From dbd89f44c326f5d9917f8d62acdcf86cf6d35b5f Mon Sep 17 00:00:00 2001
From: Eric Fischer <enf@pobox.com>
Date: Tue, 26 Sep 2017 17:12:03 -0700
Subject: [PATCH] Start trying to validate .prj files for shapefiles

---
 projection.cpp | 139 ++++++++++++++++++++++++++++++++++++++++++++++++-
 projection.hpp |   3 ++
 shapefile.cpp  |   6 +++
 3 files changed, 146 insertions(+), 2 deletions(-)

diff --git a/projection.cpp b/projection.cpp
index f1d84f6..d5ad259 100644
--- a/projection.cpp
+++ b/projection.cpp
@@ -2,11 +2,146 @@
 #include <string.h>
 #include <stdlib.h>
 #include <math.h>
+#include <vector>
+#include <string>
+#include <ctype.h>
 #include "projection.hpp"
 
+struct wkt {
+	std::string s;
+	std::vector<wkt> sublist;
+
+	wkt(std::string s_) {
+		s = s_;
+	}
+};
+
+static int peekc(FILE *f) {
+	int c = getc(f);
+	ungetc(c, f);
+	return c;
+}
+
+static void parsewkt(FILE *f, std::vector<wkt> &w) {
+	int c;
+
+	while ((c = getc(f)) != EOF) {
+		if (c == ' ' || c == '\t' || c == '\n' || c == ',') {
+			continue;
+		}
+
+		if (c == '"') {
+			std::string s;
+
+			while ((c = getc(f)) != EOF) {
+				if (c == '"' && peekc(f) == '"') {
+					getc(f);
+					s.push_back('"');
+					continue;
+				}
+
+				if (c == '"') {
+					break;
+				}
+
+				s.push_back(c);
+			}
+
+			w.push_back(wkt(s));
+			continue;
+		}
+
+		if (isalpha(c)) {
+			std::string s;
+			s.push_back(tolower(c));
+
+			while ((c = peekc(f)) != EOF && isalpha(c)) {
+				s.push_back(tolower(c));
+				c = getc(f);
+			}
+
+			w.push_back(wkt(s));
+			continue;
+		}
+
+		if (isdigit(c) || c == '.' || c == '-') {
+			std::string s;
+			s.push_back(c);
+
+			while ((c = peekc(f)) != EOF && (isdigit(c) || c == '.')) {
+				s.push_back(c);
+				c = getc(f);
+			}
+
+			w.push_back(wkt(s));
+			continue;
+		}
+
+		if (c == ')' || c == ']') {
+			return;
+		}
+
+		if (c == '(' || c == '[') {
+			if (w.size() == 0) {
+				fprintf(stderr, "[ or ( in projection without a leading token\n");
+				exit(EXIT_FAILURE);
+			}
+
+			parsewkt(f, w[w.size() - 1].sublist);
+			continue;
+		}
+
+		fprintf(stderr, "Unexpected character %c in projection\n", c);
+		exit(EXIT_FAILURE);
+	}
+}
+
+wkt *find(std::vector<wkt> *w, std::string s) {
+	for (size_t i = 0; i < w->size(); i++) {
+		if ((*w)[i].s == s) {
+			return &(*w)[i];
+		}
+	}
+
+	return NULL;
+}
+
+void warn_wgs84(FILE *f) {
+	std::vector<wkt> w;
+	parsewkt(f, w);
+
+	wkt *geogcs = find(&w, "geogcs");
+	if (geogcs == NULL) {
+		fprintf(stderr, "Warning: expected geographic coordinate system for projection\n");
+		return;
+	}
+
+	wkt *datum = find(&(geogcs->sublist), "datum");
+	if (datum == NULL) {
+		fprintf(stderr, "Warning: expected datum in projection\n");
+		return;
+	}
+
+	wkt *spheroid = find(&(datum->sublist), "spheroid");
+	if (spheroid == NULL || spheroid->sublist.size() == 0) {
+		fprintf(stderr, "Warning: expected spheroid in projection\n");
+		return;
+	}
+
+	if (spheroid->sublist[0].s != "WGS_1984") {
+		fprintf(stderr, "Warning: expected WGS_1984 spheroid in projection, not %s\n", spheroid->sublist[0].s.c_str());
+		return;
+	}
+}
+
+void warn_epsg3857(FILE *f) {
+	std::vector<wkt> w;
+	parsewkt(f, w);
+}
+
 struct projection projections[] = {
-	{"EPSG:4326", lonlat2tile, tile2lonlat, "urn:ogc:def:crs:OGC:1.3:CRS84"},
-	{"EPSG:3857", epsg3857totile, tiletoepsg3857, "urn:ogc:def:crs:EPSG::3857"},
+	{"EPSG:4326", lonlat2tile, tile2lonlat, "urn:ogc:def:crs:OGC:1.3:CRS84", warn_wgs84},
+	{"EPSG:3857", epsg3857totile, tiletoepsg3857, "urn:ogc:def:crs:EPSG::3857", warn_epsg3857},
 	{NULL, NULL},
 };
 
diff --git a/projection.hpp b/projection.hpp
index 4915f83..0f34ba8 100644
--- a/projection.hpp
+++ b/projection.hpp
@@ -1,6 +1,8 @@
 #ifndef PROJECTION_HPP
 #define PROJECTION_HPP
 
+#include <stdio.h>
+
 void lonlat2tile(double lon, double lat, int zoom, long long *x, long long *y);
 void epsg3857totile(double ix, double iy, int zoom, long long *x, long long *y);
 void tile2lonlat(long long x, long long y, int zoom, double *lon, double *lat);
@@ -14,6 +16,7 @@ struct projection {
 	void (*project)(double ix, double iy, int zoom, long long *ox, long long *oy);
 	void (*unproject)(long long ix, long long iy, int zoom, double *ox, double *oy);
 	const char *alias;
+	void (*warn)(FILE *prj);
 };
 
 extern struct projection *projection;
diff --git a/shapefile.cpp b/shapefile.cpp
index 1be484d..3ea05b9 100644
--- a/shapefile.cpp
+++ b/shapefile.cpp
@@ -198,6 +198,7 @@ drawvec decode_geometry(unsigned char *data, size_t len, int *type) {
 
 void parse_shapefile(struct serialization_state *sst, std::string fname, int layer, std::string layername) {
 	std::string dbfname = fname.substr(0, fname.size() - 3) + "dbf";
+	std::string prjname = fname.substr(0, fname.size() - 3) + "prj";
 
 	FILE *shp = fopen(fname.c_str(), "rb");
 	if (shp == NULL) {
@@ -209,6 +210,11 @@ void parse_shapefile(struct serialization_state *sst, std::string fname, int lay
 		perror(dbfname.c_str());
 		exit(EXIT_FAILURE);
 	}
+	FILE *prj = fopen(prjname.c_str(), "r");
+	if (prj != NULL) {
+		projection->warn(prj);
+		fclose(prj);
+	}
 
 	unsigned char shpheader[100];
 	if (fread(shpheader, 1, 100, shp) != 100) {