根据之前文章的说明，使用 ZGLanguage 将以下SQL解析输出扁平化、可标注的编程语言解析结果 ：
SQL代码 :
sql
CREATE TABLE TMP_O_SS_CL_1
AS
SELECT CL.OBJECTNO
, PP.PAYDATE
FROM NYBDP.O_SS_CL_LOAN_ACCT_STATIS CL
LEFT JOIN (SELECT AL.PAYDATE
, BC.BUSS_NO
FROM O_CWWS_ACCT_LOAN AL
INNER JOIN NYBDP.O_CWWS_BUSINESS_DUEBILL BD
ON 1=1
LEFT JOIN O_CWWS_BUSINESS_CONTRACT BC
ON 1=1
) PP
ON 1=1
;解析结果 :
sql
1 P 2 1 __CREATE_TABLE_SELECT__ 1 x1 CREATE
2 P 2 1 __CREATE_TABLE_SELECT__ 1 x2 TABLE
3 P 2 1 __CREATE_TABLE_SELECT__ 1 目标表 TMP_O_SS_CL_1
4 P 3 1 __CREATE_TABLE_SELECT__ 1 x3 AS
5 P 4 bbbb 2 __SELECT__ 1 x1 SELECT
6 P 4 bbbb 3 __OTH_COL__ 1 x1 CL
7 P 4 bbbb 3 __OTH_COL__ 1 x2 .
8 P 4 bbbb 3 __OTH_COL__ 1 x3 OBJECTNO
9 P 3 __OTH_COL__ 6
10 P 5 bbbbbbbbb 2 __SELECT__ 1 x3 ,
11 P 5 bbbbbbbbb 4 __OTH_COL__ 1 x1 PP
12 P 5 bbbbbbbbb 4 __OTH_COL__ 1 x2 .
13 P 5 bbbbbbbbb 4 __OTH_COL__ 1 x3 PAYDATE
14 P 4 __OTH_COL__ 11
15 P 6 bbbbbb 2 __SELECT__ 1 x4 FROM
16 P 6 bbbbbb 5 __TABLE_NAME__ 1 schema NYBDP
17 P 6 bbbbbb 5 __TABLE_NAME__ 1 pp .
18 P 6 bbbbbb 5 __TABLE_NAME__ 1 源表_2 O_SS_CL_LOAN_ACCT_STATIS
19 P 6 bbbbbb 5 __TABLE_NAME__ 1 源表别名 CL
20 P 5 __TABLE_NAME__ 16
21 P 7 b 6 __JOIN_TABLE__ 1 x5 LEFT
22 P 7 b 6 __JOIN_TABLE__ 1 x7 JOIN
23 P 7 b 7 __SUB_SELECT__ 1 x1 (
24 P 7 b 8 __SELECT__ 1 x1 SELECT
25 P 7 b 9 __OTH_COL__ 1 x1 AL
26 P 7 b 9 __OTH_COL__ 1 x2 .
27 P 7 b 9 __OTH_COL__ 1 x3 PAYDATE
28 P 9 __OTH_COL__ 25
29 P 8 bbbbbbbbbbbbbbbbb 8 __SELECT__ 1 x3 ,
30 P 8 bbbbbbbbbbbbbbbbb 10 __OTH_COL__ 1 x1 BC
31 P 8 bbbbbbbbbbbbbbbbb 10 __OTH_COL__ 1 x2 .
32 P 8 bbbbbbbbbbbbbbbbb 10 __OTH_COL__ 1 x3 BUSS_NO
33 P 10 __OTH_COL__ 30
34 P 9 bbbbbbbbbbbbbb 8 __SELECT__ 1 x4 FROM
35 P 9 bbbbbbbbbbbbbb 11 __TABLE_NAME__ 1 源表 O_CWWS_ACCT_LOAN
36 P 9 bbbbbbbbbbbbbb 11 __TABLE_NAME__ 1 源表别名 AL
37 P 11 __TABLE_NAME__ 35
38 P 10 bbbbbbbb 12 __JOIN_TABLE__ 1 x1 INNER
39 P 10 bbbbbbbb 12 __JOIN_TABLE__ 1 x2 JOIN
40 P 10 bbbbbbbb 13 __TABLE_NAME__ 1 schema NYBDP
41 P 10 bbbbbbbb 13 __TABLE_NAME__ 1 pp .
42 P 10 bbbbbbbb 13 __TABLE_NAME__ 1 源表_2 O_CWWS_BUSINESS_DUEBILL
43 P 10 bbbbbbbb 13 __TABLE_NAME__ 1 源表别名 BD
44 P 13 __TABLE_NAME__ 40
45 P 11 bbbbbbbbbbbbbbbb 12 __JOIN_TABLE__ 1 x22 ON
46 P 11 bbbbbbbbbbbbbbbb 12 __JOIN_TABLE__ 1 x23 1=1
47 P 12 __JOIN_TABLE__ 38
48 P 12 bbbbbbbbb 14 __JOIN_TABLE__ 1 x5 LEFT
49 P 12 bbbbbbbbb 14 __JOIN_TABLE__ 1 x7 JOIN
50 P 12 bbbbbbbbb 15 __TABLE_NAME__ 1 源表 O_CWWS_BUSINESS_CONTRACT
51 P 12 bbbbbbbbb 15 __TABLE_NAME__ 1 源表别名 BC
52 P 15 __TABLE_NAME__ 50
53 P 13 bbbbbbbbbbbbbbbb 14 __JOIN_TABLE__ 1 x22 ON
54 P 13 bbbbbbbbbbbbbbbb 14 __JOIN_TABLE__ 1 x23 1=1
55 P 14 __JOIN_TABLE__ 48
56 P 8 __SELECT__ 24
57 P 14 bbbbbbbbbbb 7 __SUB_SELECT__ 1 x3 )
58 P 14 bbbbbbbbbbb 7 __SUB_SELECT__ 1 x5 PP
59 P 7 __SUB_SELECT__ 23
60 P 15 bbbbbbbb 6 __JOIN_TABLE__ 1 x22 ON
61 P 15 bbbbbbbb 6 __JOIN_TABLE__ 1 x23 1=1
62 P 6 __JOIN_TABLE__ 21
63 P 2 __SELECT__ 5
64 P 16 bbbbbbbbb 1 __CREATE_TABLE_SELECT__ 1 x5 ;
65 P 1 __CREATE_TABLE_SELECT__ 1根据 ZGLanguage 解析结果，我们便可以轻松地提取出SQL的表级血缘树信息
这里我们使用Python进行加工，代码如下：
python
data_info = [{'level':'0', 'node_type':'', 'value':'', 'children':[]}]
def go2OneLevel(fs, curr_level_node, one_row) :
while one_row[0] :
# 层级结束判断
if not one_row[2] and one_row[4] == curr_level_node['level'] :
return
print('2222=', one_row)
if one_row[5] not in ('__CREATE_TABLE_SELECT__', '__TABLE_NAME__', '__SUB_SELECT__') :
one_row = fs.readline().strip().split('\t')
continue
if curr_level_node['node_type'] == '__CREATE_TABLE_SELECT__' and one_row[7] == '目标表' \
or curr_level_node['node_type'] == '__TABLE_NAME__' and one_row[7] in ('源表','源表_2') :
curr_level_node['value'] = one_row[8]
if curr_level_node['level'] != one_row[4] :
new_level_node = {'level':one_row[4], 'node_type':one_row[5], 'value':'', 'children':[]}
go2OneLevel(fs, new_level_node, one_row)
curr_level_node['children'].append(new_level_node)
one_row = fs.readline().strip().split('\t')
return
with open('out.zgl', 'r', encoding='utf-8') as fs :
one_row = fs.readline().strip().split('\t')
while one_row[0] :
print('1111=', one_row)
if one_row[5] not in ('__CREATE_TABLE_SELECT__', '__TABLE_NAME__', '__SUB_SELECT__') :
one_row = fs.readline().strip().split('\t')
continue
new_level_node = {'level':one_row[4], 'node_type':one_row[5], 'value':'', 'children':[]}
print(new_level_node)
go2OneLevel(fs, new_level_node, one_row)
data_info[0]['children'].append(new_level_node)
one_row = fs.readline().strip().split('\t')
print(data_info)执行代码后，可以得到这段 SQL 表级血缘树信息：
python
[{'level': '0', 'node_type': '', 'value': '',
'children': [{
'level': '1', 'node_type': '__CREATE_TABLE_SELECT__', 'value': 'TMP_O_SS_CL_1',
'children': [{
'level': '5', 'node_type': '__TABLE_NAME__', 'value': 'O_SS_CL_LOAN_ACCT_STATIS',
'children': []}, {
'level': '7', 'node_type': '__SUB_SELECT__', 'value': '',
'children': [{
'level': '11', 'node_type': '__TABLE_NAME__', 'value': 'O_CWWS_ACCT_LOAN',
'children': []}, {
'level': '13', 'node_type': '__TABLE_NAME__', 'value': 'O_CWWS_BUSINESS_DUEBILL',
'children': []}, {
'level': '15', 'node_type': '__TABLE_NAME__', 'value': 'O_CWWS_BUSINESS_CONTRACT',
'children': []}]}]}]}]简化以上表(value)的血缘树即是 ：
目标表 ： TMP_O_SS_CL_1
来源表 ： O_SS_CL_LOAN_ACCT_STATIS
子查询 ： SUB_SELECT
来源表 ： O_CWWS_ACCT_LOAN
来源表 ： O_CWWS_BUSINESS_DUEBILL
来源表 ： O_CWWS_BUSINESS_CONTRACT