Description: build with pcre2
Origin: other, https://helperbyte.com/questions/457338/how-to-make-pcre2-support-for-apache-24
Bug-Debian: https://bugs.debian.org/1000114
Forwarded: not-needed
Reviewed-By: Yadd <yadd@debian.org>
Last-Update: 2021-12-28

--- a/build/NWGNUmakefile
+++ b/build/NWGNUmakefile
@@ -20,7 +20,7 @@
 	$(SRC)/include/ap_config_layout.h \
 	$(NWOS)/test_char.h \
 	$(PCRE)/config.h \
-	$(PCRE)/pcre.h \
+	$(PCRE)/pcre2.h \
 	$(EOLIST) 
     
 nlms :: libs $(NWOS)/httpd.imp $(DAV)/main/dav.imp $(STDMOD)/cache/mod_cache.imp
@@ -77,7 +77,7 @@
 	@echo $(DL)GEN  $@$(DL)
 	$< $@
 
-%.exe: $(PCRE)/%.c $(PCRE)/config.h $(PCRE)/pcre.h
+%.exe: $(PCRE)/%.c $(PCRE)/config.h $(PCRE)/pcre2.h
 	@echo $(DL)Creating Build Helper $@$(DL)
 	$(CC_FOR_BUILD) $(CFLAGS_FOR_BUILD) -DHAVE_CONFIG_H $< -o $@
 
@@ -117,7 +117,7 @@
 clean ::
 	$(call DEL,$(SRC)/include/ap_config_layout.h)
 	$(call DEL,$(PCRE)/config.h)
-	$(call DEL,$(PCRE)/pcre.h)
+	$(call DEL,$(PCRE)/pcre2.h)
 	$(call DEL,$(STDMOD)/cache/mod_cache.imp)
 	$(call DEL,$(DAV)/main/dav.imp)
 	$(call DEL,$(NWOS)/httpd.imp)
--- a/include/ap_config_auto.h.in
+++ b/include/ap_config_auto.h.in
@@ -393,3 +393,6 @@
 
 /* Define to 'int' if <sys/resource.h> doesn't define it for us */
 #undef rlim_t
+
+/* Load PCRE2 */
+#define HAVE_PCRE2 1
--- a/include/ap_regex.h
+++ b/include/ap_regex.h
@@ -70,19 +70,22 @@
 
 /* Options for ap_regcomp, ap_regexec, and ap_rxplus versions: */
 
-#define AP_REG_ICASE    0x01 /** use a case-insensitive match */
-#define AP_REG_NEWLINE  0x02 /** don't match newlines against '.' etc */
-#define AP_REG_NOTBOL   0x04 /** ^ will not match against start-of-string */
-#define AP_REG_NOTEOL   0x08 /** $ will not match against end-of-string */
-
-#define AP_REG_EXTENDED (0)  /** unused */
-#define AP_REG_NOSUB    (0)  /** unused */
-
-#define AP_REG_MULTI 0x10    /* perl's /g (needs fixing) */
-#define AP_REG_NOMEM 0x20    /* nomem in our code */
-#define AP_REG_DOTALL 0x40   /* perl's /s flag */
+#define AP_REG_ICASE    0x01 /**< use a case-insensitive match */
+#define AP_REG_NEWLINE  0x02 /**< don't match newlines against '.' etc */
+#define AP_REG_NOTBOL   0x04 /**< ^ will not match against start-of-string */
+#define AP_REG_NOTEOL   0x08 /**< $ will not match against end-of-string */
+
+#define AP_REG_EXTENDED (0)  /**< unused */
+#define AP_REG_NOSUB    (0)  /**< unused */
+
+#define AP_REG_MULTI    0x10 /**< perl's /g (needs fixing) */
+#define AP_REG_NOMEM    0x20 /**< nomem in our code */
+#define AP_REG_DOTALL   0x40 /**< perl's /s flag */
+ 
+#define AP_REG_NOTEMPTY 0x080 /**< Empty match not valid */
+#define AP_REG_ANCHORED 0x100 /**< Match at the first position */
 
-#define AP_REG_DOLLAR_ENDONLY 0x200 /* '$' matches at end of subject string only */
+#define AP_REG_DOLLAR_ENDONLY 0x200 /**< '$' matches at end of subject string only */
 
 #define AP_REG_NO_DEFAULT 0x400 /**< Don't implicitely add AP_REG_DEFAULT options */
 
@@ -90,6 +93,12 @@
 
 #define AP_REG_DEFAULT (AP_REG_DOTALL|AP_REG_DOLLAR_ENDONLY)
 
+/* Arguments for ap_pcre_version_string */
+enum {
+  AP_REG_PCRE_COMPILED = 0, /** PCRE version used during program compilation */
+  AP_REG_PCRE_LOADED        /** PCRE version loaded at runtime */
+};
+
 /* Error values: */
 enum {
   AP_REG_ASSERT = 1,  /** internal error ? */
@@ -114,6 +123,15 @@
 /* The functions */
 
 /**
+ * Return PCRE version string.
+ * @param which Either AP_REG_PCRE_COMPILED (PCRE version used
+ *              during program compilation) or AP_REG_PCRE_LOADED
+ *              (PCRE version used at runtime)
+ * @return The PCRE version string
+ */
+AP_DECLARE(const char *) ap_pcre_version_string(int which);
+
+/**
  * Get default compile flags
  * @return Bitwise OR of AP_REG_* flags
  */
@@ -277,5 +295,4 @@
 }   /* extern "C" */
 #endif
 
-#endif /* AP_REGEX_T */
-
+#endif /* AP_REGEX_H */
--- a/server/util_pcre.c
+++ b/server/util_pcre.c
@@ -55,19 +55,17 @@
 #include "httpd.h"
 #include "apr_strings.h"
 #include "apr_tables.h"
-#include "pcre.h"
 
-/* PCRE_DUPNAMES is only present since version 6.7 of PCRE */
-#ifndef PCRE_DUPNAMES
-#error PCRE Version 6.7 or later required!
-#else
+#define PCRE2_CODE_UNIT_WIDTH 8
+#include "pcre2.h"
+#define PCREn(x) PCRE2_ ## x
 
+/* PCRE_DUPNAMES is only present since version 6.7 of PCRE */
 #define APR_WANT_STRFUNC
 #include "apr_want.h"
 
 #ifndef POSIX_MALLOC_THRESHOLD
 #define POSIX_MALLOC_THRESHOLD (10)
-#endif
 
 /* Table of error strings corresponding to POSIX error codes; must be
  * kept in synch with include/ap_regex.h's AP_REG_E* definitions.
@@ -81,6 +79,20 @@
     "match failed"              /* AP_REG_NOMATCH */
 };
 
+AP_DECLARE(const char *) ap_pcre_version_string(int which)
+{
+    static char buf[80];
+    switch (which) {
+    case AP_REG_PCRE_COMPILED:
+        return APR_STRINGIFY(PCREn(MAJOR)) "." APR_STRINGIFY(PCREn(MINOR)) " " APR_STRINGIFY(PCREn(DATE));
+    case AP_REG_PCRE_LOADED:
+        pcre2_config(PCRE2_CONFIG_VERSION, buf);
+        return buf;
+    default:
+        return "Unknown";
+    }
+}
+
 AP_DECLARE(apr_size_t) ap_regerror(int errcode, const ap_regex_t *preg,
                                    char *errbuf, apr_size_t errbuf_size)
 {
@@ -115,7 +127,7 @@
 
 AP_DECLARE(void) ap_regfree(ap_regex_t *preg)
 {
-    (pcre_free)(preg->re_pcre);
+    pcre2_code_free(preg->re_pcre);
 }
 
 
@@ -168,39 +180,38 @@
 */
 AP_DECLARE(int) ap_regcomp(ap_regex_t * preg, const char *pattern, int cflags)
 {
-    const char *errorptr;
-    int erroffset;
+    uint32_t capcount;
+    size_t erroffset;
     int errcode = 0;
-    int options = PCRE_DUPNAMES;
+    int options = PCREn(DUPNAMES);
 
     if ((cflags & AP_REG_NO_DEFAULT) == 0)
         cflags |= default_cflags;
 
     if ((cflags & AP_REG_ICASE) != 0)
-        options |= PCRE_CASELESS;
+        options |= PCREn(CASELESS);
     if ((cflags & AP_REG_NEWLINE) != 0)
-        options |= PCRE_MULTILINE;
+        options |= PCREn(MULTILINE);
     if ((cflags & AP_REG_DOTALL) != 0)
-        options |= PCRE_DOTALL;
+        options |= PCREn(DOTALL);
     if ((cflags & AP_REG_DOLLAR_ENDONLY) != 0)
-        options |= PCRE_DOLLAR_ENDONLY;
+        options |= PCREn(DOLLAR_ENDONLY);
 
-    preg->re_pcre =
-        pcre_compile2(pattern, options, &errcode, &errorptr, &erroffset, NULL);
-    preg->re_erroffset = erroffset;
+    preg->re_pcre = pcre2_compile((const unsigned char *)pattern,
+                                  PCRE2_ZERO_TERMINATED, options, &errcode,
+                                  &erroffset, NULL);
 
+    preg->re_erroffset = erroffset;
     if (preg->re_pcre == NULL) {
-        /*
-         * There doesn't seem to be constants defined for compile time error
-         * codes. 21 is "failed to get memory" according to pcreapi(3).
-         */
+        /* Internal ERR21 is "failed to get memory" according to pcreapi(3) */
         if (errcode == 21)
             return AP_REG_ESPACE;
         return AP_REG_INVARG;
     }
 
-    pcre_fullinfo((const pcre *)preg->re_pcre, NULL,
-                   PCRE_INFO_CAPTURECOUNT, &(preg->re_nsub));
+    pcre2_pattern_info((const pcre2_code *)preg->re_pcre,
+                       PCRE2_INFO_CAPTURECOUNT, &capcount);
+    preg->re_nsub = capcount;
     return 0;
 }
 
@@ -232,74 +243,77 @@
 {
     int rc;
     int options = 0;
-    int *ovector = NULL;
-    int small_ovector[POSIX_MALLOC_THRESHOLD * 3];
-    int allocated_ovector = 0;
+    apr_size_t nlim;
+    pcre2_match_data *matchdata;
+    size_t *ovector;
 
     if ((eflags & AP_REG_NOTBOL) != 0)
-        options |= PCRE_NOTBOL;
+        options |= PCREn(NOTBOL);
     if ((eflags & AP_REG_NOTEOL) != 0)
-        options |= PCRE_NOTEOL;
-
-    ((ap_regex_t *)preg)->re_erroffset = (apr_size_t)(-1);    /* Only has meaning after compile */
-
-    if (nmatch > 0) {
-        if (nmatch <= POSIX_MALLOC_THRESHOLD) {
-            ovector = &(small_ovector[0]);
-        }
-        else {
-            ovector = (int *)malloc(sizeof(int) * nmatch * 3);
-            if (ovector == NULL)
-                return AP_REG_ESPACE;
-            allocated_ovector = 1;
-        }
-    }
-
-    rc = pcre_exec((const pcre *)preg->re_pcre, NULL, buff, (int)len,
-                   0, options, ovector, nmatch * 3);
-
+        options |= PCREn(NOTEOL);
+    if ((eflags & AP_REG_NOTEMPTY) != 0)
+        options |= PCREn(NOTEMPTY);
+    if ((eflags & AP_REG_ANCHORED) != 0)
+        options |= PCREn(ANCHORED);
+
+    /* TODO: create a generic TLS matchdata buffer of some nmatch limit,
+     * e.g. 10 matches, to avoid a malloc-per-call. If it must be allocated,
+     * implement a general context using palloc and no free implementation.
+     */
+    nlim = ((apr_size_t)preg->re_nsub + 1) > nmatch
+         ? ((apr_size_t)preg->re_nsub + 1) : nmatch;
+    matchdata = pcre2_match_data_create(nlim, NULL);
+    if (matchdata == NULL)
+        return AP_REG_ESPACE;
+    ovector = pcre2_get_ovector_pointer(matchdata);
+    rc = pcre2_match((const pcre2_code *)preg->re_pcre,
+                     (const unsigned char *)buff, len,
+                     0, options, matchdata, NULL);
     if (rc == 0)
-        rc = nmatch;            /* All captured slots were filled in */
+        rc = nlim;            /* All captured slots were filled in */
 
     if (rc >= 0) {
         apr_size_t i;
-        for (i = 0; i < (apr_size_t)rc; i++) {
+        nlim = (apr_size_t)rc < nmatch ? (apr_size_t)rc : nmatch;
+        for (i = 0; i < nlim; i++) {
             pmatch[i].rm_so = ovector[i * 2];
             pmatch[i].rm_eo = ovector[i * 2 + 1];
         }
-        if (allocated_ovector)
-            free(ovector);
         for (; i < nmatch; i++)
             pmatch[i].rm_so = pmatch[i].rm_eo = -1;
-        return 0;
     }
 
+    pcre2_match_data_free(matchdata);
+
+    if (rc >= 0) {
+        return 0;
+    }
     else {
-        if (allocated_ovector)
-            free(ovector);
+        if (rc <= PCRE2_ERROR_UTF8_ERR1 && rc >= PCRE2_ERROR_UTF8_ERR21)
+            return AP_REG_INVARG;
         switch (rc) {
-        case PCRE_ERROR_NOMATCH:
+        case PCREn(ERROR_NOMATCH):
             return AP_REG_NOMATCH;
-        case PCRE_ERROR_NULL:
+        case PCREn(ERROR_NULL):
             return AP_REG_INVARG;
-        case PCRE_ERROR_BADOPTION:
+        case PCREn(ERROR_BADOPTION):
             return AP_REG_INVARG;
-        case PCRE_ERROR_BADMAGIC:
+        case PCREn(ERROR_BADMAGIC):
             return AP_REG_INVARG;
-        case PCRE_ERROR_UNKNOWN_NODE:
-            return AP_REG_ASSERT;
-        case PCRE_ERROR_NOMEMORY:
+        case PCREn(ERROR_NOMEMORY):
             return AP_REG_ESPACE;
-#ifdef PCRE_ERROR_MATCHLIMIT
-        case PCRE_ERROR_MATCHLIMIT:
+        case PCREn(ERROR_MATCHLIMIT):
             return AP_REG_ESPACE;
+#if defined(PCRE_ERROR_UNKNOWN_NODE)
+        case PCRE_ERROR_UNKNOWN_NODE:
+            return AP_REG_ASSERT;
 #endif
-#ifdef PCRE_ERROR_BADUTF8
-        case PCRE_ERROR_BADUTF8:
+#if defined(PCRE_ERROR_BADUTF8)
+        case PCREn(ERROR_BADUTF8):
             return AP_REG_INVARG;
 #endif
-#ifdef PCRE_ERROR_BADUTF8_OFFSET
-        case PCRE_ERROR_BADUTF8_OFFSET:
+#if defined(PCRE_ERROR_BADUTF8_OFFSET)
+        case PCREn(ERROR_BADUTF8_OFFSET):
             return AP_REG_INVARG;
 #endif
         default:
@@ -312,17 +326,17 @@
                            apr_array_header_t *names, const char *prefix,
                            int upper)
 {
-    int namecount;
-    int nameentrysize;
-    int i;
     char *nametable;
 
-    pcre_fullinfo((const pcre *)preg->re_pcre, NULL,
-                       PCRE_INFO_NAMECOUNT, &namecount);
-    pcre_fullinfo((const pcre *)preg->re_pcre, NULL,
-                       PCRE_INFO_NAMEENTRYSIZE, &nameentrysize);
-    pcre_fullinfo((const pcre *)preg->re_pcre, NULL,
-                       PCRE_INFO_NAMETABLE, &nametable);
+    uint32_t namecount;
+    uint32_t nameentrysize;
+    uint32_t i;
+    pcre2_pattern_info((const pcre2_code *)preg->re_pcre,
+                       PCRE2_INFO_NAMECOUNT, &namecount);
+    pcre2_pattern_info((const pcre2_code *)preg->re_pcre,
+                       PCRE2_INFO_NAMEENTRYSIZE, &nameentrysize);
+    pcre2_pattern_info((const pcre2_code *)preg->re_pcre,
+                       PCRE2_INFO_NAMETABLE, &nametable);
 
     for (i = 0; i < namecount; i++) {
         const char *offset = nametable + i * nameentrysize;