Update handler.py

handler.py

@@ -18,7 +18,7 @@ from contextlib import contextmanager
 import requests
 logging.basicConfig(format="%(asctime)s - %(message)s", level=logging.INFO)
 # Git clone command
-git_clone_command = "git clone https://github.com/OpenPecha/tibetan-aligner"
+git_clone_command = "git clone https://github.com/TenzinGayche/bertalign.git"
 
 # Run the command using subprocess
 try:
@@ -26,13 +26,9 @@ try:
     print("Git clone successful!")
 except subprocess.CalledProcessError as e:
     print(f"Error while running Git clone command: {e}")
+from bertalign import Bertalign
 
 
-ALIGNER_SCRIPT_DIR = Path("./tibetan-aligner").resolve()
-ALIGNER_SCRIPT_NAME = "align_tib_en.sh"
-ALIGNER_SCRIPT_PATH = ALIGNER_SCRIPT_DIR / ALIGNER_SCRIPT_NAME
-assert ALIGNER_SCRIPT_PATH.is_file()
-
 import requests
 
 GITHUB_USERNAME = "pechawa"
@@ -46,18 +42,8 @@ GITHUB_API_ENDPOINT = f"https://api.github.com/orgs/{GITHUB_ORG}/repos"
 DEBUG =  True
 
 quiet = "-q" if DEBUG else ""
-def make_dir_executable(dir_path: Path):
-    for fn in dir_path.iterdir():
-        st = os.stat(fn)
-        os.chmod(fn, st.st_mode | stat.S_IEXEC)
-        st = os.stat(fn)
-        os.chmod(fn, st.st_mode | stat.S_IXGRP)
-        st = os.stat(fn)
-        os.chmod(fn, st.st_mode | stat.S_IXOTH)
-
-
-make_dir_executable(ALIGNER_SCRIPT_DIR)
-
+import subprocess
+import logging
 
 def create_github_repo(repo_path: Path, repo_name: str, version: str, realign: bool):
     logging.info("[INFO] Creating GitHub repo...")
@@ -114,6 +100,7 @@ def create_github_repo(repo_path: Path, repo_name: str, version: str, realign: b
     subprocess.run(f"git push -u origin {branch_name}".split(), cwd=str(repo_path))
 
     return f"Branch '{branch_name}' updated in {repo_name}" if realign else response.json()["html_url"]
+
 def convert_raw_align_to_tm(align_fn: Path, tm_path: Path):
     if DEBUG:
         logging.debug("[INFO] Conerting raw alignment to TM repo...")
@@ -273,47 +260,27 @@ def download_file(s3_public_url: str, output_fn) -> Path:
 def _run_align_script(bo_fn, en_fn, output_dir):
     start = time.time()
 
-    # Execute the alignment script
-    cmd = [str(ALIGNER_SCRIPT_PATH), str(bo_fn), str(en_fn), str(output_dir)]
-    try:
-        output = subprocess.run(
-            cmd,
-            check=True,
-            capture_output=True,
-            text=True,
-            cwd=str(ALIGNER_SCRIPT_DIR),
-        )
-    except subprocess.CalledProcessError as e:
-        logging.error(f"Alignment script failed with error: {e}")
-        return None
-
-    # Extract the output file path
-    try:
-        output_fn = re.search(r"\[OUTPUT\] (.*)", output.stdout).group(1)
-        output_fn = "/" + output_fn.split("//")[-1]  # Correcting the duplicated path
-    except IndexError as e:
-        logging.error(f"Error processing file path: {e}")
-        return None
-    except AttributeError as e:
-        logging.error("Failed to find output file path in script output.")
-        return None
-
-    output_fn = Path(output_fn)
-
-    # Check if the file exists and read its content
-    if output_fn.exists() and output_fn.is_file():
-        content = output_fn.read_text()
-        if content:
-            logging.info(f"Length of content in bo.tx.clean{output_fn}: {len(content)}") 
-        else:
-            logging.warning(f"The file {output_fn} is empty.")
-    else:
-        logging.error(f"The file {output_fn} does not exist.")
-
+    # Configure logging
+    logging.basicConfig(level=logging.INFO)
+    
+    # Read the text from the files
+    bo_text = Path(bo_fn).read_text(encoding='utf-8')
+    en_text = Path(en_fn).read_text(encoding='utf-8')
+    
+    # Initialize the aligner and align sentences
+    aligner = Bertalign(bo_text, en_text, "bo", "en")
+    aligner.align_sents()
+    result = aligner.return_tsv()
+    
+    # Prepare the output filename and write the result
+    output_fn = Path(output_dir).joinpath('result.txt')  # Ensures correct path handling
+    with open(output_fn, "w", encoding='utf-8') as f:
+        f.write(result)
+    
+    # Calculate and log the time taken
     end = time.time()
     total_time = round((end - start) / 60, 2)
     logging.info(f"Total time taken for Aligning: {total_time} mins")
-
     return output_fn
 def align(text_pair):
     
@@ -350,7 +317,6 @@ def align(text_pair):
         except Exception as e:
             logging.error(f"Error in creating TM repository: {e}")
             return {"error": f"Error in repository creation: {e}"}
-
 class EndpointHandler():
     def __init__(self, path=""):
         self.path = path
@@ -364,5 +330,4 @@ class EndpointHandler():
             A :obj:`list`:. The list contains the embeddings of the inference inputs
         """
         data = data.pop("inputs",data)
-        return align(data)
-
+        return align(data)