Merge pull request #75 from sasjs/mp_getcols

Mp getcols macro

all.sas

@@ -2216,6 +2216,14 @@ Usage:
   select count(*) into: orig from &lib..&ds;
   quit;
 
+  %local notfound;
+  proc sql outobs=10 noprint;
+  select distinct &col  into: notfound separated by ' '
+    from &lib..&ds
+    where &col not in (
+      select &ccol from &clib..&cds
+    );
+
   %mp_abort(iftrue= (&syscc ne 0)
     ,mac=&sysmacroname
     ,msg=%str(syscc=&syscc after macro query)
@@ -2226,7 +2234,7 @@ Usage:
     test_description=symget('desc');
     test_result='FAIL';
     test_comments="&sysmacroname: &lib..&ds..&col has &result values "
-      !!"not in &clib..&cds..&ccol ";
+      !!"not in &clib..&cds..&ccol.. First 10 vals:"!!symget('notfound');
   %if &test=ANYVAL %then %do;
     if &result < &orig then test_result='PASS';
   %end;
@@ -4230,6 +4238,72 @@ filename &fref1 clear;
 
 %mend mp_filtervalidate;
 /**
+  @file
+  @brief Creates a dataset with column metadata.
+  @details This macro takes the `proc contents` output and "tidies it up" in the
+  following ways:
+
+    @li Blank labels are filled in with column names
+    @li Formats are reconstructed with default values
+    @li Types such as DATE / TIME / DATETIME are inferred from the formats
+
+  Example usage:
+
+      %mp_getcols(sashelp.airline,outds=work.myds)
+
+  @param ds The dataset from which to obtain column metadata
+  @param outds= (work.cols) The output dataset to create. Sample data:
+  |NAME $|LENGTH 8|VARNUM 8|LABEL $|FORMAT $49|TYPE $1 |DDTYPE $|
+  |---|---|---|---|---|---|---|
+  |AIR|8|2|international airline travel (thousands)|8.|N|NUMERIC|
+  |DATE|8|1|DATE|MONYY.|N|DATE|
+  |REGION|3|3|REGION|$3.|C|CHARACTER|
+
+  <h4> Related Macros </h4>
+  @li mf_getvarlist.sas
+  @li mm_getcols.sas
+
+  @version 9.2
+  @author Allan Bowe
+  @copyright Macro People Ltd - this is a licensed product and
+    NOT FOR RESALE OR DISTRIBUTION.
+
+**/
+
+%macro mp_getcols(ds, outds=work.cols);
+
+proc contents noprint data=&ds
+  out=_data_ (keep=name type length label varnum format:);
+run;
+data &outds(keep=name type length varnum format label ddtype);
+  set &syslast(rename=(format=format2 type=type2));
+  name=upcase(name);
+  if type2=2 then do;
+    length format $49.;
+    if format2='' then format=cats('$',length,'.');
+    else if formatl=0 then format=cats(format2,'.');
+    else format=cats(format2,formatl,'.');
+    type='C';
+    ddtype='CHARACTER';
+  end;
+  else do;
+    if format2='' then format=cats(length,'.');
+    else if formatl=0 then format=cats(format2,'.');
+    else if formatd=0 then format=cats(format2,formatl,'.');
+    else format=cats(format2,formatl,'.',formatd);
+    type='N';
+    if format=:'DATETIME' then ddtype='DATETIME';
+    else if format=:'DATE' or format=:'DDMMYY' or format=:'MMDDYY'
+      or format=:'YYMMDD' or format=:'E8601DA' or format=:'B8601DA'
+      or format=:'MONYY'
+    then ddtype='DATE';
+    else if format=:'TIME' then ddtype='TIME';
+    else ddtype='NUMERIC';
+  end;
+  if label='' then label=name;
+run;
+
+%mend mp_getcols;/**
   @file mp_getconstraints.sas
   @brief Get constraint details at column level
   @details Useful for capturing constraints before they are dropped / reapplied
@@ -18831,7 +18905,7 @@ data _null_;
   put ' ';
   put '-- open file and perform the substitution ';
   put 'file = io.open(fpath,"r") ';
-  put 'fcontent = file:read() ';
+  put 'fcontent = file:read("*all") ';
   put 'file:close() ';
   put 'fcontent = string.gsub( ';
   put '  fcontent, ';
@@ -18846,7 +18920,7 @@ data _null_;
   put 'io.close(file) ';
 run;
 
-%inc "%sysfunc(pathname(work))/ml_gsubfile.lua";
+%inc "%sysfunc(pathname(work))/ml_gsubfile.lua" /source2;
 
 %mend ml_gsubfile;
 /**
@@ -19240,7 +19314,7 @@ data _null_;
   put '-- JSON.LUA ENDS HERE ';
 run;
 
-%inc "%sysfunc(pathname(work))/ml_json.lua";
+%inc "%sysfunc(pathname(work))/ml_json.lua" /source2;
 
 %mend ml_json;
 /**

base/mp_assertcolvals.sas

@@ -115,6 +115,14 @@
   select count(*) into: orig from &lib..&ds;
   quit;
 
+  %local notfound;
+  proc sql outobs=10 noprint;
+  select distinct &col  into: notfound separated by ' '
+    from &lib..&ds
+    where &col not in (
+      select &ccol from &clib..&cds
+    );
+
   %mp_abort(iftrue= (&syscc ne 0)
     ,mac=&sysmacroname
     ,msg=%str(syscc=&syscc after macro query)
@@ -125,7 +133,7 @@
     test_description=symget('desc');
     test_result='FAIL';
     test_comments="&sysmacroname: &lib..&ds..&col has &result values "
-      !!"not in &clib..&cds..&ccol ";
+      !!"not in &clib..&cds..&ccol.. First 10 vals:"!!symget('notfound');
   %if &test=ANYVAL %then %do;
     if &result < &orig then test_result='PASS';
   %end;

base/mp_getcols.sas

@@ -0,0 +1,67 @@
+/**
+  @file
+  @brief Creates a dataset with column metadata.
+  @details This macro takes the `proc contents` output and "tidies it up" in the
+  following ways:
+
+    @li Blank labels are filled in with column names
+    @li Formats are reconstructed with default values
+    @li Types such as DATE / TIME / DATETIME are inferred from the formats
+
+  Example usage:
+
+      %mp_getcols(sashelp.airline,outds=work.myds)
+
+  @param ds The dataset from which to obtain column metadata
+  @param outds= (work.cols) The output dataset to create. Sample data:
+  |NAME $|LENGTH 8|VARNUM 8|LABEL $|FORMAT $49|TYPE $1 |DDTYPE $|
+  |---|---|---|---|---|---|---|
+  |AIR|8|2|international airline travel (thousands)|8.|N|NUMERIC|
+  |DATE|8|1|DATE|MONYY.|N|DATE|
+  |REGION|3|3|REGION|$3.|C|CHARACTER|
+
+  <h4> Related Macros </h4>
+  @li mf_getvarlist.sas
+  @li mm_getcols.sas
+
+  @version 9.2
+  @author Allan Bowe
+  @copyright Macro People Ltd - this is a licensed product and
+    NOT FOR RESALE OR DISTRIBUTION.
+
+**/
+
+%macro mp_getcols(ds, outds=work.cols);
+
+proc contents noprint data=&ds
+  out=_data_ (keep=name type length label varnum format:);
+run;
+data &outds(keep=name type length varnum format label ddtype);
+  set &syslast(rename=(format=format2 type=type2));
+  name=upcase(name);
+  if type2=2 then do;
+    length format $49.;
+    if format2='' then format=cats('$',length,'.');
+    else if formatl=0 then format=cats(format2,'.');
+    else format=cats(format2,formatl,'.');
+    type='C';
+    ddtype='CHARACTER';
+  end;
+  else do;
+    if format2='' then format=cats(length,'.');
+    else if formatl=0 then format=cats(format2,'.');
+    else if formatd=0 then format=cats(format2,formatl,'.');
+    else format=cats(format2,formatl,'.',formatd);
+    type='N';
+    if format=:'DATETIME' then ddtype='DATETIME';
+    else if format=:'DATE' or format=:'DDMMYY' or format=:'MMDDYY'
+      or format=:'YYMMDD' or format=:'E8601DA' or format=:'B8601DA'
+      or format=:'MONYY'
+    then ddtype='DATE';
+    else if format=:'TIME' then ddtype='TIME';
+    else ddtype='NUMERIC';
+  end;
+  if label='' then label=name;
+run;
+
+%mend mp_getcols;

build.py

@@ -22,7 +22,7 @@ for file in files:
         for line in infile:
           ml.write("  put '" + line.rstrip().replace("'","''") + " ';\n")
     ml.write("run;\n\n")
-    ml.write("%inc \"%sysfunc(pathname(work))/" + name + ".lua\";\n\n")
+    ml.write("%inc \"%sysfunc(pathname(work))/" + name + ".lua\" /source2;\n\n")
     ml.write("%mend " + name + ";\n")
 
 ml.close()

lua/gsubfile.lua

@@ -10,7 +10,7 @@ end
 
 -- open file and perform the substitution
 file = io.open(fpath,"r")
-fcontent = file:read()
+fcontent = file:read("*all")
 file:close()
 fcontent = string.gsub(
   fcontent,

lua/ml_gsubfile.sas

@@ -24,7 +24,7 @@ data _null_;
   put ' ';
   put '-- open file and perform the substitution ';
   put 'file = io.open(fpath,"r") ';
-  put 'fcontent = file:read() ';
+  put 'fcontent = file:read("*all") ';
   put 'file:close() ';
   put 'fcontent = string.gsub( ';
   put '  fcontent, ';
@@ -39,6 +39,6 @@ data _null_;
   put 'io.close(file) ';
 run;
 
-%inc "%sysfunc(pathname(work))/ml_gsubfile.lua";
+%inc "%sysfunc(pathname(work))/ml_gsubfile.lua" /source2;
 
 %mend ml_gsubfile;

lua/ml_json.sas

@@ -389,6 +389,6 @@ data _null_;
   put '-- JSON.LUA ENDS HERE ';
 run;
 
-%inc "%sysfunc(pathname(work))/ml_json.lua";
+%inc "%sysfunc(pathname(work))/ml_json.lua" /source2;
 
 %mend ml_json;

tests/crossplatform/mp_getcols.test.sas

@@ -0,0 +1,33 @@
+/**
+  @file
+  @brief Testing mp_getcols macro
+
+  <h4> SAS Macros </h4>
+  @li mp_getcols.sas
+  @li mp_assertcolvals.sas
+  @li mp_assertdsobs.sas
+
+**/
+
+
+/* valid filter */
+%mp_getcols(sashelp.airline,outds=work.info)
+
+
+%mp_assertdsobs(work.info,
+  desc=Has 3 records,
+  test=EQUALS 3,
+  outds=work.test_results
+)
+
+data work.check;
+  length val $10;
+  do val='NUMERIC','DATE','CHARACTER';
+    output;
+  end;
+run;
+%mp_assertcolvals(work.info.ddtype,
+  checkvals=work.check.val,
+  desc=All values have a match,
+  test=ALLVALS
+)

tests/crossplatform/mp_gsubfile.test.sas

@@ -30,4 +30,37 @@ run;
   iftrue=("&str1"="&str"),
   desc=Check that simple replacement was successful,
   outds=work.test_results
+)
+
+/**
+  * test 2 - replace from additional line
+  */
+%global str2 strcheck2 strcheck2b;
+%let file2=%sysfunc(pathname(work))/file2.txt;
+%let pat2=replace/me;
+%let str2=with/this;
+data _null_;
+  file "&file2";
+  put 'line1';output;
+  put "&pat2";output;
+  put "&pat2";output;
+run;
+%mp_gsubfile(file=&file2, patternvar=pat2, replacevar=str2)
+data _null_;
+  infile "&file2";
+  input;
+  if _n_=2 then call symputx('strcheck2',_infile_);
+  if _n_=3 then call symputx('strcheck2b',_infile_);
+  putlog _infile_;
+run;
+
+%mp_assert(
+  iftrue=("&strcheck2"="&str2"),
+  desc=Check that multi line replacement was successful (line2),
+  outds=work.test_results
+)
+%mp_assert(
+  iftrue=("&strcheck2b"="&str2"),
+  desc=Check that multi line replacement was successful (line3),
+  outds=work.test_results
 )