SAS宏list_intersection报错排查及优化方案咨询

Let's break down your issues step by step, then fix the code and share more robust, efficient implementations for your list intersection macro.

Why You're Getting Those Errors #

First, let's address the mistakes in your initial code, then the sneaky issue in your revised version:

1. Initial Code Syntax Blunders

   • You tried to assign values directly in %local statements (like %local output =;) — %local only declares variable scope, you need separate %let statements to initialize values.
   • Your %if statement was missing a closing parenthesis: %if (%scan(...) = %scan(...) %then ... threw off the macro parser entirely.
   • Re-declaring output with %local inside the %then branch is unnecessary (and invalid) once you've already declared it at the top of the macro.

2. Revised Code's Hidden Problem
   SAS 9.3's macro parser has trouble with /* ... */ comments embedded directly in the parameter list, especially when those comments contain commas (like your space separated list, or unique term notes). It misinterprets the commas as parameter separators, leading to the weird "macro keyword appears as text" errors. Moving comments outside the parameter list fixes this.

Fixed Basic Version of Your Macro #

This fixes all syntax issues, cleans up the comment placement, and adds small improvements like trimming leading spaces:

%macro list_intersection(list1=, list2=);
    /* 
     * Returns a space-separated list of elements common to both input lists
     * Parameters:
     *   list1: First space-separated list (or single term)
     *   list2: Second space-separated list (or single term)
     */
    %local output i j item1 item2;
    %let output =;
    %let i = 1;
    
    %do %while (%length(%scan(&list1, &i)));
        %let item1 = %scan(&list1, &i);
        %let j = 1;
        %do %while (%length(%scan(&list2, &j)));
            %let item2 = %scan(&list2, &j);
            %if &item1 = &item2 %then %let output = &output &item1;
            %let j = %eval(&j + 1);
        %end;
        %let i = %eval(&i + 1);
    %end;
    
    /* Trim leading space from the final output */
    %sysfunc(trim(&output))
%mend list_intersection;

More Efficient & Robust Implementations #

The double-loop approach works for small lists, but it's slow for long lists and doesn't handle duplicate elements. Here are two better options:

Option 1: Data Step with Hash Table (Best for Long Lists)

Hash tables provide O(1) lookups, making this way faster than nested loops. It also automatically deduplicates the intersection:

%macro list_intersection(list1=, list2=);
    /* 
     * Returns a deduplicated, space-separated intersection of two lists
     * Parameters:
     *   list1: First space-separated list (or single term)
     *   list2: Second space-separated list (or single term)
     */
    %local output;
    
    data _null_;
        /* Adjust item length to match your longest possible list element */
        length item $200;
        declare hash h();
        h.defineKey('item');
        h.defineDone();
        
        /* Load all elements from list1 into the hash table */
        do i = 1 to countw("&list1", ' ');
            item = scan("&list1", i, ' ');
            h.add();
        end;
        
        /* Check list2 elements against the hash table and collect matches */
        length output $10000; /* Adjust based on expected output size */
        output = '';
        do i = 1 to countw("&list2", ' ');
            item = scan("&list2", i, ' ');
            if h.find() = 0 then do;
                output = catx(' ', output, item);
                h.remove(); /* Prevent duplicate matches in output */
            end;
        end;
        
        /* Pass the result back to a macro variable */
        call symputx('output', output, 'L');
    run;
    
    &output
%mend list_intersection;

Option 2: Pure Macro Function Implementation (No Data Step)

If you prefer staying entirely in macro code, use regular expressions to check for element existence and avoid duplicates:

%macro list_intersection(list1=, list2=);
    /* 
     * Returns a deduplicated, space-separated intersection of two lists
     * Parameters:
     *   list1: First space-separated list (or single term)
     *   list2: Second space-separated list (or single term)
     */
    %local output i item;
    %let output =;
    
    %do i = 1 %to %sysfunc(countw(&list1, %str( )));
        %let item = %scan(&list1, &i, %str( ));
        /* Check if item exists in list2 AND isn't already in the output */
        %if %sysfunc(prxmatch(/(^|\s)&item(\s|$)/, &list2)) 
            and not %sysfunc(prxmatch(/(^|\s)&item(\s|$)/, &output)) %then
            %let output = &output &item;
    %end;
    
    %sysfunc(trim(&output))
%mend list_intersection;

Test It Out #

Here's how to verify the macro works:

%let fruits1 = apple banana cherry date;
%let fruits2 = banana date elderberry apple;

%put Intersection Result: %list_intersection(list1=&fruits1, list2=&fruits2);

You'll get this output in the log: Intersection Result: apple banana date

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