Implement enough to read bytecode instructions

microLua.lua

@@ -2,4 +2,5 @@ local readbc = require('readbc')
 
 fp = io.open('tests/test1.luo')
 header = readbc.read_header(fp)
-print(header.name)
+proto = readbc.read_proto(fp)
+print(#proto.bcins)

readbc.lua

@@ -5,13 +5,53 @@ local bit=require('bit')
 local HEADER_MAGIC = {27, string.byte('L'), string.byte('J')}
 local HEADER_FLAG_STRIP = 0x02
 
+local function read_u8(fp)
+    return string.byte(fp:read(1))
+end
+
+local function read_u16(fp)
+    -- TODO: endianess
+    local b1 = read_u8(fp)
+    local b2 = read_u8(fp)
+
+    return bit.bor(b2*0x100, b1)
+end
+
+local function read_u32(fp)
+    -- TODO: endianess
+    local h1 = read_u16(fp)
+    local h2 = read_u16(fp)
+
+    return bit.bor(h2*0x10000, h1)
+end
+
+local function read_u8n(fp, n)
+    return { string.byte(fp:read(n), 1, n) }
+end
+
+local function read_u16n(fp, n)
+    local res = {}
+    for i = 1, n do
+        res[i] = read_u16(fp)
+    end
+    return res
+end
+
+local function read_u32n(fp, n)
+    local res = {}
+    for i = 1, n do
+        res[i] = read_u32(fp)
+    end
+    return res
+end
+
 local function read_uleb128(fp)
     local last_byte = false
     local res = 0
     local shift = 0
     -- the bytes are read from the LSB to the MSB
     while not last_byte do
-        local uleb_byte = string.byte(fp:read(1))
+        local uleb_byte = read_u8(fp)
         res = bit.bor(res, bit.lshift(bit.band(uleb_byte, 0x7f), shift))
 
         if bit.band(uleb_byte, 0x80) == 0 then
@@ -22,8 +62,8 @@ local function read_uleb128(fp)
 end
 
 local function read_header(fp)
-    local magic = { string.byte(fp:read(3), 1, 3) }
-    local version = string.byte(fp:read(1))
+    local magic = read_u8n(fp, 3)
+    local version = read_u8(fp)
     local flags = read_uleb128(fp)
 
     local name
@@ -43,6 +83,57 @@ local function read_header(fp)
 end
 
 local function read_proto(fp)
+    local length = read_uleb128(fp)
+    -- if the the last read proto was the last one
+    -- we have reached the end marker, which is a zero
+    -- this overlaps with a uleb128 zero representation
+    if length == 0 then
+        return
+    end
+
+    -- read proto header
+    local flags = read_u8(fp)
+    local numparams = read_u8(fp)
+    local framesize = read_u8(fp)
+    local numuv = read_u8(fp)
+    local numkgc = read_uleb128(fp)
+    local numkn = read_uleb128(fp)
+    local numbc = read_uleb128(fp)
+
+    local debugLen = 0
+    local debugInfo = nil
+
+    if bit.band(flags, HEADER_FLAG_STRIP) ~= 0 then
+        debugLen = read_uleb128(fp)
+
+        -- TODO: is this correct?
+        if debugLen ~= 0 then
+            local firstLine = read_uleb128(fp)
+            local numLine = read_uleb128(fp)
+            debugInfo = {
+                firstLine = firstLine,
+                numLine = numLine,
+            }
+        end
+    end
+
+    local bcins = read_u32n(fp, numbc)
+    local uvdata = read_u16n(fp, numuv)
+    -- local kgc
+    -- local numkn
+
+    return {
+        flags = flags,
+        numparams = numparams,
+        framsize = framesize,
+        numuv = numuv,
+        numkgc = numkgc,
+        numkn = numkn,
+        numbc = numbc,
+        debugInfo = debugInfo,
+        bcins = bcins,
+        uvdata = uvdata,
+    }
 end
 
 return {