如何读取含变量名=值格式的SAS输入数据并生成目标数据集？

Got it, let's break down how to read this variable=value formatted data in SAS to get exactly the dataset you want (with just name and math columns).

First, your input data is structured as key-value pairs where each person's attributes are listed in a single line. The tricky part is handling names with spaces (like "Mary Louise") and only extracting the variables you care about.

Method 1: Extract Values Directly from the Full Line #

This is the simplest approach if you only need name and math—we'll read the entire line as text, then pull out the values we need using string manipulation functions:

data want;
    /* Define variable lengths upfront to avoid truncating names */
    length name $20 math 8;
    infile datalines dlm=' ' missover;
    
    /* Hold the entire line in the input buffer so we can access it with _infile_ */
    input @;
    
    /* Extract name: Replace "name=" with a delimiter, then grab the first segment (including spaces) */
    name = scan(tranwrd(_infile_, 'name=', '|'), 1, ' |');
    
    /* Extract math: Replace "math=" with a delimiter, convert the string to a number */
    math = input(scan(tranwrd(_infile_, 'math=', '|'), 1, ' |'), 8.);
    
    /* Keep only the variables we need */
    keep name math;
    
    /* Your raw data */
    datalines;
name=Linda english=95 math=94 science=90
name=Susan math=88 english=91 science=90
name=Mary Louise math=90 english=84 science=81
;
run;

/* Verify the output */
proc print data=want;
run;

How this works:

• input @; keeps the entire line in the input buffer, so we can use the special _infile_ variable to read the full text of the line.
• tranwrd(_infile_, 'name=', '|') replaces the "name=" prefix with a pipe (|), making it easy to split the line with the scan function. We split on both | and spaces to capture the full name (even with spaces).
• For math, we do the same but convert the extracted string to a numeric value using input().

Method 2: Generalized Key-Value Parsing (Great for Future Expansion) #

If you might need to extract other variables later (like english or science), this loop-based method is more flexible. It reads each key-value pair one by one and assigns values to variables:

data want;
    length key $10 value $20 name $20 math 8;
    infile datalines dlm=' ' dsd missover;
    
    /* Loop through all key-value pairs in the line */
    do until (eof);
        input key value @;
        
        /* Assign values to variables based on the key */
        if key = 'name=' then name = value;
        else if key = 'math=' then math = input(value, 8.);
        /* Handle names with spaces: if we haven't finished the name, append the next segment */
        else if missing(name) then name = catx(' ', name, key);
        
        /* Check if we've reached the last pair in the line */
        eof = (scan(_infile_, -1, ' ') = key);
    end;
    
    keep name math;
    datalines;
name=Linda english=95 math=94 science=90
name=Susan math=88 english=91 science=90
name=Mary Louise math=90 english=84 science=81
;
run;

How this works:

• We loop through each key-value pair (split by spaces). When we hit name=, we start storing the name. If the next "key" isn't a known variable (like "Louise"), we append it to the name since it's part of the full name.
• When we hit math=, we convert the value to a number and assign it to the math variable.

Expected Output #

Either method will produce exactly the dataset you wanted:

内容的提问来源于stack exchange，提问作者Sujit Patnaik