        cmd = self.cmd("cat-file")
         cmd.raw(True)
-        cmd.batch = '%(objectsize)'
+        cmd.batch = "%(objectsize)"
 
-        if isinstance(ref, unicode):
-            ref = ref.encode('utf8')
-        cmd.stdin('%s:%s' % (ref, path))
+        # Format: <ref>:<path>
+        # Construct it in binary since the path might not be utf8.
+        cmd.stdin(ref.encode("utf8") + b":" + path)
 
         out = cmd.run()
         head = out.readline()
-        if not head or head.strip().endswith('missing'):
+        if not head or head.strip().endswith(b"missing"):
             return None
 
-        return Blob(out.read()[:int(head)])
+        return Blob(out.read()[: int(head)])
 
     def last_commit_timestamp(self):
         """Return the timestamp of the last commit."""
-        refs = self.for_each_ref(pattern = 'refs/heads/',
-                sort = '-committerdate', count = 1)
+        refs = self.for_each_ref(
+            pattern="refs/heads/", sort="-committerdate", count=1
+        )
         for obj_id, _, _ in refs:
             commit = self.commit(obj_id)
             return commit.committer_epoch
         return -1
 
 
-class Commit (object):
+class Commit(object):
     """A git commit."""
 
-    def __init__(self, repo,
-            commit_id, parents, tree,
-            author, author_epoch, author_tz,
-            committer, committer_epoch, committer_tz,
-            message):
+    def __init__(
+        self,
+        repo,
+        commit_id,
+        parents,
+        tree,
+        author,
+        author_epoch,
+        author_tz,
+        committer,
+        committer_epoch,
+        committer_tz,
+        message,
+    ):
         self._repo = repo
         self.id = commit_id
         self.parents = parents
@@ -378,28 +369,30 @@ class Commit (object):
         self.committer_tz = committer_tz
         self.message = message
 
-        self.author_name, self.author_email = \
-                email.utils.parseaddr(self.author)
+        self.author_name, self.author_email = email.utils.parseaddr(
+            self.author
+        )
 
-        self.committer_name, self.committer_email = \
-                email.utils.parseaddr(self.committer)
+        self.committer_name, self.committer_email = email.utils.parseaddr(
+            self.committer
+        )
 
-        self.subject, self.body = self.message.split('\n', 1)
+        self.subject, self.body = self.message.split("\n", 1)
 
         self.author_date = Date(self.author_epoch, self.author_tz)
         self.committer_date = Date(self.committer_epoch, self.committer_tz)
 
-
         # Only get this lazily when we need it; most of the time it's not
         # required by the caller.
         self._diff = None
 
     def __repr__(self):
-        return '<C %s p:%s a:%s s:%r>' % (
-                self.id[:7],
-                ','.join(p[:7] for p in self.parents),
-                self.author_email,
-                self.subject[:20])
+        return "<C %s p:%s a:%s s:%r>" % (
+            self.id[:7],
+            ",".join(p[:7] for p in self.parents),
+            self.author_email,
+            self.subject[:20],
+        )
 
     @property
     def diff(self):
@@ -411,57 +404,68 @@ class Commit (object):
     @staticmethod
     def from_str(repo, buf):
         """Parses git rev-list output, returns a commit object."""
-        if '\n\n' in buf:
+        if "\n\n" in buf:
             # Header, commit message
-            header, raw_message = buf.split('\n\n', 1)
+            header, raw_message = buf.split("\n\n", 1)
         else:
             # Header only, no commit message
-            header, raw_message = buf.rstrip(), '    '
+            header, raw_message = buf.rstrip(), "    "
 
-        header_lines = header.split('\n')
+        header_lines = header.split("\n")
         commit_id = header_lines.pop(0)
 
         header_dict = defaultdict(list)
         for line in header_lines:
-            k, v = line.split(' ', 1)
+            k, v = line.split(" ", 1)
             header_dict[k].append(v)
 
-        tree = header_dict['tree'][0]
-        parents = set(header_dict['parent'])
-        author, author_epoch, author_tz = \
-                header_dict['author'][0].rsplit(' ', 2)
-        committer, committer_epoch, committer_tz = \
-                header_dict['committer'][0].rsplit(' ', 2)
+        tree = header_dict["tree"][0]
+        parents = set(header_dict["parent"])
+
+        authorhdr = header_dict["author"][0]
+        author, author_epoch, author_tz = authorhdr.rsplit(" ", 2)
+
+        committerhdr = header_dict["committer"][0]
+        committer, committer_epoch, committer_tz = committerhdr.rsplit(" ", 2)
 
         # Remove the first four spaces from the message's lines.
-        message = ''
-        for line in raw_message.split('\n'):
-            message += line[4:] + '\n'
+        message = ""
+        for line in raw_message.split("\n"):
+            message += line[4:] + "\n"
+
+        return Commit(
+            repo,
+            commit_id=commit_id,
+            tree=tree,
+            parents=parents,
+            author=author,
+            author_epoch=author_epoch,
+            author_tz=author_tz,
+            committer=committer,
+            committer_epoch=committer_epoch,
+            committer_tz=committer_tz,
+            message=message,
+        )
 
-        return Commit(repo,
-                commit_id = commit_id, tree = tree, parents = parents,
-                author = author,
-                author_epoch = author_epoch, author_tz = author_tz,
-                committer = committer,
-                committer_epoch = committer_epoch, committer_tz = committer_tz,
-                message = message)
 
 class Date:
     """Handy representation for a datetime from git."""
+
     def __init__(self, epoch, tz):
         self.epoch = int(epoch)
         self.tz = tz
         self.utc = datetime.datetime.utcfromtimestamp(self.epoch)
 
         self.tz_sec_offset_min = int(tz[1:3]) * 60 + int(tz[4:])
-        if tz[0] == '-':
+        if tz[0] == "-":
             self.tz_sec_offset_min = -self.tz_sec_offset_min
 
         self.local = self.utc + datetime.timedelta(
-                                    minutes = self.tz_sec_offset_min)
+            minutes=self.tz_sec_offset_min
+        )
 
-        self.str = self.utc.strftime('%a, %d %b %Y %H:%M:%S +0000 ')
-        self.str += '(%s %s)' % (self.local.strftime('%H:%M'), self.tz)
+        self.str = self.utc.strftime("%a, %d %b %Y %H:%M:%S +0000 ")
+        self.str += "(%s %s)" % (self.local.strftime("%H:%M"), self.tz)
 
     def __str__(self):
         return self.str
@@ -469,6 +473,7 @@ class Date:
 
 class Diff:
     """A diff between two trees."""
+
     def __init__(self, ref, changes, body):
         """Constructor.
 
@@ -488,23 +493,23 @@ class Diff:
             ref_id = next(lines)
         except StopIteration:
             # No diff; this can happen in merges without conflicts.
-            return Diff(None, [], '')
+            return Diff(None, [], "")
 
         # First, --numstat information.
         changes = []
         l = next(lines)
-        while l != '\n':
-            l = l.rstrip('\n')
-            added, deleted, fname = l.split('\t', 2)
-            added = added.replace('-', '0')
-            deleted = deleted.replace('-', '0')
+        while l != "\n":
+            l = l.rstrip("\n")
+            added, deleted, fname = l.split("\t", 2)
+            added = added.replace("-", "0")
+            deleted = deleted.replace("-", "0")
             fname = smstr(unquote(fname))
             changes.append((int(added), int(deleted), fname))
             l = next(lines)
 
         # And now the diff body. We just store as-is, we don't really care for
         # the contents.
-        body = ''.join(lines)
+        body = "".join(lines)
 
         return Diff(ref_id, changes, body)
 
@@ -512,13 +517,15 @@ class Diff:
 class Tree:
     """ A git tree."""
 
-    def __init__(self, repo, ref):
+    def __init__(self, repo: Repo, ref: str):
         self.repo = repo
         self.ref = ref
 
-    def ls(self, path, recursive = False):
+    def ls(
+        self, path, recursive=False
+    ) -> Iterable[Tuple[str, smstr, Optional[int]]]:
         """Generates (type, name, size) for each file in path."""
-        cmd = self.repo.cmd('ls-tree')
+        cmd = self.repo.cmd("ls-tree")
         cmd.long = None
         if recursive:
             cmd.r = None
@@ -532,17 +539,17 @@ class Tree:
 
         for l in cmd.run():
             _mode, otype, _oid, size, name = l.split(None, 4)
-            if size == '-':
+            if size == "-":
                 size = None
             else:
                 size = int(size)
 
             # Remove the quoting (if any); will always give us a str.
-            name = unquote(name.strip('\n'))
+            name = unquote(name.strip("\n"))
 
             # Strip the leading path, the caller knows it and it's often
             # easier to work with this way.
-            name = name[len(path):]
+            name = name[len(path) :]
 
             # We use a smart string for the name, as it's often tricky to
             # manipulate otherwise.
@@ -552,12 +559,12 @@ class Tree:
 class Blob:
     """A git blob."""
 
-    def __init__(self, raw_content):
+    def __init__(self, raw_content: bytes):
         self.raw_content = raw_content
         self._utf8_content = None
 
     @property
     def utf8_content(self):
         if not self._utf8_content:
-            self._utf8_content = self.raw_content.decode('utf8', 'replace')
+            self._utf8_content = self.raw_content.decode("utf8", "replace")
         return self._utf8_content

pyproject.toml

@@ -0,0 +1,3 @@
+[tool.black]
+line-length = 79
+include = "(git-arr|git.py|utils.py)$"

utils.py

@@ -5,16 +5,16 @@ These are mostly used in templates, for presentation purposes.
 """
 
 try:
-    import pygments
-    from pygments import highlight
-    from pygments import lexers
-    from pygments.formatters import HtmlFormatter
+    import pygments  # type: ignore
+    from pygments import highlight  # type: ignore
+    from pygments import lexers  # type: ignore
+    from pygments.formatters import HtmlFormatter  # type: ignore
 except ImportError:
     pygments = None
 
 try:
-    import markdown
-    import markdown.treeprocessors
+    import markdown  # type: ignore
+    import markdown.treeprocessors  # type: ignore
 except ImportError:
     markdown = None
 
@@ -23,12 +23,16 @@ import mimetypes
 import string
 import os.path
 
-def shorten(s, width = 60):
+import git
+
+
+def shorten(s: str, width=60):
     if len(s) < 60:
         return s
     return s[:57] + "..."
 
-def can_colorize(s):
+
+def can_colorize(s: str):
     """True if we can colorize the string, False otherwise."""
     if pygments is None:
         return False
@@ -41,7 +45,7 @@ def can_colorize(s):
     # If any of the first 5 lines is over 300 characters long, don't colorize.
     start = 0
     for i in range(5):
-        pos = s.find('\n', start)
+        pos = s.find("\n", start)
         if pos == -1:
             break
 
@@ -51,7 +55,8 @@ def can_colorize(s):
 
     return True
 
-def can_markdown(repo, fname):
+
+def can_markdown(repo: git.Repo, fname: str):
     """True if we can process file through markdown, False otherwise."""
     if markdown is None:
         return False
@@ -61,73 +66,86 @@ def can_markdown(repo, fname):
 
     return fname.endswith(".md")
 
+
 def can_embed_image(repo, fname):
     """True if we can embed image file in HTML, False otherwise."""
     if not repo.info.embed_images:
         return False
 
-    return (('.' in fname) and
-            (fname.split('.')[-1].lower() in [ 'jpg', 'jpeg', 'png', 'gif' ]))
+    return ("." in fname) and (
+        fname.split(".")[-1].lower() in ["jpg", "jpeg", "png", "gif"]
+    )
 
-def colorize_diff(s):
-    lexer = lexers.DiffLexer(encoding = 'utf-8')
-    formatter = HtmlFormatter(encoding = 'utf-8',
-                    cssclass = 'source_code')
+
+def colorize_diff(s: str) -> str:
+    lexer = lexers.DiffLexer(encoding="utf-8")
+    formatter = HtmlFormatter(encoding="utf-8", cssclass="source_code")
 
     return highlight(s, lexer, formatter)
 
-def colorize_blob(fname, s):
+
+def colorize_blob(fname, s: str) -> str:
     try:
-        lexer = lexers.guess_lexer_for_filename(fname, s, encoding = 'utf-8')
+        lexer = lexers.guess_lexer_for_filename(fname, s, encoding="utf-8")
     except lexers.ClassNotFound:
         # Only try to guess lexers if the file starts with a shebang,
         # otherwise it's likely a text file and guess_lexer() is prone to
         # make mistakes with those.
-        lexer = lexers.TextLexer(encoding = 'utf-8')
-        if s.startswith('#!'):
+        lexer = lexers.TextLexer(encoding="utf-8")
+        if s.startswith("#!"):
             try:
-                lexer = lexers.guess_lexer(s[:80], encoding = 'utf-8')
+                lexer = lexers.guess_lexer(s[:80], encoding="utf-8")
             except lexers.ClassNotFound:
                 pass
 
-    formatter = HtmlFormatter(encoding = 'utf-8',
-                    cssclass = 'source_code',
-                    linenos = 'table',
-                    anchorlinenos = True,
-                    lineanchors = 'line')
+    formatter = HtmlFormatter(
+        encoding="utf-8",
+        cssclass="source_code",
+        linenos="table",
+        anchorlinenos=True,
+        lineanchors="line",
+    )
 
     return highlight(s, lexer, formatter)
 
-def markdown_blob(s):
+
+def markdown_blob(s: str) -> str:
     extensions = [
         "markdown.extensions.fenced_code",
         "markdown.extensions.tables",
         RewriteLocalLinksExtension(),
     ]
-    return markdown.markdown(s, extensions = extensions)
+    return markdown.markdown(s, extensions=extensions)
 
-def embed_image_blob(fname, image_data):
+
+def embed_image_blob(fname: str, image_data: bytes) -> str:
     mimetype = mimetypes.guess_type(fname)[0]
-    return '<img style="max-width:100%;" src="data:{0};base64,{1}" />'.format( \
-                                    mimetype, base64.b64encode(image_data))
+    b64img = base64.b64encode(image_data).decode("ascii")
+    return '<img style="max-width:100%;" src="data:{0};base64,{1}" />'.format(
+        mimetype, b64img
+    )
 
-def is_binary(s):
+
+def is_binary(b: bytes):
     # Git considers a blob binary if NUL in first ~8KB, so do the same.
-    return '\0' in s[:8192]
+    return b"\0" in b[:8192]
 
-def hexdump(s):
-    graph = string.ascii_letters + string.digits + string.punctuation + ' '
+
+def hexdump(s: bytes):
+    graph = string.ascii_letters + string.digits + string.punctuation + " "
+    b = s.decode("latin1")
     offset = 0
-    while s:
-        t = s[:16]
-        hexvals = ['%.2x' % ord(c) for c in t]
-        text = ''.join(c if c in graph else '.' for c in t)
-        yield offset, ' '.join(hexvals[:8]), ' '.join(hexvals[8:]), text
+    while b:
+        t = b[:16]
+        hexvals = ["%.2x" % ord(c) for c in t]
+        text = "".join(c if c in graph else "." for c in t)
+        yield offset, " ".join(hexvals[:8]), " ".join(hexvals[8:]), text
         offset += 16
-        s = s[16:]
+        b = b[16:]
 
 
 if markdown:
+
     class RewriteLocalLinks(markdown.treeprocessors.Treeprocessor):
         """Rewrites relative links to files, to match git-arr's links.
 
@@ -137,6 +155,7 @@ if markdown:
         Note that we're already assuming a degree of sanity in the HTML, so we
         don't re-check that the path is reasonable.
         """
+
         def run(self, root):
             for child in root:
                 if child.tag == "a":
@@ -157,9 +176,8 @@ if markdown:
             new_target = os.path.join(head, "f=" + tail + ".html")
             tag.set("href", new_target)
 
-
     class RewriteLocalLinksExtension(markdown.Extension):
         def extendMarkdown(self, md, md_globals):
             md.treeprocessors.add(
-                    "RewriteLocalLinks", RewriteLocalLinks(), "_end")
-
+                "RewriteLocalLinks", RewriteLocalLinks(), "_end"
+            )

views/blob.html

@@ -10,7 +10,7 @@
 % relroot = reltree + '../' * (len(branch.split('/')) - 1)
 
 <title>git &raquo; {{repo.name}} &raquo;
-    {{branch}} &raquo; {{dirname.unicode}}{{fname.unicode}}</title>
+    {{branch}} &raquo; {{dirname.raw}}{{fname.raw}}</title>
 <link rel="stylesheet" type="text/css"
     href="{{relroot}}../../../../../static/git-arr.css"/>
 <link rel="stylesheet" type="text/css"
@@ -33,7 +33,7 @@
 %   if not c.raw:
 %     continue
 %   end
-    <a href="{{base.url}}{{c.url}}/">{{c.unicode}}</a> /
+    <a href="{{base.url}}{{c.url}}/">{{c.raw}}</a> /
 %   base += c + '/'
 % end
     <a href="">{{!fname.html}}</a>
@@ -45,7 +45,7 @@
         <td>empty &mdash; 0 bytes</td>
     </tr>
 </table>
-% elif can_embed_image(repo, fname.unicode):
+% elif can_embed_image(repo, fname.raw):
 {{!embed_image_blob(fname.raw, blob.raw_content)}}
 % elif is_binary(blob.raw_content):
 <table class="nice blob-binary">
@@ -72,12 +72,12 @@
     </tr>
 %   end
 </table>
-% elif can_markdown(repo, fname.unicode):
+% elif can_markdown(repo, fname.raw):
 <div class="markdown">
 {{!markdown_blob(blob.utf8_content)}}
 </div>
 % elif can_colorize(blob.utf8_content):
-{{!colorize_blob(fname.unicode, blob.utf8_content)}}
+{{!colorize_blob(fname.raw, blob.utf8_content)}}
 % else:
 <pre class="blob-body">
 {{blob.utf8_content}}

views/tree-list.html

@@ -1,5 +1,5 @@
 <table class="nice toggable ls" id="ls">
-% key_func = lambda (t, n, s): (t != 'tree', n.raw)
+% key_func = lambda x: (x[0] != 'tree', x[1].raw)
 % for type, name, size in sorted(tree.ls(dirname.raw), key = key_func):
     <tr class="{{type}}">
 %   if type == "blob":

views/tree.html

@@ -10,7 +10,7 @@
 % relroot = reltree + '../' * (len(branch.split('/')) - 1)
 
 <title>git &raquo; {{repo.name}} &raquo;
-    {{branch}} &raquo; {{dirname.unicode if dirname.unicode else '/'}}</title>
+    {{branch}} &raquo; {{dirname.raw if dirname.raw else '/'}}</title>
 <link rel="stylesheet" type="text/css"
     href="{{relroot}}../../../../../static/git-arr.css"/>
 <meta http-equiv="content-type" content="text/html; charset=utf-8"/>
@@ -31,7 +31,7 @@
 %   if not c.raw:
 %     continue
 %   end
-    <a href="{{base.url}}{{c.url}}/">{{c.unicode}}</a> /
+    <a href="{{base.url}}{{c.url}}/">{{c.raw}}</a> /
 %   base += c + '/'
 % end
 </h3>