Изменить логику дискаверинга исходников для TS_*

### Принцип работы:

1. В переменную с помощью `_GLOB` записываем все файлы (*.ts *.tsx *.js *.jsx *.json) проекта, кроме дефолтного списка исключений (типа `node_modules`, `dist/build`, `a.yaml`, `ya.make`, `package.json`, `pnpm-lock.yaml` и подобные)
2. Вызываем плагин (`_TS_CONFIGURE`), где удаляем из этого списка файлы, не покрываемые `tsconfig.json`
3. Получившееся значение используем в качестве источника инпутов для таргета

11 files changed, 278 insertions, 31 deletions

diff --git a/build/conf/ts/ts.conf b/build/conf/ts/ts.conf
index 2812035788..dbf05d48b0 100644
--- a/build/conf/ts/ts.conf
+++ b/build/conf/ts/ts.conf
@@ -6,6 +6,8 @@ NOTS_TOOL_BASE_ARGS=--build-root $ARCADIA_BUILD_ROOT --bindir $BINDIR --curdir $
 NOTS_TOOL_NODE_MODULES_BUNDLE=$BINDIR/node_modules.tar
 ERM_PACKAGES_PATH=devtools/frontend_build_platform/erm/erm-packages.json
 
+TS_CONFIG_PATH=tsconfig.json
+
 module _TS_BASE_UNIT: _BARE_UNIT {
     # Propagates peers to related modules
     .PEERDIR_POLICY=as_build_from
@@ -44,6 +46,38 @@ macro _TS_CONFIG_EPILOGUE() {
     _SETUP_EXTRACT_NODE_MODULES_RECIPE(${MODDIR})
 }
 
+
+# List of the inputs, filled in _TS_GLOB. Will be reduced in _TS_CONFIGURE macro.
+# Used as inputs in TS_COMPILE through `$_TS_GLOB_AS_INPUTS(IN $TS_GLOB_FILES)`
+TS_GLOB_FILES=
+
+# Hardcoded "include" list (all other files will be ignored)
+TS_GLOB_INCLUDE=**/*
+
+# Hardcoded "exclude" list (reasonable default).
+TS_GLOB_EXCLUDE=$TS_CONFIG_PATH \
+  ya.make a.yaml \
+  (.code|.idea)/**/* \
+  (build|dist|bundle|.*|$TS_NEXT_OUTPUT_DIR)/**/* \
+  node_modules/**/* package.json pnpm-lock.yaml .* \
+  tests/**/* **/*.(test|spec).(ts|tsx|js|jsx)
+
+
+### _TS_GLOB() # internal
+###
+### Fill $TS_GLOB_FILES with potential inputs.
+### It will be reduced later in _TS_CONFIGURE based on `tsconfig.json` rules.
+### So it is important to call _TS_CONFIGURE() with _TS_GLOB()!
+macro _TS_GLOB() {
+    _GLOB(TS_GLOB_FILES $TS_GLOB_INCLUDE EXCLUDE $TS_GLOB_EXCLUDE)
+}
+
+# Ugly hack for using inputs from the variable
+macro _TS_GLOB_AS_INPUTS(IN{input}[]) {
+  .CMD=${input;hide:IN}
+}
+
+
 _TS_FILES_COPY_CMD=
 
 ### TS_FILES(Files...)
diff --git a/build/conf/ts/ts_bundle.conf b/build/conf/ts/ts_bundle.conf
index 370e5be750..0c455cff20 100644
--- a/build/conf/ts/ts_bundle.conf
+++ b/build/conf/ts/ts_bundle.conf
@@ -6,12 +6,17 @@ WEBPACK_CONFIG_PATH=webpack.config.js
 TS_BUNDLE_WEBPACK=$TOUCH_UNIT \
     && $_TS_FILES_COPY_CMD \
     && $ADD_VCS_INFO_FILE_CMD \
-    && ${cwd:BINDIR} $NOTS_TOOL bundle-webpack $NOTS_TOOL_BASE_ARGS --webpack-resource $WEBPACK_ROOT \
-    --moddir $MODDIR --webpack-config ${input:WEBPACK_CONFIG_PATH} --ts-config ${input:TS_CONFIG_PATH} \
-    --vcs-info "${VCS_INFO_FILE}" --output-dir ${WEBPACK_OUTPUT_DIR} \
-    --node-modules-bundle $NOTS_TOOL_NODE_MODULES_BUNDLE $NODE_MODULES_BUNDLE_AS_OUTPUT ${hide:PEERS} \
-    ${input;hide:"./package.json"} ${TS_CONFIG_FILES} ${output;hide:"output.tar"} \
-    ${output;hide:"package.json"} ${kv;hide:"p TSWP"} ${kv;hide:"pc magenta"}
+    && ${cwd:BINDIR} $NOTS_TOOL bundle-webpack $NOTS_TOOL_BASE_ARGS \
+      --moddir $MODDIR \
+      --ts-config ${input:TS_CONFIG_PATH} \
+      --node-modules-bundle $NOTS_TOOL_NODE_MODULES_BUNDLE $NODE_MODULES_BUNDLE_AS_OUTPUT ${hide:PEERS} \
+      --webpack-resource $WEBPACK_ROOT \
+      --webpack-config ${input:WEBPACK_CONFIG_PATH} \
+      --vcs-info "${VCS_INFO_FILE}" \
+      --output-dir ${WEBPACK_OUTPUT_DIR} \
+    ${input;hide:"package.json"} ${TS_CONFIG_FILES} $_TS_GLOB_AS_INPUTS(IN $TS_GLOB_FILES) \
+    ${output;hide:"package.json"} ${output;hide:"output.tar"} \
+    ${kv;hide:"p TSWP"} ${kv;hide:"pc magenta"}
 
 ### @usage: WEBPACK_OUTPUT(DirName)
 ###
@@ -48,6 +53,7 @@ multimodule TS_BUNDLE {
         SET_APPEND(_MAKEFILE_INCLUDE_LIKE_DEPS ${CURDIR}/${TS_CONFIG_PATH} ${CURDIR}/package.json)
         SET(TS_CONFIG_DEDUCE_OUT no)
 
+        _TS_GLOB()
         _TS_CONFIGURE($TS_CONFIG_PATH)
 
         # we should set NODE_MODULES_BUNDLE_AS_OUTPUT conditionally,
diff --git a/build/conf/ts/ts_library.conf b/build/conf/ts/ts_library.conf
index d5675daf29..ed3a59fe96 100644
--- a/build/conf/ts/ts_library.conf
+++ b/build/conf/ts/ts_library.conf
@@ -1,12 +1,16 @@
 # TYPESCRIPT_ROOT is defined by _PEERDIR_TS_RESOURCE(typescript)
 TYPESCRIPT_ROOT=
-TS_CONFIG_PATH=tsconfig.json
+
 TS_COMPILE=$TOUCH_UNIT \
     && $_TS_FILES_COPY_CMD \
-    && ${cwd:BINDIR} $NOTS_TOOL compile-ts $NOTS_TOOL_BASE_ARGS --tsc-resource $TYPESCRIPT_ROOT \
-    --moddir $MODDIR --ts-config ${input:TS_CONFIG_PATH} --node-modules-bundle $NOTS_TOOL_NODE_MODULES_BUNDLE \
-    $NODE_MODULES_BUNDLE_AS_OUTPUT ${hide:PEERS} ${input;hide:"./package.json"} ${TS_CONFIG_FILES} \
-    ${output;hide:"package.json"} ${output;hide:"output.tar"} ${kv;hide:"p TSC"} ${kv;hide:"pc magenta"}
+    && ${cwd:BINDIR} $NOTS_TOOL compile-ts $NOTS_TOOL_BASE_ARGS \
+      --moddir $MODDIR \
+      --ts-config ${input:TS_CONFIG_PATH} \
+      --node-modules-bundle $NOTS_TOOL_NODE_MODULES_BUNDLE $NODE_MODULES_BUNDLE_AS_OUTPUT ${hide:PEERS} \
+      --tsc-resource $TYPESCRIPT_ROOT \
+    ${input;hide:"package.json"} ${TS_CONFIG_FILES} $_TS_GLOB_AS_INPUTS(IN $TS_GLOB_FILES) \
+    ${output;hide:"package.json"} ${output;hide:"output.tar"} \
+    ${kv;hide:"p TSC"} ${kv;hide:"pc magenta"}
 
 ### @usage: TS_LIBRARY([name])
 ###
@@ -40,6 +44,8 @@ multimodule TS_LIBRARY {
         # based on whether module has deps or doesn't have
         _SET_NODE_MODULES_BUNDLE_AS_OUTPUT()
 
+        _TS_GLOB()
+
         _TS_CONFIGURE($TS_CONFIG_PATH)
     }
 
diff --git a/build/conf/ts/ts_next.conf b/build/conf/ts/ts_next.conf
index c9158d773c..bc66e176dc 100644
--- a/build/conf/ts/ts_next.conf
+++ b/build/conf/ts/ts_next.conf
@@ -2,18 +2,21 @@
 NEXT_ROOT=
 TS_NEXT_OUTPUT_DIR=.next
 TS_NEXT_CONFIG_PATH=next.config.js
-TS_NEXT_SRC_FILES=
 
 TS_NEXT_CMD=$TOUCH_UNIT \
     && $_TS_FILES_COPY_CMD \
     && $ADD_VCS_INFO_FILE_CMD \
     && ${cwd:BINDIR} $NOTS_TOOL build-nextjs $NOTS_TOOL_BASE_ARGS \
-    --moddir $MODDIR --nextjs-resource $NEXT_ROOT \
-    --nextjs-config ${input:TS_NEXT_CONFIG_PATH} --ts-config ${input:TS_CONFIG_PATH} \
-    --vcs-info "${VCS_INFO_FILE}" --output-dir ${TS_NEXT_OUTPUT_DIR} \
+    --moddir $MODDIR \
+    --ts-config ${input:TS_CONFIG_PATH} \
     --node-modules-bundle $NOTS_TOOL_NODE_MODULES_BUNDLE $NODE_MODULES_BUNDLE_AS_OUTPUT ${hide:PEERS} \
-    ${input;hide:"./package.json"} ${TS_CONFIG_FILES} ${output;hide:"output.tar"} \
-    ${output;hide:"package.json"} ${kv;hide:"p TSNEXT"} ${kv;hide:"pc magenta"}
+    --nextjs-resource $NEXT_ROOT \
+    --nextjs-config ${input:TS_NEXT_CONFIG_PATH} \
+    --vcs-info "${VCS_INFO_FILE}" \
+    --output-dir ${TS_NEXT_OUTPUT_DIR} \
+    ${input;hide:"package.json"} ${TS_CONFIG_FILES} $_TS_GLOB_AS_INPUTS(IN $TS_GLOB_FILES) \
+    ${output;hide:"output.tar"} ${output;hide:"package.json"} \
+    ${kv;hide:"p TSNEXT"} ${kv;hide:"pc magenta"}
 
 ### @usage: TS_NEXT()
 ###
@@ -31,7 +34,7 @@ multimodule TS_NEXT {
         .CMD=TS_NEXT_CMD
         .PEERDIRSELF=NODE_MODULES
         # epilogue is not inherited from TS_LIBRARY
-        .EPILOGUE=_TS_NEXT_EPILOG
+        .EPILOGUE=_TS_CONFIG_EPILOGUE
 
         # by default multimodule overrides inherited MODULE_TAG to submodule name (BUILD in this case)
         # but we have to set it to TS for include processor to work
@@ -41,6 +44,7 @@ multimodule TS_NEXT {
         SET_APPEND(_MAKEFILE_INCLUDE_LIKE_DEPS ${CURDIR}/${TS_CONFIG_PATH} ${CURDIR}/package.json)
         SET(TS_CONFIG_DEDUCE_OUT no)
 
+        _TS_GLOB()
         _TS_CONFIGURE($TS_CONFIG_PATH)
 
         # we should set NODE_MODULES_BUNDLE_AS_OUTPUT conditionally,
@@ -56,11 +60,6 @@ macro TS_NEXT_CONFIG(Path) {
     SET(TS_NEXT_CONFIG_PATH $Path)
 }
 
-macro _TS_NEXT_EPILOG() {
-    _TS_CONFIG_EPILOGUE()
-    _GLOB(TS_NEXT_SRC_FILES ${CURDIR}/src/pages/**/* ${CURDIR}/pages/**/* ${CURDIR}/src/app/**/* ${CURDIR}/app/**/* ${CURDIR}/public/**/*)
-    SRCS($TS_NEXT_SRC_FILES)
-}
 
 ### @usage: TS_NEXT_OUTPUT(DirName)
 ###
diff --git a/build/conf/ts/ts_vite_bundle.conf b/build/conf/ts/ts_vite_bundle.conf
index 74feb6cfd6..4c6590b687 100644
--- a/build/conf/ts/ts_vite_bundle.conf
+++ b/build/conf/ts/ts_vite_bundle.conf
@@ -6,12 +6,17 @@ VITE_CONFIG_PATH=vite.config.ts
 TS_BUNDLE_VITE=$TOUCH_UNIT \
     && $_TS_FILES_COPY_CMD \
     && $ADD_VCS_INFO_FILE_CMD \
-    && ${cwd:BINDIR} $NOTS_TOOL bundle-vite $NOTS_TOOL_BASE_ARGS --vite-resource $VITE_ROOT \
-    --moddir $MODDIR --vite-config ${input:VITE_CONFIG_PATH} --ts-config ${input:TS_CONFIG_PATH} \
-    --vcs-info "${VCS_INFO_FILE}" --output-dir ${VITE_OUTPUT_DIR} \
-    --node-modules-bundle $NOTS_TOOL_NODE_MODULES_BUNDLE $NODE_MODULES_BUNDLE_AS_OUTPUT ${hide:PEERS} \
-    ${input;hide:"./package.json"} ${TS_CONFIG_FILES} ${output;hide:"output.tar"} \
-    ${output;hide:"package.json"} ${kv;hide:"p VB"} ${kv;hide:"pc magenta"}
+    && ${cwd:BINDIR} $NOTS_TOOL bundle-vite $NOTS_TOOL_BASE_ARGS \
+      --moddir $MODDIR \
+      --ts-config ${input:TS_CONFIG_PATH} \
+      --node-modules-bundle $NOTS_TOOL_NODE_MODULES_BUNDLE $NODE_MODULES_BUNDLE_AS_OUTPUT ${hide:PEERS} \
+      --vite-resource $VITE_ROOT \
+      --vite-config ${input:VITE_CONFIG_PATH} \
+      --vcs-info "${VCS_INFO_FILE}" \
+      --output-dir ${VITE_OUTPUT_DIR} \
+    ${input;hide:"package.json"} ${TS_CONFIG_FILES} $_TS_GLOB_AS_INPUTS(IN $TS_GLOB_FILES) \
+    ${output;hide:"package.json"} ${output;hide:"output.tar"} \
+    ${kv;hide:"p TSVB"} ${kv;hide:"pc magenta"}
 
 ### @usage: VITE_OUTPUT(DirName)
 ###
@@ -48,6 +53,7 @@ multimodule TS_VITE_BUNDLE {
         SET_APPEND(_MAKEFILE_INCLUDE_LIKE_DEPS ${CURDIR}/${TS_CONFIG_PATH} ${CURDIR}/package.json)
         SET(TS_CONFIG_DEDUCE_OUT no)
 
+        _TS_GLOB()
         _TS_CONFIGURE($TS_CONFIG_PATH)
 
         # we should set NODE_MODULES_BUNDLE_AS_OUTPUT conditionally,
diff --git a/build/plugins/lib/nots/typescript/tests/test_ts_glob.py b/build/plugins/lib/nots/typescript/tests/test_ts_glob.py
new file mode 100644
index 0000000000..9d38a8d80f
--- /dev/null
+++ b/build/plugins/lib/nots/typescript/tests/test_ts_glob.py
@@ -0,0 +1,75 @@
+from build.plugins.lib.nots.typescript.ts_glob import ts_glob, TsGlobConfig
+
+
+class TestTsGlobIncluding:
+    ts_glob_config = TsGlobConfig(
+        root_dir="src", out_dir="build", include=["src/module_a/**/*", "src/module_b/**/*", "src/module_x"]
+    )
+
+    def test_dir_include(self):
+        # arrange
+        all_files = [
+            "src/module_x/index.ts",
+        ]
+
+        # act + arrange
+        assert ts_glob(self.ts_glob_config, all_files) == [
+            "src/module_x/index.ts",
+        ]
+
+    def test_deep_include(self):
+        # arrange
+        all_files = [
+            "src/module_a/index.ts",
+            "src/module_b/index.ts",
+            "src/module_c/index.ts",
+        ]
+
+        # act + arrange
+        assert ts_glob(self.ts_glob_config, all_files) == [
+            "src/module_a/index.ts",
+            "src/module_b/index.ts",
+        ]
+
+
+class TestTsGlobExcluding:
+    ts_glob_config = TsGlobConfig(root_dir="src", out_dir="build", include=["src/**/*"])
+
+    def test_only_in_root_dir(self):
+        # arrange
+        all_files = [
+            "CHANGELOG.md",
+            "fake-src/one-more-src.ts",
+            "src/index.ts",
+        ]
+
+        # act + assert
+        assert ts_glob(self.ts_glob_config, all_files) == ["src/index.ts"]
+
+    def test_exclude_out_dir(self):
+        # arrange
+        all_files = ["build/index.js"]
+
+        # act + assert
+        assert ts_glob(self.ts_glob_config, all_files) == []
+
+    def test_exclude_out_dir_none(self):
+        # arrange
+        all_files = ["build/index.js"]
+        ts_glob_config = TsGlobConfig(root_dir=".", out_dir=None)
+
+        # act + assert
+        assert ts_glob(ts_glob_config, all_files) == ["build/index.js"]
+
+    def test_complex(self):
+        # arrange
+        all_files = [
+            "CHANGELOG.md",
+            "fake-src/one-more-src.ts",
+            "src/baz.ts",
+            "src/index.ts",
+            "src/required_file.ts",
+        ]
+
+        # act + assert
+        assert ts_glob(self.ts_glob_config, all_files) == ["src/baz.ts", "src/index.ts", "src/required_file.ts"]
diff --git a/build/plugins/lib/nots/typescript/tests/ya.make b/build/plugins/lib/nots/typescript/tests/ya.make
index 44798138bc..2e038f6c96 100644
--- a/build/plugins/lib/nots/typescript/tests/ya.make
+++ b/build/plugins/lib/nots/typescript/tests/ya.make
@@ -4,6 +4,7 @@ OWNER(g:frontend-build-platform)
 
 TEST_SRCS(
     ts_config.py
+    test_ts_glob.py
 )
 
 PEERDIR(
diff --git a/build/plugins/lib/nots/typescript/ts_config.py b/build/plugins/lib/nots/typescript/ts_config.py
index c54121a9d1..b4ad9c3d3f 100644
--- a/build/plugins/lib/nots/typescript/ts_config.py
+++ b/build/plugins/lib/nots/typescript/ts_config.py
@@ -1,9 +1,9 @@
 import copy
-import os
 import json
+import os
 
 from .ts_errors import TsError, TsValidationError
-
+from .ts_glob import ts_glob, TsGlobConfig
 from ..package_manager.base import utils
 
 DEFAULT_TS_CONFIG_FILE = "tsconfig.json"
@@ -249,3 +249,22 @@ class TsConfig(object):
 
         with open(path, "w") as f:
             json.dump(self.data, f, indent=indent)
+
+    def filter_files(self, all_files):
+        # type: (list[str]) -> list[str]
+        """
+        Filters all the files by the rules from this tsconig.json. The result will be used as input entries in `ya make`.
+
+        Known limits:
+
+        - `files` nots supported, use `include` (see `self.validate()`)
+        - `exclude` not implemented, because `tsc` still uses "excluded" files as declaration files (for typing and referencing)
+        """
+
+        ts_glob_config = TsGlobConfig(
+            root_dir=self.compiler_option("rootDir"),
+            out_dir=self.compiler_option("outDir"),
+            include=self.data.get("include"),
+        )
+
+        return ts_glob(ts_glob_config, all_files)
diff --git a/build/plugins/lib/nots/typescript/ts_glob.py b/build/plugins/lib/nots/typescript/ts_glob.py
new file mode 100644
index 0000000000..b81e8774a5
--- /dev/null
+++ b/build/plugins/lib/nots/typescript/ts_glob.py
@@ -0,0 +1,77 @@
+import fnmatch
+import os.path
+
+
+class TsGlobConfig:
+    def __init__(self, root_dir, out_dir=None, include=None):
+        # type: (TsGlobConfig, str, str, list[str], list[str]) -> None
+
+        self.root_dir = root_dir  # Required
+        self.out_dir = out_dir
+
+        self.include = include or ["**/*"]
+
+
+def __path_to_match_rule(path):
+    # type: (str) -> str
+
+    # already a rule
+
+    # convert "**/*" to "*" (python compatible with fnmatch)
+    if path.endswith('**/*'):
+        return path[:-3]  # /**/* -> /*
+
+    if path.endswith("*") or ('*' in path or '?' in path):
+        return path
+
+    # special cases
+    if path == ".":
+        return "*"
+
+    # filename
+    root, ext = os.path.splitext(path)
+    if ext:
+        return path
+
+    # dirname ?
+    return os.path.join(path, '*')
+
+
+def __filter_files(files, path_or_rule):
+    # type: (set[str], str) -> set[str]
+
+    rule = __path_to_match_rule(path_or_rule)
+
+    result = set()
+    for path in files:
+        py_rule = __path_to_match_rule(rule)
+        if path == rule or fnmatch.fnmatch(path, py_rule):
+            result.add(path)
+
+    return result
+
+
+def ts_glob(glob_config, all_files):
+    # type: (TsGlobConfig, list[str]) -> list[str]
+
+    result = set(all_files)
+
+    # only in `root_dir`
+    result &= __filter_files(result, glob_config.root_dir)
+
+    # only listed by `include` option
+    include_only = set()
+    for include_path in glob_config.include:
+        include_only |= __filter_files(result, include_path)
+
+    result &= include_only  # keep only intersection (common in both sets)
+
+    skip_files = set()
+
+    # exclude out_dir
+    if glob_config.out_dir:
+        skip_files |= __filter_files(result, glob_config.out_dir)
+
+    result -= skip_files  # keep only differences (the elements in `result` that not exist in `skip_files`)
+
+    return sorted(result)
diff --git a/build/plugins/lib/nots/typescript/ya.make b/build/plugins/lib/nots/typescript/ya.make
index 8847f9bbd3..ac31196472 100644
--- a/build/plugins/lib/nots/typescript/ya.make
+++ b/build/plugins/lib/nots/typescript/ya.make
@@ -5,6 +5,7 @@ OWNER(g:frontend-build-platform)
 PY_SRCS(
     __init__.py
     ts_errors.py
+    ts_glob.py
     ts_config.py
 )
 
diff --git a/build/plugins/nots.py b/build/plugins/nots.py
index 740b042376..f6363dfd6f 100644
--- a/build/plugins/nots.py
+++ b/build/plugins/nots.py
@@ -200,6 +200,29 @@ def on_ts_configure(unit, tsconfig_path):
     unit.set(["TS_CONFIG_PRESERVE_JSX", to_yesno(tsconfig.compiler_option("jsx") == "preserve")])
 
     _setup_eslint(unit)
+    _filter_inputs_by_rules_from_tsconfig(unit, tsconfig)
+
+
+def __strip_prefix(prefix, line):
+    # type: (str, str) -> str
+    if line.startswith(prefix):
+        prefix_len = len(prefix)
+        return line[prefix_len:]
+
+    return line
+
+
+def _filter_inputs_by_rules_from_tsconfig(unit, tsconfig):
+    """
+    Reduce file list from the TS_GLOB_FILES variable following tsconfig.json rules
+    """
+    mod_dir = unit.get("MODDIR")
+    target_path = "${ARCADIA_ROOT}/" + mod_dir + "/"
+
+    all_files = [__strip_prefix(target_path, f) for f in unit.get("TS_GLOB_FILES").split(" ")]
+    filtered_files = tsconfig.filter_files(all_files)
+
+    unit.set(["TS_GLOB_FILES", ' '.join([target_path + f for f in filtered_files])])
 
 
 def _get_ts_test_data_dirs(unit):