Changeset: 05f19810601b for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=05f19810601b
Modified Files:
monetdb5/modules/atoms/Tests/json04.mal
monetdb5/modules/atoms/Tests/json04.stable.out
monetdb5/modules/atoms/json_atom.c
monetdb5/modules/atoms/json_atom.h
monetdb5/modules/atoms/json_atom.mal
Branch: default
Log Message:
Add list unnesting
diffs (265 lines):
diff --git a/monetdb5/modules/atoms/Tests/json04.mal
b/monetdb5/modules/atoms/Tests/json04.mal
--- a/monetdb5/modules/atoms/Tests/json04.mal
+++ b/monetdb5/modules/atoms/Tests/json04.mal
@@ -17,3 +17,10 @@ io.print(j);
io.print(s);
redo (i,j) := iterator.next(w);
exit (i,j);
+
+js := calc.json("[1,true, [2,false]]");
+io.print(js);
+(l1,l2) := json.unnest(js);
+io.print(l2);
+l3 := json.unnest(js);
+io.print(l3);
diff --git a/monetdb5/modules/atoms/Tests/json04.stable.out
b/monetdb5/modules/atoms/Tests/json04.stable.out
--- a/monetdb5/modules/atoms/Tests/json04.stable.out
+++ b/monetdb5/modules/atoms/Tests/json04.stable.out
@@ -33,6 +33,12 @@ barrier (i,j) := iterator.new(w);
io.print(s);
redo (i,j) := iterator.next(w);
exit (i,j);
+ js := calc.json("[1,true, [2,false]]");
+ io.print(js);
+ (l1,l2) := json.unnest(js);
+ io.print(l2);
+ l3 := json.unnest(js);
+ io.print(l3);
end main;
[ "[{ \"category\": \"reference\", \"author\": \"Nigel Rees\", \"title\":
\"Sayings of the Century\", \"price\": 8.95 }, { \"category\": \"fiction\",
\"author\": \"Evelyn Waugh\", \"title\": \"Sword of Honour\", \"price\": 12.99
}, { \"category\": \"fiction\", \"author\": \"Herman Melville\", \"title\":
\"Moby Dick\", \"isbn\": \"0-553-21311-3\", \"price\": 8.99 }, { \"category\":
\"fiction\", \"author\": \"J. R. R. Tolkien\", \"title\": \"The Lord of the
Rings\", \"isbn\": \"0-395-19395-8\", \"price\": 22.99 }, { \"color\": \"red\",
\"price\": 19.95 }]" ]
#---------------------------------------------------------#
@@ -124,6 +130,21 @@ end main;
[ "red" ]
[ "19.95 " ]
[ "19.95 " ]
+[ "[1,true, [2,false]]" ]
+#-------------------------#
+# h t # name
+# void json # type
+#-------------------------#
+[ 0@0, "1" ]
+[ 1@0, "true" ]
+[ 2@0, "[2,false]" ]
+#-------------------------#
+# h t # name
+# void json # type
+#-------------------------#
+[ 0@0, "1" ]
+[ 1@0, "true" ]
+[ 2@0, "[2,false]" ]
# 22:43:20 >
# 22:43:20 > "Done."
diff --git a/monetdb5/modules/atoms/json_atom.c
b/monetdb5/modules/atoms/json_atom.c
--- a/monetdb5/modules/atoms/json_atom.c
+++ b/monetdb5/modules/atoms/json_atom.c
@@ -418,7 +418,7 @@ str JSONlength(int *ret, json *js)
}
//the access functions assume a valid json object or
-//single nested array of objects ([[[{object},..]]]
+//single nested array of objects ([{object},..]
//any structure violation leads to an early abort
//The keys should be unique in an object
static str
@@ -601,7 +601,7 @@ str JSONunnest(int *key, int *val, json
char *msg= MAL_SUCCEED;
char *result = NULL;
size_t l,lim;
- int nesting=0;
+ int nesting=0, valuelist=0;
char *j = *js;
bk = BATnew(TYPE_void,TYPE_str,64);
@@ -629,12 +629,14 @@ str JSONunnest(int *key, int *val, json
bv->T->nonil = 1;
skipblancs;
- while( *j == '['){
+ // unnest {} or [{},...] or [val,...]
+ // in the latter case the name table remains empty
+ if( *j == '['){
nesting++;
j++;
+ skipblancs;
}
- if ( *j != '{' )
- throw(MAL,"json.unnest","JSON object expected");
+ valuelist = *j != '{';
// the result is an array of values
result = (char *) GDKmalloc(BUFSIZ);
@@ -645,36 +647,38 @@ str JSONunnest(int *key, int *val, json
}
lim = BUFSIZ;
- for( j++; *j && *j != '}'; j++){
+ for( (valuelist?j: j++); *j && *j != '}'; j++){
skipblancs;
if (*j == ']'){
break;
}
if (*j == '}')
break;
- if (*j != '"'){
- msg = createException(MAL,"json.unnest","Name
expected");
- goto wrapup;
+ if ( !valuelist){
+ if (*j != '"'){
+ msg = createException(MAL,"json.unnest","Name
expected");
+ goto wrapup;
+ }
+ namebegin = j+1;
+ msg = JSONstringParser(j+1, &j);
+ if ( msg)
+ goto wrapup;
+ nameend = j-1;
+ l = nameend - namebegin;
+ if ( l + 2 > lim )
+ result = GDKrealloc(result, lim += BUFSIZ);
+ strncpy(result,namebegin,nameend-namebegin);
+ result[l] = 0;
+ BUNappend(bk, result, FALSE);
+
+ skipblancs;
+ if ( *j != ':'){
+ msg = createException(MAL,"json.unnest","Value
expected");
+ goto wrapup;
+ }
+ j++;
+ skipblancs;
}
- namebegin = j+1;
- msg = JSONstringParser(j+1, &j);
- if ( msg)
- goto wrapup;
- nameend = j-1;
- l = nameend - namebegin;
- if ( l + 2 > lim )
- result = GDKrealloc(result, lim += BUFSIZ);
- strncpy(result,namebegin,nameend-namebegin);
- result[l] = 0;
- BUNappend(bk, result, FALSE);
-
- skipblancs;
- if ( *j != ':'){
- msg = createException(MAL,"json.unnest","Value
expected");
- goto wrapup;
- }
- j++;
- skipblancs;
valuebegin = j;
msg = JSONvalueParser(j,&j);
if ( msg)
@@ -689,13 +693,13 @@ str JSONunnest(int *key, int *val, json
skipblancs;
if (*j == '}'){
- if(nesting ){
- while (*j && *j != '{' && *j != ']') j++;
+ if(!valuelist ){
+ while (*j && *j != '{' ) j++;
if ( *j != '{') j--;
}
continue;
}
- if (*j != ',')
+ if (*j != ',' && !(valuelist && *j == ']'))
msg = createException(MAL,"json.unnest","',' expected");
}
wrapup:;
@@ -705,6 +709,69 @@ wrapup:;
return msg;
}
+str JSONunnestOne(int *val, json *js)
+{
+ BAT *bv;
+ char *valuebegin,*valueend;
+ char *msg= MAL_SUCCEED;
+ char *result = NULL;
+ size_t l,lim;
+ char *j = *js;
+
+ bv = BATnew(TYPE_void,TYPE_json,64);
+ if ( bv == NULL){
+ throw(MAL,"json.unnest",MAL_MALLOC_FAIL);
+ }
+ BATseqbase(bv,0);
+ bv->hsorted = 1;
+ bv->hrevsorted = 0;
+ bv->H->nonil =1;
+ bv->tsorted = 1;
+ bv->trevsorted = 0;
+ bv->T->nonil = 1;
+
+ skipblancs;
+ // unnest a list
+ if( *j != '[')
+ throw(MAL,"json.unnest","JSON list expected");
+
+ // the result is an array of values
+ result = (char *) GDKmalloc(BUFSIZ);
+ if ( result == 0){
+ BBPreleaseref(bv->batCacheid);
+ throw(MAL,"json.unnest",MAL_MALLOC_FAIL);
+ }
+ lim = BUFSIZ;
+
+ for( j++; *j && *j != ']'; j++){
+ skipblancs;
+ if (*j == ']'){
+ break;
+ }
+ valuebegin = j;
+ msg = JSONvalueParser(j,&j);
+ if ( msg)
+ goto wrapup;
+ valueend = j;
+ l= valueend - valuebegin;
+ if ( l + 2 > lim )
+ result = GDKrealloc(result, lim += BUFSIZ);
+ strncpy(result,valuebegin,l);
+ result[l] = 0;
+ BUNappend(bv, result, FALSE);
+
+ skipblancs;
+ if (*j == ']')
+ continue;
+ if ( *j && *j != ',' )
+ msg = createException(MAL,"json.unnest","',' expected");
+ }
+wrapup:;
+ BBPkeepref(*val= bv->batCacheid);
+ GDKfree(result);
+ return msg;
+}
+
str JSONunnestGrouped(int *grp, int *key, int *val, json *js)
{
BAT *bk, *bv, *bg;
diff --git a/monetdb5/modules/atoms/json_atom.h
b/monetdb5/modules/atoms/json_atom.h
--- a/monetdb5/modules/atoms/json_atom.h
+++ b/monetdb5/modules/atoms/json_atom.h
@@ -56,6 +56,7 @@ json_export str JSONisarray(int *ret, js
json_export str JSONlength(int *ret, json *j);
json_export str JSONunnest(int *key, int *val, json *j);
+json_export str JSONunnestOne(int *val, json *j);
json_export str JSONunnestGrouped(int *grp, int *key, int *val, json *j);
json_export str JSONnest(Client cntxt, MalBlkPtr mb, MalStkPtr stk, InstrPtr
pci);
json_export str JSONnames(int *ret, json *j);
diff --git a/monetdb5/modules/atoms/json_atom.mal
b/monetdb5/modules/atoms/json_atom.mal
--- a/monetdb5/modules/atoms/json_atom.mal
+++ b/monetdb5/modules/atoms/json_atom.mal
@@ -73,6 +73,10 @@ command unnest(val:json)(k:bat[:oid,:str
address JSONunnest
comment "Expands the outermost JSON object into key-value pairs.";
+command unnest(val:json):bat[:oid,:json]
+address JSONunnestOne
+comment "Expands a JSON list to its elements.";
+
pattern nest(o:bat[:oid,:oid],k:bat[:oid,:str],v:bat[:oid,:any]):json
address JSONnest
comment "Nest the key-value pairs with object identity into a list of JSON
objects ";
_______________________________________________
checkin-list mailing list
[email protected]
http://mail.monetdb.org/mailman/listinfo/checkin-list
