Merge pull request #6872 from xmake-io/opti

Improve hash

11 files changed, 144 insertions, 36 deletions

diff --git a/xmake/actions/config/configfiles.lua b/xmake/actions/config/configfiles.lua
index ce435d774..60e5ef378 100644
--- a/xmake/actions/config/configfiles.lua
+++ b/xmake/actions/config/configfiles.lua
@@ -292,7 +292,7 @@ function _generate_configfile(srcfile, dstfile, fileinfo, targets, preprocessors
         end
     else
         -- generate to the temporary file first
-        local dstfile_tmp = path.join(os.tmpdir(), hash.uuid4(srcfile))
+        local dstfile_tmp = os.tmpfile(srcfile)
         os.tryrm(dstfile_tmp)
         os.cp(srcfile, dstfile_tmp)
 
diff --git a/xmake/core/base/hash.lua b/xmake/core/base/hash.lua
index 362eb67e5..e846ba302 100644
--- a/xmake/core/base/hash.lua
+++ b/xmake/core/base/hash.lua
@@ -25,11 +25,15 @@ local hash  = hash or {}
 local io    = require("base/io")
 local utils = require("base/utils")
 local bytes = require("base/bytes")
+local libc  = require("base/libc")
 
 -- save metatable and builtin functions
 hash._md5 = hash._md5 or hash.md5
 hash._sha = hash._sha or hash.sha
 hash._xxhash = hash._xxhash or hash.xxhash
+hash._rand32 = hash._rand32 or hash.rand32
+hash._rand64 = hash._rand64 or hash.rand64
+hash._rand128 = hash._rand128 or hash.rand128
 
 -- generate md5 from the given file or data
 function hash.md5(file_or_data)
@@ -70,6 +74,24 @@ function hash.sha256(file_or_data)
     return hashstr, errors
 end
 
+-- generate uuid, e.g "91E8ECF1-417F-4EDF-A574-E22D7D8D204A"
+function hash.uuid(str)
+    return hash.uuid4(str)
+end
+
+-- generate xxhash32 from the given file or data
+function hash.xxhash32(file_or_data)
+    local hashstr, errors
+    if bytes.instance_of(file_or_data) then
+        local datasize = file_or_data:size()
+        local dataaddr = file_or_data:caddr()
+        hashstr, errors = hash._xxhash(32, dataaddr, datasize)
+    else
+        hashstr, errors = hash._xxhash(32, file_or_data)
+    end
+    return hashstr, errors
+end
+
 -- generate xxhash64 from the given file or data
 function hash.xxhash64(file_or_data)
     local hashstr, errors
@@ -96,20 +118,52 @@ function hash.xxhash128(file_or_data)
     return hashstr, errors
 end
 
--- generate uuid, e.g "91E8ECF1-417F-4EDF-A574-E22D7D8D204A"
-function hash.uuid(str)
-    return hash.uuid4(str)
-end
-
--- TODO, we should optimize it
 -- generate hash32 from string, e.g. "91e8ecf1"
 function hash.strhash32(str)
-    return hash.uuid4(str):split("-", {plain = true})[1]:lower()
+    local data = libc.ptraddr(libc.dataptr(str))
+    local size = #str
+    return hash._xxhash(32, data, size)
+end
+
+-- generate hash64 from string, e.g. "91e8ecf191e8ecf1"
+function hash.strhash64(str)
+    local data = libc.ptraddr(libc.dataptr(str))
+    local size = #str
+    return hash._xxhash(64, data, size)
 end
 
 -- generate hash128 from string, e.g. "91e8ecf1417f4edfa574e22d7d8d204a"
 function hash.strhash128(str)
-    return hash.uuid4(str):replace("-", "", {plain = true}):lower()
+    local data = libc.ptraddr(libc.dataptr(str))
+    local size = #str
+    return hash._xxhash(128, data, size)
+end
+
+-- generate random32 hash
+function hash.rand32()
+    if hash._rand32 then
+        return hash._rand32()
+    else
+        return hash.strhash32(tostring(math.random()))
+    end
+end
+
+-- generate random64 hash
+function hash.rand64()
+    if hash._rand64 then
+        return hash._rand64()
+    else
+        return hash.strhash64(tostring(math.random()))
+    end
+end
+
+-- generate random128 hash
+function hash.rand128()
+    if hash._rand128 then
+        return hash._rand128()
+    else
+        return hash.strhash128(tostring(math.random()))
+    end
 end
 
 -- return module: hash
diff --git a/xmake/core/base/os.lua b/xmake/core/base/os.lua
index 41fe07d22..988d9e4bf 100644
--- a/xmake/core/base/os.lua
+++ b/xmake/core/base/os.lua
@@ -761,7 +761,8 @@ function os.tmpfile(opt_or_key)
         key = opt_or_key.key
         opt = opt_or_key
     end
-    return path.join(os.tmpdir(opt), "_" .. (hash.uuid4(key):gsub("-", "")))
+    local filename = "_" .. (key and hash.strhash128(key) or (hash.rand128()))
+    return path.join(os.tmpdir(opt), filename)
 end
 
 -- exit program
diff --git a/xmake/core/base/scheduler.lua b/xmake/core/base/scheduler.lua
index c96b32441..63c9fe009 100644
--- a/xmake/core/base/scheduler.lua
+++ b/xmake/core/base/scheduler.lua
@@ -299,7 +299,7 @@ function scheduler:_co_curenvs_update(envs)
     for _, key in ipairs(table.orderkeys(envs)) do
         envs_hash = envs_hash .. key:upper() .. envs[key]
     end
-    envs_hash = hash.uuid4(envs_hash):sub(1, 8)
+    envs_hash = hash.strhash32(envs_hash)
     self._CO_CURENVS_HASH = envs_hash
 
     -- save the current directory for each coroutine
diff --git a/xmake/core/package/package.lua b/xmake/core/package/package.lua
index 7a1d8e24b..db943815a 100644
--- a/xmake/core/package/package.lua
+++ b/xmake/core/package/package.lua
@@ -1691,6 +1691,12 @@ function _instance:_compute_buildhash()
     self:buildhash()
 end
 
+-- hash.strhash128 has been switched to xxhash.
+-- For compatibility, the old hash algorithm is still used here.
+function _instance:_strhash128(str)
+    return hash.uuid4(str):replace("-", "", {plain = true}):lower()
+end
+
 -- get the build hash
 function _instance:buildhash()
     local buildhash = self._BUILDHASH
@@ -1752,7 +1758,7 @@ function _instance:buildhash()
                 table.sort(toolchains)
                 str = str .. "_" .. table.concat(toolchains, "_")
             end
-            return hash.strhash128(str)
+            return self:_strhash128(str)
         end
         local function _get_installdir(...)
             local name = self:name():lower():gsub("::", "_")
diff --git a/xmake/core/project/project.lua b/xmake/core/project/project.lua
index 86050ebec..5af05c21f 100644
--- a/xmake/core/project/project.lua
+++ b/xmake/core/project/project.lua
@@ -25,6 +25,7 @@ local project = project or {}
 local os                    = require("base/os")
 local io                    = require("base/io")
 local path                  = require("base/path")
+local hash                  = require("base/hash")
 local task                  = require("base/task")
 local utils                 = require("base/utils")
 local table                 = require("base/table")
@@ -1380,7 +1381,8 @@ function project.tmpfile(opt_or_key)
         key = opt_or_key.key
         opt = opt_or_key
     end
-    return path.join(project.tmpdir(opt), "_" .. (hash.uuid4(key):gsub("-", "")))
+    local filename = "_" .. (key and hash.strhash128(key) or (hash.rand128()))
+    return path.join(project.tmpdir(opt), "_" .. filename)
 end
 
 -- get all modes
diff --git a/xmake/core/sandbox/modules/hash.lua b/xmake/core/sandbox/modules/hash.lua
index 47235bcd6..b5927b029 100644
--- a/xmake/core/sandbox/modules/hash.lua
+++ b/xmake/core/sandbox/modules/hash.lua
@@ -47,7 +47,7 @@ end
 function sandbox_hash.sha1(file_or_data)
     local sha1, errors = hash.sha1(file_or_data)
     if not sha1 then
-        raise("cannot generate sha1 for %s, %s", file_or_data, errors or "unknown errors")
+        raise("cannot generate sha1, %s", errors or "unknown errors")
     end
     return sha1
 end
@@ -56,7 +56,7 @@ end
 function sandbox_hash.sha256(file_or_data)
     local sha256, errors = hash.sha256(file_or_data)
     if not sha256 then
-        raise("cannot generate sha256 for %s, %s", file_or_data, errors or "unknown errors")
+        raise("cannot generate sha256, %s", errors or "unknown errors")
     end
     return sha256
 end
@@ -65,45 +65,90 @@ end
 function sandbox_hash.md5(file_or_data)
     local md5, errors = hash.md5(file_or_data)
     if not md5 then
-        raise("cannot generate md5 for %s, %s", file_or_data, errors or "unknown errors")
+        raise("cannot generate md5, %s", errors or "unknown errors")
     end
     return md5
 end
 
+-- generate xxhash32 from the given file or data
+function sandbox_hash.xxhash32(file_or_data)
+    local result, errors = hash.xxhash32(file_or_data)
+    if not result then
+        raise("cannot generate xxhash32, %s", errors or "unknown errors")
+    end
+    return result
+end
+
 -- generate xxhash64 from the given file or data
 function sandbox_hash.xxhash64(file_or_data)
-    local xxhash64, errors = hash.xxhash64(file_or_data)
-    if not xxhash64 then
-        raise("cannot generate xxhash64 for %s, %s", file_or_data, errors or "unknown errors")
+    local result, errors = hash.xxhash64(file_or_data)
+    if not result then
+        raise("cannot generate xxhash64, %s", errors or "unknown errors")
     end
-    return xxhash64
+    return result
 end
 
 -- generate xxhash128 from the given file or data
 function sandbox_hash.xxhash128(file_or_data)
-    local xxhash128, errors = hash.xxhash128(file_or_data)
-    if not xxhash128 then
-        raise("cannot generate xxhash128 for %s, %s", file_or_data, errors or "unknown errors")
+    local result, errors = hash.xxhash128(file_or_data)
+    if not result then
+        raise("cannot generate xxhash128, %s", errors or "unknown errors")
     end
-    return xxhash128
+    return result
 end
 
 -- generate hash32 from string
 function sandbox_hash.strhash32(str)
-    local hash32, errors = hash.strhash32(str)
-    if not hash32 then
+    local result, errors = hash.strhash32(str)
+    if not result then
+        raise("cannot generate hash32 for %s, %s", str, errors or "unknown errors")
+    end
+    return result
+end
+
+-- generate hash64 from string
+function sandbox_hash.strhash64(str)
+    local result, errors = hash.strhash64(str)
+    if not result then
         raise("cannot generate hash32 for %s, %s", str, errors or "unknown errors")
     end
-    return hash32
+    return result
 end
 
 -- generate hash128 from string
 function sandbox_hash.strhash128(str)
-    local hash128, errors = hash.strhash128(str)
-    if not hash128 then
+    local result, errors = hash.strhash128(str)
+    if not result then
         raise("cannot generate hash128 for %s, %s", str, errors or "unknown errors")
     end
-    return hash128
+    return result
+end
+
+-- generate random32
+function sandbox_hash.rand32()
+    local result, errors = hash.rand32()
+    if not result then
+        raise("cannot generate random32, %s", errors or "unknown errors")
+    end
+    return result
+end
+
+-- generate random64
+function sandbox_hash.rand64()
+    local result, errors = hash.rand64()
+    if not result then
+        raise("cannot generate random64, %s", errors or "unknown errors")
+    end
+    return result
+end
+
+-- generate random128
+function sandbox_hash.rand128()
+    local result, errors = hash.rand128()
+    if not result then
+        raise("cannot generate random128, %s", errors or "unknown errors")
+    end
+    return result
 end
 
 -- return module
diff --git a/xmake/modules/private/cache/build_cache.lua b/xmake/modules/private/cache/build_cache.lua
index 3b82578e5..905426ef5 100644
--- a/xmake/modules/private/cache/build_cache.lua
+++ b/xmake/modules/private/cache/build_cache.lua
@@ -128,7 +128,7 @@ function cachekey(program, cppinfo, envs)
             end
         end
     end
-    return hash.xxhash128(bytes(table.concat(items, "")))
+    return hash.strhash128(table.concat(items, ""))
 end
 
 -- get cache root directory
diff --git a/xmake/plugins/project/cmake/cmakelists.lua b/xmake/plugins/project/cmake/cmakelists.lua
index a5ec312cb..a339a654e 100644
--- a/xmake/plugins/project/cmake/cmakelists.lua
+++ b/xmake/plugins/project/cmake/cmakelists.lua
@@ -1074,7 +1074,7 @@ function _add_target_link_libraries(cmakelists, target, outputdir)
 
 
     local has_links = #target:objectfiles() > objectfiles_set:size()
-    local key = target:name() .. "_" .. hash.uuid():split("-", {plain = true})[1]
+    local key = target:name() .. "_" .. hash.rand32()
     if has_links then
         cmakelists:print("add_library(target_objectfiles_%s OBJECT IMPORTED GLOBAL)", key)
         cmakelists:print("set_property(TARGET target_objectfiles_%s PROPERTY IMPORTED_OBJECTS", key)
@@ -1244,7 +1244,7 @@ function _add_target_custom_commands_for_batchcmds(cmakelists, target, outputdir
         --
         -- @see https://gitlab.kitware.com/cmake/cmake/-/issues/17802
         --
-        local key = target:name() .. "_" .. hash.uuid():split("-", {plain = true})[1]
+        local key = target:name() .. "_" .. hash.rand32()
         cmakelists:print("add_custom_command(OUTPUT output_%s", key)
         for _, cmd in ipairs(cmds) do
             local command = _get_command_string(cmd, outputdir)
diff --git a/xmake/rules/c++/modules/support.lua b/xmake/rules/c++/modules/support.lua
index 8d1c39531..cee0f1ab8 100644
--- a/xmake/rules/c++/modules/support.lua
+++ b/xmake/rules/c++/modules/support.lua
@@ -336,7 +336,7 @@ function modules_cachedir(target, opt)
         moduletype = "interfaces"
     elseif opt.scan then
         moduletype = "scans"
-    else 
+    else
         moduletype = "implementation"
     end
     local cachedir = path.join(target:autogendir(), "rules", "bmi", "cache", moduletype)
@@ -347,7 +347,7 @@ function modules_cachedir(target, opt)
 end
 
 function get_modulehash(sourcefile)
-    return hash.uuid(sourcefile):split("-", {plain = true})[1]:lower()
+    return hash.strhash32(sourcefile)
 end
 
 function get_metafile(target, module)
diff --git a/xmake/rules/c++/unity_build/unity_build.lua b/xmake/rules/c++/unity_build/unity_build.lua
index e8f37f868..3df2d75b4 100644
--- a/xmake/rules/c++/unity_build/unity_build.lua
+++ b/xmake/rules/c++/unity_build/unity_build.lua
@@ -36,7 +36,7 @@ function _merge_unityfile(target, sourcefile_unity, sourcefiles, opt)
             sourcefile_unity = path.absolute(sourcefile_unity)
             sourcefile = path.relative(sourcefile, path.directory(sourcefile_unity))
             if uniqueid then
-                unityfile:print("#define %s %s", uniqueid, "unity_" .. hash.uuid():split("-", {plain = true})[1])
+                unityfile:print("#define %s %s", uniqueid, "unity_" .. hash.rand32())
             end
             unityfile:print("#include \"%s\"", sourcefile)
             if uniqueid then