Build all Flex files standalone

The proposed Meson build system will need a way to ignore certain
generated files in order to coexist with the autoconf build system,
and C files generated by Flex which are #include'd into .y files make
this more difficult. In similar vein to 72b1e3a, arrange for all Flex
C files to compile to their own .o targets.

Reviewed by Andres Freund

Discussion: https://www.postgresql.org/message-id/20220810171935.7k5zgnjwqzalzmtm%40awork3.anarazel.de
Discussion: https://www.postgresql.org/message-id/CAFBsxsF8Gc2StS3haXofshHCzqNMRXiSxvQEYGwnFsTmsdwNeg@mail.gmail.com

Author: j-naylor

contrib/cube/.gitignore

@@ -1,3 +1,4 @@
+/cubeparse.h
 /cubeparse.c
 /cubescan.c
 # Generated subdirectories

contrib/cube/Makefile

@@ -4,7 +4,8 @@ MODULE_big = cube
 OBJS = \
 	$(WIN32RES) \
 	cube.o \
-	cubeparse.o
+	cubeparse.o \
+	cubescan.o
 
 EXTENSION = cube
 DATA = cube--1.2.sql cube--1.2--1.3.sql cube--1.3--1.4.sql cube--1.4--1.5.sql \
@@ -15,8 +16,6 @@ HEADERS = cubedata.h
 
 REGRESS = cube cube_sci
 
-EXTRA_CLEAN = y.tab.c y.tab.h
-
 SHLIB_LINK += $(filter -lm, $(LIBS))
 
 ifdef USE_PGXS
@@ -30,11 +29,16 @@ include $(top_builddir)/src/Makefile.global
 include $(top_srcdir)/contrib/contrib-global.mk
 endif
 
+# See notes in src/backend/parser/Makefile about the following two rules
+cubeparse.h: cubeparse.c
+	touch $@
+
+cubeparse.c: BISONFLAGS += -d
 
-# cubescan is compiled as part of cubeparse
-cubeparse.o: cubescan.c
+# Force these dependencies to be known even without dependency info built:
+cubeparse.o cubescan.o: cubeparse.h
 
 distprep: cubeparse.c cubescan.c
 
 maintainer-clean:
-	rm -f cubeparse.c cubescan.c
+	rm -f cubeparse.h cubeparse.c cubescan.c

contrib/cube/cube.c

@@ -119,11 +119,11 @@ cube_in(PG_FUNCTION_ARGS)
 {
 	char	   *str = PG_GETARG_CSTRING(0);
 	NDBOX	   *result;
+	Size		scanbuflen;
 
-	cube_scanner_init(str);
+	cube_scanner_init(str, &scanbuflen);
 
-	if (cube_yyparse(&result) != 0)
-		cube_yyerror(&result, "cube parser failed");
+	cube_yyparse(&result, scanbuflen);
 
 	cube_scanner_finish();
 

contrib/cube/cubedata.h

@@ -61,9 +61,9 @@ typedef struct NDBOX
 
 /* in cubescan.l */
 extern int	cube_yylex(void);
-extern void cube_yyerror(NDBOX **result, const char *message) pg_attribute_noreturn();
-extern void cube_scanner_init(const char *str);
+extern void cube_yyerror(NDBOX **result, Size scanbuflen, const char *message) pg_attribute_noreturn();
+extern void cube_scanner_init(const char *str, Size *scanbuflen);
 extern void cube_scanner_finish(void);
 
 /* in cubeparse.y */
-extern int	cube_yyparse(NDBOX **result);
+extern int	cube_yyparse(NDBOX **result, Size scanbuflen);

contrib/cube/cubeparse.y

@@ -23,9 +23,6 @@
 #define YYMALLOC palloc
 #define YYFREE   pfree
 
-static char *scanbuf;
-static int	scanbuflen;
-
 static int item_count(const char *s, char delim);
 static NDBOX *write_box(int dim, char *str1, char *str2);
 static NDBOX *write_point_as_box(int dim, char *str);
@@ -34,6 +31,7 @@ static NDBOX *write_point_as_box(int dim, char *str);
 
 /* BISON Declarations */
 %parse-param {NDBOX **result}
+%parse-param {Size scanbuflen}
 %expect 0
 %name-prefix="cube_yy"
 
@@ -265,5 +263,3 @@ write_point_as_box(int dim, char *str)
 
 	return bp;
 }
-
-#include "cubescan.c"

contrib/cube/cubescan.l

@@ -1,9 +1,21 @@
-%{
+%top{
 /*
  * A scanner for EMP-style numeric ranges
  * contrib/cube/cubescan.l
  */
 
+#include "postgres.h"
+
+/*
+ * NB: include cubeparse.h only AFTER defining YYSTYPE (to match cubeparse.y)
+ * and cubedata.h for NDBOX.
+ */
+#include "cubedata.h"
+#define YYSTYPE char *
+#include "cubeparse.h"
+}
+
+%{
 /* LCOV_EXCL_START */
 
 /* No reason to constrain amount of data slurped */
@@ -21,9 +33,7 @@ fprintf_to_ereport(const char *fmt, const char *msg)
 
 /* Handles to the buffer that the lexer uses internally */
 static YY_BUFFER_STATE scanbufhandle;
-/* this is now declared in cubeparse.y: */
-/* static char *scanbuf; */
-/* static int	scanbuflen; */
+static char *scanbuf;
 %}
 
 %option 8bit
@@ -45,24 +55,24 @@ NaN          [nN][aA][nN]
 
 %%
 
-{float}      yylval = yytext; return CUBEFLOAT;
-{infinity}   yylval = yytext; return CUBEFLOAT;
-{NaN}        yylval = yytext; return CUBEFLOAT;
-\[           yylval = "("; return O_BRACKET;
-\]           yylval = ")"; return C_BRACKET;
-\(           yylval = "("; return O_PAREN;
-\)           yylval = ")"; return C_PAREN;
-\,           yylval = ","; return COMMA;
+{float}      cube_yylval = yytext; return CUBEFLOAT;
+{infinity}   cube_yylval = yytext; return CUBEFLOAT;
+{NaN}        cube_yylval = yytext; return CUBEFLOAT;
+\[           cube_yylval = "("; return O_BRACKET;
+\]           cube_yylval = ")"; return C_BRACKET;
+\(           cube_yylval = "("; return O_PAREN;
+\)           cube_yylval = ")"; return C_PAREN;
+\,           cube_yylval = ","; return COMMA;
 [ \t\n\r\f]+ /* discard spaces */
 .            return yytext[0]; /* alert parser of the garbage */
 
 %%
 
 /* LCOV_EXCL_STOP */
 
-/* result is not used, but Bison expects this signature */
+/* result and scanbuflen are not used, but Bison expects this signature */
 void
-yyerror(NDBOX **result, const char *message)
+cube_yyerror(NDBOX **result, Size scanbuflen, const char *message)
 {
 	if (*yytext == YY_END_OF_BUFFER_CHAR)
 	{
@@ -87,9 +97,9 @@ yyerror(NDBOX **result, const char *message)
  * Called before any actual parsing is done
  */
 void
-cube_scanner_init(const char *str)
+cube_scanner_init(const char *str, Size *scanbuflen)
 {
-	Size	slen = strlen(str);
+	Size		slen = strlen(str);
 
 	/*
 	 * Might be left over after ereport()
@@ -100,7 +110,7 @@ cube_scanner_init(const char *str)
 	/*
 	 * Make a scan buffer with special termination needed by flex.
 	 */
-	scanbuflen = slen;
+	*scanbuflen = slen;
 	scanbuf = palloc(slen + 2);
 	memcpy(scanbuf, str, slen);
 	scanbuf[slen] = scanbuf[slen + 1] = YY_END_OF_BUFFER_CHAR;

contrib/seg/.gitignore

@@ -1,3 +1,4 @@
+/segparse.h
 /segparse.c
 /segscan.c
 # Generated subdirectories

contrib/seg/Makefile

@@ -4,7 +4,8 @@ MODULE_big = seg
 OBJS = \
 	$(WIN32RES) \
 	seg.o \
-	segparse.o
+	segparse.o \
+	segscan.o
 
 EXTENSION = seg
 DATA = seg--1.1.sql seg--1.1--1.2.sql seg--1.2--1.3.sql seg--1.3--1.4.sql \
@@ -29,10 +30,16 @@ include $(top_srcdir)/contrib/contrib-global.mk
 endif
 
 
-# segscan is compiled as part of segparse
-segparse.o: segscan.c
+# See notes in src/backend/parser/Makefile about the following two rules
+segparse.h: segparse.c
+	touch $@
+
+segparse.c: BISONFLAGS += -d
+
+# Force these dependencies to be known even without dependency info built:
+segparse.o segscan.o: segparse.h
 
 distprep: segparse.c segscan.c
 
 maintainer-clean:
-	rm -f segparse.c segscan.c
+	rm -f segparse.h segparse.c segscan.c

contrib/seg/segparse.y

@@ -160,6 +160,3 @@ seg_atof(const char *value)
 	datum = DirectFunctionCall1(float4in, CStringGetDatum(value));
 	return DatumGetFloat4(datum);
 }
-
-
-#include "segscan.c"

contrib/seg/segscan.l

@@ -1,8 +1,18 @@
-%{
+%top{
 /*
  * A scanner for EMP-style numeric ranges
  */
+#include "postgres.h"
+
+/*
+ * NB: include segparse.h only AFTER including segdata.h, because segdata.h
+ * contains the definition for SEG.
+ */
+#include "segdata.h"
+#include "segparse.h"
+}
 
+%{
 /* LCOV_EXCL_START */
 
 /* No reason to constrain amount of data slurped */
@@ -21,7 +31,6 @@ fprintf_to_ereport(const char *fmt, const char *msg)
 /* Handles to the buffer that the lexer uses internally */
 static YY_BUFFER_STATE scanbufhandle;
 static char *scanbuf;
-static int	scanbuflen;
 %}
 
 %option 8bit
@@ -42,12 +51,12 @@ float        ({integer}|{real})([eE]{integer})?
 
 %%
 
-{range}      yylval.text = yytext; return RANGE;
-{plumin}     yylval.text = yytext; return PLUMIN;
-{float}      yylval.text = yytext; return SEGFLOAT;
-\<           yylval.text = "<"; return EXTENSION;
-\>           yylval.text = ">"; return EXTENSION;
-\~           yylval.text = "~"; return EXTENSION;
+{range}      seg_yylval.text = yytext; return RANGE;
+{plumin}     seg_yylval.text = yytext; return PLUMIN;
+{float}      seg_yylval.text = yytext; return SEGFLOAT;
+\<           seg_yylval.text = "<"; return EXTENSION;
+\>           seg_yylval.text = ">"; return EXTENSION;
+\~           seg_yylval.text = "~"; return EXTENSION;
 [ \t\n\r\f]+ /* discard spaces */
 .            return yytext[0]; /* alert parser of the garbage */
 
@@ -56,7 +65,7 @@ float        ({integer}|{real})([eE]{integer})?
 /* LCOV_EXCL_STOP */
 
 void
-yyerror(SEG *result, const char *message)
+seg_yyerror(SEG *result, const char *message)
 {
 	if (*yytext == YY_END_OF_BUFFER_CHAR)
 	{
@@ -94,7 +103,6 @@ seg_scanner_init(const char *str)
 	/*
 	 * Make a scan buffer with special termination needed by flex.
 	 */
-	scanbuflen = slen;
 	scanbuf = palloc(slen + 2);
 	memcpy(scanbuf, str, slen);
 	scanbuf[slen] = scanbuf[slen + 1] = YY_END_OF_BUFFER_CHAR;

src/backend/Makefile

@@ -174,10 +174,10 @@ utils/probes.o: utils/probes.d $(SUBDIROBJS)
 # Be sure that these files get removed by the maintainer-clean target
 distprep:
 	$(MAKE) -C parser	gram.c gram.h scan.c
-	$(MAKE) -C bootstrap	bootparse.c bootscanner.c
+	$(MAKE) -C bootstrap	bootparse.c bootparse.h bootscanner.c
 	$(MAKE) -C catalog	distprep
 	$(MAKE) -C nodes	distprep
-	$(MAKE) -C replication	repl_gram.c repl_scanner.c syncrep_gram.c syncrep_scanner.c
+	$(MAKE) -C replication	repl_gram.c repl_gram.h repl_scanner.c syncrep_gram.c syncrep_gram.h syncrep_scanner.c
 	$(MAKE) -C storage/lmgr	lwlocknames.h lwlocknames.c
 	$(MAKE) -C utils	distprep
 	$(MAKE) -C utils/adt	jsonpath_gram.c jsonpath_scan.c
@@ -292,13 +292,16 @@ maintainer-clean: distclean
 	$(MAKE) -C nodes $@
 	$(MAKE) -C utils $@
 	rm -f bootstrap/bootparse.c \
+	      bootstrap/bootparse.h \
 	      bootstrap/bootscanner.c \
 	      parser/gram.c \
 	      parser/gram.h \
 	      parser/scan.c \
 	      replication/repl_gram.c \
+	      replication/repl_gram.h \
 	      replication/repl_scanner.c \
 	      replication/syncrep_gram.c \
+	      replication/syncrep_gram.h \
 	      replication/syncrep_scanner.c \
 	      storage/lmgr/lwlocknames.c \
 	      storage/lmgr/lwlocknames.h \

src/backend/bootstrap/.gitignore

@@ -1,2 +1,3 @@
+/bootparse.h
 /bootparse.c
 /bootscanner.c

src/backend/bootstrap/Makefile

@@ -14,12 +14,19 @@ override CPPFLAGS := -I. -I$(srcdir) $(CPPFLAGS)
 
 OBJS = \
 	bootparse.o \
+	bootscanner.o \
 	bootstrap.o
 
 include $(top_srcdir)/src/backend/common.mk
 
-# bootscanner is compiled as part of bootparse
-bootparse.o: bootscanner.c
+# See notes in src/backend/parser/Makefile about the following two rules
+bootparse.h: bootparse.c
+	touch $@
+
+bootparse.c: BISONFLAGS += -d
+
+# Force these dependencies to be known even without dependency info built:
+bootparse.o bootscanner.o: bootparse.h
 
 # bootparse.c and bootscanner.c are in the distribution tarball, so
 # they are not cleaned here.

src/backend/bootstrap/bootparse.y

@@ -488,5 +488,3 @@ boot_ident:
 		| XNULL			{ $$ = pstrdup($1); }
 		;
 %%
-
-#include "bootscanner.c"