Merge pull request #9 from FSoft-AI4Code/dev/v0.0.9

Dev/v0.0.9

.gitignore

@@ -11,4 +11,6 @@ src/codetext.egg-info/*
 *.pyc
 *.so
 *.whl
-
+.idea
+.vscode
+*.iml

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "codetext"
-version = "0.0.8"
+version = "0.0.9"
 authors = [
   { name="Dung Manh Nguyen", email="dungnm.workspace@gmail.com" },
 ]
@@ -17,11 +17,12 @@ classifiers = [
     "Operating System :: OS Independent",
 ]
 dependencies = [
-    "tree-sitter>=0.20",
+    "tree-sitter==0.20.4",
     "Levenshtein>=0.20",
     "langdetect>=1.0.0",
     "bs4>=0.0.1",
-    "tabulate>=0.9.0"
+    "tabulate>=0.9.0",
+    "tree_sitter_languages>=1.10.0"
 ]
 
 [project.urls]

requirements.txt

@@ -1,6 +1,7 @@
 # for preprocessing
-tree-sitter
+tree-sitter==0.20.4
 tabulate
 Levenshtein
 langdetect
 bs4
+tree_sitter_languages==1.10.2

src/codetext/utils/utils.py

@@ -92,13 +92,19 @@ def parse_code(raw_code: str, language: str='Auto', tree_sitter_path: str=None)
         calling_script_path = Path(inspect.getframeinfo(sys._getframe(1)).filename)
         load_path = str(calling_script_path.parent)
 
-    ts_lang_path = os.path.join(load_path, 'tree-sitter', f'{language}.so')
-    if not os.path.exists(ts_lang_path):
-        logger.warning(f"Not found `{language}.so` in `{load_path}/tree-sitter/`, attemp to build language")
-        build_language(language, load_path)
-
+    # Get parser from languages
     parser = Parser()
-    language = Language(load_path + f"/tree-sitter/{language}.so", language)
+    try:
+        from tree_sitter_languages import get_language, get_parser
+        language = get_language(language)
+    except ImportError:
+        # Work-around when pre-built binaries wheels for tree-sitter-languages are not available
+        logger.warning(f"Troubled importing 'tree-sitter-languages', attemp to look for pre-built binaries in the workspace")    
+        ts_lang_path = os.path.join(load_path, 'tree-sitter', f'{language}.so')
+        if not os.path.exists(ts_lang_path):
+            logger.warning(f"Not found `{language}.so` in `{load_path}/tree-sitter/`, attemp to build language")
+            build_language(language, load_path)    
+        language = Language(load_path + f"/tree-sitter/{language}.so", language)
     parser.set_language(language)
 
     if isinstance(raw_code, str):

tests/setup.py

@@ -1,8 +1,12 @@
 from ..src.codetext.utils import build_language
-
+from tree_sitter_languages import get_language, get_parser
 
 if __name__ == '__main__':
     lang_list = ['python', 'cpp', 'java', 'c-sharp', 'ruby', 'rust', 'javascript', 'php', 'go']
 
     for lang in lang_list:
-        build_language(lang)
+        # build_language(lang)
+        try:
+            get_parser(get_language(lang))
+        except:
+            build_language(lang)

tests/test_utils/test_utils.py

@@ -14,8 +14,6 @@ def test_parse_code(self):
         def sum_2_num(a, b):
             return a + b
         """
-
-        build_language(language='python')
         parse_code(sample, 'python')