Search: Add plain text match (#13)

Chaoses-Ib · Nov 7, 2021 · 78705b5 · 78705b5
1 parent fcd67e9
commit 78705b5
Show file tree

Hide file tree

Showing 2 changed files with 60 additions and 6 deletions.
diff --git a/Hijacker/match.cpp b/Hijacker/match.cpp
@@ -1,7 +1,6 @@
 #include "pch.h"
 #include "match.hpp"
 #include <functional>
-#include <IbWinCppLib/WinCppLib.hpp>
 
 char32_t read_char32(const char8_t* str, int* length) {
     char c = str[0];
@@ -15,31 +14,72 @@ char32_t read_char32(const char8_t* str, int* length) {
 
 Pattern* compile(const char8_t* pattern, PatternFlag::Value flags, std::vector<pinyin::PinyinFlagValue>* pinyin_flags) {
     size_t length = 1;  // '\0'
+    size_t length_u8 = 0;
     {
         const char8_t* p = pattern;
         int char_len;
-        for (char32_t c = read_char32(p, &char_len); c; c = read_char32(p += char_len, &char_len))
+        for (char32_t c = read_char32(p, &char_len); c; c = read_char32(p += char_len, &char_len)) {
             length++;
+            length_u8 += char_len;
+        }
     }
     //Pattern* pat = ib::Addr(new ib::Byte[sizeof Pattern + length * sizeof(char32_t)]);
-    Pattern* pat = ib::Addr(HeapAlloc(GetProcessHeap(), 0, sizeof Pattern + length * sizeof(char32_t)));
+    Pattern* pat = ib::Addr(HeapAlloc(GetProcessHeap(), 0, sizeof Pattern + length * sizeof(char32_t) + length_u8 * sizeof(char8_t)));
 
     pat->flags = flags;
     pat->pinyin_flags = pinyin_flags;
 
+    pat->pattern_len = length - 1;
+    pat->pattern_u8_len = length_u8;
+
     const char8_t* p = pattern;
     int char_len;
     for (size_t i = 0; i < length; i++) {
-        pat->pattern[i] = read_char32(p, &char_len);
+        pat->pattern()[i] = read_char32(p, &char_len);
         p += char_len;
     }
 
+    memcpy(pat->pattern_u8(), pattern, length_u8);
+
     return pat;
 }
 
 int exec(Pattern* pattern, const char8_t* subject, int length, size_t nmatch, int pmatch[], PatternFlag::Value flags)
 {
     const char8_t* subject_end = subject + length;
+
+    // plain text match
+    bool plain = true;
+    {
+        const char8_t* s = subject;
+        int char_len;
+        for (char32_t c = read_char32(s, &char_len); s != subject_end; c = read_char32(s += char_len, &char_len)) {
+            if (c >= 0x3007) {
+                plain = false;
+                break;
+            }
+        }
+    }
+    if (plain) {
+        std::u8string_view sv(subject, length);
+        std::u8string_view pt = pattern->pattern_u8_sv();
+        auto it = std::search(sv.begin(), sv.end(), pt.begin(), pt.end(),
+            [](char8_t c1, char8_t c2) {
+                return std::toupper(c1) == std::toupper(c2);
+            });
+
+        if (it == sv.end()) {
+            return -1;
+        } else {
+            if (nmatch) {
+                pmatch[0] = it - sv.begin();
+                pmatch[1] = it - sv.begin() + pt.size();
+                return 1;
+            } else {
+                return 0;
+            }
+        }
+    }
 
     // DFA?
     auto char_match = [pattern](char32_t c, const char32_t* pat) -> std::vector<size_t> {
@@ -85,7 +125,7 @@ int exec(Pattern* pattern, const char8_t* subject, int length, size_t nmatch, in
     const char8_t* sub = subject;
     int char_len;
     while (sub != subject_end) {
-        if (const char8_t* s = subject_match(sub, pattern->pattern)) {
+        if (const char8_t* s = subject_match(sub, pattern->pattern())) {
             if (nmatch) {
                 pmatch[0] = sub - subject;
                 pmatch[1] = s - subject;

diff --git a/Hijacker/match.hpp b/Hijacker/match.hpp
@@ -1,5 +1,6 @@
 #pragma once
 #include <vector>
+#include <IbWinCppLib/WinCppLib.hpp>
 #define IB_PINYIN_ENCODING 32
 #include <IbPinyinLib/Pinyin.hpp>
 
@@ -14,7 +15,20 @@ struct PatternFlag {
 struct Pattern {
     PatternFlag::Value flags;
     std::vector<pinyin::PinyinFlagValue>* pinyin_flags;
-    char32_t pattern[];
+    unsigned int pattern_len;
+    unsigned int pattern_u8_len;
+    //char32_t pattern[];
+    //char8_t pattern_u8[];
+
+    char32_t* pattern() {
+        return ib::Addr(this) + sizeof(Pattern);
+    }
+    char8_t* pattern_u8() {
+        return ib::Addr(this) + sizeof(Pattern) + (pattern_len + 1) * sizeof(char32_t);
+    }
+    std::u8string_view pattern_u8_sv() {
+        return { pattern_u8(), pattern_u8_len };
+    }
 };
 
 Pattern* compile(const char8_t* pattern, PatternFlag::Value flags, std::vector<pinyin::PinyinFlagValue>* pinyin_flags);