From c3b4686082bc92c70886ee848064009c8f628193 Mon Sep 17 00:00:00 2001 From: Zher Huei Lee Date: Mon, 24 Oct 2016 22:13:26 +0100 Subject: [PATCH] Added global sub and bounds checking to RegEx --- doc/base/classes.xml | 14 +++++----- modules/regex/regex.cpp | 57 ++++++++++++++++++++++++++++++++--------- modules/regex/regex.h | 2 +- 3 files changed, 54 insertions(+), 19 deletions(-) diff --git a/doc/base/classes.xml b/doc/base/classes.xml index cafb1449158..47fa1deb551 100644 --- a/doc/base/classes.xml +++ b/doc/base/classes.xml @@ -32582,7 +32582,7 @@ - Searches the text for the compiled pattern. Returns a [RegExMatch] container of the first matching reult if found, otherwise null. The starting point of the serch could be specified without moving the string start anchor. + Searches the text for the compiled pattern. Returns a [RegExMatch] container of the first matching reult if found, otherwise null. The region to search within can be specified without modifying where the start and end anchor would be. @@ -32590,14 +32590,16 @@ - + - + - + + + - Searches the specified text for the compiled pattern and returns the text with the result replaced. Escapes and backreferences such as [code]\1[/code] and [code]\g<name>[/code] are automatically expanded and resolved. If no change was found the unmodified text is returned instead. + Searches the text for the compiled pattern and replaces it with the specified string. Escapes and backreferences such as [code]\1[/code] and [code]\g<name>[/code] expanded and resolved. By default only the first instance is replaced but it can be changed for all instances (global replacement). The region to search within can be specified without modifying where the start and end anchor would be. @@ -32616,7 +32618,7 @@ - Using results from the search, returns the specified string with escapes and backreferences such as [code]\1[/code] and [code]\g<name>[/code] expanded and resolved + Using results from the search, returns the specified string with escapes and backreferences such as [code]\1[/code] and [code]\g<name>[/code] expanded and resolved. diff --git a/modules/regex/regex.cpp b/modules/regex/regex.cpp index 8f26d764c42..388e6dfdecc 100644 --- a/modules/regex/regex.cpp +++ b/modules/regex/regex.cpp @@ -1340,6 +1340,12 @@ Error RegEx::compile(const String& p_pattern) { Ref RegEx::search(const String& p_text, int p_start, int p_end) const { + ERR_FAIL_COND_V(!is_valid(), NULL); + ERR_FAIL_COND_V(p_start < 0, NULL); + ERR_FAIL_COND_V(p_start >= p_text.length(), NULL); + ERR_FAIL_COND_V(p_end > p_text.length(), NULL); + ERR_FAIL_COND_V(p_end != -1 && p_end < p_start, NULL); + Ref res = memnew(RegExMatch()); for (int i = 0; i < group_names.size(); ++i) { @@ -1350,7 +1356,7 @@ Ref RegEx::search(const String& p_text, int p_start, int p_end) cons res->string = p_text; - if (p_end < p_start || p_end > p_text.length()) + if (p_end == -1) p_end = p_text.length(); RegExSearch s(res, p_end, lookahead_depth); @@ -1369,18 +1375,45 @@ Ref RegEx::search(const String& p_text, int p_start, int p_end) cons return NULL; } -String RegEx::sub(const String& p_text, const String& p_template, int p_start, int p_end) const { +String RegEx::sub(const String& p_text, const String& p_replacement, bool p_all, int p_start, int p_end) const { - Ref m = search(p_text, p_start, p_end); - RegExMatch::Group& s = m->captures[0]; - if (s.start >= 0) { - String res = p_text.substr(0, s.start) + m->expand(p_template); - int end = s.start + s.length; - if (end < p_text.length()) - res += p_text.substr(end, p_text.length() - end); - return res; + ERR_FAIL_COND_V(!is_valid(), p_text); + ERR_FAIL_COND_V(p_start < 0, p_text); + ERR_FAIL_COND_V(p_start >= p_text.length(), p_text); + ERR_FAIL_COND_V(p_end > p_text.length(), p_text); + ERR_FAIL_COND_V(p_end != -1 && p_end < p_start, p_text); + + String text = p_text; + int start = p_start; + + if (p_end == -1) + p_end = p_text.length(); + + while (start < text.length() && (p_all || start == p_start)) { + + Ref m = search(text, start, p_end); + + RegExMatch::Group& s = m->captures[0]; + + if (s.start < 0) + break; + + String res = text.substr(0, s.start) + m->expand(p_replacement); + + start = res.length(); + + if (s.length == 0) + ++start; + + int sub_end = s.start + s.length; + if (sub_end < text.length()) + res += text.substr(sub_end, text.length() - sub_end); + + p_end += res.length() - text.length(); + + text = res; } - return p_text; + return text; } void RegEx::clear() { @@ -1456,7 +1489,7 @@ void RegEx::_bind_methods() { ObjectTypeDB::bind_method(_MD("clear"),&RegEx::clear); ObjectTypeDB::bind_method(_MD("compile","pattern"),&RegEx::compile); ObjectTypeDB::bind_method(_MD("search","text","start","end"),&RegEx::search, DEFVAL(0), DEFVAL(-1)); - ObjectTypeDB::bind_method(_MD("sub","text","template","start","end"),&RegEx::sub, DEFVAL(0), DEFVAL(-1)); + ObjectTypeDB::bind_method(_MD("sub","text","replacement","all","start","end"),&RegEx::sub, DEFVAL(false), DEFVAL(0), DEFVAL(-1)); ObjectTypeDB::bind_method(_MD("is_valid"),&RegEx::is_valid); ObjectTypeDB::bind_method(_MD("get_pattern"),&RegEx::get_pattern); ObjectTypeDB::bind_method(_MD("get_group_count"),&RegEx::get_group_count); diff --git a/modules/regex/regex.h b/modules/regex/regex.h index 283368c34fc..8d31b847732 100644 --- a/modules/regex/regex.h +++ b/modules/regex/regex.h @@ -97,7 +97,7 @@ public: Error compile(const String& p_pattern); Ref search(const String& p_text, int p_start = 0, int p_end = -1) const; - String sub(const String& p_text, const String& p_template, int p_start = 0, int p_end = -1) const; + String sub(const String& p_text, const String& p_replacement, bool p_all = false, int p_start = 0, int p_end = -1) const; bool is_valid() const; String get_pattern() const;