Use argparse in plain_bpmf_compiler

Source/Data/Makefile

@@ -7,7 +7,10 @@ all: data.txt data-plain-bpmf.txt associated-phrases-v2.txt
 install: all
 
 data-plain-bpmf.txt: curation/compilers/plain_bpmf_compiler.py BPMFBase.txt BPMFPunctuations.txt
-	$(PYTHON) -m curation.compilers.plain_bpmf_compiler BPMFBase.txt BPMFPunctuations.txt data-plain-bpmf.txt
+	$(PYTHON) -m curation.compilers.plain_bpmf_compiler \
+		--bpmf_base BPMFBase.txt \
+		--punctuations BPMFPunctuations.txt \
+		--output data-plain-bpmf.txt
 
 data.txt: curation/compilers/main_compiler.py BPMFBase.txt BPMFMappings.txt BPMFPunctuations.txt \
           PhraseFreq.txt phrase.occ Symbols.txt Macros.txt\

Source/Data/curation/compilers/plain_bpmf_compiler.py

@@ -1,50 +1,62 @@
 #!/usr/bin/env python3
 # -*- coding: utf-8 -*-
 from .compiler_utils import HEADER, convert_vks_rows_to_sorted_kvs_rows
+import argparse
 import re
-import sys
 
 __author__ = "Lukhnos Liu and The McBopomofo Authors"
 __copyright__ = "Copyright 2012 and onwards The McBopomofo Authors"
 __license__ = "MIT"
 
-skip = re.compile(r"．\s+_punctuation.*_>")
+# Punctuation fix for Plain Bopomofo mode
+SKIP_RE = re.compile(r"．\s+_punctuation.*_>")
+PUNCTUATION_FIX = ["．", '_punctuation_"', "0.0"]
 
-insert = ["．", "_punctuation_\"", "0.0"]
 
-
-def main():
-    if len(sys.argv) < 4:
-        sys.exit('Usage: cook-plain-bpmf.py bpmf-base punctuation-list output')
-
-    bpmf_base = open(sys.argv[1], "r")
-    punctuation_list = open(sys.argv[2], "r")
+def cook(bpmf_base_path, punctuations_path, output_path):
     output = []
 
-    while True:
-        line = bpmf_base.readline()
-        if not line: break
-        kv = line.split(" ")
-        output.append((kv[0], kv[1], "0.0"))
-
-    while True:
-        line = punctuation_list.readline()
-        if not line:
-            break
-        if skip.search(line):
-            continue
-        row = line.rstrip().split(" ")
-        assert len(row) == 3
-        output.append(row)
-
-    output.append(insert)
+    with open(bpmf_base_path, "r") as bpmf_base:
+        for line in bpmf_base:
+            line = line.strip()
+            if not line:
+                continue
+            elements = line.split(" ")
+            assert len(elements) >= 2
+            output.append((elements[0], elements[1], "0.0"))
+
+    with open(punctuations_path, "r") as punctuation_list:
+        for line in punctuation_list:
+            if SKIP_RE.search(line):
+                continue
+            row = line.rstrip().split(" ")
+            assert len(row) == 3
+            output.append(row)
+
+    output.append(PUNCTUATION_FIX)
 
     output = convert_vks_rows_to_sorted_kvs_rows(output)
-    with open(sys.argv[3], "w") as fout:
-        fout.write(HEADER)
+    with open(output_path, "w") as output_file:
+        output_file.write(HEADER)
         for row in output:
-            fout.write("%s %s %s\n" % tuple(row))
+            output_file.write("%s %s %s\n" % tuple(row))
 
 
-if __name__ == '__main__':
+def main():
+    parser = argparse.ArgumentParser(
+        description="compile plain Bopomofo phrases database"
+    )
+    parser.add_argument("--bpmf_base", required=True)
+    parser.add_argument("--punctuations", required=True)
+    parser.add_argument("--output", required=True)
+
+    args = parser.parse_args()
+    cook(
+        bpmf_base_path=args.bpmf_base,
+        punctuations_path=args.punctuations,
+        output_path=args.output,
+    )
+
+
+if __name__ == "__main__":
     main()