From ee6d1b3fb499e13306e1963d56794264320aa7a4 Mon Sep 17 00:00:00 2001 From: Georgi Gerganov Date: Sun, 28 Apr 2024 18:36:57 +0300 Subject: [PATCH] unicode : simplify --- unicode.cpp | 33 +++++++++++++++------------------ 1 file changed, 15 insertions(+), 18 deletions(-) diff --git a/unicode.cpp b/unicode.cpp index 55fafb6c5..214f78658 100644 --- a/unicode.cpp +++ b/unicode.cpp @@ -620,25 +620,22 @@ std::vector unicode_regex_split(const std::string & text, const std continue; } - if (regex_expr[i] == '\\' && i + 1 < regex_expr.size()) { - if (regex_expr[i + 1] == 'p') { - if (i + 3 < regex_expr.size() && regex_expr[i + 2] == '{') { - if (regex_expr[i + 4] == '}') { - const std::string pat = regex_expr.substr(i, 5); - if (k_ucat_enum.find(pat) != k_ucat_enum.end()) { - if (!inside) { - regex_expr_collapsed += '['; - } - regex_expr_collapsed += k_ucat_cpt.at(k_ucat_enum.at(pat)); - regex_expr_collapsed += k_ucat_map.at(k_ucat_enum.at(pat)); - if (!inside) { - regex_expr_collapsed += ']'; - } - i += 4; - continue; - } - } + if (regex_expr[i + 0] == '\\' && i + 4 < regex_expr.size() && + regex_expr[i + 1] == 'p' && + regex_expr[i + 2] == '{' && + regex_expr[i + 4] == '}') { + const std::string pat = regex_expr.substr(i, 5); + if (k_ucat_enum.find(pat) != k_ucat_enum.end()) { + if (!inside) { + regex_expr_collapsed += '['; } + regex_expr_collapsed += k_ucat_cpt.at(k_ucat_enum.at(pat)); + regex_expr_collapsed += k_ucat_map.at(k_ucat_enum.at(pat)); + if (!inside) { + regex_expr_collapsed += ']'; + } + i += 4; + continue; } }