/src/gdal/ogr/ogr_expat.cpp

Source
/******************************************************************************
 *
 * Project:  OGR
 * Purpose:  Convenience function for parsing with Expat library
 * Author:   Even Rouault, even dot rouault at spatialys.com
 *
 ******************************************************************************
 * Copyright (c) 2009-2012, Even Rouault <even dot rouault at spatialys.com>
 *
 * SPDX-License-Identifier: MIT
 ****************************************************************************/

#ifdef HAVE_EXPAT

#include "cpl_port.h"
#include "cpl_conv.h"
#include "cpl_string.h"
#include "ogr_expat.h"

#include <cstddef>
#include <cstdlib>

#include "cpl_error.h"

constexpr size_t OGR_EXPAT_MAX_ALLOWED_ALLOC = 10000000;

static void *OGRExpatMalloc(size_t size) CPL_WARN_UNUSED_RESULT;
static void *OGRExpatRealloc(void *ptr, size_t size) CPL_WARN_UNUSED_RESULT;

/************************************************************************/
/*                              CanAlloc()                              */
/************************************************************************/

static bool CanAlloc(size_t size)
{
    if (size < OGR_EXPAT_MAX_ALLOWED_ALLOC)
        return true;

    if (CPLTestBool(CPLGetConfigOption("OGR_EXPAT_UNLIMITED_MEM_ALLOC", "NO")))
        return true;

    CPLError(
        CE_Failure, CPLE_OutOfMemory,
        "Expat tried to malloc %d bytes. File probably corrupted. "
        "This may also happen in case of a very big XML comment, in which case "
        "you may define the OGR_EXPAT_UNLIMITED_MEM_ALLOC configuration "
        "option to YES to remove that protection.",
        static_cast<int>(size));
    return false;
}

/************************************************************************/
/*                           OGRExpatMalloc()                           */
/************************************************************************/

static void *OGRExpatMalloc(size_t size)
{
    if (CanAlloc(size))
        return malloc(size);

    return nullptr;
}

/************************************************************************/
/*                          OGRExpatRealloc()                           */
/************************************************************************/

// Caller must replace the pointer with the returned pointer.
static void *OGRExpatRealloc(void *ptr, size_t size)
{
    if (CanAlloc(size))
        return realloc(ptr, size);

    return nullptr;
}

/************************************************************************/
/*                          FillWINDOWS1252()                           */
/************************************************************************/

static void FillWINDOWS1252(XML_Encoding *info)
{
    // Map CP1252 bytes to Unicode values.
    for (int i = 0; i < 0x80; ++i)
        info->map[i] = i;

    info->map[0x80] = 0x20AC;
    info->map[0x81] = -1;
    info->map[0x82] = 0x201A;
    info->map[0x83] = 0x0192;
    info->map[0x84] = 0x201E;
    info->map[0x85] = 0x2026;
    info->map[0x86] = 0x2020;
    info->map[0x87] = 0x2021;
    info->map[0x88] = 0x02C6;
    info->map[0x89] = 0x2030;
    info->map[0x8A] = 0x0160;
    info->map[0x8B] = 0x2039;
    info->map[0x8C] = 0x0152;
    info->map[0x8D] = -1;
    info->map[0x8E] = 0x017D;
    info->map[0x8F] = -1;
    info->map[0x90] = -1;
    info->map[0x91] = 0x2018;
    info->map[0x92] = 0x2019;
    info->map[0x93] = 0x201C;
    info->map[0x94] = 0x201D;
    info->map[0x95] = 0x2022;
    info->map[0x96] = 0x2013;
    info->map[0x97] = 0x2014;
    info->map[0x98] = 0x02DC;
    info->map[0x99] = 0x2122;
    info->map[0x9A] = 0x0161;
    info->map[0x9B] = 0x203A;
    info->map[0x9C] = 0x0153;
    info->map[0x9D] = -1;
    info->map[0x9E] = 0x017E;
    info->map[0x9F] = 0x0178;

    for (int i = 0xA0; i <= 0xFF; ++i)
        info->map[i] = i;
}

/************************************************************************/
/*                           FillISO885915()                            */
/************************************************************************/

static void FillISO885915(XML_Encoding *info)
{
    // Map ISO-8859-15 bytes to Unicode values.
    // Generated by generate_encoding_table.c.
    for (int i = 0x00; i < 0xA4; ++i)
        info->map[i] = i;
    info->map[0xA4] = 0x20AC;
    info->map[0xA5] = 0xA5;
    info->map[0xA6] = 0x0160;
    info->map[0xA7] = 0xA7;
    info->map[0xA8] = 0x0161;
    for (int i = 0xA9; i < 0xB4; ++i)
        info->map[i] = i;
    info->map[0xB4] = 0x017D;
    for (int i = 0xB5; i < 0xB8; ++i)
        info->map[i] = i;
    info->map[0xB8] = 0x017E;
    for (int i = 0xB9; i < 0xBC; ++i)
        info->map[i] = i;
    info->map[0xBC] = 0x0152;
    info->map[0xBD] = 0x0153;
    info->map[0xBE] = 0x0178;
    for (int i = 0xBF; i < 0x100; ++i)
        info->map[i] = i;
}

/************************************************************************/
/*                   OGRExpatUnknownEncodingHandler()                   */
/************************************************************************/

static int
OGRExpatUnknownEncodingHandler(void * /* unused_encodingHandlerData */,
                               const XML_Char *name, XML_Encoding *info)
{
    if (EQUAL(name, "WINDOWS-1252"))
        FillWINDOWS1252(info);
    else if (EQUAL(name, "ISO-8859-15"))
        FillISO885915(info);
    else
    {
        CPLDebug("OGR", "Unhandled encoding %s", name);
        return XML_STATUS_ERROR;
    }

    info->data = nullptr;
    info->convert = nullptr;
    info->release = nullptr;

    return XML_STATUS_OK;
}

/************************************************************************/
/*                      OGRCreateExpatXMLParser()                       */
/************************************************************************/

XML_Parser OGRCreateExpatXMLParser()
{
    XML_Memory_Handling_Suite memsuite;
    memsuite.malloc_fcn = OGRExpatMalloc;
    memsuite.realloc_fcn = OGRExpatRealloc;
    memsuite.free_fcn = free;
    XML_Parser hParser = XML_ParserCreate_MM(nullptr, &memsuite, nullptr);

    XML_SetUnknownEncodingHandler(hParser, OGRExpatUnknownEncodingHandler,
                                  nullptr);

    return hParser;
}

#endif  // HAVE_EXPAT

Coverage Report

Created: 2026-06-30 08:33

Line	Count	Source
1		/******************************************************************************
2		*
3		* Project: OGR
4		* Purpose: Convenience function for parsing with Expat library
5		* Author: Even Rouault, even dot rouault at spatialys.com
6		*
7		******************************************************************************
8		* Copyright (c) 2009-2012, Even Rouault <even dot rouault at spatialys.com>
9		*
10		* SPDX-License-Identifier: MIT
11		****************************************************************************/
12
13		#ifdef HAVE_EXPAT
14
15		#include "cpl_port.h"
16		#include "cpl_conv.h"
17		#include "cpl_string.h"
18		#include "ogr_expat.h"
19
20		#include <cstddef>
21		#include <cstdlib>
22
23		#include "cpl_error.h"
24
25		constexpr size_t OGR_EXPAT_MAX_ALLOWED_ALLOC = 10000000;
26
27		static void *OGRExpatMalloc(size_t size) CPL_WARN_UNUSED_RESULT;
28		static void OGRExpatRealloc(void ptr, size_t size) CPL_WARN_UNUSED_RESULT;
29
30		/************************************************************************/
31		/* CanAlloc() */
32		/************************************************************************/
33
34		static bool CanAlloc(size_t size)
35	8.41M	{
36	8.41M	if (size < OGR_EXPAT_MAX_ALLOWED_ALLOC)
37	8.41M	return true;
38
39	0	if (CPLTestBool(CPLGetConfigOption("OGR_EXPAT_UNLIMITED_MEM_ALLOC", "NO")))
40	0	return true;
41
42	0	CPLError(
43	0	CE_Failure, CPLE_OutOfMemory,
44	0	"Expat tried to malloc %d bytes. File probably corrupted. "
45	0	"This may also happen in case of a very big XML comment, in which case "
46	0	"you may define the OGR_EXPAT_UNLIMITED_MEM_ALLOC configuration "
47	0	"option to YES to remove that protection.",
48	0	static_cast<int>(size));
49	0	return false;
50	0	}
51
52		/************************************************************************/
53		/* OGRExpatMalloc() */
54		/************************************************************************/
55
56		static void *OGRExpatMalloc(size_t size)
57	8.19M	{
58	8.19M	if (CanAlloc(size))
59	8.19M	return malloc(size);
60
61	0	return nullptr;
62	8.19M	}
63
64		/************************************************************************/
65		/* OGRExpatRealloc() */
66		/************************************************************************/
67
68		// Caller must replace the pointer with the returned pointer.
69		static void OGRExpatRealloc(void ptr, size_t size)
70	218k	{
71	218k	if (CanAlloc(size))
72	218k	return realloc(ptr, size);
73
74	0	return nullptr;
75	218k	}
76
77		/************************************************************************/
78		/* FillWINDOWS1252() */
79		/************************************************************************/
80
81		static void FillWINDOWS1252(XML_Encoding *info)
82	18	{
83		// Map CP1252 bytes to Unicode values.
84	2.32k	for (int i = 0; i < 0x80; ++i)
85	2.30k	info->map[i] = i;
86
87	18	info->map[0x80] = 0x20AC;
88	18	info->map[0x81] = -1;
89	18	info->map[0x82] = 0x201A;
90	18	info->map[0x83] = 0x0192;
91	18	info->map[0x84] = 0x201E;
92	18	info->map[0x85] = 0x2026;
93	18	info->map[0x86] = 0x2020;
94	18	info->map[0x87] = 0x2021;
95	18	info->map[0x88] = 0x02C6;
96	18	info->map[0x89] = 0x2030;
97	18	info->map[0x8A] = 0x0160;
98	18	info->map[0x8B] = 0x2039;
99	18	info->map[0x8C] = 0x0152;
100	18	info->map[0x8D] = -1;
101	18	info->map[0x8E] = 0x017D;
102	18	info->map[0x8F] = -1;
103	18	info->map[0x90] = -1;
104	18	info->map[0x91] = 0x2018;
105	18	info->map[0x92] = 0x2019;
106	18	info->map[0x93] = 0x201C;
107	18	info->map[0x94] = 0x201D;
108	18	info->map[0x95] = 0x2022;
109	18	info->map[0x96] = 0x2013;
110	18	info->map[0x97] = 0x2014;
111	18	info->map[0x98] = 0x02DC;
112	18	info->map[0x99] = 0x2122;
113	18	info->map[0x9A] = 0x0161;
114	18	info->map[0x9B] = 0x203A;
115	18	info->map[0x9C] = 0x0153;
116	18	info->map[0x9D] = -1;
117	18	info->map[0x9E] = 0x017E;
118	18	info->map[0x9F] = 0x0178;
119
120	1.74k	for (int i = 0xA0; i <= 0xFF; ++i)
121	1.72k	info->map[i] = i;
122	18	}
123
124		/************************************************************************/
125		/* FillISO885915() */
126		/************************************************************************/
127
128		static void FillISO885915(XML_Encoding *info)
129	22.9k	{
130		// Map ISO-8859-15 bytes to Unicode values.
131		// Generated by generate_encoding_table.c.
132	3.78M	for (int i = 0x00; i < 0xA4; ++i)
133	3.75M	info->map[i] = i;
134	22.9k	info->map[0xA4] = 0x20AC;
135	22.9k	info->map[0xA5] = 0xA5;
136	22.9k	info->map[0xA6] = 0x0160;
137	22.9k	info->map[0xA7] = 0xA7;
138	22.9k	info->map[0xA8] = 0x0161;
139	275k	for (int i = 0xA9; i < 0xB4; ++i)
140	252k	info->map[i] = i;
141	22.9k	info->map[0xB4] = 0x017D;
142	91.6k	for (int i = 0xB5; i < 0xB8; ++i)
143	68.7k	info->map[i] = i;
144	22.9k	info->map[0xB8] = 0x017E;
145	91.6k	for (int i = 0xB9; i < 0xBC; ++i)
146	68.7k	info->map[i] = i;
147	22.9k	info->map[0xBC] = 0x0152;
148	22.9k	info->map[0xBD] = 0x0153;
149	22.9k	info->map[0xBE] = 0x0178;
150	1.51M	for (int i = 0xBF; i < 0x100; ++i)
151	1.48M	info->map[i] = i;
152	22.9k	}
153
154		/************************************************************************/
155		/* OGRExpatUnknownEncodingHandler() */
156		/************************************************************************/
157
158		static int
159		OGRExpatUnknownEncodingHandler(void * /* unused_encodingHandlerData */,
160		const XML_Char name, XML_Encoding info)
161	23.3k	{
162	23.3k	if (EQUAL(name, "WINDOWS-1252"))
163	18	FillWINDOWS1252(info);
164	23.3k	else if (EQUAL(name, "ISO-8859-15"))
165	22.9k	FillISO885915(info);
166	391	else
167	391	{
168	391	CPLDebug("OGR", "Unhandled encoding %s", name);
169	391	return XML_STATUS_ERROR;
170	391	}
171
172	22.9k	info->data = nullptr;
173	22.9k	info->convert = nullptr;
174	22.9k	info->release = nullptr;
175
176	22.9k	return XML_STATUS_OK;
177	23.3k	}
178
179		/************************************************************************/
180		/* OGRCreateExpatXMLParser() */
181		/************************************************************************/
182
183		XML_Parser OGRCreateExpatXMLParser()
184	79.9k	{
185	79.9k	XML_Memory_Handling_Suite memsuite;
186	79.9k	memsuite.malloc_fcn = OGRExpatMalloc;
187	79.9k	memsuite.realloc_fcn = OGRExpatRealloc;
188	79.9k	memsuite.free_fcn = free;
189	79.9k	XML_Parser hParser = XML_ParserCreate_MM(nullptr, &memsuite, nullptr);
190
191	79.9k	XML_SetUnknownEncodingHandler(hParser, OGRExpatUnknownEncodingHandler,
192	79.9k	nullptr);
193
194	79.9k	return hParser;
195	79.9k	}
196
197		#endif // HAVE_EXPAT