download PaConvert to get api_alias_mapping.json when build docs (#6580)

docs/guides/model_convert/convert_from_pytorch/.gitignore

@@ -1 +1,2 @@
 docs_mappings.json
+api_alias_mapping.json

docs/guides/model_convert/convert_from_pytorch/apply_reference_from_api_difference.py

@@ -3,11 +3,14 @@
 import sys
 
 script_path = os.path.abspath(__file__)
-script_dir = os.path.dirname(__file__)
-sys.path.append(script_dir)
-print(script_dir)
+
+# convert from pytorch basedir
+CFP_BASEDIR = os.path.dirname(__file__)
+sys.path.append(CFP_BASEDIR)
+print(CFP_BASEDIR)
 
 from validate_mapping_in_api_difference import (
+    download_file_by_git,
     get_meta_from_diff_file,
     process_mapping_index as reference_mapping_item,
 )
@@ -189,12 +192,14 @@ def reference_mapping_item_processer(line, line_idx, state, output, context):
 
 
 if __name__ == "__main__":
-    # convert from pytorch basedir
-    cfp_basedir = os.path.dirname(__file__)
+    api_alias_source = "paconvert/api_alias_mapping.json"
+    # api_alias_mapping.json
+    api_alias_file = download_file_by_git(api_alias_source)
+
     # pytorch_api_mapping_cn
-    mapping_index_file = os.path.join(cfp_basedir, "pytorch_api_mapping_cn.md")
+    mapping_index_file = os.path.join(CFP_BASEDIR, "pytorch_api_mapping_cn.md")
 
-    api_difference_basedir = os.path.join(cfp_basedir, "api_difference")
+    api_difference_basedir = os.path.join(CFP_BASEDIR, "api_difference")
 
     mapping_file_pattern = re.compile(r"^torch\.(?P<api_name>.+)\.md$")
     # get all diff files (torch.*.md)

docs/guides/model_convert/convert_from_pytorch/validate_mapping_in_api_difference.py

@@ -1,10 +1,16 @@
 import json
 import os
 import re
+import shutil
+import subprocess
+import sys
+import tempfile
 import typing
 from enum import IntEnum
 from typing import TypedDict
 
+CFP_BASEDIR = os.path.dirname(__file__)
+
 mapping_type_set = {
     # type 1
     "无参数",
@@ -503,11 +509,58 @@ def process_mapping_index(index_path, item_processer, context={}):
     return 0
 
 
+PACONVERT_REPO_URL = "https://github.com/PaddlePaddle/PaConvert.git"
+
+
+def download_file_by_git(
+    source, destination=None, repo_url=PACONVERT_REPO_URL, overwrite=False
+):
+    if destination is None:
+        destination = os.path.basename(source)
+
+    dest_path = os.path.join(CFP_BASEDIR, destination)
+
+    if os.path.exists(dest_path) and not overwrite:
+        print(f"File {dest_path} already exists, skip fetching.")
+        return
+
+    try:
+        # create temp dir
+        temp_dir = tempfile.mkdtemp()
+        # clone repo to temp dir
+        subprocess.run(
+            [
+                "git",
+                "clone",
+                "--depth=1",
+                "--single-branch",
+                repo_url,
+                temp_dir,
+            ],
+            check=True,
+        )
+        print(f"Succeeded to clone repo '{repo_url}'.")
+
+        # copy file from repo to dest
+        src_file = os.path.join(temp_dir, source)
+        shutil.copy(src_file, dest_path)
+        print(f"Succeeded to copy file {source} from repo to {dest_path}.")
+    except subprocess.CalledProcessError as e:
+        print(f"Failed to fetch file by git: {e}")
+        sys.exit(-1)
+    finally:
+        shutil.rmtree(temp_dir)
+
+    return dest_path
+
+
 if __name__ == "__main__":
-    # convert from pytorch basedir
-    cfp_basedir = os.path.dirname(__file__)
+    api_alias_source = "paconvert/api_alias_mapping.json"
+    # api_alias_mapping.json
+    api_alias_file = download_file_by_git(api_alias_source)
+
     # pytorch_api_mapping_cn
-    mapping_index_file = os.path.join(cfp_basedir, "pytorch_api_mapping_cn.md")
+    mapping_index_file = os.path.join(CFP_BASEDIR, "pytorch_api_mapping_cn.md")
 
     if not os.path.exists(mapping_index_file):
         raise Exception(f"Cannot find mapping index file: {mapping_index_file}")
@@ -521,7 +574,7 @@ def process_mapping_index(index_path, item_processer, context={}):
     )
     # index_data_dict = {i['torch_api'].replace('\_', '_'): i for i in index_data}
 
-    api_difference_basedir = os.path.join(cfp_basedir, "api_difference")
+    api_difference_basedir = os.path.join(CFP_BASEDIR, "api_difference")
 
     mapping_file_pattern = re.compile(r"^torch\.(?P<api_name>.+)\.md$")
     # get all diff files (torch.*.md)
@@ -551,7 +604,7 @@ def process_mapping_index(index_path, item_processer, context={}):
     meta_dict = {m["torch_api"].replace(r"\_", "_"): m for m in metas}
 
     # 该文件用于 PaConvert 的文档对齐工作
-    api_diff_output_path = os.path.join(cfp_basedir, "docs_mappings.json")
+    api_diff_output_path = os.path.join(CFP_BASEDIR, "docs_mappings.json")
 
     with open(api_diff_output_path, "w", encoding="utf-8") as f:
         json.dump(metas, f, ensure_ascii=False, indent=4)