repobrowse 0.0.0

data/lib/repobrowse/git_commit_html.rb

@@ -0,0 +1,320 @@
+# -*- encoding: utf-8 -*-
+# Copyright (C) 2017-2018 all contributors <repobrowse-public@80x24.org>
+# License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
+# frozen_string_literal: true
+require_relative 'html'
+
+class Repobrowse::GitCommitHTML < Repobrowse::HTML
+
+  # rugged doesn't seem to have a way to show diffstats, decorations
+  # or combined diffs (--cc/--combined) for merges, so use git-show here
+  CMT_CMD = %w(show -z --numstat -p --encoding=UTF-8 --pretty=format:%p%n%d%x00)
+
+  def commit_header(env, repo, cmt)
+    msg = Rugged.prettify_message(cmt.message)
+    subject, body = msg.split(/\r?\n\r?\n/, 2)
+    subject.strip!
+    ht(subject)
+    start(subject, repo)
+    parents = cmt.parents
+    oid = @commit = cmt.oid
+    @buf << '   commit ' + oid
+    @mhelp = nil
+    case parents.size
+    when 0
+    when 1
+      @buf << %Q[ (<a\nhref="#{oid}.patch">patch</a>)]
+      pfx = '   parent'
+    else
+      @mhelp = "\n This is a merge, showing combined diff:\n\n"
+      pfx = '  parents'
+    end
+
+    @buf << "\n     "
+    @buf << %Q(tree <a\nrel=nofollow\nhref="src/#{oid}">#{cmt.tree_oid}</a>\n)
+    pad = 0
+    idents = [ '   author', 'committer' ].map do |field|
+      x = cmt.__send__(field.strip)
+      name_email = ht(+"#{x[:name]} <#{x[:email]}>")
+      len = name_email.size
+      pad = len if len > pad
+      [ field, name_email, x[:time].strftime('%Y-%m-%d %k:%M:%S %z') ]
+    end
+    idents.each do |field, name_email, time|
+      @buf << "#{field} #{[name_email].pack("A#{pad}")}\t#{time}\n"
+    end
+    @repo = repo
+    cmd = CMT_CMD.dup
+    cmd << @commit
+    cmd << '--'
+    @rd = repo.driver.popen(cmd, encoding: Encoding::UTF_8)
+    abbr = @rd.gets(chomp: true).split(' ')
+    parents.each_with_index do |pt, i|
+      title = Rugged.prettify_message(pt.summary)
+      title.strip!
+      @buf << %Q(#{pfx} <a id=P#{i}\nhref="#{pt.oid}">#{abbr[i]}</a> #{ht(title)}\n)
+      pfx = '         '
+    end
+    refnames = @rd.gets("\0", chomp: true)
+    @buf << "\n<b>#{subject}</b>#{ht(refnames)}\n\n"
+    @buf << ht(body) if body
+    @buf << "<a\nid=D>---</a>\n"
+    @anchors = {}
+    @parents = parents
+    @nchg = @nadd = @ndel = 0
+    @state = :stat_begin
+  end
+
+  # do not break anchor links if the combined diff doesn't show changes:
+  def show_unchanged
+    unchanged = @anchors.keys.sort!
+    unchanged[0] or return ''
+    buf = +<<EOS
+
+ There are uninteresting changes from this merge.
+ See the <a\nhref="#P0">parents</a>, or view final state(s) below:
+
+EOS
+
+    unchanged.each do |anchor|
+      fn = @repo.driver.git_unquote(@anchors[anchor])
+      href = ha(+"src/#@commit:#{fn}")
+      buf << "\t<a\nrel=nofollow\nid=#{
+        anchor}
+        }\nhref=#{href}>#{ht(+fn)}</a>\n"
+    end
+    buf
+  end
+
+  def diffstat_rename_line(from, to)
+    anchor = -to_anchor(to)
+    @anchors[anchor] = to
+    from_parts = from.split('/')
+    to_parts = to.split('/')
+    base = []
+    while to_parts[0] && to_parts[0] == from_parts[0]
+      base << to_parts.shift
+      from_parts.shift
+    end
+    from = from_parts.join('/')
+    to = to_parts.join('/')
+    to = %Q(<a\nhref="##{anchor}">#{ht(to)}</a>)
+    if base[0]
+      base = ht(base.join('/'))
+      "#{base}/{#{from} =&gt; #{to}}"
+    else
+      "#{from} =&gt; #{to}"
+    end
+  end
+
+  def diffstat_line(line)
+    line =~ /\A(\S+)\t+(\S+)\t+(.*)/ or die("bad stat line: #{line.inspect}")
+    add = -$1
+    del = -$2
+    fn = -$3
+    if fn != '' # normal modification
+      anchor = -to_anchor(fn)
+      @anchors[anchor] = -fn
+      line = %Q(<a\nhref="##{anchor}">#{ht(fn.dup)}</a>)
+    else # rename
+      from = @rd.gets("\0", chomp: true) or die('EOF rename (from)')
+      to = @rd.gets("\0", chomp: true) or die('EOF rename (to)')
+      line = diffstat_rename_line(from, to);
+    end
+
+    # text changes show numerically, Binary does not
+    if add =~ /\A\d+\z/ && del =~ /\A\d+\z/
+      @nadd += add.to_i
+      @ndel += del.to_i
+      add = "+#{add}"
+      del = "-#{del}"
+    else # just in case...
+      ht(add)
+      ht(del)
+    end
+    @nchg += 1
+    " #{sprintf('% 6s/%-6s', del, add)}\t#{line}\n"
+  end
+
+  def diffstat_end
+    ret = +"\n #@nchg "
+    ret << (@nchg == 1 ? 'file changed, ' : 'files changed, ')
+    ret << @nadd.to_s
+    ret << (@nadd == 1 ? ' insertion(+), ' : ' insertions(+), ')
+    ret << @ndel.to_s
+    ret << (@ndel == 1 ? " deletion(-)\n\n" : " deletions(-)\n\n")
+  end
+
+  def die(msg)
+    raise RuntimeError, "#{msg} (#@commit)", []
+  end
+
+  # diff --git a/foo.c b/bar.c
+  def git_diff_ab_hdr(fa, fb)
+    html_a = ht(fa.dup)
+    html_b = ht(fb.dup)
+    fa = @repo.driver.git_unquote(fa)
+    fb = @repo.driver.git_unquote(fb)
+    fa.sub!(%r{\Aa/}, '')
+    fb.sub!(%r{\Ab/}, '')
+    anchor = -to_anchor(fb)
+    @anchors.delete(anchor)
+    @fa = fa
+    @fb = fb
+    # not wasting bandwidth on links here
+    # links in hunk headers are far more useful with line offsets
+    %Q(<a\nid="#{anchor}">diff</a> --git #{html_a} #{html_b}\n)
+  end
+
+    # diff (--cc|--combined)
+  def git_diff_cc_hdr(combined, path)
+    html_path = ht(path.dup)
+    path = @repo.driver.git_unquote(path)
+    anchor = to_anchor(path)
+    @anchors.delete(anchor)
+    @path_cc = path
+    %Q(<a\nid="#{anchor}">diff</a> --#{combined} #{html_path}\n)
+  end
+
+  # index abcdef89..01234567
+  def git_diff_ab_index(da, db, tail)
+    # not wasting bandwidth on links here, yet
+    # links in hunk headers are far more useful with line offsets
+    "index #{da}..#{db}#{ht(tail)}\n"
+  end
+
+  def git_diff_src_link(ref, file, lineno, text)
+    fragment = lineno ? "#n#{lineno}" : ''
+    href = ha(+"src/#{ref}:#{file}#{fragment}")
+    %Q(<a\nrel=nofollow\nhref=#{href}>#{text}</a>)
+  end
+
+  # @@ -1,2 +3,4 @@ (regular diff)
+  def git_diff_ab_hunk(ca, cb, ctx)
+    na = ca.match(/\A-(\d+)/)[1]
+    nb = cb.match(/\A\+(\d+)/)[1]
+
+    # we add "rel=nofollow" here to reduce load on search engines, here
+    rv = +'@@ '
+    rv << (na == '0' ? ca : git_diff_src_link(@parents[0], @fa, na, ca))
+    rv << ' '
+    rv << (nb == '0' ? cb : git_diff_src_link(@commit, @fb, nb, cb))
+    rv << " @@#{ht(ctx)}\n"
+  end
+
+  # index abcdef09,01234567..76543210
+  def git_diff_cc_index(before, last, tail)
+    ht(tail)
+    @parent_objs_cc = before.split(',')
+
+    # not wasting bandwidth on links here, yet
+    # links in hunk headers are far more useful with line offsets
+    "index #{before}..#{last}#{tail}\n"
+  end
+
+  # @@@ -1,2 -3,4 +5,6 @@@ (combined diff)
+  def git_diff_cc_hunk(at, offs, ctx)
+    offs = offs.split(' ')
+    last = offs.pop
+    rv = at.dup
+
+    offs.each_with_index do |off, i|
+      parent = @parents[i]
+      blob = @parent_objs_cc[i]
+      lineno = off.match(/\A-(\d+)/)[1]
+
+      if lineno == '0' # new file (does this happen with --cc?)
+        rv << " #{off}"
+      else
+        href = ha(+"src/#{parent}?id=#{blob}#n#{lineno}")
+        rv << %Q( <a\nhref=#{href}>#{off}</a>)
+      end
+    end
+
+    lineno = last.match(/\A\+(\d+)/)[1]
+    rv << ' '
+    if lineno == '0' # deleted file (does this happen with --cc?)
+      rv << last
+    else
+      rv << git_diff_src_link(@commit, @path_cc, lineno, last)
+    end
+    rv << " #{at}#{ht(ctx)}\n"
+  end
+
+  def diff_line(line)
+    # dfa and dfb class names match public-inbox search term prefix
+    case line
+    when /\A\+/
+      %Q{<span\nclass="dfa">#{ht(line.chomp!)}</span>\n}
+    when /\A\-/
+      %Q{<span\nclass="dfb">#{ht(line.chomp!)}</span>\n}
+    when %r{\Adiff --git ("?a/.+) ("?b/.+)\n\z} # regular
+      git_diff_ab_hdr($1, $2)
+    when /\Adiff --(cc|combined) (.+)\n\z/ # merge
+      git_diff_cc_hdr($1, $2)
+    when /\Aindex ([a-f0-9]+)\.\.([a-f0-9]+)(.*)\n\z/ # regular
+      git_diff_ab_index($1, $2, $3)
+    when /\A@@ ([\d,\+\-]+) ([\d,\+\-]+) @@(.*)\n\z/ # regular
+      git_diff_ab_hunk($1, $2, $3)
+    when /\Aindex ([a-f0-9]+,[^\.]+)\.\.([a-f0-9]+)(.*)\n\z/ # --cc
+      git_diff_cc_index($1, $2, $3)
+    when /\A(@@@+) (\S+.*\S+) @@@+(.*)\n\z/  # --cc
+      git_diff_cc_hunk($1, $2, $3)
+    when nil
+      diff_done
+    else
+      ht(line)
+    end
+  end
+
+  def diff_done
+    buf, @mhelp = @mhelp, nil
+    @state = :done
+    "#{buf}#{show_unchanged}</pre></body></html>"
+  end
+
+  # called by the Rack server
+  def each
+    buf = @buf
+    @buf = nil
+    yield buf
+    buf.clear
+    while buf = each_i
+      yield buf
+      buf.clear unless buf.frozen?
+    end
+  end
+
+  def each_i
+    case @state
+    when :stat_begin
+      # merges start with an extra '\0' before the diffstat
+      # non-merge commits start with an extra '\n', instead
+      sep = @mhelp ? "\0" : "\n"
+      @rd.gets(sep) == sep or die('diffstat line not empty')
+      @state = :stat
+    when :stat
+      case line = @rd.gets("\0", chomp: true)
+      when nil
+        if @mhelp
+          @mhelp = "\n This is a merge, and the combined diff is empty.\n"
+          return diff_done
+        end
+        die('premature EOF')
+      when ''
+        @state = :diff
+        return diffstat_end
+      else
+        return diffstat_line(line)
+      end
+    when :diff
+      return diff_line(@rd.gets)
+    when :done
+      return
+    end while true
+  end
+
+  def close
+    @rd = @rd&.close
+  end
+end

data/lib/repobrowse/git_disambiguate.rb

@@ -0,0 +1,21 @@
+# -*- encoding: binary -*-
+# Copyright (C) 2017-2018 all contributors <repobrowse-public@80x24.org>
+# License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
+# frozen_string_literal: true
+module Repobrowse::GitDisambiguate
+  def git_disambiguate(r, repo, pfx, ref, path)
+    ambiguous = false
+    ambiguous = true if ref.squeeze!('/')
+    ambiguous = true if ref.chomp!('/')
+    if path
+      ambiguous = true if path.chomp!('/')
+      ambiguous = true if path.squeeze!('/')
+    end
+    if ambiguous
+      loc = r.base_url
+      loc << "#{r.script_name}/#{repo.name}/#{pfx}/#{ref}"
+      loc << ":#{path}" if path
+      r.redirect(loc) # throws
+    end
+  end
+end

data/lib/repobrowse/git_http_backend.rb

@@ -0,0 +1,109 @@
+# Copyright (C) 2017-2018 all contributors <repobrowse-public@80x24.org>
+# License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
+# frozen_string_literal: true
+
+require 'time'
+
+# provides smart HTTP cloning for git repos
+module Repobrowse::GitHTTPBackend
+
+  # used to provide streaming Rack response body
+  class EachWrap
+    def initialize(io)
+      @io = io
+    end
+
+    def each
+      begin
+        buf = @io.readpartial(16384, buf)
+        yield buf
+      rescue EOFError
+        return
+      ensure
+        buf&.clear
+      end while true
+    end
+
+    def close
+      @io.close
+    end
+  end
+
+  def input_to_file(env)
+    tmp = Tempfile.new('git-http-backend-in')
+    tmp.unlink
+    tmp.sync = true
+    IO.copy_stream(env['rack.input'], tmp.to_io)
+    tmp.rewind
+    tmp
+  end
+
+  def run_http_backend(r, repo, path)
+    penv = {
+      'GIT_HTTP_EXPORT_ALL' => '1',
+      'PATH_TRANSLATED' => -"#{repo.path}/#{path}",
+    }
+    env = r.env
+    # GIT_COMMITTER_NAME, GIT_COMMITTER_EMAIL
+    # may be set in the server-process and are passed as-is
+    %w(QUERY_STRING REMOTE_USER REMOTE_ADDR HTTP_CONTENT_ENCODING
+       CONTENT_TYPE SERVER_PROTOCOL REQUEST_METHOD).each do |k|
+      v = env[k] and penv[k] = v
+    end
+    IO.popen(penv, %W(git http-backend), in: input_to_file(env))
+  rescue => e
+    r_err(r, "E: #{e.message} (#{e.class}) on #{repo.path}")
+  end
+
+  # returns undef if 403 so it falls back to dumb HTTP
+  def serve_smart(r, repo, path)
+    rd = run_http_backend(r, repo, path)
+    code = 200
+    h = {}
+
+    # parse CGI response headers
+    case rd.gets
+    when %r{\AStatus:\s+(\d+)}i
+      code = $1.to_i
+    when %r{\A([^:]+):\s*(.*)\r\n\z}
+      h[-$1] = -$2
+    when "\r\n"
+      break # headers done onto the body
+    when nil
+      rd.close
+      r_err(r, "unexpected EOF on git http-backend in #{repo.path}")
+    end while true
+
+    r.halt [ code, h, EachWrap.new(rd) ]
+  end
+
+  def git_http_backend_routes(r, repo)
+    r.is(path = 'git-upload-pack') { r.post { serve_smart(r, repo, path) } }
+    r.is(path = 'info/refs') {
+      # QUERY_STRING has exactly one parameter
+      # See https://80x24.org/git/src/v2.16.1:Documentation/technical/http-protocol.txt
+      if /\Aservice=git-\w+-pack\z/ =~ r.env['QUERY_STRING']
+        r.get { serve_smart(r, repo, path) }
+      else
+        static(r, -"#{repo.path}/#{path}", 'text/plain', nil)
+      end
+    }
+    %w(HEAD cloneurl description
+       objects/info/http-alternates
+       objects/info/alternates
+       objects/info/packs).each { |txt|
+      r.is(txt) { static(r, -"#{repo.path}/#{txt}", 'text/plain', nil) }
+    }
+    r.is('objects', :git_x2, :git_x38) { |x2, x38|
+      static(r, -"#{repo.path}/objects/#{x2}/#{x38}",
+             'application/x-git-loose-object')
+    }
+    [ :git_pack, 'application/x-git-packed-objects',
+      :git_pack_idx, 'application/x-git-packed-objects-toc'
+    ].each_slice(2) { |sym, type|
+      r.is('objects/pack', :sym) { |o|
+        static(r, -"#{repo.path}/objects/pack/#{o}", type)
+      }
+    }
+  end
+end