aboutsummaryrefslogtreecommitdiff
path: root/backend/tolData/enwiki/genDumpIndexDb.py
diff options
context:
space:
mode:
Diffstat (limited to 'backend/tolData/enwiki/genDumpIndexDb.py')
-rwxr-xr-xbackend/tolData/enwiki/genDumpIndexDb.py13
1 files changed, 5 insertions, 8 deletions
diff --git a/backend/tolData/enwiki/genDumpIndexDb.py b/backend/tolData/enwiki/genDumpIndexDb.py
index 3955885..1bffb27 100755
--- a/backend/tolData/enwiki/genDumpIndexDb.py
+++ b/backend/tolData/enwiki/genDumpIndexDb.py
@@ -4,14 +4,11 @@ import sys, os, re
import bz2
import sqlite3
-usageInfo = f"""
-Usage: {sys.argv[0]}
-
-Adds data from the wiki dump index-file into a database.
-"""
-if len(sys.argv) > 1:
- print(usageInfo, file=sys.stderr)
- sys.exit(1)
+import argparse
+parser = argparse.ArgumentParser(description="""
+Adds data from the wiki dump index-file into a database
+""", formatter_class=argparse.RawDescriptionHelpFormatter)
+parser.parse_args()
indexFile = "enwiki-20220501-pages-articles-multistream-index.txt.bz2" # Had about 22e6 lines
indexDb = "dumpIndex.db"