Add tests for compile success and failure with the utf8 option.

This requires a test in configure.ac for whether UTF8 is enabled in
the PCRE2 lib against which the VMOD is built.

configure.ac

@@ -99,6 +99,24 @@ else
    AC_SUBST([LOCALE_FR_VTC], [locale_fr.vtc.no])
 fi
 
+AC_MSG_CHECKING([whether pcre2 supports Unicode])
+my_save_cflags="$CFLAGS"
+CFLAGS="$PCRE2_CFLAGS -L $PCRE2_LIBS"
+AC_RUN_IFELSE([AC_LANG_SOURCE([[
+	#define PCRE2_CODE_UNIT_WIDTH 8
+	#include "pcre2.h"
+	#include <stdint.h>
+	int main() {
+		uint32_t where;
+		int ret = pcre2_config(PCRE2_CONFIG_UNICODE, &where);
+		if (ret < 0) return(-1);
+		return(where == 0);
+	}]])],
+	[ac_pcre2_unicode=yes], [ac_pcre2_unicode=no])
+AC_MSG_RESULT([$ac_pcre2_unicode])
+CFLAGS="$my_save_cflags"
+AC_SUBST([PCRE2_UTF8], [$ac_pcre2_unicode])
+
 # --enable-stack-protector
 AC_ARG_ENABLE(stack-protector,
        AS_HELP_STRING([--enable-stack-protector],[enable stack protector (default is YES)]),

src/Makefile.am

@@ -21,18 +21,22 @@ vcc_if.c: vcc_if.h
 vcc_if.h vmod_pcre2.man.rst: @VMODTOOL@ $(top_srcdir)/src/vmod_pcre2.vcc
 	@VMODTOOL@ $(top_srcdir)/src/vmod_pcre2.vcc
 
-VMOD_TESTS = $(top_srcdir)/src/tests/*.vtc $(top_srcdir)/src/tests/locale_fr.vtc
+VMOD_TESTS = $(builddir)/tests/*.vtc $(builddir)/tests/locale_fr.vtc $(builddir)/tests/compile_utf8.vtc
 .PHONY: $(VMOD_TESTS)
 
 locale_fr:
 	cp $(top_srcdir)/src/tests/@LOCALE_FR_VTC@ $(top_srcdir)/src/tests/locale_fr.vtc
 
-$(top_srcdir)/src/tests/*.vtc $(top_srcdir)/src/tests/locale_fr.vtc: locale_fr libvmod_pcre2.la
+utf8:
+	cp $(builddir)/tests/compile_utf8.vtc.@PCRE2_UTF8@ $(builddir)/tests/compile_utf8.vtc
+
+$(builddir)/tests/*.vtc $(builddir)/tests/locale_fr.vtc $(builddir)/tests/compile_utf8.vtc: locale_fr utf8 libvmod_pcre2.la
 	PATH=@LIBVARNISHAPI_SBINDIR@:$$PATH \
 	@VARNISHTEST@ -Dvmod_topbuild=$(abs_top_builddir) $@
 
 check: $(VMOD_TESTS)
 	rm $(top_srcdir)/src/tests/locale_fr.vtc
+	rm $(builddir)/tests/compile_utf8.vtc
 
 EXTRA_DIST = \
 	vmod_pcre2.vcc \

src/tests/compile_utf8.vtc.no

+# -*-mode: vcl; coding: utf-8;-*-
+
+# Those instructions tell emacs to use utf-8 encoding for this source;
+# you might want to set up your editor similarly.
+
+varnishtest "compilation success and failure when pcre2 unicode is disabled"
+
+# compiles, but is not interpreted as utf8
+varnish v1 -vcl {
+	import pcre2 from "${vmod_topbuild}/src/.libs/libvmod_pcre2.so";
+	backend b { .host = "${bad_ip}"; }
+
+	sub vcl_init {
+		new r1 = pcre2.regex(pattern=
+"[^ABCDEFGHIJKLMNOPQRSTUVWXYZÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞĀĂĄĆĈĊČĎĐĒĔĖĘĚĜĞĠĢĤĦĨĪĬĮİIJĴĶĹĻĽĿŁŃŅŇŊŌŎŐŒŔŖŘŚŜŞŠŢŤŦŨŪŬŮŰŲŴŶŸŹŻŽƁƂƄƆƇƉƊƋƎƏƐƑƓƔƖƗƘƜƝƟƠƢƤƦƧƩƬƮƯƱƲƳƵƷƸƼDŽLJNJǍǏǑǓǕǗǙǛǞǠǢǤǦǨǪǬǮDZǴǶǷǸǺǼǾȀȂȄȆȈȊȌȎȐȒȔȖȘȚȜȞȠȢȤȦȨȪȬȮȰȲȺȻȽȾɁΆΈΉΊΌΎΏΑΒΓΔΕΖΗΘΙΚΛΜΝΞΟΠΡΣΤΥΦΧΨΩΪΫϒϓϔϘϚϜϞϠϢϤϦϨϪϬϮϴϷϹϺϽϾϿЀЁЂЃЄЅІЇЈЉЊЋЌЍЎЏАБВГДЕЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯѠѢѤѦѨѪѬѮѰѲѴѶѸѺѼѾҀҊҌҎҐҒҔҖҘҚҜҞҠҢҤҦҨҪҬҮҰҲҴҶҸҺҼҾӀӁӃӅӇӉӋӍӐӒӔӖӘӚӜӞӠӢӤӦӨӪӬӮӰӲӴӶӸԀԂԄԆԈԊԌԎԱԲԳԴԵԶԷԸԹԺԻԼԽԾԿՀՁՂՃՄՅՆՇՈՉՊՋՌՍՎՏՐՑՒՓՔՕՖႠႡႢႣႤႥႦႧႨႩႪႫႬႭႮႯႰႱႲႳႴႵႶႷႸႹႺႻႼႽႾႿჀჁჂჃჄჅḀḂḄḆḈḊḌḎḐḒḔḖḘḚḜḞḠḢḤḦḨḪḬḮḰḲḴḶḸḺḼḾṀṂṄṆṈṊṌṎṐṒṔṖṘṚṜṞṠṢṤṦṨṪṬṮṰṲṴṶṸṺṼṾẀẂẄẆẈẊẌẎẐẒẔẠẢẤẦẨẪẬẮẰẲẴẶẸẺẼẾỀỂỄỆỈỊỌỎỐỒỔỖỘỚỜỞỠỢỤỦỨỪỬỮỰỲỴỶỸἈἉἊἋἌἍἎἏἘἙἚἛἜἝἨἩἪἫἬἭἮἯἸἹἺἻἼἽἾἿὈὉὊὋὌὍὙὛὝὟὨὩὪὫὬὭὮὯᾸᾹᾺΆῈΈῊΉῘῙῚΊῨῩῪΎῬῸΌῺΏabcdefghijklmnopqrstuvwxyzªµºßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿāăąćĉċčďđēĕėęěĝğġģĥħĩīĭįıijĵķĸĺļľŀłńņňʼnŋōŏőœŕŗřśŝşšţťŧũūŭůűųŵŷźżžſƀƃƅƈƌƍƒƕƙƚƛƞơƣƥƨƪƫƭưƴƶƹƺƽƾƿdžljnjǎǐǒǔǖǘǚǜǝǟǡǣǥǧǩǫǭǯǰdzǵǹǻǽǿȁȃȅȇȉȋȍȏȑȓȕȗșțȝȟȡȣȥȧȩȫȭȯȱȳȴȵȶȷȸȹȼȿɀɐɑɒɓɔɕɖɗɘəɚɛɜɝɞɟɠɡɢɣɤɥɦɧɨɩɪɫɬɭɮɯɰɱɲɳɴɵɶɷɸɹɺɻɼɽɾɿʀʁʂʃʄʅʆʇʈʉʊʋʌʍʎʏʐʑʒʓʔʕʖʗʘʙʚʛʜʝʞʟʠʡʢʣʤʥʦʧʨʩʪʫʬʭʮʯΐάέήίΰαβγδεζηθικλμνξοπρςστυφχψωϊϋόύώϐϑϕϖϗϙϛϝϟϡϣϥϧϩϫϭϯϰϱϲϳϵϸϻϼабвгдежзийклмнопрстуфхцчшщъыьэюяѐёђѓєѕіїјљњћќѝўџѡѣѥѧѩѫѭѯѱѳѵѷѹѻѽѿҁҋҍҏґғҕҗҙқҝҟҡңҥҧҩҫҭүұҳҵҷҹһҽҿӂӄӆӈӊӌӎӑӓӕӗәӛӝӟӡӣӥӧөӫӭӯӱӳӵӷӹԁԃԅԇԉԋԍԏաբգդեզէըթժիլխծկհձղճմյնշոչպջռսվտրցւփքօֆևᴀᴁᴂᴃᴄᴅᴆᴇᴈᴉᴊᴋᴌᴍᴎᴏᴐᴑᴒᴓᴔᴕᴖᴗᴘᴙᴚᴛᴜᴝᴞᴟᴠᴡᴢᴣᴤᴥᴦᴧᴨᴩᴪᴫᵢᵣᵤᵥᵦᵧᵨᵩᵪᵫᵬᵭᵮᵯᵰᵱᵲᵳᵴᵵᵶᵷᵹᵺᵻᵼᵽᵾᵿᶀᶁᶂᶃᶄᶅᶆᶇᶈᶉᶊᶋᶌᶍᶎᶏᶐᶑᶒᶓᶔᶕᶖᶗᶘᶙᶚḁḃḅḇḉḋḍḏḑḓḕḗḙḛḝḟḡḣḥḧḩḫḭḯḱḳḵḷḹḻḽḿṁṃṅṇṉṋṍṏṑṓṕṗṙṛṝṟṡṣṥṧṩṫṭṯṱṳṵṷṹṻṽṿẁẃẅẇẉẋẍẏẑẓẕẖẗẘẙẚẛạảấầẩẫậắằẳẵặẹẻẽếềểễệỉịọỏốồổỗộớờởỡợụủứừửữựỳỵỷỹἀἁἂἃἄἅἆἇἐἑἒἓἔἕἠἡἢἣἤἥἦἧἰἱἲἳἴἵἶἷὀὁὂὃὄὅὐὑὒὓὔὕὖὗὠὡὢὣὤὥὦὧὰάὲέὴήὶίὸόὺύὼώᾀᾁᾂᾃᾄᾅᾆᾇᾐᾑᾒᾓᾔᾕᾖᾗᾠᾡᾢᾣᾤᾥᾦᾧᾰᾱᾲᾳᾴᾶᾷιῂῃῄῆῇῐῑῒΐῖῗῠῡῢΰῤῥῦῧῲῳῴῶῷⲁⲃⲅⲇⲉⲋⲍⲏⲑⲓⲕⲗⲙⲛⲝⲟⲡⲣⲥⲧⲩⲫⲭⲯⲱⲳⲵⲷⲹⲻⲽⲿⳁⳃⳅⳇⳉⳋⳍⳏⳑⳓⳕⳗⳙⳛⳝⳟⳡⳣⳤⴀⴁⴂⴃⴄⴅⴆⴇⴈⴉⴊⴋⴌⴍⴎⴏⴐⴑⴒⴓⴔⴕⴖⴗⴘⴙⴚⴛⴜⴝⴞⴟⴠⴡⴢⴣⴤⴥfffiflffifflſtstﬓﬔﬕﬖﬗ\d-_^]");
+	}
+} -start
+
+varnish v1 -errvcl {vmod pcre2 error: Cannot compile '.' in r constructor: this version of PCRE2 does not have Unicode support at offset 0} {
+	import pcre2 from "${vmod_topbuild}/src/.libs/libvmod_pcre2.so";
+	backend b { .host = "${bad_ip}"; }
+
+	sub vcl_init {
+		new r = pcre2.regex(".", utf=true);
+	}
+}
+
+varnish v1 -errvcl {vmod pcre2 error: Cannot compile '.' in r constructor: this version of PCRE2 does not have Unicode support at offset 0} {
+	import pcre2 from "${vmod_topbuild}/src/.libs/libvmod_pcre2.so";
+	backend b { .host = "${bad_ip}"; }
+
+	sub vcl_init {
+		new r = pcre2.regex(".", ucp=true);
+	}
+}
+
+# These options are legal when Unicode is not supported.
+varnish v1 -vcl {
+	import pcre2 from "${vmod_topbuild}/src/.libs/libvmod_pcre2.so";
+	backend b { .host = "${bad_ip}"; }
+
+	sub vcl_init {
+		new r = pcre2.regex(".", never_ucp=true, never_utf=true);
+	}
+}