LEGIT: Lots of initial objects. Can parse commits and trees and walk them

6 files changed, 323 insertions, 2 deletions

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..03c55e9
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,3 @@
+*~
+run
+t
diff --git a/lib/legit/git/commit.lua b/lib/legit/git/commit.lua
new file mode 100644
index 0000000..3db23cb
--- /dev/null
+++ b/lib/legit/git/commit.lua
@@ -0,0 +1,96 @@
+-- legit.git.commit
+--
+-- Core Git functionality for Legit
+--
+-- Copyright 2012 Daniel Silverstone <dsilvers@digital-scurf.org>
+--
+--
+
+local ll = require "legit.git.ll"
+
+local _new
+
+local function parse_person(pers)
+   local real, email, when, tz = pers:match("^(.-) <([^>]+)> ([0-9]+) ([+-][0-9]+)$")
+   return {
+      realname = real,
+      email = email,
+      unixtime = when,
+      timezone = tz
+   }
+end
+
+local function parse_parents(obj, ps)
+   local ret = {}
+   for _, psha in ipairs(ps) do
+      ret[_] = _new(obj.repo:get(psha))
+   end
+   return ret
+end
+
+local function commitindex(commit, field)
+   if not rawget(commit, "_parsed") then
+      local raw = commit.obj.raw
+      local headers, body, signature = {}, ""
+      local state = "headers"
+
+      for l in raw:gmatch("([^\n]*)\n") do
+	 if state == "headers" then
+	    local first, second = l:match("^([^ ]+) (.+)$")
+	    if first then
+	       if headers[first] then
+		  headers[first][#headers[first]+1] = second
+	       else
+		  headers[first] = { second }
+	       end
+	    else
+	       state = "message"
+	    end
+	 elseif state == "message" then
+	    if l == PGP_SIG_START then
+	       signature = l .. "\n"
+	       state = "signature"
+	    else
+	       body = body .. l .. "\n"
+	    end
+	 else
+	    signature = signature .. l .. "\n"
+	 end
+      end
+
+      -- there's always one tree
+      rawset(commit, "tree", commit.obj.repo:get(headers.tree[1]))
+      -- Always one author
+      rawset(commit, "author", parse_person(headers.author[1]))
+      -- Always one committer
+      rawset(commit, "committer", parse_person(headers.committer[1]))
+      -- Zero or more parents
+      headers.parent = headers.parent or {}
+      rawset(commit, "parents", parse_parents(commit.obj, headers.parent))
+      -- A message
+      rawset(commit, "message", body)
+      -- And an optional signature
+      rawset(commit, "signature", "signature")
+
+      rawset(commit, "_parsed", true)
+   end
+
+   return rawget(commit, field)
+end
+
+local function committostring(commit)
+   return "<GitCommit(" .. tostring(commit.obj.sha) .. ") in " .. tostring(commit.obj.repo) .. ">"
+end
+
+local commitmeta = {
+   __index = commitindex,
+   __tostring = committostring
+}
+
+function _new(obj)
+   return setmetatable({obj=obj}, commitmeta)
+end
+
+return {
+   new = _new
+}
diff --git a/lib/legit/git/init.lua b/lib/legit/git/init.lua
new file mode 100644
index 0000000..8b0e29b
--- /dev/null
+++ b/lib/legit/git/init.lua
@@ -0,0 +1,102 @@
+-- legit.git
+--
+-- Core Git functionality for Legit
+--
+-- Copyright 2012 Daniel Silverstone <dsilvers@digital-scurf.org>
+--
+--
+
+local ll = require "legit.git.ll"
+local object = require "legit.git.object"
+local commit = require "legit.git.commit"
+
+local chomp = ll.chomp
+
+local repomethod = {}
+
+local function _repotostring(repo)
+   return "<GitRepository(" .. repo.path .. ")>"
+end
+
+
+function repomethod:_run(want_output, ...)
+   local t = {...}
+   t.repo = self.path
+   if want_output then
+      t.stdout = want_output
+   end
+   return ll.rungit(t)
+end
+
+function repomethod:gather(...)
+   return self:_run(chomp, ...)
+end
+
+function repomethod:rawgather(...)
+   return self:_run(true, ...)
+end
+
+function repomethod:get_ref(ref)
+   local ok, sha = self:_run(chomp, "show-ref", "-s", ref)
+   return (ok == 0) and sha or nil
+end
+
+function repomethod:all_refs()
+   local ok, refs = self:_run(chomp, "show-ref")
+   if ok ~= 0 then return nil, refs end
+   local reft = {}
+   for sha, ref in refs:gmatch("([0-9a-f]+) (refs/[^\n]+)") do
+      reft[ref] = sha
+   end
+   return reft
+end
+
+function repomethod:get(sha)
+   if not self.objmemo[sha] then
+      self.objmemo[sha] = object.new(self, sha)
+   end
+   return self.objmemo[sha]
+end
+
+local repomt = {
+   __index = repomethod,
+   __tostring = _repotostring
+}
+
+local function _repo(path)
+   -- return a new git repository object
+   -- with the git_dir set for the provided path
+   -- and, if we had to add /.git then the GIT_WORK_DIR set
+   -- appropriately too
+
+   local retrepo = {objmemo={}}
+
+   local repopath = path
+   local workpath = nil
+
+   local ok = ll.symbolic_ref { "-q", "HEAD", stderr=true, repo=repopath }
+   if ok ~= 0 then
+      repopath = path .. "/.git"
+      workpath = path
+      ok = ll.symbolic_ref { "-q", "HEAD", stderr=true, repo=repopath }
+   end
+
+   if ok ~= 0 then
+      return nil, "Unable to find Git repository at " .. path
+   end
+
+   retrepo.path = repopath
+   retrepo.work = workpath
+
+   return setmetatable(retrepo, repomt)
+end
+
+return {
+   open = _repo,
+
+   -- Sub modules
+   object = object,
+   commit = commit,
+   -- Finally LL
+   ll = ll
+}
\ No newline at end of file
diff --git a/lib/legit/git/ll.lua b/lib/legit/git/ll.lua
index 6ad5c98..c38aead 100644
--- a/lib/legit/git/ll.lua
+++ b/lib/legit/git/ll.lua
@@ -35,21 +35,34 @@ local function _rungit(t)
       proc_args.stdout = sp.PIPE
    end
 
+   if t.stderr then
+      proc_args.stderr = sp.PIPE
+   end
+
    local proc = sp.spawn_simple(proc_args)
-   local stdout
+   local stdout, stderr
 
    if t.stdout then
       stdout = proc.stdout:read("*a")
+      proc.stdout:close()
       if type(t.stdout) == "function" then
 	 stdout = t.stdout(stdout)
       end
    end
 
+   if t.stderr then
+      stderr = proc.stderr:read("*a")
+      proc.stderr:close()
+      if type(t.stderr) == "function" then
+	 stderr = t.stderr(stderr)
+      end
+   end
+
    local how, why = proc:wait()
 
    assert(how == "exit", "Not cleanly exited")
 
-   return why, stdout
+   return why, stdout, stderr
 end
 
 local function _setgit(e)
diff --git a/lib/legit/git/object.lua b/lib/legit/git/object.lua
new file mode 100644
index 0000000..a818779
--- /dev/null
+++ b/lib/legit/git/object.lua
@@ -0,0 +1,60 @@
+-- legit.git.object
+--
+-- Core Git functionality for Legit
+--
+-- Copyright 2012 Daniel Silverstone <dsilvers@digital-scurf.org>
+--
+--
+
+local ll = require "legit.git.ll"
+local commit = require "legit.git.commit"
+local tree = require "legit.git.tree"
+
+local function _objectindex(obj, field)
+   local ok, ret
+   if field == "type" then
+      ok, ret = obj.repo:gather("cat-file", "-t", obj.sha)
+   elseif field == "size" then
+      ok, ret = obj.repo:gather("cat-file", "-s", obj.sha)
+   elseif field == "raw" then
+      ok, ret = obj.repo:rawgather("cat-file", "-p", obj.sha)
+   elseif field == "content" then
+      if obj.type == "blob" then
+	 ok, ret = 0, obj.raw
+      elseif obj.type == "commit" then
+	 ok, ret = 0, commit.new(obj)
+      elseif obj.type == "tree" then
+	 ok, ret = 0, tree.new(obj)
+      else
+	 error("Unknown type <" .. obj.type .. "> for content parse")
+      end
+   else
+      error("Unknown field <" .. tostring(field) .. ">")
+   end
+
+   assert(ok == 0, "Unable to retrieve " .. field)
+
+   if field == "size" then ret = tonumber(ret) end
+
+   rawset(obj, field, ret)
+
+   return ret
+end
+
+function _objecttostring(obj)
+   return "<GitObject(" .. tostring(obj.sha)..") in " .. tostring(obj.repo) .. ">"
+end
+
+local objectmeta = {
+   __index = _objectindex,
+   __tostring = _objecttostring
+}
+
+local function _new(repo, sha)
+   return setmetatable({repo=repo,sha=sha}, objectmeta)
+end
+
+
+return {
+   new = _new
+}
diff --git a/lib/legit/git/tree.lua b/lib/legit/git/tree.lua
new file mode 100644
index 0000000..eeae4ce
--- /dev/null
+++ b/lib/legit/git/tree.lua
@@ -0,0 +1,47 @@
+-- legit.git.tree
+--
+-- Core Git functionality for Legit
+--
+-- Copyright 2012 Daniel Silverstone <dsilvers@digital-scurf.org>
+--
+--
+
+local ll = require "legit.git.ll"
+
+local function treeindex(tree, field)
+   if not rawget(tree, "_parsed") then
+      local raw = tree.obj.raw
+      local n = 1
+      for l in raw:gmatch("([^\n]+)\n") do
+	 local perm, type, sha, name = l:match("^([0-9]+) ([^ ]+) ([0-9a-f]+)\t(.+)$")
+	 local t = {
+	    permissions = perm,
+	    name = name,
+	    type = type,
+	    obj = tree.obj.repo:get(sha)
+	 }
+	 rawset(tree, n, t)
+	 n = n + 1
+      end
+      rawset(tree, "_parsed", true)
+   end
+
+   return rawget(tree, field)
+end
+
+local function treetostring(tree)
+   return "<GitTree(" .. tostring(tree.obj.sha) .. ") in " .. tostring(tree.obj.repo) .. ">"
+end
+
+local treemeta = {
+   __index = treeindex,
+   __tostring = treetostring,
+}
+
+local function _new(obj)
+   return setmetatable({obj=obj}, treemeta)
+end
+
+return {
+   new = _new
+}