(github dnsmkl#18) Fix spacing around operators

Issue was registered only regarding paranthesis,
but problem was wider - also ',' (comma) was impacted.

There is downside: now we get strange formatting for '(x)as col_alias'.

lib_fsqlf/CMakeLists.txt

@@ -17,6 +17,7 @@ set(libfsqlf_sources
     formatter/lex_wrapper.c
     formatter/print_keywords.c
     formatter/tokque.c
+    kw/is_word.c
     kw/kw.c
     kw/kwmap.c
     lex/token.c)

lib_fsqlf/formatter/print_keywords.c

@@ -71,7 +71,6 @@ static struct fsqlf_spacing calculate_spacing_pure(
             , beforespacing_of_current.space
             , beforespacing_of_current.new_line || beforespacing_of_current.indent);
 
-
     // adjacent words MUST have some spacing
     if (!r.new_line && !r.indent && !r.space &&
         isword_of_prev && isword_of_current) {

lib_fsqlf/formatter/tokque.c

@@ -1,6 +1,7 @@
 #include <stdio.h> // fprintf, fputs
 #include <assert.h> // assert
 #include <ctype.h> // isalnum
+#include "../kw/is_word.h"
 #include "../lex/token.h" // struct FSQLF_token, FSQLF_clear_token, FSQLF_set_token
 #include "print_keywords.h" // FSQLF_print
 #include "tokque.h"
@@ -32,7 +33,7 @@ static struct fsqlf_kw_conf * dummy_kw_for_txt(char *txt, size_t length)
 
     // Word-vs-operator check.
     // Ensure that two adjacent words have spacing inbetween.
-    kw->is_word = !(length == 1 && !isalnum(txt[0]));
+    kw->is_word = FSQLF_is_word(txt, length);
     return kw;
 }
 

lib_fsqlf/kw/is_word.c

@@ -0,0 +1,22 @@
+#include <ctype.h>
+#include <stddef.h>
+
+
+// "word" as opposed to "operator"/"separator"/"delimiter".
+// "SELECT" is a word, "<=" is not a word.
+// It is necessary to separate "SELECT" and "column1" with space.
+//
+// Side-note:
+// "<" and "=" also need space preserved between them.
+// (in case such malformed SQL is formatted)
+int FSQLF_is_word(char* text, size_t length) {
+    if (length <= 2) {
+        for (size_t i = 0; i < length; ++i) {
+            if (!isalnum(text[i])) {
+                return 0;
+            }
+        }
+    }
+
+    return 1;
+}

lib_fsqlf/kw/is_word.h

@@ -0,0 +1,8 @@
+#ifndef IS_WORD_H
+#define IS_WORD_H
+
+
+int FSQLF_is_word(char* txt, size_t length);
+
+
+#endif

lib_fsqlf/kw/kwmap.c

@@ -1,4 +1,5 @@
 #include <lib_fsqlf.h>
+#include "is_word.h"
 
 
 // Functions that operate on kw-map as a whole.
@@ -30,19 +31,22 @@ void fsqlf_kwmap_init(fsqlf_kwmap_t *kwmap)
     *kwmap = NULL;
     #define XMACRO(NAME, gib, nlb, tb, sb, gia, nla, ta, sa, TEXT) \
     do { \
+        size_t length = strlen(TEXT); \
         FSQLF_kw_create(kwmap, #NAME); \
         fsqlf_kw_get(*kwmap, #NAME)->before.global_indent_change = gib; \
-        fsqlf_kw_get(*kwmap, #NAME)->before.new_line    = nlb; \
-        fsqlf_kw_get(*kwmap, #NAME)->before.indent      = tb; \
-        fsqlf_kw_get(*kwmap, #NAME)->before.space       = sb; \
+        fsqlf_kw_get(*kwmap, #NAME)->before.new_line = nlb; \
+        fsqlf_kw_get(*kwmap, #NAME)->before.indent = tb; \
+        fsqlf_kw_get(*kwmap, #NAME)->before.space = sb; \
         fsqlf_kw_get(*kwmap, #NAME)->after.global_indent_change = gia; \
-        fsqlf_kw_get(*kwmap, #NAME)->after.new_line     = nla; \
-        fsqlf_kw_get(*kwmap, #NAME)->after.indent       = ta; \
-        fsqlf_kw_get(*kwmap, #NAME)->after.space        = sa; \
-        fsqlf_kw_get(*kwmap, #NAME)->print_original_text = FSQLF_KWSPELLING_USE_HARDCODED_DEFAULT; \
-        fsqlf_kw_get(*kwmap, #NAME)->print_case         = FSQLF_KWCASE_UPPER; \
-        fsqlf_kw_get(*kwmap, #NAME)->text               = TEXT; \
-        fsqlf_kw_get(*kwmap, #NAME)->is_word            = 1;  \
+        fsqlf_kw_get(*kwmap, #NAME)->after.new_line = nla; \
+        fsqlf_kw_get(*kwmap, #NAME)->after.indent = ta; \
+        fsqlf_kw_get(*kwmap, #NAME)->after.space = sa; \
+        fsqlf_kw_get(*kwmap, #NAME)->print_original_text = \
+            FSQLF_KWSPELLING_USE_HARDCODED_DEFAULT; \
+        fsqlf_kw_get(*kwmap, #NAME)->print_case = FSQLF_KWCASE_UPPER; \
+        fsqlf_kw_get(*kwmap, #NAME)->text = TEXT; \
+        fsqlf_kw_get(*kwmap, #NAME)->is_word = \
+            FSQLF_is_word(TEXT, length); \
     } while (0);
     #include "kwmap_defaults.def"
     #undef XMACRO

lib_fsqlf/kw/kwmap_defaults.def

@@ -51,7 +51,7 @@ XMACRO( kw_right_p_sub            , 0, 1, 0, 0,  0, 0, 0, 1, ")"            )
 XMACRO( kw_create                 , 0, 2, 0, 0,  0, 0, 0, 0, "CREATE"       )
 XMACRO( kw_left_p_create          , 0, 1, 0, 0,  0, 1, 0, 0, "("            )
 XMACRO( kw_right_p_create         , 0, 1, 0, 0,  0, 1, 0, 0, ")"            )
-XMACRO( kw_comma_create           , 0, 1, 0, 0,  0, 0, 0, 0, ","            )
+XMACRO( kw_comma_create           , 0, 1, 0, 0,  0, 0, 0, 1, ","            )
 XMACRO( kw_drop                   , 0, 2, 0, 0,  0, 0, 0, 0, "DROP"         )
 XMACRO( kw_table                  , 0, 0, 0, 1,  0, 0, 0, 0, "TABLE"        )
 XMACRO( kw_ifexists               , 0, 0, 0, 0,  0, 0, 0, 0, "IF EXISTS"    )

makefile

@@ -81,6 +81,7 @@ LCOBJ += $(BLD)/lib_fsqlf/formatter/print_keywords.o
 LCOBJ += $(BLD)/lib_fsqlf/formatter/tokque.o
 LCOBJ += $(BLD)/lib_fsqlf/kw/kw.o
 LCOBJ += $(BLD)/lib_fsqlf/kw/kwmap.o
+LCOBJ += $(BLD)/lib_fsqlf/kw/is_word.o
 LCOBJ += $(BLD)/lib_fsqlf/lex/token.o
 LCOBJ += $(BLD)/utils/queue/queue.o
 LCOBJ += $(BLD)/utils/stack/stack.o

tests/cases/bigquery_expected.sql

@@ -1,18 +1,18 @@
 
 SELECT
   asbaba AS g
-, cast ( bfbfb AS date ) + 1 as x
+, cast(bfbfb AS date) + 1 as x
 , 2 - 1 as b1
 , 2 * 1 as c1
 , 2 / 1 as d1
 , "str" || "cat"as concatted /*/*/
-, min ( a ) over ( partition by x,c order by r rows between 1 preceeding and 1 preceeding )
+, min(a)over(partition by x,c order by r rows between 1 preceeding and 1 preceeding)
 , 'dfgdf/*gdfgd*/fg''dfgdf/*gdfgd*/fg''dfgdf/*gdfgd*/fg' as c /*a*a*a*/
-, cast ( 19.2 AS decimal ( 18,2 ) ) /******/
+, cast(19.2 AS decimal(18,2))/******/
 , ' */ ' /**/
-, cast ( 19.2 AS decimal ( 18,2 ) ) /* sdsd  *****---*/
-, asextract ( day FROM t ) as _toto -- comment case select from
-FROM ( bobo as gogo
+, cast(19.2 AS decimal(18,2))/* sdsd  *****---*/
+, asextract(day FROM t)as _toto -- comment case select from
+FROM (bobo as gogo
 LEFT JOIN
 (
     SELECT
@@ -36,13 +36,13 @@ LEFT JOIN
           2
     ) T
 ) baba
- ON 1 IN ( 1,2,3 )
+ ON 1 IN (1,2,3)
 OR x <> 1
 AND 1 = 1
 AND 1 = 1
 AND "a" LIKE "b")
-LEFT JOIN ( baba
-CROSS JOIN gaga )
+LEFT JOIN (baba
+CROSS JOIN gaga)
  ON baba.g = T.z
 WHERE toto
 AND EXISTS
@@ -70,7 +70,7 @@ AND a IN
       1
     FROM hsd
 )
-AND b IN ( 19,2,3 )
+AND b IN (19,2,3)
 ;
 
 

tests/cases/create_table_expected.sql

@@ -3,8 +3,8 @@
 CREATE TABLE schema_name.table_name
 (
 id int
-, col2 varchar ( 30 )
-, col3 non standart keywords char ( 1 )
+, col2 varchar(30)
+, col3 non standart keywords char(1)
 )
 ;
 
@@ -14,7 +14,7 @@ id int
 CREATE volatile multiset TABLE schema_name.table_name
 (
 id int
-, col2 varchar ( 30 )
-, col3 non standart keywords char ( 1 )
+, col2 varchar(30)
+, col3 non standart keywords char(1)
 )
 ;

tests/cases/group_order_expected.sql

@@ -3,6 +3,6 @@ SELECT
   1
 , 2
 FROM x
-GROUP BY 1 , 2 , 3 , 4
-ORDER BY 1 , 2
+GROUP BY 1, 2, 3, 4
+ORDER BY 1, 2
 ;

tests/cases/using_expected.sql

@@ -3,4 +3,4 @@ SELECT
   *
 FROM table1
 JOIN table2 USING
-( nicekey )
+(nicekey)