Regex: Add regex_match_all_c and regex_match_all natives (by Nextra)

dlls/regex/CRegEx.cpp

@@ -42,7 +42,7 @@ RegEx::RegEx()
 	re = NULL;
 	mFree = true;
 	subject = NULL;
-	mSubStrings = 0;
+	mSubStrings.clear();
 }
 
 void RegEx::Clear()
@@ -54,9 +54,9 @@ void RegEx::Clear()
 	re = NULL;
 	mFree = true;
 	if (subject)
-		delete [] subject;
+		delete[] subject;
 	subject = NULL;
-	mSubStrings = 0;
+	mSubStrings.clear();
 }
 
 RegEx::~RegEx()
@@ -152,11 +152,11 @@ int RegEx::Match(const char *str)
 
 	if (mFree || re == NULL)
 		return -1;
-		
+
 	this->ClearMatch();
 
 	//save str
-	subject = new char[strlen(str)+1];
+	subject = new char[strlen(str) + 1];
 	strcpy(subject, str);
 
 	rc = pcre_exec(re, NULL, subject, (int)strlen(subject), 0, 0, ovector, 30);
@@ -166,37 +166,110 @@ int RegEx::Match(const char *str)
 		if (rc == PCRE_ERROR_NOMATCH)
 		{
 			return 0;
-		} else {
+		}
+		else {
 			mErrorOffset = rc;
 			return -1;
 		}
 	}
 
-	mSubStrings = rc;
+	RegExSub res;
+	mSubStrings.ensure(rc);
+
+	for (int s = 0; s < rc; ++s)
+	{
+		res.start = ovector[2 * s];
+		res.end = ovector[2 * s + 1];
+		mSubStrings.append(res);
+	}
 
 	return 1;
 }
+
+int RegEx::MatchAll(const char *str)
+{
+	int rc = 0;
+	int rr = 0;
+	int offset = 0;
+
+	if (mFree || re == NULL)
+		return -1;
+
+	this->ClearMatch();
+
+	//save str
+	subject = new char[strlen(str) + 1];
+	strcpy(subject, str);
+
+	RegExSub sub, whole;
+	while ((rr = pcre_exec(re, NULL, subject, (int)strlen(subject), offset, 0, ovector, 30)))
+	{
+		if (rr < 0)
+		{
+			if (rr == PCRE_ERROR_NOMATCH)
+			{
+				break;
+			}
+			else
+			{
+				mErrorOffset = rr;
+
+				if (rc)
+					this->ClearMatch();
+
+				return -1;
+			}
+		}
+
+		rc += rr;
+		mSubStrings.ensure(rc);
+
+		for (int s = 1; s < rr; ++s)
+		{
+			sub.start = ovector[2 * s];
+			sub.end = ovector[2 * s + 1];
+			mSubStrings.append(sub);
+		}
+
+		offset = ovector[1];
+	}
+
+	if (!rc)
+		return 0;
+
+	sub = mSubStrings.at(0);
+	whole.start = sub.start;
+	sub = mSubStrings.back();
+	whole.end = sub.end;
+
+	mSubStrings.insert(0, whole);
+
+	return 1;
+}
+
 void RegEx::ClearMatch()
 {
 	// Clears match results
 	mErrorOffset = 0;
 	mError = NULL;
 	if (subject)
-		delete [] subject;
+		delete[] subject;
 	subject = NULL;
-	mSubStrings = 0;
+	mSubStrings.clear();
 }
 
 const char *RegEx::GetSubstring(int s, char buffer[], int max)
 {
 	int i = 0;
-	if (s >= mSubStrings || s < 0)
+	if ((size_t)s >= mSubStrings.length() || s < 0)
 		return NULL;
 
-	char *substr_a = subject + ovector[2*s];
+	RegExSub sub = mSubStrings.at(s);
-	int substr_l = ovector[2*s+1] - ovector[2*s];
 
-	for (i = 0; i<substr_l; i++)
+	char *substr_a = subject + sub.start;
+	int substr_l = sub.end - sub.start;
+
+	for (i = 0; i < substr_l; i++)
 	{
 		if (i >= max)
 			break;
@@ -206,5 +279,4 @@ const char *RegEx::GetSubstring(int s, char buffer[], int max)
 	buffer[i] = '\0';
 
 	return buffer;
-}
+}
-

dlls/regex/CRegEx.h

@@ -32,29 +32,40 @@
  */
 #ifndef _INCLUDE_CREGEX_H
 #define _INCLUDE_CREGEX_H
+ 
+#include <am-vector.h>
 
 class RegEx
 {
 public:
+	struct RegExSub {
+		int start, end;
+	};
+
 	RegEx();
 	~RegEx();
+
 	bool isFree(bool set=false, bool val=false);
 	void Clear();
 
 	int Compile(const char *pattern, const char* flags = NULL);
 	int Compile(const char *pattern, int iFlags);
 	int Match(const char *str);
+	int MatchAll(const char *str);
 	void ClearMatch();
 	const char *GetSubstring(int s, char buffer[], int max);
+
 public:
 	int mErrorOffset;
 	const char *mError;
-	int mSubStrings;
+	int Count() { return mSubStrings.length(); }
+
 private:
 	pcre *re;
 	bool mFree;
 	int ovector[30];
 	char *subject;
+	ke::Vector<RegExSub> mSubStrings;
 };
 
 #endif //_INCLUDE_CREGEX_H

dlls/regex/module.cpp

@@ -79,7 +79,7 @@ static cell AMX_NATIVE_CALL regex_compile_ex(AMX *amx, cell *params)
 {
 	int len;
 	const char *regex = MF_GetAmxString(amx, params[1], 0, &len);
-	
+
 	int id = GetPEL();
 	RegEx *x = PEL[id];
 
@@ -95,9 +95,7 @@ static cell AMX_NATIVE_CALL regex_compile_ex(AMX *amx, cell *params)
 	return id + 1;
 }
 
-// 1.8 includes the last parameter
+cell match(AMX *amx, cell *params, bool all)
-// Regex:regex_match(const string[], const pattern[], &ret, error[], maxLen, const flags[] = "");
-static cell AMX_NATIVE_CALL regex_match(AMX *amx, cell *params)
 {
 	int len;
 	const char *str = MF_GetAmxString(amx, params[1], 0, &len);
@@ -105,24 +103,29 @@ static cell AMX_NATIVE_CALL regex_match(AMX *amx, cell *params)
 
 	int id = GetPEL();
 	RegEx *x = PEL[id];
-	
+
 	char* flags = NULL;
-	
+
 	if ((params[0] / sizeof(cell)) >= 6) // compiled with 1.8's extra parameter
 	{
 		flags = MF_GetAmxString(amx, params[6], 2, &len);
 	}
-	
+
 	if (x->Compile(regex, flags) == 0)
 	{
 		cell *eOff = MF_GetAmxAddr(amx, params[3]);
 		const char *err = x->mError;
 		*eOff = x->mErrorOffset;
-		MF_SetAmxString(amx, params[4], err?err:"unknown", params[5]);
+		MF_SetAmxString(amx, params[4], err ? err : "unknown", params[5]);
 		return -1;
 	}
 
-	int e = x->Match(str);
+	int e;
+	if (all)
+		e = x->MatchAll(str);
+	else
+		e = x->Match(str);
+
 	if (e == -1)
 	{
 		/* there was a match error.  destroy this and move on. */
@@ -130,24 +133,38 @@ static cell AMX_NATIVE_CALL regex_match(AMX *amx, cell *params)
 		*res = x->mErrorOffset;
 		x->Clear();
 		return -2;
-	} else if (e == 0) {
+	}
+	else if (e == 0) {
 		cell *res = MF_GetAmxAddr(amx, params[3]);
 		*res = 0;
 		x->Clear();
 		return 0;
-	} else {
+	}
+	else {
 		cell *res = MF_GetAmxAddr(amx, params[3]);
-		*res = x->mSubStrings;
+		*res = x->Count();
 	}
 
-	return id+1;
+	return id + 1;
 }
 
-// native regex_match_c(const string[], Regex:id, &ret);
+// 1.8 includes the last parameter
-static cell AMX_NATIVE_CALL regex_match_c(AMX *amx, cell *params)
+// Regex:regex_match(const string[], const pattern[], &ret, error[], maxLen, const flags[] = "");
+static cell AMX_NATIVE_CALL regex_match(AMX *amx, cell *params)
+{
+	return match(amx, params, false);
+}
+
+// Regex:regex_match_all(const string[], const pattern[], &ret, error[], maxLen, const flags[] = "");
+static cell AMX_NATIVE_CALL regex_match_all(AMX *amx, cell *params)
+{
+	return match(amx, params, true);
+}
+
+cell match_c(AMX *amx, cell *params, bool all)
 {
 	int len;
-	int id = params[2]-1;
+	int id = params[2] - 1;
 	const char *str = MF_GetAmxString(amx, params[1], 0, &len);
 
 	if (id >= (int)PEL.length() || id < 0 || PEL[id]->isFree())
@@ -155,31 +172,50 @@ static cell AMX_NATIVE_CALL regex_match_c(AMX *amx, cell *params)
 		MF_LogError(amx, AMX_ERR_NATIVE, "Invalid regex handle %d", id);
 		return 0;
 	}
-	
+
 	RegEx *x = PEL[id];
 
-	int e = x->Match(str);
+	int e;
+	if (all)
+		e = x->MatchAll(str);
+	else
+		e = x->Match(str);
+
 	if (e == -1)
 	{
 		/* there was a match error.  move on. */
 		cell *res = MF_GetAmxAddr(amx, params[3]);
 		*res = x->mErrorOffset;
 		/* only clear the match results, since the regex object
-		   may still be referenced later */
+		may still be referenced later */
 		x->ClearMatch();
 		return -2;
-	} else if (e == 0) {
+	}
+	else if (e == 0) {
 		cell *res = MF_GetAmxAddr(amx, params[3]);
 		*res = 0;
 		/* only clear the match results, since the regex object
-		   may still be referenced later */
+		may still be referenced later */
 		x->ClearMatch();
 		return 0;
-	} else {
-		cell *res = MF_GetAmxAddr(amx, params[3]);
-		*res = x->mSubStrings;
-		return x->mSubStrings;
 	}
+	else {
+		cell *res = MF_GetAmxAddr(amx, params[3]);
+		*res = x->Count();
+		return x->Count();
+	}
+}
+
+// native regex_match_c(const string[], Regex:id, &ret);
+static cell AMX_NATIVE_CALL regex_match_c(AMX *amx, cell *params)
+{
+	return match_c(amx, params, false);
+}
+
+// native regex_match_all_c(const string[], Regex:id, &ret);
+static cell AMX_NATIVE_CALL regex_match_all_c(AMX *amx, cell *params)
+{
+	return match_c(amx, params, true);
 }
 
 // native regex_match_ex(Regex:id, const string[], &RegexError:ret = REGEX_ERROR_NONE);
@@ -219,7 +255,7 @@ static cell AMX_NATIVE_CALL regex_match_ex(AMX *amx, cell *params)
 	}
 	else 
 	{
-		return x->mSubStrings;
+		return x->Count();
 	}
 }
 
@@ -270,6 +306,8 @@ AMX_NATIVE_INFO regex_Natives[] = {
 	{"regex_match",				regex_match},
 	{"regex_match_c",			regex_match_c},
 	{"regex_match_ex",			regex_match_ex},
+	{"regex_match_all",			regex_match_all},
+	{"regex_match_all_c",		regex_match_all_c},
 	{"regex_substr",			regex_substr},
 	{"regex_free",				regex_free},
 	{NULL,						NULL},

plugins/include/regex.inc

@@ -266,6 +266,61 @@ native Regex:regex_compile_ex(const pattern[], flags = 0, error[]= "", maxLen =
  */
 native regex_match_ex(Handle:regex, const str[], &RegexError:ret = REGEX_ERROR_NONE);
 
+/**
+ * Matches a string against a pre-compiled regular expression pattern, matching all 
+ * occurances of the pattern inside the string. This is similar to using the "g" flag 
+ * in perl regex.
+ *
+ *
+ * @param pattern		The regular expression pattern.
+ * @param string		The string to check.
+ * @param ret			Error code, if applicable, or number of results on success.
+ *
+ * @return				-2 = Matching error (error code is stored in ret)
+ *						 0 = No match.
+ *						>1 = Number of results.
+ *
+ * @note				You should free the returned handle (with regex_free())
+ *						when you are done with this pattern.
+ *
+ * @note				Use the regex handle passed to this function to extract
+ *						matches with regex_substr().
+ */
+native regex_match_all_c(const string[], Regex:pattern, &ret);
+
+/**
+ * Matches a string against a regular expression pattern, matching all occurances of the
+ * pattern inside the string. This is similar to using the "g" flag in perl regex.
+ *
+ * @note				If you intend on using the same regular expression pattern
+ *						multiple times, consider using regex_compile and regex_match_c 
+ *						instead of making this function reparse the expression each time.
+ *
+ * @param string		The string to check.
+ * @param pattern		The regular expression pattern.
+ * @param ret			Error code, or result state of the match.
+ * @param error		Error message, if applicable.
+ * @param maxLen		Maximum length of the error buffer.
+ * @param flags		General flags for the regular expression.
+ *						i = Ignore case
+ *						m = Multilines (affects ^ and $ so that they match 
+ *							the start/end of a line rather than matching the 
+ *							start/end of the string).
+ *						s = Single line (affects . so that it matches any character, 
+ *							even new line characters).
+ *						x = Pattern extension (ignore whitespace and # comments).
+ *
+ * @return				-2 = Matching error (error code is stored in ret)
+ *						-1 = Error in pattern (error message and offset # in error and ret)
+ *						 0 = No match.
+ *						>1 = Handle for getting more information (via regex_substr)
+ *
+ * @note				Flags only exist in amxmodx 1.8 and later.
+ * @note				You should free the returned handle (with regex_free())
+ *						when you are done extracting all of the substrings.
+ */
+native Regex:regex_match_all(const string[], const pattern[], &ret, error[], maxLen, const flags[] = "");
+
 /**
  * Matches a string against a regular expression pattern.
  *
@@ -298,4 +353,4 @@ stock regex_match_simple(const str[], const pattern[], flags = 0, error[]="", ma
 	regex_free(regex);
 
 	return substrings;
-}
+}