filename fileURL url 'https://raw.githubusercontent.com/brendanjodowd/SAS/master/brendan_macros.sas';
%include fileURL;
filename fileURL; Some macros produce datasets or html output as a standalone statement. These are introduced with a semicolon below to indicate that they work on their own, e.g. show_datatsets;
Other macros return some kind of a string which is intended for use as code within some other kind of statement like a data step. These are not presented with a semicolon below, e.g. %list_vars() . You can usually check what they return by preceding with a %put statement.
A list here refers to a sequence of words that will be used as code. It could be a list of datasets, variables, years, etc. Macros designed for use with lists are useful when you building a series of datasets or when you're trying to write flexible code. They are not designed for working with character type variables, since they will probably pass the name of the variable rather than the string contained within it.
- Generating Lists
- Joining Lists
- Manipulating Words in Lists
- Manipulating String Variables
- Checking variables in datasets
- Managing Datasets and Variables
- Managing macros and macro variables
- Handy abbreviations
- Checking counts and duplicates
- running_thru_list
- tell_me_about
%seq(x, y)Generates an ascending list from one number to another.
%rep(<string>, <integer>)Returns a string repeated a number of times, separated by spaces.
%add_prefix(<some list here>, prefix)
%add_sufffix(<some list here>, suffix)Returns a list but with a string concatenated to start (or end) of each word. Useful for converting one kind of list into another. E.g.
%put %add_prefix(2010 2011 2012, accounts_);... returns: accounts_2010 accounts_2011 accounts_2012
%list_vars(some_dataset)
%list_num_vars(some_dataset)
%list_char_vars(some_dataset)Returns a list of variables in a dataset that you might use as part of your code. You might choose to manipulate the string that gets returned, for example by adding or removing certain words. Can be useful in certain proc sql commands where the full list of variables is required.
%add_commas(<some list here>)Returns a list but with commas between each word. Handy for proc sql and certain other functions.
%add_keep(<some list here> , var_1 var_2)Returns a list but with (keep = var_1 var_2) after each word. Handy for when you're using set with a list of datasets. A bit more efficient than specifying these on the output dataset.
E.g.
set %add_keep(team_1 team_2 team_3 , name age );... becomes: set team_1 (keep = name age) team_2 (keep = name age) team_3 (keep = name age);
Boolean operations on multiple lists. These are especially powerful when used in combination with %list_vars() and other macros for modifying lists.
%union(<first list> <second list> <third list>)Returns the combined set of lists with duplicates removed. E.g.
%put %union(id age sales id job name id name height);... returns: id age sales job name height
%intersect_lists(<first list> , <second list>)Returns elements which appear in both lists. Very useful for keep and drop statements, or where a variable has an inconsistent name across datasets.
E.g.
%put %intersect_lists(id age sales job name , id name height);... returns: id name
%left_anti_join(<first list> , <second list>)Returns the first list with the exclusion of any elements which appear in the second. Again, useful for keep, drop and set statements.
E.g.
%put %left_anti_join(id age job name , id name height);... returns: age job
%pairwise_join(<first list> , <second list> , sep=)Returns the lists joined together in a pairwise fashion, with each pair separated by sep (null by default). Lists must be the same length.
E.g. %pairwise_join(hand foot tree , bag ball house) returns handbag football treehouse.
And %pairwise_join(library_2010 library_2012 , file_2010 file_2012 , sep=. ) returns library_2010.file2010 library_2012.file_2012.
Tools for querying and adjusting lists.
%find_word( <some list> , word)Returns n, where word is the nth word in <some list>, and 0 if word does not appear in <some list>. Differs from FINDW() which returns character position. Finds whole words and is case insensitive.
%first_word(<some list>)
%last_word(<some list>)Simply returns the first or last word in a list.
%words_beginning_with(<some list> , phrase)
%words_ending_with(<some list> , phrase)
%words_containing(<some list> , phrase)Returns all words from a list which begin/end/contain phrase. Useful for dropping or keeping a raft of similar variables.
%remove_word(<some list> , word)Returns the list with word removed. Case insensitive and matches whole word only.
%remove_nth_word(<some list> , n)Returns the list with the nth word removed.
%replace_word(<some list> , old_word , new_word)Returns the list with old_word replaced by new_word. Case insensitive and matches whole word only.
%list_substr(<some list> , <start number> , <length number>)Returns a list of substrings of each word in <some list>, with each substring defined by a starting position (<start number>) and length (<length number>).
E.g. %list_substr(file_2010_a file_2011_a file_2012_b , 6 , 4) would return 2010 2011 2012.
%as_num( <some character variable> )
%as_char( <some numeric variable> )A neat little expression for quickly making a new numeric/character variable using a character/numeric variable as input. Negates need for put/input in majority of cases. E.g.
/*inside data step*/
numeric_age = %as_num(character_age);%extract_num( <some string with a number in it> )A stronger version of %as_num(), extracting a number from among other characters in a variable. E.g.:
code = %extract_num( "(2) Arts and Humanities" )...gives code a value of 2.
%squish( <some string variable> )Removes leading and trailing blanks, and replaces multiple spaces with a single space. Inspired by str_squish() in the stringr package for R.
%crop_left( string_variable , "phrase" )
%crop_right( string_variable , "phrase" )Similar to LEFT() and RIGHT() in Excel. crop_left returns a string up to (and excluding) the first instance of "phrase" in string_variable.
crop_right returns the string after the first appearance of "phrase".
%print_Vars(dataset_name);Prints out a list of variables (inc. type and length) in a short, handy command.
%var_exist(dataset_name , variable_name)Returns '1' or '0' depending on whether variable_name appears in dataset_name. Useful for case-based handling of datasets. Also used in some other macros.
%var_type(dataset_name , variable_name)Returns 'C' or 'N' depending on whether variable_name is of type character or numeric. Useful for case-based handling of datasets.
%var_type(dataset_name , variable_name)Returns (memory) length of variable_name. Useful for case-based handling of datasets.
%var_fmt(dataset_name , variable_name)Returns the format of variable_name in dataset_name.
%right_case(dataset_name , variable_name)Returns the word variable_name again, but with letters capitalised in the same way that they are in the dataset. Used within some other macros to ensure that case-insensitive inputs do not alter the original capitalisation of variable names.
%max_length(dataset_name , some_char_variable);This outputs a line to the log which tells you the maximum string length of the variable some_char_variable. Useful when you want to reduce the memory available to a variable, but you're not sure if doing so will crop strings.
%show_datasets;Produces a handy table showing all the datasets in the WORK library, including their size, and numbers of rows and columns.
%dataset_exist(dataset_name)Logical test to see if a dataset exists. Used in many other macros.
%datasets_exist( <a list of dataset names> )Passed a list of dataset names, returns a list of those which actually exist. Not a logical test like dataset_exist(). Useful for setting a list of datasets in 'greedy' fashion.
E.g. %datasets_exist(sashelp.cars something_else sashelp.baseball) should return sashelp.cars sashelp.baseball.
%delete_dataset(dataset_name);Deletes a dataset. You can also pass a list of datasets, or use in conjunction with colon (:) to delete datasets starting with a common string.
%rename_var(dataset_name , old_var= , new_var= , warn=YES);A handy macro for changing variable names. Uses proc datasets and so is fairly efficient. Includes an optional variable, WARN. If set to YES (default), you get a warning in the log if old_var does not exist. Or you can set warn to NO, which will make no changes to the dataset and will generate no warning, and this is useful for managing inconsistent datasets.
%add_label(dataset_name , variable_name , label , warn=YES);Adds a label to a variable. WARN is set to yes by default, meaning that you get a warning in the log if the variable doesn't exist. Set it to NO if you don't want a warning (I.e., apply label in the case that the variable exists).
%drop_format(dataset_name , variable_name);Removes a format from a variable. Handy if datasets come in with a weird format.
%show_macros;Prints a list of all the macros that have been defined so far.
%is_int(<macro variable>)Logical test to check if a macro variable can be used as an integer.
%is_blank(<macro variable>)Logical test to check if a macro variable is blank or not. Useful for creating checks within your own macros. I think I got this from Lex Jansen
%pf(dataset_name , var_1);
%pf(dataset_name , var_1*var_2);Abbreviation for proc freq. Displays only counts when a 2D table is requested (nocol norow nopercent options).
%ps(dataset_name , var_1 var_2);Abbreviation for proc sort. Pass any number of variables.
%unique(dataset_name , variable_name);Based on the unique() function in R. Outputs frequency counts for each variable_name to the log. Also states number of unique entries for variable_name. Does not produce frequency counts if there more than 100 unique values.
%check_duplicates(dataset_name , variable_name);This outputs a two-column table to log. It gives a frequency count OF a frequency count for variable_name. Why would you want this? Well you can see easily whether every value for variable_name appears once, for example, or you can see whether every value for variable_name appears ten times. I find it useful when applied to some kind of id variable, and I expect each id to appear exactly n times.
%abort_if_duplicates(dataset_name , variable_name);This returns %abort(); if any value for variable_name appears more than once. A useful failsafe.
%abort_if_duplicates(dataset_name , variable_name);This returns %abort(); if a frequency count over variable_name returns more than one unique frequency. A useful failsafe if you're building a dataset that should have exactly ten entries (for example) for each unique value of variable_name.
%running_thru_list(2010 2011 2012 2013 , some_macro);
/* ...is equivalent to... */
%some_macro(2010);
%some_macro(2011);
%some_macro(2012);
%some_macro(2013);Repeats a macro over a list, and produces a progress indicator to the log which includes an estimate of the remaining time.
Example of log output:
Current time is 9:26:31
Remaining time is 3 minutes and 50 seconds , will be finished at 9:30:21
Duration of last run was 94 % of the average
|#######################75%############------------|The list that you use could be the name of a series of datasets, it could be a list of years, names, months, etc. A good idea is to define the list as a macro variable so that you can operate a series of macros to a list that is defined in only one location. You can use a series of years in combination with %add_prefix(), for example, to operate on a series of datasets. E.g.
%let years_to_process = 2010 2011 2012 2013;
%running_thru_list( %add_prefix( &years_to_process, SALES_ ) , some_macro);
/* ...is equivalent to... */
%some_macro(SALES_2010);
%some_macro(SALES_2011);
%some_macro(SALES_2012);
%some_macro(SALES_2013);%tell_me_about( <some list of datasets> );This macro is useful for investigating a series of inconsistently structured datasets, e.g. the length or name of a variable changes from one month to the next. It produces three tables which compare the stucture of the series of datasets.
- name_record - gives the variable names in each dataset.
- type_record - gives the type (C or N) of each variable in each dataset.
- length_record - gives the length of each variable in each dataset.
It will name the columns in each of these three tables according to the names of the datasets that you provide. It uses filenames only provided they are different, or library names otherwise.
If you provide a single dataset, then only one table is created with all information in three columns.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent