cpg2py 1.0.3__py3-none-any.whl → 1.0.4__py3-none-any.whl

cpg2py/__init__.py

@@ -1,35 +1,20 @@
-import os
-from pathlib import Path
 from csv import DictReader
+from pathlib import Path
 from .cpg import _Graph
 from .abc import *
 
-def __remove_null_bytes(file_path: Path) -> Path:
-    """Create a new clean file without NULL bytes and return its path."""
-    clean_path = file_path.with_suffix(".clean.csv")  # Create a temp clean file
-    if clean_path.exists(): os.remove(clean_path)
-    with open(file_path, "rb") as f:
-        content = f.read().replace(b"\x00", b"")  # Remove NULL bytes
-    with open(clean_path, "wb") as f:
-        f.write(content)
-    return clean_path
-
-def cpg_graph(node_csv: Path, edge_csv: Path) -> _Graph: 
-    # Clean files before reading
-    tmp_node_csv = __remove_null_bytes(node_csv)
+def cpg_graph(node_csv: Path, edge_csv: Path) -> "_Graph":
     storage = Storage()
-    with open(tmp_node_csv, 'r') as n_file: 
+    with open(node_csv, 'r') as n_file: 
         reader = DictReader(n_file, delimiter='\t')
         for node_props in reader: 
             nid = node_props.get("id:int", None) 
             if nid is None: node_props.get("id") 
             if not storage.add_node(nid): 
-                print(f"Node {nid} already exists in the graph")
+                print(f"WARN: Node {nid} already exists in the graph")
             if not storage.set_node_props(nid, node_props): 
-                print(f"Failed to set properties for node {nid}")
-    if tmp_node_csv.exists(): os.remove(tmp_node_csv)
-    tmp_edge_csv = __remove_null_bytes(edge_csv)
-    with open(tmp_edge_csv, 'r') as f:
+                print(f"WARN: Failed to set properties for node {nid}")
+    with open(edge_csv, 'r') as f:
         reader = DictReader(f, delimiter='\t')
         for edge_props in reader: 
             f_nid = str(edge_props.get("start", None) )
@@ -39,12 +24,17 @@ def cpg_graph(node_csv: Path, edge_csv: Path) -> _Graph:
             e_type = str(edge_props.get("type", None))
             if e_type is None: e_type = str(edge_props.get("type:str"))
             edge_id = (f_nid, t_nid, e_type)
+            if not storage.contains_node(edge_id[0]): 
+                storage.add_node(edge_id[0])
+                print(f"WARN: node {edge_id[0]} does not exists")
+            if not storage.contains_node(edge_id[1]): 
+                storage.add_node(edge_id[1])
+                print(f"WARN: node {edge_id[1]} does not exists")
             if not storage.add_edge(edge_id): 
-                print(f"Edge {f_nid} -> {t_nid} already exists in the graph")
+                print(f"WARN: Edge {f_nid} -> {t_nid} already exists in the graph")
             if not storage.set_edge_props(edge_id, edge_props): 
-                print(f"Failed to set properties for edge {edge_id}")
-    if tmp_edge_csv.exists(): os.remove(tmp_edge_csv)
-    return _Graph(storage)
+                print(f"WARN: Failed to set properties for edge {edge_id}")
+        return _Graph(storage)
 
 
 __all__ = ['cpg_graph', 'AbcGraphQuerier', 'AbcNodeQuerier', 'AbcEdgeQuerier', 'Storage']

{cpg2py-1.0.3.dist-info → cpg2py-1.0.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: cpg2py
-Version: 1.0.3
+Version: 1.0.4
 Summary: A graph-based data structure designed for querying CSV files in Joern format in Python
 Home-page: https://github.com/YichaoXu/cpg2py
 Author: Yichao Xu

{cpg2py-1.0.3.dist-info → cpg2py-1.0.4.dist-info}/RECORD

@@ -1,4 +1,4 @@
-cpg2py/__init__.py,sha256=InP4wevhbz1-Yp8cy-K3m1Bo8vJFaHfQLDEjeBrXVIg,2262
+cpg2py/__init__.py,sha256=4GXaMrMxlqWKpjrwAZlSwBCG-9_osMeChuFS-WRqtKg,1918
 cpg2py/abc/__init__.py,sha256=HgDXsJkGcPQvr6ac3hMrUU9TuQibMaP7-oXbXLs_iLI,207
 cpg2py/abc/edge.py,sha256=zLlRebDKT2qK5XOJq23_UUCESgGsvgalbXdx_uUcKks,1240
 cpg2py/abc/graph.py,sha256=DbeHN6qsNaqcqL9cUquJWC16BihDxy1MEhyuL-qVMBI,3902
@@ -8,8 +8,8 @@ cpg2py/cpg/__init__.py,sha256=fO59Yd7OISyxdjdZDJ5zFM41r4cYKawOsvpqV4gWrGM,48
 cpg2py/cpg/edge.py,sha256=brG7cQl7tjihbUCBYyjNUnMLsKIif-XQc_TFdGz8ZYo,1007
 cpg2py/cpg/graph.py,sha256=n4EjAhhjzl0XRJBwlAmK5AcSrECiqhjA6dXt2Ucf7IM,3448
 cpg2py/cpg/node.py,sha256=-2xl8c-eSbe4Kv4xPAEf6POlCYOV4j0voxJAKDZj3Cc,2037
-cpg2py-1.0.3.dist-info/LICENSE,sha256=vTjbt7iL1hUilI8E87FoQerEDa9nbpeip26iA6bguHI,1066
-cpg2py-1.0.3.dist-info/METADATA,sha256=PTXtPmIDRjlyIZBOZJ_fXWrYMwK73WHlVoG6jHZFBMM,7077
-cpg2py-1.0.3.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-cpg2py-1.0.3.dist-info/top_level.txt,sha256=xDY8faKh5Rczvsqb5Jt9Sq-Y7EOImh7jh-m1oVTnH5k,7
-cpg2py-1.0.3.dist-info/RECORD,,
+cpg2py-1.0.4.dist-info/LICENSE,sha256=vTjbt7iL1hUilI8E87FoQerEDa9nbpeip26iA6bguHI,1066
+cpg2py-1.0.4.dist-info/METADATA,sha256=Xhx-JFz1RECxW-aLxPhfy4elJgFoaTjxxiXR28YD-c8,7077
+cpg2py-1.0.4.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+cpg2py-1.0.4.dist-info/top_level.txt,sha256=xDY8faKh5Rczvsqb5Jt9Sq-Y7EOImh7jh-m1oVTnH5k,7
+cpg2py-1.0.4.dist-info/RECORD,,