diff options
Diffstat (limited to 'sse-feed')
-rwxr-xr-x | sse-feed | 62 |
1 files changed, 44 insertions, 18 deletions
@@ -1,36 +1,52 @@ #!/usr/bin/python -import sys, os, MYSQLdb +import sys, os, MySQLdb, stat from popen2 import Popen3 -LEXER_PATH="." +def last_insert_id(cursor): + cursor.execute("SELECT LAST_INSERT_ID()"); + return cursor.fetchone()[0] -db = MySQLdb.connect(host = "localhost", user = "sse_web", passwd = "ece6Yoli", db = "sse") - -def process_file(root, path): +def process_file(package_id, root, path): + global cursor print "Processing %s" % path - + cursor.execute("INSERT INTO file (package_id, path, language_id) VALUES (%i, '%s', '0')" % (package_id, path)); - p = Popen3("%s/lex-c %s" % (LEXER_PATH, os.path.join(root, path))) + file_id = last_insert_id(cursor); + + p = Popen3("lex-c %s" % (os.path.join(root, path))) for identifier in p.fromchild: - print "ID:", identifier.strip() + text = identifier.strip() + + cursor.execute("INSERT IGNORE INTO word (text, type, file_id) VALUES ('%s', 'word', '%i')" % (text, file_id)) + cursor.execute("UPDATE word SET cnt=cnt+1 WHERE text='%s' AND type='word' AND file_id=%i" % (text, file_id)) - if p.wait() != 0: - print "WARNING: Subprocess failed!" + if p.wait() != 0: + print "WARNING: Subprocess failed!" del p -def handle_file(root, path, filename): +def handle_file(package_id, root, path, filename): + + t = sys.lstat(os.path.join(path, filename)) - extension = filename.split(".")[-1] + if stat.F_ISREG(t.st_mode): - if extension in ("c", "h"): - process_file(root, os.path.join(path, filename)) + extension = filename.split(".")[-1] + if extension in ("c", "h"): + process_file(package_id, root, os.path.join(path, filename)) + return -def handle_tree(path): + os.unlink(os.path.join(root, path, filename)) + +def handle_tree(path, name, url, md): + global cursor + + cursor.execute("INSERT INTO package (path, name, url, timestamp, md) VALUES ('%s', '%s', '%s', NOW(), '%s')" % (path + "/%s", name, url, md)); + package_id = last_insert_id(cursor); path = os.path.realpath(path) @@ -38,7 +54,17 @@ def handle_tree(path): for f in files: assert path + "/" == (dirpath + "/") [:len(path)+1] - handle_file(path, dirpath[len(path)+1:], f) + handle_file(package_id, path, dirpath[len(path)+1:], f) + +db = MySQLdb.connect(host = "localhost", user = "sse_web", passwd = "ece6Yoli", db = "sse") +cursor = db.cursor(); +cursor.execute("SET AUTOCOMMIT=0") +cursor.execute("START TRANSACTION") + +assert len(sys.argv) == 5 + +handle_tree(sys.argv[1], sys.argv[2], sys.argv[3], sys.argv[4]) -for a in sys.argv[1:]: - handle_tree(a) +cursor.execute("COMMIT") +cursor.close() +db.close() |