Fixes (#155)

* Remove unused OpenAI code conversion script

The script was a one-off helper for porting SDK examples between
languages. It was never wired into any generate script or workflow and
required an openai package that the repo never declared.

* Add "make mypy" to type check own Python scripts

Run mypy in strict mode over everything except submodules, which are
generated SDKs checked in their own repos, and annotate the scripts that
were still untyped so the target passes.

Switch black to --extend-exclude, since --exclude replaces its default
exclude list and would make it walk the new .mypy_cache directory.

* Fix version bump leaving the SDKs on mixed versions

Match changelog headers on whole lines. "## 4.26.4" is a substring of
both "## 4.26.4+1" and "## 4.26.41", so a patch release or a
double-digit month silently suppressed the next entry, leaving a Dart
package version with no matching CHANGELOG entry.

Validate year, month and patch in argparse and check every file the
release touches exists before the first write. "new-version.py 26 13"
used to rewrite two configs and then die on ValueError, leaving the
remaining seven SDKs on the old version.

Update the Swift changelog alongside the Dart one. It had already
fallen a release behind the Swift config.

* Fail file splitters when no parts match

The C# and Java splitters exited 0 having split nothing when the
copyright banner regex found no matches, so a changed banner would ship
un-split API files and surface much later as a compile error.

Raise instead, and replace the equivalent assert in the PHP splitter so
python -O cannot strip it.

* Drive the version bump from a single SDK table

The config paths, the release file list and the nine set_*_version
functions repeated the same nine SDKs four times over. Adding an SDK
meant touching all four.

Collapse them into one SDKS tuple. The config is always
codegen/config-<name>.json and the changelog, where there is one, is
always submodules/<name>/CHANGELOG.md, so the name derives both, and the
version format defaults to the plain YY.M.P that seven of the nine use.
Adding an SDK is now one line.

Author: Denis-Averin

.gitignore

@@ -10,3 +10,4 @@ bin/
 codegen/.generated/
 codegen/debug*.json
 obj/
+.mypy_cache/

Makefile

@@ -6,7 +6,12 @@ format: format-black
 
 .PHONY: format-black
 format-black:
-	python -m black --line-length=120 --exclude submodules -v .
+	python -m black --line-length=120 --extend-exclude submodules -v .
+
+# Type check our own scripts, SDKs in submodules have their own checks
+.PHONY: mypy
+mypy:
+	python -m mypy --config-file $(CURDIR)/mypy.ini .
 
 .PHONY: update
 update:

codegen/Tools/split-cs-file.py

@@ -4,14 +4,15 @@
 import errno
 import os
 import re
+import typing
 
 SPLIT_RE = re.compile(
     r'//\s+[-]+\n//\s+<copyright company="Aspose" file="(?P<file>.+?\.cs)">',
     re.MULTILINE,
 )
 
 
-def main(src_file, dst_dir):
+def main(src_file: typing.IO[str], dst_dir: str) -> None:
     remaining = src_file.read()
     try:
         os.makedirs(dst_dir)
@@ -20,7 +21,10 @@ def main(src_file, dst_dir):
         if e.errno != errno.EEXIST:
             raise
 
-    for match in reversed(list(SPLIT_RE.finditer(remaining))[1:]):
+    found = list(reversed(list(SPLIT_RE.finditer(remaining))[1:]))
+    if not found:
+        raise SystemExit("No parts matching regex '%s' found in %s" % (SPLIT_RE.pattern, src_file.name))
+    for match in found:
         start_pos = match.span()[0]
 
         with open(os.path.join(dst_dir, match.groupdict()["file"]), "wt") as out_f:
@@ -34,14 +38,14 @@ def main(src_file, dst_dir):
     src_file.close()
 
 
-def parse_args():
+def parse_args() -> argparse.Namespace:
     parser = argparse.ArgumentParser()
     parser.add_argument("src_file", type=argparse.FileType("rt+"))
     parser.add_argument("dst_dir", type=str)
-    args = parser.parse_args()
 
-    return vars(args)
+    return parser.parse_args()
 
 
 if __name__ == "__main__":
-    main(**parse_args())
+    parsed_args = parse_args()
+    main(parsed_args.src_file, parsed_args.dst_dir)

codegen/Tools/split-java-file.py

@@ -4,14 +4,15 @@
 import errno
 import os
 import re
+import typing
 
 SPLIT_RE = re.compile(
     r'//\s+[-]+\n//\s+<copyright company="Aspose" file="(?P<file>.+?\.java)">',
     re.MULTILINE,
 )
 
 
-def main(src_file, dst_dir):
+def main(src_file: typing.IO[str], dst_dir: str) -> None:
     remaining = src_file.read()
     try:
         os.makedirs(dst_dir)
@@ -20,7 +21,10 @@ def main(src_file, dst_dir):
         if e.errno != errno.EEXIST:
             raise
 
-    for match in reversed(list(SPLIT_RE.finditer(remaining))[1:]):
+    found = list(reversed(list(SPLIT_RE.finditer(remaining))[1:]))
+    if not found:
+        raise SystemExit("No parts matching regex '%s' found in %s" % (SPLIT_RE.pattern, src_file.name))
+    for match in found:
         start_pos = match.span()[0]
 
         with open(os.path.join(dst_dir, match.groupdict()["file"]), "wt") as out_f:
@@ -34,14 +38,14 @@ def main(src_file, dst_dir):
     src_file.close()
 
 
-def parse_args():
+def parse_args() -> argparse.Namespace:
     parser = argparse.ArgumentParser()
     parser.add_argument("src_file", type=argparse.FileType("rt+"))
     parser.add_argument("dst_dir", type=str)
-    args = parser.parse_args()
 
-    return vars(args)
+    return parser.parse_args()
 
 
 if __name__ == "__main__":
-    main(**parse_args())
+    parsed_args = parse_args()
+    main(parsed_args.src_file, parsed_args.dst_dir)

codegen/Tools/split-php-file.py

@@ -4,14 +4,15 @@
 import errno
 import os
 import re
+import typing
 
 SPLIT_RE = re.compile(
     r'/[\*]+\s*\n\s\*\s+[-]+\n\s*\*\s+<copyright company="Aspose" file="(?P<file>.+?\.php)">',
     re.MULTILINE,
 )
 
 
-def main(src_file, dst_dir):
+def main(src_file: typing.IO[str], dst_dir: str) -> None:
     remaining = src_file.read()
     try:
         os.makedirs(dst_dir)
@@ -21,7 +22,8 @@ def main(src_file, dst_dir):
             raise
 
     found = list(reversed(list(SPLIT_RE.finditer(remaining))[1:]))
-    assert found, "No parts matching regex '%s' found" % SPLIT_RE.pattern
+    if not found:
+        raise SystemExit("No parts matching regex '%s' found in %s" % (SPLIT_RE.pattern, src_file.name))
     for match in found:
         filename = match.groupdict()["file"]
         classname = os.path.splitext(filename)[0]
@@ -44,14 +46,14 @@ def main(src_file, dst_dir):
     src_file.close()
 
 
-def parse_args():
+def parse_args() -> argparse.Namespace:
     parser = argparse.ArgumentParser()
     parser.add_argument("src_file", type=argparse.FileType("rt+"))
     parser.add_argument("dst_dir", type=str)
-    args = parser.parse_args()
 
-    return vars(args)
+    return parser.parse_args()
 
 
 if __name__ == "__main__":
-    main(**parse_args())
+    parsed_args = parse_args()
+    main(parsed_args.src_file, parsed_args.dst_dir)

mypy.ini

@@ -0,0 +1,5 @@
+[mypy]
+python_version = 3.12
+strict = True
+# SDKs are generated code with their own type checks in their own repos
+exclude = (^|/)submodules/