Implement type coercion

Makefile.am

@@ -12,12 +12,12 @@ if USE_PANDOC
 jo.1: jo.pandoc
 	@test -n "$(PANDOC)" || \
 	  { echo 'pandoc' not found during configure.; exit 1; }
-	$(PANDOC) -s -w man -o $@ $<
+	$(PANDOC) -s -w man+simple_tables -o $@ $<
 
 jo.md: jo.pandoc
 	@test -n "$(PANDOC)" || \
 	  { echo 'pandoc' not found during configure.; exit 1; }
-	$(PANDOC) -s -w markdown -o $@ $<
+	$(PANDOC) -s -w markdown+simple_tables -o $@ $<
 
 endif
 

jo.1

@@ -1,4 +1,5 @@
-.\" Automatically generated by Pandoc 1.16.0.2
+.\"t
+.\" Automatically generated by Pandoc 1.19.2.1
 .\"
 .TH "JO" "1" "" "User Manuals" ""
 .hy
@@ -7,7 +8,7 @@
 jo \- JSON output from a shell
 .SH SYNOPSIS
 .PP
-jo [\-p] [\-a] [\-B] [\-v] [\-V] [word ...]
+jo [\-p] [\-a] [\-B] [\-v] [\-V] [\-\-] [ [\-s|\-n|\-b] word ...]
 .SH DESCRIPTION
 .PP
 \f[I]jo\f[] creates a JSON string on \f[I]stdout\f[] from _word_s given
@@ -32,6 +33,119 @@ specified.
 When the \f[C]:=\f[] operator is used in a \f[I]word\f[], the name to
 the right of \f[C]:=\f[] is a file containing JSON which is parsed and
 assigned to the key left of the operator.
+.SH TYPE COERCION
+.PP
+\f[I]jo\f[]\[aq]s type guesses can be overridden on a per\-word basis by
+prefixing \f[I]word\f[] with \f[C]\-s\f[] for \f[I]string\f[],
+\f[C]\-n\f[] for \f[I]number\f[], or \f[C]\-b\f[] for \f[I]boolean\f[].
+The list of _word_s \f[I]must\f[] be prefixed with \f[C]\-\-\f[], to
+indicate to \f[I]jo\f[] that there are no more global options.
+.PP
+Type coercion works as follows:
+.PP
+.TS
+tab(@);
+l l l l l.
+T{
+word
+T}@T{
+\-s
+T}@T{
+\-n
+T}@T{
+\-b
+T}@T{
+default
+T}
+_
+T{
+a=
+T}@T{
+"a":""
+T}@T{
+"a":0
+T}@T{
+"a":false
+T}@T{
+"a":null
+T}
+T{
+a=string
+T}@T{
+"a":"string"
+T}@T{
+"a":6
+T}@T{
+"a":true
+T}@T{
+"a":"string"
+T}
+T{
+a="quoted"
+T}@T{
+"a":""quoted""
+T}@T{
+"a":8
+T}@T{
+"a":true
+T}@T{
+"a":""quoted""
+T}
+T{
+a=12345
+T}@T{
+"a":"12345"
+T}@T{
+"a":12345
+T}@T{
+"a":true
+T}@T{
+"a":12345
+T}
+T{
+a=true
+T}@T{
+"a":"true"
+T}@T{
+"a":1
+T}@T{
+"a":true
+T}@T{
+"a":true
+T}
+T{
+a=false
+T}@T{
+"a":"false"
+T}@T{
+"a":0
+T}@T{
+"a":false
+T}@T{
+"a":false
+T}
+T{
+a=null
+T}@T{
+"a":""
+T}@T{
+"a":0
+T}@T{
+"a":false
+T}@T{
+"a":null
+T}
+.TE
+.PP
+Coercing a non\-number string to number outputs the \f[I]length\f[] of
+the string.
+.PP
+Coercing a non\-boolean string to boolean outputs \f[C]false\f[] if the
+string is empty, \f[C]true\f[] otherwise.
+.PP
+Type coercion only applies to \f[C]key=value\f[] words, and individual
+words in a \f[C]\-a\f[] array.
+Coercing other words has no effect.
 .SH EXAMPLES
 .PP
 Create an object.
@@ -127,6 +241,27 @@ $\ jo\ \-p\ name=Jane\ point[]=1\ point[]=2\ geo[lat]=10\ geo[lon]=20
 \f[]
 .fi
 .PP
+Type coercion:
+.IP
+.nf
+\f[C]
+$\ jo\ \-p\ \-\-\ \-s\ a=true\ b=true\ \-s\ c=123\ d=123\ \-b\ e="1"\ \-b\ f="true"\ \-n\ g="This\ is\ a\ test"\ \-b\ h="This\ is\ a\ test"
+{
+\ \ \ "a":\ "true",
+\ \ \ "b":\ true,
+\ \ \ "c":\ "123",
+\ \ \ "d":\ 123,
+\ \ \ "e":\ true,
+\ \ \ "f":\ true,
+\ \ \ "g":\ 14,
+\ \ \ "h":\ true
+}
+
+$\ jo\ \-a\ \-\-\ \-s\ 123\ \-n\ "This\ is\ a\ test"\ \-b\ C_Rocks\ 456
+["123",14,true,456]
+\f[]
+.fi
+.PP
 Read element values from files: a value which starts with \f[C]\@\f[] is
 read in plain whereas if it begins with a \f[C]%\f[] it will be
 base64\-encoded:
@@ -153,7 +288,7 @@ $\ jo\ files:=child.json
 .fi
 .SH OPTIONS
 .PP
-\f[I]jo\f[] understands the following options.
+\f[I]jo\f[] understands the following global options.
 .TP
 .B \-a
 Interpret the list of \f[I]words\f[] as array values and produce an

jo.c

@@ -35,6 +35,7 @@
 #define FLAG_PRETTY	0x02
 #define FLAG_NOBOOL	0x04
 #define FLAG_BOOLEAN	0x08
+#define FLAG_MASK	(FLAG_ARRAY | FLAG_PRETTY | FLAG_NOBOOL | FLAG_BOOLEAN)
 
 static JsonNode *pile;		/* pile of nested objects/arrays */
 
@@ -45,6 +46,14 @@ static JsonNode *pile;		/* pile of nested objects/arrays */
 # define ftello	ftell
 #endif
 
+JsonTag flags_to_tag(int flags) {
+	return flags / (FLAG_MASK + 1);
+}
+
+int tag_to_flags(JsonTag tag) {
+	return (FLAG_MASK + 1) * tag;
+}
+
 void json_copy_to_object(JsonNode * obj, JsonNode * object_or_array, int clobber)
 {
 	JsonNode *node;
@@ -105,15 +114,83 @@ char *slurp_file(FILE *fp, size_t *out_len, bool fold_newlines)
 	return (buf);
 }
 
+JsonNode *jo_mknull(JsonTag type) {
+	switch (type) {
+		case JSON_STRING:
+			return json_mkstring("");
+			break;
+		case JSON_NUMBER:
+			return json_mknumber(0);
+			break;
+		case JSON_BOOL:
+			return json_mkbool(false);
+			break;
+		default:
+			return json_mknull();
+			break;
+	}
+}
+
+JsonNode *jo_mkbool(bool b, JsonTag type) {
+	switch (type) {
+		case JSON_STRING:
+			return json_mkstring(b ? "true" : "false");
+			break;
+		case JSON_NUMBER:
+			return json_mknumber(b ? 1 : 0);
+			break;
+		default:
+			return json_mkbool(b);
+			break;
+	}
+}
+
+JsonNode *jo_mkstring(char *str, JsonTag type) {
+	switch (type) {
+		case JSON_NUMBER:
+			/* Length of string */
+			return json_mknumber(strlen(str));
+			break;
+		case JSON_BOOL:
+			/* True if not empty */
+			return json_mkbool(strlen(str) > 0);
+			break;
+		default:
+			return json_mkstring(str);
+			break;
+	}
+}
+
+JsonNode *jo_mknumber(char *str, JsonTag type) {
+	/* ASSUMPTION: str already tested as valid number */
+	double n = strtod(str, NULL);
+
+	switch (type) {
+		case JSON_STRING:
+			/* Just return the original representation */
+			return json_mkstring(str);
+			break;
+		case JSON_BOOL:
+			return json_mkbool(n != 0);
+			break;
+		default:
+			/* ASSUMPTION: str already tested as valid number */
+			return json_mknumber(n);
+			break;
+	}
+}
+
 /*
  * Attempt to "sniff" the type of data in `str' and return
  * a JsonNode of the correct JSON type.
  */
 
 JsonNode *vnode(char *str, int flags)
 {
+	JsonTag type = flags_to_tag(flags);
+
 	if (strlen(str) == 0) {
-		return json_mknull();
+		return jo_mknull(type);
 	}
 
 	/* If str begins with a double quote, keep it a string */
@@ -126,23 +203,23 @@ JsonNode *vnode(char *str, int flags)
 			*bp = 0;		/* Chop closing double quote */
 		return json_mkstring(str + 1);
 #endif
-		return json_mkstring(str);
+		return jo_mkstring(str, type);
 	}
 
 	char *endptr;
 	double num = strtod(str, &endptr);
 
 	if (!*endptr && isfinite(num)) {
-		return json_mknumber(num);
+		return jo_mknumber(str, type);
 	}
 
 	if (!(flags & FLAG_NOBOOL)) {
 		if (strcmp(str, "true") == 0) {
-			return json_mkbool(true);
+			return jo_mkbool(true, type);
 		} else if (strcmp(str, "false") == 0) {
-			return json_mkbool(false);
+			return jo_mkbool(false, type);
 		} else if (strcmp(str, "null") == 0) {
-			return json_mknull();
+			return jo_mknull(type);
 		}
 	}
 
@@ -196,7 +273,7 @@ JsonNode *vnode(char *str, int flags)
 		return (obj);
 	}
 
-	return json_mkstring(str);
+	return jo_mkstring(str, type);
 }
 
 /*
@@ -523,9 +600,28 @@ int main(int argc, char **argv)
 		}
 	} else {
 		while ((kv = *argv++)) {
-			p = utf8_from_locale(kv, -1);
-			append_kv(json, flags, p);
-			utf8_free(p);
+			if (kv[0] == '-') {
+				/* Set one-shot coerce flag */
+				switch (kv[1]) {
+					case 'b':
+						flags |= tag_to_flags(JSON_BOOL);
+						break;
+					case 's':
+						flags |= tag_to_flags(JSON_STRING);
+						break;
+					case 'n':
+						flags |= tag_to_flags(JSON_NUMBER);
+						break;
+					default:
+						exit(usage(progname));
+				}
+			} else {
+				p = utf8_from_locale(kv, -1);
+				append_kv(json, flags, p);
+				utf8_free(p);
+				/* Reset any one-shot coerce flags */
+				flags &= FLAG_MASK;
+			}
 		}
 	}