Remove undocumented command-line options (#361)

* Remove --accumulate-numeric-attributes

* Remove join-sqlite, etc.

* Remove --accumulate-numeric-attributes from overzoom

* Remove --assign-to-bins and --bin-by-id-list

* Remove --clip-polygon and --clip-bounding-box

* Remove FSL expressions

* Update version and changelog
This commit is contained in:
Erica Fischer
2025-07-31 17:03:01 -07:00
committed by GitHub
parent 68ab8dcc22
commit 533e000faa
14 changed files with 35 additions and 2920 deletions
-321
View File
@@ -9,170 +9,6 @@
#include "milo/dtoa_milo.h"
#include "text.hpp"
static std::string mvt_value_to_string(mvt_value const &one, bool &fail, std::vector<std::string> const &unidecode_data) {
switch (one.type) {
case mvt_string:
if (unidecode_data.size() > 0) {
return unidecode_smash(unidecode_data, one.c_str());
} else {
return one.get_string_value();
}
case mvt_float:
return milo::dtoa_milo(one.numeric_value.float_value);
case mvt_double:
return milo::dtoa_milo(one.numeric_value.double_value);
case mvt_int:
return std::to_string(one.numeric_value.int_value);
case mvt_uint:
return std::to_string(one.numeric_value.uint_value);
case mvt_sint:
return std::to_string(one.numeric_value.sint_value);
case mvt_bool:
return one.numeric_value.bool_value ? "true" : "false";
case mvt_null:
fail = true; // null op string => null
return "";
case mvt_no_such_key:
default:
fprintf(stderr, "unhandled mvt_type %d\n", one.type);
exit(EXIT_IMPOSSIBLE);
}
}
// Alter the JSON parse tree in place to replace the original match string
// with its unidecode-smashed version.
//
// To avoid repeated re-smashings of the same JSON object, objects that have
// already been smashed are marked by setting their refcon to the unidecode data.
static void smash(std::vector<std::string> const &unidecode_data, json_object *j) {
if (j->value.string.refcon == (void *) &unidecode_data) {
return;
}
std::string s = unidecode_smash(unidecode_data, j->value.string.string);
j->value.string.string = (char *) realloc(j->value.string.string, s.size() + 1);
if (j->value.string.string == NULL) {
perror("realloc for unidecode_smash");
exit(EXIT_MEMORY);
}
strcpy(j->value.string.string, s.c_str());
j->value.string.refcon = (void *) &unidecode_data;
}
int compare_fsl(mvt_value const &one, json_object *two, bool &fail, std::vector<std::string> const &unidecode_data) {
// In FSL expressions, the attribute value is coerced to the type
// of the JSON literal value it is being compared to.
//
// If it cannot be converted, the comparison returns null
// (which is distinct from true and false but is falsy if
// it is the final output value).
if (two->type == JSON_NULL) {
fail = true; // anything op null => null
return 0;
}
if (two->type == JSON_NUMBER) {
double lhs;
switch (one.type) {
case mvt_string: {
char *endptr = NULL;
const char *s = one.c_str();
lhs = strtod(s, &endptr);
if (endptr == s) {
fail = true; // non-numeric-string op number => null
return 0;
}
} break;
case mvt_float:
lhs = one.numeric_value.float_value;
break;
case mvt_double:
lhs = one.numeric_value.double_value;
break;
case mvt_int:
lhs = one.numeric_value.int_value;
break;
case mvt_uint:
lhs = one.numeric_value.uint_value;
break;
case mvt_sint:
lhs = one.numeric_value.sint_value;
break;
case mvt_bool:
lhs = one.numeric_value.bool_value;
break;
case mvt_null:
fail = true; // null op number => null
return 0;
default:
fprintf(stderr, "unhandled mvt_type %d\n", one.type);
exit(EXIT_IMPOSSIBLE);
}
if (lhs < two->value.number.number) {
return -1;
} else if (lhs > two->value.number.number) {
return 1;
} else {
return 0;
}
}
if (two->type == JSON_STRING) {
std::string lhs = mvt_value_to_string(one, fail, unidecode_data);
if (unidecode_data.size() > 0) {
smash(unidecode_data, two);
}
return strcmp(lhs.c_str(), two->value.string.string);
}
if (two->type == JSON_TRUE || two->type == JSON_FALSE) {
bool lhs;
switch (one.type) {
case mvt_string:
lhs = one.get_string_view().size() > 0;
break;
case mvt_float:
lhs = one.numeric_value.float_value != 0;
break;
case mvt_double:
lhs = one.numeric_value.double_value != 0;
break;
case mvt_int:
lhs = one.numeric_value.int_value != 0;
break;
case mvt_uint:
lhs = one.numeric_value.uint_value != 0;
break;
case mvt_sint:
lhs = one.numeric_value.sint_value != 0;
break;
case mvt_bool:
lhs = one.numeric_value.bool_value;
break;
case mvt_null:
fail = true; // null op bool => null
return 0;
case mvt_no_such_key:
default:
fprintf(stderr, "unhandled mvt_type %d\n", one.type);
exit(EXIT_IMPOSSIBLE);
}
bool rhs = two->type == JSON_TRUE;
return lhs - rhs;
}
fprintf(stderr, "unhandled JSON type %d\n", two->type);
exit(EXIT_IMPOSSIBLE);
}
int compare(mvt_value const &one, json_object *two, bool &fail) {
switch (one.type) {
case mvt_string:
@@ -292,163 +128,6 @@ static int eval(std::function<mvt_value(std::string const &)> feature, json_obje
exit(EXIT_FILTER);
}
// FSL conjunctions
if (f->value.array.length == 3 &&
f->value.array.array[1]->type == JSON_STRING &&
(strcmp(f->value.array.array[1]->value.string.string, "or") == 0 ||
strcmp(f->value.array.array[1]->value.string.string, "and") == 0)) {
int lhs;
if (f->value.array.array[0]->type == JSON_STRING) {
// if LHS of a boolean conjunction is a string, it is an attribute reference
mvt_value ff = feature(std::string(f->value.array.array[0]->value.string.string));
if (ff.type != mvt_no_such_key) {
if (ff.type == mvt_bool) {
lhs = ff.numeric_value.bool_value;
} else {
lhs = -1; // not boolean: null
}
} else {
lhs = -1; // not found: null
}
} else {
lhs = eval(feature, f->value.array.array[0], exclude_attributes, unidecode_data);
}
int rhs = eval(feature, f->value.array.array[2], exclude_attributes, unidecode_data);
if (lhs < 0 && rhs < 0) {
return -1; // null op null => null
}
if (strcmp(f->value.array.array[1]->value.string.string, "or") == 0) {
return (lhs > 0) || (rhs > 0);
} else {
return (lhs > 0) && (rhs > 0);
}
}
// FSL comparators
if (f->value.array.length == 3 &&
f->value.array.array[0]->type == JSON_STRING &&
f->value.array.array[1]->type == JSON_STRING &&
(strcmp(f->value.array.array[1]->value.string.string, "lt") == 0 ||
strcmp(f->value.array.array[1]->value.string.string, "gt") == 0 ||
strcmp(f->value.array.array[1]->value.string.string, "le") == 0 ||
strcmp(f->value.array.array[1]->value.string.string, "ge") == 0 ||
strcmp(f->value.array.array[1]->value.string.string, "eq") == 0 ||
strcmp(f->value.array.array[1]->value.string.string, "ne") == 0 ||
strcmp(f->value.array.array[1]->value.string.string, "cn") == 0 ||
strcmp(f->value.array.array[1]->value.string.string, "nc") == 0 ||
strcmp(f->value.array.array[1]->value.string.string, "in") == 0 ||
strcmp(f->value.array.array[1]->value.string.string, "ni") == 0 ||
strcmp(f->value.array.array[1]->value.string.string, "is") == 0 ||
strcmp(f->value.array.array[1]->value.string.string, "isnt") == 0 ||
false)) {
mvt_value lhs = feature(std::string(f->value.array.array[0]->value.string.string));
if (lhs.type == mvt_no_such_key) {
lhs.type = mvt_null;
}
if (f->value.array.array[2]->type == JSON_NULL && strcmp(f->value.array.array[1]->value.string.string, "is") == 0) {
return lhs.type == mvt_null; // null is null => true, anything is null => false
}
if (f->value.array.array[2]->type == JSON_NULL && strcmp(f->value.array.array[1]->value.string.string, "isnt") == 0) {
return lhs.type != mvt_null; // null isnt null => false, anything isnt null => true
}
bool fail = false;
if (f->value.array.array[2]->type == JSON_STRING &&
(strcmp(f->value.array.array[1]->value.string.string, "cn") == 0 ||
strcmp(f->value.array.array[1]->value.string.string, "nc") == 0)) {
if (lhs.type == mvt_null) {
return -1; // null cn/nc anything => null
}
std::string s = mvt_value_to_string(lhs, fail, unidecode_data);
if (fail) {
return -1; // null cn anything => false
}
bool contains;
if (unidecode_data.size() > 0) {
smash(unidecode_data, f->value.array.array[2]);
}
contains = strstr(s.c_str(), f->value.array.array[2]->value.string.string);
if (strcmp(f->value.array.array[1]->value.string.string, "cn") == 0) {
return contains;
} else {
return !contains;
}
}
if (f->value.array.array[2]->type == JSON_ARRAY &&
(strcmp(f->value.array.array[1]->value.string.string, "in") == 0 ||
strcmp(f->value.array.array[1]->value.string.string, "ni") == 0)) {
// no null check here, since null can be in or not in
bool contains = false;
for (size_t i = 0; i < f->value.array.array[2]->value.array.length; i++) {
if (lhs.type == mvt_null) {
if (f->value.array.array[2]->value.array.array[i]->type == JSON_NULL) {
contains = true; // null is null
break;
} else {
contains = false; // null isnt non-null
}
} else {
fail = false;
static std::vector<std::string> no_unidecode_data;
int cmp = compare_fsl(lhs, f->value.array.array[2]->value.array.array[i], fail, no_unidecode_data);
if (fail) {
contains = false; // non-null isnt null
} else if (cmp == 0) {
contains = true;
break;
}
}
}
if (strcmp(f->value.array.array[1]->value.string.string, "in") == 0) {
return contains;
} else {
return !contains;
}
}
if (lhs.type == mvt_null) {
return -1; // null compared to anything => null
}
int cmp = compare_fsl(lhs, f->value.array.array[2], fail, unidecode_data);
if (fail) {
return -1; // null
}
if (strcmp(f->value.array.array[1]->value.string.string, "eq") == 0) {
return cmp == 0;
}
if (strcmp(f->value.array.array[1]->value.string.string, "ne") == 0) {
return cmp != 0;
}
if (strcmp(f->value.array.array[1]->value.string.string, "gt") == 0) {
return cmp > 0;
}
if (strcmp(f->value.array.array[1]->value.string.string, "ge") == 0) {
return cmp >= 0;
}
if (strcmp(f->value.array.array[1]->value.string.string, "lt") == 0) {
return cmp < 0;
}
if (strcmp(f->value.array.array[1]->value.string.string, "le") == 0) {
return cmp <= 0;
}
fprintf(stderr, "expression fsl comparison: can't happen %s\n", f->value.array.array[1]->value.string.string);
exit(EXIT_IMPOSSIBLE);
}
if (f->value.array.array[0]->type != JSON_STRING) {
fprintf(stderr, "Filter operation is not a string: %s\n", json_stringify(f));
exit(EXIT_FILTER);