biocommons#661 - Improve Parser startup time by storing generated Pyt…

…hon code

sbin/generate_parser.py

@@ -0,0 +1,54 @@
+#!/usr/bin/env python
+# This generates the hgvs.parser.Parser Python code from the OMeta grammar file
+# We generate it offline then keep the statically generated file to reduce startup time
+# @see https://github.com/biocommons/hgvs/issues/661
+
+import hashlib
+import os
+import sys
+
+import parsley
+from ometa.grammar import OMeta
+from ometa.builder import writePython
+
+if __name__ == "__main__":
+    script_path = os.path.realpath(__file__)
+    script_dir = os.path.dirname(script_path)
+    hgvs_base_dir = os.path.dirname(script_dir)
+    grammar_file = os.path.join(hgvs_base_dir, "src/hgvs/_data/hgvs.pymeta")
+    generated_code_dir = os.path.join(hgvs_base_dir, "src/hgvs/generated")
+
+    grammar_hash = hashlib.md5(open(grammar_file, 'rb').read()).hexdigest()
+    prefix_length = len(hgvs_base_dir) + 1  # extra to also remove slash
+
+    header_template = """# --------------------------------------------------
+# THIS IS A GENERATED FILE. DO NOT MODIFY.
+# Changes will be overwritten by the generation script.
+#    Generated by: {generate_script}
+#    Grammar file: {grammar_file}
+#    Grammar hash: {grammar_hash}
+# Parsley version: {parsley_version}
+#  Python version: {python_version}
+# --------------------------------------------------
+"""
+    header = header_template.format(generate_script=script_path[prefix_length:],
+                                    grammar_file=grammar_file[prefix_length:],
+                                    grammar_hash=grammar_hash,
+                                    parsley_version=parsley.__version__,
+                                    python_version=sys.version)
+
+    g = OMeta(open(grammar_file).read(), name="Grammar")
+    tree = g.parseGrammar("Grammar")
+    source = writePython(tree, "Grammar")
+
+    # Create a package for the new file for Python2 compatability
+    if not os.path.exists(generated_code_dir):
+        os.mkdir(generated_code_dir)
+        package_filename = os.path.join(generated_code_dir, "__init__.py")
+        with open(package_filename, "w") as _:
+            pass
+
+    source_filename = os.path.join(generated_code_dir, "hgvs_grammar.py")
+    with open(source_filename, "w") as source_file:
+        source_file.write(header + "\n")
+        source_file.write(source)