1 // Copyright (C) 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
4 **********************************************************************
5 * Copyright (C) 2005-2016, International Business Machines
6 * Corporation and others. All Rights Reserved.
7 **********************************************************************
10 #include "unicode/utypes.h"
12 #if !UCONFIG_NO_CONVERSION
23 #define BUFFER_SIZE 8192
25 #define NEW_ARRAY(type,count) (type *) uprv_malloc((count) * sizeof(type))
26 #define DELETE_ARRAY(array) uprv_free((void *) (array))
28 InputText::InputText(UErrorCode &status)
29 : fInputBytes(NEW_ARRAY(uint8_t, BUFFER_SIZE)), // The text to be checked. Markup will have been
30 // removed if appropriate.
31 fByteStats(NEW_ARRAY(int16_t, 256)), // byte frequency statistics for the input text.
32 // Value is percent, not absolute.
37 if (fInputBytes == NULL || fByteStats == NULL) {
38 status = U_MEMORY_ALLOCATION_ERROR;
42 InputText::~InputText()
44 DELETE_ARRAY(fDeclaredEncoding);
45 DELETE_ARRAY(fByteStats);
46 DELETE_ARRAY(fInputBytes);
49 void InputText::setText(const char *in, int32_t len)
53 fRawInput = (const uint8_t *) in;
54 fRawLength = len == -1? (int32_t)uprv_strlen(in) : len;
57 void InputText::setDeclaredEncoding(const char* encoding, int32_t len)
61 len = (int32_t)uprv_strlen(encoding);
64 len += 1; // to make place for the \0 at the end.
65 uprv_free(fDeclaredEncoding);
66 fDeclaredEncoding = NEW_ARRAY(char, len);
67 uprv_strncpy(fDeclaredEncoding, encoding, len);
71 UBool InputText::isSet() const
73 return fRawInput != NULL;
77 * MungeInput - after getting a set of raw input data to be analyzed, preprocess
78 * it by removing what appears to be html markup.
82 void InputText::MungeInput(UBool fStripTags) {
86 bool inMarkup = FALSE;
91 // html / xml markup stripping.
92 // quick and dirty, not 100% accurate, but hopefully good enough, statistically.
93 // discard everything within < brackets >
94 // Count how many total '<' and illegal (nested) '<' occur, so we can make some
95 // guess as to whether the input was actually marked up at all.
96 // TODO: Think about how this interacts with EBCDIC charsets that are detected.
98 for (srci = 0; srci < fRawLength && dsti < BUFFER_SIZE; srci += 1) {
101 if (b == (uint8_t)0x3C) { /* Check for the ASCII '<' */
111 fInputBytes[dsti++] = b;
114 if (b == (uint8_t)0x3E) { /* Check for the ASCII '>' */
123 // If it looks like this input wasn't marked up, or if it looks like it's
124 // essentially nothing but markup abandon the markup stripping.
125 // Detection will have to work on the unstripped input.
127 if (openTags<5 || openTags/5 < badTags ||
128 (fInputLen < 100 && fRawLength>600))
130 int32_t limit = fRawLength;
132 if (limit > BUFFER_SIZE) {
136 for (srci=0; srci<limit; srci++) {
137 fInputBytes[srci] = fRawInput[srci];
144 // Tally up the byte occurence statistics.
145 // These are available for use by the various detectors.
148 uprv_memset(fByteStats, 0, (sizeof fByteStats[0]) * 256);
150 for (srci = 0; srci < fInputLen; srci += 1) {
151 fByteStats[fInputBytes[srci]] += 1;
154 for (int32_t i = 0x80; i <= 0x9F; i += 1) {
155 if (fByteStats[i] != 0) {