JSONのファイルから32KBを超えるSASプログラムを抽出する。

LIBNAMEのJSONエンジンの欠点は、32KBを超える長さのデータを抽出できない点です。できないものは仕方ないので、32KBを超えてもSASプログラムを抽出するコードを書きました。 字句解析の状態遷移を考えて "code": の値を抽出してバックスラッシュのエンコードを解いてファイルの保存します。入力のJSONファイルは整形されていないコンパクトや書式の前提です。昔、Lex, Yaccを使っていた記憶を掘り起こした。

view plainprint?

1. /* Code extraction from JSON file of job definition */  
2. %macro extractJobCode(infile=, outfile=, debug=0);  
3.     %local/readonly DQ='22'x;  
4.     %local/readonly BSL='5c'x;  
5.     %local/readonly BS='08'x;  
6.     %local/readonly FF='0c'x;  
7.     %local/readonly NL='0a'x;  
8.     %local/readonly CR='0d'x;  
9.     %local/readonly TB='09'x;  
10.     %local/readonly SP='20'x;  
11.   
12.     data _NULL_;  
13.         attrib filein length=8 label='input file id';  
14.         attrib fileid length=8 label='output file id';  
15.         attrib stat length=8 label='State of lexical analysis';  
16.         attrib token length=$256 label='Token';  
17.         attrib ucode length=$4 label='Unicode';  
18.         attrib count length=8;  
19.   
20.         /* Open input/output files */  
21.         filein=fopen("&infile" , 'I', 1, 'B');  
22.         fileid=fopen("&outfile", 'O', 1, 'B');  
23.         count=0;  
24.         stat=0;  
25.         rec=&SP;  
26.   
27.         do while(fread(filein)=0);  
28.             rc=fget(filein, rec, 1);  
29.   
30.             %if &debug=1 %then  
31.                 %do;  
32.   
33.                     if count < 512 then  
34.                         put stat=rec=token=;  
35.                     count=count + 1;  
36.                 %end;  
37.   
38.             if stat eq 0 and rec eq &DQ then  
39.                 stat=1;  
40.             else if stat eq 1 then  
41.                 do;  
42.   
43.                     if rec eq &DQ then  
44.                         do;  
45.   
46.                             /* Double quotes */  
47.                             if token eq 'code' then  
48.                                 stat=2;  
49.                             else  
50.                                 stat=0;  
51.                             token='';  
52.                         end;  
53.                     else  
54.                         do;  
55.                             token=catt(token, rec);  
56.                             stat=1;  
57.                         end;  
58.                 end;  
59.             else if stat eq 2 then  
60.                 do;  
61.   
62.                     if rec in (&SP, &TB, &NL) then  
63.                         stat=2;  
64.                     else if rec eq ':' then  
65.                         stat=3;  
66.                     else  
67.                         stat=0;  
68.                 end;  
69.             else if stat eq 3 then  
70.                 do;  
71.   
72.                     if rec in (&SP, &TB, &NL) then  
73.                         stat=3;  
74.                     else if rec eq &DQ then  
75.                         stat=4;  
76.                     else  
77.                         stat=0;  
78.                 end;  
79.             else if stat eq 4 then  
80.                 do;  
81.   
82.                     if rec eq &DQ then  
83.                         stat=0;  
84.                     else if rec eq &BSL then  
85.                         stat=5;  
86.                     else  
87.                         do;  
88.                             rc=fput(fileid, rec);  
89.                             rc=fwrite(fileid);  
90.                         end;  
91.                 end;  
92.             else if stat eq 5 then  
93.                 do;  
94.   
95.                     /* Handling of backslash-escaped characters */  
96.                     if rec eq 'b' then  
97.                         do;  
98.                             rc=fput(fileid, &BS);  
99.                             rc=fwrite(fileid);  
100.                             stat=4;  
101.                         end;  
102.                     else if rec eq &DQ then  
103.                         do;  
104.                             rc=fput(fileid, &DQ);  
105.                             rc=fwrite(fileid);  
106.                             stat=4;  
107.                         end;  
108.                     else if rec eq &BSL then  
109.                         do;  
110.                             rc=fput(fileid, &BSL);  
111.                             rc=fwrite(fileid);  
112.                             stat=4;  
113.                         end;  
114.                     else if rec eq 'f' then  
115.                         do;  
116.                             rc=fput(fileid, &FF);  
117.                             rc=fwrite(fileid);  
118.                             stat=4;  
119.                         end;  
120.                     else if rec eq 'n' then  
121.                         do;  
122.                             rc=fput(fileid, &NL);  
123.                             rc=fwrite(fileid);  
124.                             stat=4;  
125.                         end;  
126.                     else if rec eq 'r' then  
127.                         do;  
128.                             rc=fput(fileid, &CR);  
129.                             rc=fwrite(fileid);  
130.                             stat=4;  
131.                         end;  
132.                     else if rec eq 't' then  
133.                         do;  
134.                             rc=fput(fileid, &TB);  
135.                             rc=fwrite(fileid);  
136.                             stat=4;  
137.                         end;  
138.                     else if rec eq 'u' then  
139.                         do;  
140.                             ucode='';  
141.                             do i=1 to 4;  
142.                                 rc=fread(filein);  
143.                                 rc=fget(filein, rec, 1);  
144.                                 ucode=catt(ucode, rec);  
145.                             end;  
146.                             if ucode eq '0026' then  
147.                                 do;  
148.                                     rc=fput(fileid, '&');  
149.                                     rc=fwrite(fileid);  
150.                                 end;  
151.                             else if ucode eq '003c' then  
152.                                 do;  
153.                                     rc=fput(fileid, '<');  
154.                                     rc=fwrite(fileid);  
155.                                 end;  
156.                             else if ucode eq '003e' then  
157.                                 do;  
158.                                     rc=fput(fileid, '>');  
159.                                     rc=fwrite(fileid);  
160.                                 end;  
161.                             else  
162.                                 do;  
163.                                     rc=fput(fileid, &BSL);  
164.                                     rc=fput(fileid, 'u');  
165.                                     rc=fput(fileid, ucode);  
166.                                     rc=fwrite(fileid);  
167.                                 end;  
168.                             stat=4;  
169.                         end;  
170.                 end;  
171.         end;  
172.   
173.         /* Close input/output files */  
174.         rc=fclose(filein);  
175.         rc=fclose(fileid);  
176.     run;  
177.   
178. %mend;  

/* Code extraction from JSON file of job definition */
%macro extractJobCode(infile=, outfile=, debug=0);
	%local/readonly DQ='22'x;
	%local/readonly BSL='5c'x;
	%local/readonly BS='08'x;
	%local/readonly FF='0c'x;
	%local/readonly NL='0a'x;
	%local/readonly CR='0d'x;
	%local/readonly TB='09'x;
	%local/readonly SP='20'x;

	data _NULL_;
		attrib filein length=8 label='input file id';
		attrib fileid length=8 label='output file id';
		attrib stat length=8 label='State of lexical analysis';
		attrib token length=$256 label='Token';
		attrib ucode length=$4 label='Unicode';
		attrib count length=8;

		/* Open input/output files */
		filein=fopen("&infile" , 'I', 1, 'B');
		fileid=fopen("&outfile", 'O', 1, 'B');
		count=0;
		stat=0;
		rec=&SP;

		do while(fread(filein)=0);
			rc=fget(filein, rec, 1);

			%if &debug=1 %then
				%do;

					if count < 512 then
						put stat=rec=token=;
					count=count + 1;
				%end;

			if stat eq 0 and rec eq &DQ then
				stat=1;
			else if stat eq 1 then
				do;

					if rec eq &DQ then
						do;

							/* Double quotes */
							if token eq 'code' then
								stat=2;
							else
								stat=0;
							token='';
						end;
					else
						do;
							token=catt(token, rec);
							stat=1;
						end;
				end;
			else if stat eq 2 then
				do;

					if rec in (&SP, &TB, &NL) then
						stat=2;
					else if rec eq ':' then
						stat=3;
					else
						stat=0;
				end;
			else if stat eq 3 then
				do;

					if rec in (&SP, &TB, &NL) then
						stat=3;
					else if rec eq &DQ then
						stat=4;
					else
						stat=0;
				end;
			else if stat eq 4 then
				do;

					if rec eq &DQ then
						stat=0;
					else if rec eq &BSL then
						stat=5;
					else
						do;
							rc=fput(fileid, rec);
							rc=fwrite(fileid);
						end;
				end;
			else if stat eq 5 then
				do;

					/* Handling of backslash-escaped characters */
					if rec eq 'b' then
						do;
							rc=fput(fileid, &BS);
							rc=fwrite(fileid);
							stat=4;
						end;
					else if rec eq &DQ then
						do;
							rc=fput(fileid, &DQ);
							rc=fwrite(fileid);
							stat=4;
						end;
					else if rec eq &BSL then
						do;
							rc=fput(fileid, &BSL);
							rc=fwrite(fileid);
							stat=4;
						end;
					else if rec eq 'f' then
						do;
							rc=fput(fileid, &FF);
							rc=fwrite(fileid);
							stat=4;
						end;
					else if rec eq 'n' then
						do;
							rc=fput(fileid, &NL);
							rc=fwrite(fileid);
							stat=4;
						end;
					else if rec eq 'r' then
						do;
							rc=fput(fileid, &CR);
							rc=fwrite(fileid);
							stat=4;
						end;
					else if rec eq 't' then
						do;
							rc=fput(fileid, &TB);
							rc=fwrite(fileid);
							stat=4;
						end;
					else if rec eq 'u' then
						do;
							ucode='';
							do i=1 to 4;
								rc=fread(filein);
								rc=fget(filein, rec, 1);
								ucode=catt(ucode, rec);
							end;
							if ucode eq '0026' then
								do;
									rc=fput(fileid, '&');
									rc=fwrite(fileid);
								end;
							else if ucode eq '003c' then
								do;
									rc=fput(fileid, '<');
									rc=fwrite(fileid);
								end;
							else if ucode eq '003e' then
								do;
									rc=fput(fileid, '>');
									rc=fwrite(fileid);
								end;
							else
								do;
									rc=fput(fileid, &BSL);
									rc=fput(fileid, 'u');
									rc=fput(fileid, ucode);
									rc=fwrite(fileid);
								end;
							stat=4;
						end;
				end;
		end;

		/* Close input/output files */
		rc=fclose(filein);
		rc=fclose(fileid);
	run;

%mend;