views/tree.html: Fix lambda syntax

Some versions of bottle.py don't deal well with the "if" inside the lambda, so
work around it by just using comparison and simplifying the function.

Signed-off-by: Alberto Bertogli <albertito@blitiri.com.ar>

README

@@ -19,6 +19,13 @@ 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.
 
@@ -41,9 +48,9 @@ use, by running:
 That can be useful when making changes to the software itself.
 
 
-Where to report bugs
---------------------
+Contact
+-------
 
-If you want to report bugs, or have any questions or comments, just let me
-know at albertito@blitiri.com.ar.
+If you want to report bugs, send patches, or have any questions or comments,
+just let me know at albertito@blitiri.com.ar.
 

git-arr

@@ -5,6 +5,7 @@ git-arr: A git web html generator.
 
 from __future__ import print_function
 
+import sys
 import os
 import math
 import optparse
@@ -20,6 +21,18 @@ import git
 import utils
 
 
+# Tell bottle where to find the views.
+# 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/')
+
+# 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/'
+
+
 # The list of repositories is a global variable for convenience. It will be
 # populated by load_config().
 repos = {}
@@ -41,7 +54,9 @@ def load_config(path):
         'web_url': '',
         'web_url_file': 'web_url',
         'git_url': '',
-        'git_url_file': 'git_url',
+        'git_url_file': 'cloneurl',
+        'embed_markdown': 'yes',
+        'embed_images': 'no',
     }
 
     config = configparser.SafeConfigParser(defaults)
@@ -49,14 +64,10 @@ def load_config(path):
 
     # Do a first pass for general sanity checking and recursive expansion.
     for s in config.sections():
-        if not config.has_option(s, 'path'):
-            raise configparser.NoOptionError(
-                    '%s is missing the mandatory path' % s)
-
         if config.getboolean(s, 'recursive'):
             for path in os.listdir(config.get(s, 'path')):
-                fullpath = config.get(s, 'path') + '/' + path
-                if not os.path.exists(fullpath + '/HEAD'):
+                fullpath = find_git_dir(config.get(s, 'path') + '/' + path)
+                if not fullpath:
                     continue
 
                 if os.path.exists(fullpath + '/disable_gitweb'):
@@ -76,7 +87,13 @@ def load_config(path):
             config.remove_section(s)
 
     for s in config.sections():
-        fullpath = 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')))
+
+        config.set(s, 'path', fullpath)
         config.set(s, 'name', s)
 
         desc = config.get(s, 'desc')
@@ -100,8 +117,34 @@ def load_config(path):
         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')
+
         repos[r.name] = r
 
+def find_git_dir(path):
+    """Returns the path to the git directory for the given repository.
+
+    This function takes a path to a git repository, and returns the path to
+    its git directory. If the repo is bare, it will be the same path;
+    otherwise it will be path + '.git/'.
+
+    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]
+
+    for p in [ path, path + '/.git' ]:
+        if check(p):
+            return p
+
+    return ''
+
 
 def repo_filter(unused_conf):
     """Bottle route filter for repos."""
@@ -133,9 +176,13 @@ def with_utils(f):
     """
     utilities = {
         'shorten': utils.shorten,
-        'has_colorizer': utils.has_colorizer,
+        '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,
         'abort': bottle.abort,
         'smstr': git.smstr,
     }
@@ -215,14 +262,24 @@ def blob(repo, bname, fname, dirname = ''):
 
 @bottle.route('/static/<path:path>')
 def static(path):
-    return bottle.static_file(path, root = './static/')
+    return bottle.static_file(path, root = static_path)
 
 
 #
 # Static HTML generation
 #
 
-def generate(output):
+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
+    # e.status as an integer, and newer versions make that a string, and using
+    # e.status_code for the code.
+    if isinstance(e.status, int):
+        return e.status == 404
+    else:
+        return e.status_code == 404
+
+def generate(output, skip_index = False):
     """Generate static html to the output directory."""
     def write_to(path, func_or_str, args = (), mtime = None):
         path = output + '/' + path
@@ -298,14 +355,17 @@ def generate(output):
                             (str(r.name), str(bn), oname.raw),
                         tree, (r, bn, oname.url), mtime)
 
-    write_to('index.html', index())
+    if not skip_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/git-arr.css'],
-            os.stat('static/git-arr.css').st_mtime)
-    write_to('static/syntax.css', read_f, ['static/syntax.css'],
-            os.stat('static/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)
 
     for r in sorted(repos.values(), key = lambda r: r.name):
         write_to('r/%s/index.html' % r.name, summary(r))
@@ -346,7 +406,7 @@ def generate(output):
                 # 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 e.status == 404:
+                if is_404(e):
                     print('404 in tag %s (%s)' % (tag_name, obj_id))
                 else:
                     raise
@@ -359,6 +419,7 @@ def main():
     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()
 
@@ -367,22 +428,25 @@ def main():
 
     try:
         load_config(opts.config)
-    except configparser.NoOptionError as e:
+    except (configparser.NoOptionError, ValueError) as e:
         print('Error parsing config:', e)
+        return
 
     if not args:
         parser.error('Must specify an action (serve|generate)')
 
     if opts.only:
-        global repos
-        repos = [ r for r in repos if r.name in opts.only ]
+        for rname in list(repos.keys()):
+            if rname not in opts.only:
+                del repos[rname]
 
     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)
+        generate(output = opts.output,
+                skip_index = len(opts.only) > 0)
     else:
         parser.error('Unknown action %s' % args[0])
 

git.py

@@ -41,7 +41,7 @@ class EncodeWrapper:
         return s.decode(self.encoding, errors = self.errors)
 
 
-def run_git(repo_path, params, stdin = None):
+def run_git(repo_path, params, stdin = None, silent_stderr = False):
     """Invokes git with the given parameters.
 
     This function invokes git with the given parameters, and returns a
@@ -49,11 +49,17 @@ def run_git(repo_path, params, stdin = None):
     """
     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)
+        p = subprocess.Popen(params,
+                stdin = None, stdout = subprocess.PIPE, stderr = stderr)
     else:
         p = subprocess.Popen(params,
-                stdin = subprocess.PIPE, stdout = subprocess.PIPE)
+                stdin = subprocess.PIPE, stdout = subprocess.PIPE,
+                stderr = stderr)
         p.stdin.write(stdin)
         p.stdin.close()
 
@@ -201,11 +207,13 @@ class Repo:
         """Returns a GitCommand() on our path."""
         return GitCommand(self.path, cmd)
 
-    def for_each_ref(self, pattern = None, sort = None):
+    def for_each_ref(self, pattern = None, sort = None, count = None):
         """Returns a list of references."""
         cmd = self.cmd('for-each-ref')
         if sort:
             cmd.sort = sort
+        if count:
+            cmd.count = count
         if pattern:
             cmd.arg(pattern)
 
@@ -319,7 +327,7 @@ class Repo:
             ref = self.branch
         return Tree(self, ref)
 
-    def blob(self, path, ref = None):
+    def blob(self, path, ref = None, raw = False):
         """Returns the contents of the given path."""
         if not ref:
             ref = self.branch
@@ -335,8 +343,21 @@ class Repo:
         if not head or head.strip().endswith('missing'):
             return None
 
+        # Raw option in case we need a binary blob and not a utf-8 encoded one.
+        if raw:
+            return out.fd.read()
+
         return out.read()
 
+    def last_commit_timestamp(self):
+        """Return the timestamp of the last commit."""
+        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):
     """A git commit."""
@@ -391,7 +412,12 @@ class Commit (object):
     @staticmethod
     def from_str(repo, buf):
         """Parses git rev-list output, returns a commit object."""
-        header, raw_message = buf.split('\n\n', 1)
+        if '\n\n' in buf:
+            # Header, commit message
+            header, raw_message = buf.split('\n\n', 1)
+        else:
+            # Header only, no commit message
+            header, raw_message = buf.rstrip(), '    '
 
         header_lines = header.split('\n')
         commit_id = header_lines.pop(0)

hooks/README

@@ -0,0 +1,14 @@
+
+You can use the post-receive hook to automatically generate the repository
+view after a push.
+
+To do so, configure in your target repository the following options:
+
+  $ git config hooks.git-arr-config /path/to/site.conf
+  $ git config hooks.git-arr-output /var/www/git/
+
+  # Only if the git-arr executable is not on your $PATH.
+  $ git config hooks.git-arr-path /path/to/git-arr
+
+Then copy the post-receive file to the "hooks" directory in your repository.
+

hooks/post-receive

@@ -0,0 +1,58 @@
+#!/bin/sh
+#
+# git-arr post-receive hook
+#
+# This is a script intended to be used as a post-receive hook, which updates
+# its git-arr view.
+#
+# You should place it /path/to/your/repository.git/hooks/.
+
+# Config
+# --------
+#
+# hooks.git-arr-config
+#   The git-arr configuration file to use. Mandatory.
+#   Example: /srv/git-arr/site.conf
+#
+# hooks.git-arr-output
+#   Directory for the generated output. Mandatory.
+#   Example: /srv/www/git/
+#
+# hooks.git-arr-path
+#   The path to the git-arr executable. Optional, defaults to "git-arr".
+#
+# hooks.git-arr-repo-name
+#   The git-arr repository name. Optional, defaults to the path name.
+
+git_arr_config="$(git config --path hooks.git-arr-config)"
+git_arr_output="$(git config --path hooks.git-arr-output)"
+
+git_arr_path="$(git config --path hooks.git-arr-path 2> /dev/null)"
+git_arr_repo_name="$(git config hooks.git-arr-repo-name 2> /dev/null)"
+
+if [ -z "$git_arr_config" -o -z "$git_arr_output" ]; then
+	echo "Error: missing config options."
+	echo "Both hooks.git-arr-config and hooks.git-arr-output must be set."
+	exit 1
+fi
+
+if [ -z "$git_arr_path" ]; then
+	git_arr_path=git-arr
+fi
+
+if [ -z "$git_arr_repo_name" ]; then
+	PARENT_DIR=$(cd $(dirname "$0")/..; echo "$PWD")
+	git_arr_repo_name=$(basename "$PARENT_DIR")
+fi
+
+echo "Running git-arr"
+$git_arr_path --config "$git_arr_config" generate \
+	--output "$git_arr_output" \
+	--only "$git_arr_repo_name" > /dev/null
+RESULT=$?
+
+if [ $RESULT -ne 0 ]; then
+	echo "Error running git-arr"
+	exit $RESULT
+fi
+

sample.conf

@@ -38,8 +38,8 @@ path = /srv/git/repo/
 
 # File name to get the git URLs from (optional).
 # If git_url is not set, attempt to get its value from this file.
-# Default: "git_url"
-#git_url_file = git_url
+# Default: "cloneurl" (same as gitweb).
+#git_url_file = cloneurl
 
 # Do we look for repositories within this path? (optional).
 # This option enables a recursive, 1 level search for repositories within the

static/git-arr.css

@@ -100,6 +100,26 @@ span.tag {
     background-color: #ffff88;
 }
 
+/* Age of an object.
+ * Note this is hidden by default as we rely on javascript to show it. */
+span.age {
+    display: none;
+    color: gray;
+    font-size: x-small;
+}
+
+span.age-band0 {
+    color: darkgreen;
+}
+
+span.age-band1 {
+    color: green;
+}
+
+span.age-band2 {
+    color: seagreen;
+}
+
 /* Commit message and diff. */
 pre.commit-message {
     font-size: large;

static/git-arr.js

@@ -0,0 +1,63 @@
+/* Miscellaneous javascript functions for git-arr. */
+
+/* Return the current timestamp. */
+function now() {
+    return (new Date().getTime() / 1000);
+}
+
+/* Return a human readable string telling "how long ago" for a timestamp. */
+function how_long_ago(timestamp) {
+    if (timestamp < 0)
+        return "never";
+
+	var seconds = Math.floor(now() - timestamp);
+
+	var interval = Math.floor(seconds / (365 * 24 * 60 * 60));
+	if (interval > 1)
+		return interval + " years ago";
+
+	interval = Math.floor(seconds / (30 * 24 * 60 * 60));
+	if (interval > 1)
+		return interval + " months ago";
+
+	interval = Math.floor(seconds / (24 * 60 * 60));
+
+	if (interval > 1)
+		return interval + " days ago";
+	interval = Math.floor(seconds / (60 * 60));
+
+	if (interval > 1)
+		return interval + " hours ago";
+
+	interval = Math.floor(seconds / 60);
+	if (interval > 1)
+		return interval + " minutes ago";
+
+    if (seconds > 1)
+        return Math.floor(seconds) + " seconds ago";
+
+    return "about now";
+}
+
+/* Go through the document and replace the contents of the span.age elements
+ * with a human-friendly variant, and then show them. */
+function replace_timestamps() {
+    var elements = document.getElementsByClassName("age");
+    for (var i = 0; i < elements.length; i++) {
+        var e = elements[i];
+
+        var timestamp = e.innerHTML;
+        e.innerHTML = how_long_ago(timestamp);
+        e.style.display = "inline";
+
+        if (timestamp > 0) {
+            var age = now() - timestamp;
+            if (age < (2 * 60 * 60))
+                e.className = e.className + " age-band0";
+            else if (age < (3 * 24 * 60 * 60))
+                e.className = e.className + " age-band1";
+            else if (age < (30 * 24 * 60 * 60))
+                e.className = e.className + " age-band2";
+        }
+    }
+}

utils.py

@@ -12,14 +12,59 @@ try:
 except ImportError:
     pygments = None
 
+try:
+    import markdown
+except ImportError:
+    markdown = None
+
+import base64
+import mimetypes
 
 def shorten(s, width = 60):
     if len(s) < 60:
         return s
     return s[:57] + "..."
 
-def has_colorizer():
-    return pygments is not None
+def can_colorize(s):
+    """True if we can colorize the string, False otherwise."""
+    if pygments is None:
+        return False
+
+    # Pygments can take a huge amount of time with long files, or with very
+    # long lines; these are heuristics to try to avoid those situations.
+    if len(s) > (512 * 1024):
+        return False
+
+    # 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)
+        if pos == -1:
+            break
+
+        if pos - start > 300:
+            return False
+        start = pos + 1
+
+    return True
+
+def can_markdown(repo, fname):
+    """True if we can process file through markdown, False otherwise."""
+    if markdown is None:
+        return False
+
+    if not repo.info.embed_markdown:
+        return False
+
+    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' ]))
 
 def colorize_diff(s):
     lexer = lexers.DiffLexer(encoding = 'utf-8')
@@ -30,12 +75,37 @@ def colorize_diff(s):
 
 def colorize_blob(fname, s):
     try:
-        lexer = lexers.guess_lexer_for_filename(fname, s)
+        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('#!'):
+            try:
+                lexer = lexers.guess_lexer(s[:80], encoding = 'utf-8')
+            except lexers.ClassNotFound:
+                pass
+
     formatter = HtmlFormatter(encoding = 'utf-8',
                     cssclass = 'source_code',
-                    linenos = 'table')
+                    linenos = 'table',
+                    anchorlinenos = True,
+                    lineanchors = 'line')
 
     return highlight(s, lexer, formatter)
 
+def markdown_blob(s):
+    return markdown.markdown(s)
+
+def embed_image_blob(repo, dirname, fname):
+    mimetype = mimetypes.guess_type(fname)[0]
+
+    # Unfortunately, bottle seems to require utf-8 encoded data.
+    # We have to refetch the blob with raw=True, because the utf-8 encoded
+    # version of the blob available in the bottle template discards binary data.
+    raw_blob = repo.blob(dirname + fname, raw = True)
+
+    return '<img style="max-width:100%;" src="data:{0};base64,{1}" />'.format( \
+                                    mimetype, base64.b64encode(raw_blob))
+

views/blob.html

@@ -29,14 +29,20 @@
     <a href="{{relroot}}">[{{repo.branch}}]</a> /
 % base = smstr(relroot)
 % for c in dirname.split('/'):
-%   if not c.raw: continue
+%   if not c.raw:
+%     continue
+%   end
     <a href="{{base.url}}{{c.url}}/">{{c.unicode}}</a> /
 %   base += c + '/'
 % end
     <a href="">{{!fname.html}}</a>
 </h3>
 
-% if has_colorizer():
+% if can_embed_image(repo, fname.unicode):
+{{!embed_image_blob(repo, dirname.raw, fname.raw)}}
+% elif can_markdown(repo, fname.unicode):
+{{!markdown_blob(blob)}}
+% elif can_colorize(blob):
 {{!colorize_blob(fname.unicode, blob)}}
 % else:
 <pre class="blob-body">

views/commit.html

@@ -22,10 +22,10 @@
             <span class="date" title="{{c.author_date}}">
                 {{c.author_date.utc}} UTC</span></td></tr>
     <tr><td>committer</td>
-        <td><span class="name">{{c.author_name}}</span>
-            <span class="email">&lt;{{c.author_email}}&gt;</span><br/>
-            <span class="date" title="{{c.author_date}}">
-                {{c.author_date.utc}} UTC</span></td></tr>
+        <td><span class="name">{{c.committer_name}}</span>
+            <span class="email">&lt;{{c.committer_email}}&gt;</span><br/>
+            <span class="date" title="{{c.committer_date}}">
+                {{c.committer_date.utc}} UTC</span></td></tr>
 
 % for p in c.parents:
     <tr><td>parent</td>
@@ -55,7 +55,7 @@
 
 <hr/>
 
-% if has_colorizer():
+% if can_colorize(c.diff.body):
 {{!colorize_diff(c.diff.body)}}
 % else:
 <pre class="diff-body">

views/index.html

@@ -5,9 +5,10 @@
 <title>git</title>
 <link rel="stylesheet" type="text/css" href="static/git-arr.css"/>
 <meta http-equiv="content-type" content="text/html; charset=utf-8"/>
+<script src="static/git-arr.js"></script>
 </head>
 
-<body class="index">
+<body class="index" onload="replace_timestamps()">
 <h1>git</h1>
 
 <table class="nice">
@@ -20,6 +21,7 @@
     <tr>
         <td><a href="r/{{repo.name}}/">{{repo.name}}</a></td>
         <td><a href="r/{{repo.name}}/">{{repo.info.desc}}</a></td>
+        <td><span class="age">{{repo.last_commit_timestamp()}}</span></td>
     </tr>
     %end
 </table>

views/tree.html

@@ -27,14 +27,16 @@
     <a href="{{relroot}}">[{{repo.branch}}]</a> /
 % base = smstr(relroot)
 % for c in dirname.split('/'):
-%   if not c.raw: continue
+%   if not c.raw:
+%     continue
+%   end
     <a href="{{base.url}}{{c.url}}/">{{c.unicode}}</a> /
 %   base += c + '/'
 % end
 </h3>
 
 <table class="nice ls">
-% key_func = lambda (t, n, s): (0 if t == 'tree' else 1, n.raw)
+% key_func = lambda (t, n, s): (t != 'tree', n.raw)
 % for type, name, size in sorted(tree.ls(dirname.raw), key = key_func):
     <tr class="{{type}}">
 %   if type == "blob":