rufus/src/parser.c

/*
 * Rufus: The Reliable USB Formatting Utility
 * Elementary Unicode compliant find/replace parser
 * Copyright (c) 2012 Pete Batard <pete@akeo.ie>
 *
 * This program is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program.  If not, see <http://www.gnu.org/licenses/>.
 */

/* Memory leaks detection - define _CRTDBG_MAP_ALLOC as preprocessor macro */
#ifdef _CRTDBG_MAP_ALLOC
#include <stdlib.h>
#include <crtdbg.h>
#endif

#include <windows.h>
#include <stdio.h>
#include <wchar.h>
#include <string.h>
#include <malloc.h>
#include <io.h>
#include <fcntl.h>

#include "rufus.h"
#include "msapi_utf8.h"

typedef struct {
	uint8_t version[4];
	char* type;				// "release", "beta", "notice"
	char* platform;			// target platform ("windows", "linux", etc.)
	char* platform_arch;	// "x86", "x64", "arm"
	char* platform_min;		// minimum platform version required
	char* download_url[2];
	char* release_notes;
} rufus_update;


// Parse a line of UTF-16 text and return the data if it matches the 'token'
// The parsed line is of the form: [ ]token[ ]=[ ]["]data["][ ] and the line
// is modified by the parser
static wchar_t* get_token_data_line(const wchar_t* wtoken, wchar_t* wline)
{
	const wchar_t wspace[] = L" \t";	// The only whitespaces we recognize as such
	const wchar_t weol[] = L"\r\n";
	size_t i, r;
	BOOLEAN quoteth;

	if ((wtoken == NULL) || (wline == NULL) || (wline[0] == 0))
		return NULL;

	// Eliminate trailing EOL characters
	wline[wcscspn(wline, weol)] = 0;

	i = 0;

	// Skip leading spaces
	i += wcsspn(&wline[i], wspace);

	// Our token should begin a line
	if (_wcsnicmp(&wline[i], wtoken, wcslen(wtoken)) != 0)
		return NULL;

	// Token was found, move past token
	i += wcslen(wtoken);

	// Skip spaces
	i += wcsspn(&wline[i], wspace);

	// Check for an equal sign
	if (wline[i] != L'=') 
		return NULL;
	i++;

	// Skip spaces after equal sign
	i += wcsspn(&wline[i], wspace);

	// eliminate leading quote, if it exists
	if (wline[i] == L'"') {
		quoteth = TRUE;
		i++;
	}

	// Keep the starting pos of our data
	r = i;

	// locate end of string or quote
	while ( (wline[i] != 0) && ((wline[i] != L'"') || ((wline[i] == L'"') && (!quoteth))) )
		i++;
	wline[i] = 0;

	return (wline[r] == 0)?NULL:&wline[r];
}

// Parse a file (ANSI or UTF-8 or UTF-16) and return the data for the first occurence of 'token'
// The returned string is UTF-8 and MUST be freed by the caller
char* get_token_data_file(const char* token, const char* filename)
{
	wchar_t *wtoken = NULL, *wdata= NULL, *wfilename = NULL;
	wchar_t buf[1024];
	FILE* fd = NULL;
	char *ret = NULL;

	if ((filename == NULL) || (token == NULL))
		return NULL;
	if ((filename[0] == 0) || (token[0] == 0))
		return NULL;

	wfilename = utf8_to_wchar(filename);
	if (wfilename == NULL) {
		uprintf("Could not convert '%s' to UTF-16\n", filename);
		goto out;
	}
	wtoken = utf8_to_wchar(token);
	if (wfilename == NULL) {
		uprintf("Could not convert '%s' to UTF-16\n", token);
		goto out;
	}
	fd = _wfopen(wfilename, L"r, ccs=UNICODE");
	if (fd == NULL) goto out;

	// Process individual lines. NUL is always appended.
	// Ideally, we'd check that our buffer fits the line
	while (fgetws(buf, ARRAYSIZE(buf), fd) != NULL) {
		wdata = get_token_data_line(wtoken, buf);
		if (wdata != NULL) {
			ret = wchar_to_utf8(wdata);
			break;
		}
	}

out:
	if (fd != NULL)
		fclose(fd);
	safe_free(wfilename);
	safe_free(wtoken);
	return ret;
}

// Parse a buffer (ANSI or UTF-8) and return the data for the 'n'th occurence of 'token'
// The returned string is UTF-8 and MUST be freed by the caller
char* get_token_data_buffer(const char* token, unsigned int n, const char* buffer, size_t buffer_size)
{
	unsigned int j;
	wchar_t *wtoken = NULL, *wdata = NULL, *wbuffer = NULL, *wline = NULL;
	size_t i;
	BOOL done = FALSE;
	char* ret = NULL;

	// We're handling remote data => better safe than sorry
	if ((token == NULL) || (buffer == NULL) || (buffer_size <= 4) || (buffer_size > 65536))
		goto out;

	// Ensure that our buffer is NUL terminated
	if (buffer[buffer_size-1] != 0)
		goto out;

	wbuffer = utf8_to_wchar(buffer);
	wtoken = utf8_to_wchar(token);
	if ((wbuffer == NULL) || (wtoken == NULL))
		goto out;

	// Process individual lines
	for (i=0,j=0,done=FALSE; (j!=n)&&(!done); ) {
		wline = &wbuffer[i];

		for(;(wbuffer[i]!=L'\n')&&(wbuffer[i]!=L'\r')&&(wbuffer[i]!=0);i++);
		if (wbuffer[i]==0) {
			done = TRUE;
		} else {
			wbuffer[i++] = 0;
		}
		wdata = get_token_data_line(wtoken, wline);
		if (wdata != NULL) {
			j++;
		}
	}
out:
	if (wdata != NULL)
		ret = wchar_to_utf8(wdata);
	safe_free(wbuffer);
	safe_free(wtoken);
	return ret;
}

static __inline char* get_sanitized_token_data_buffer(const char* token, unsigned int n, const char* buffer, size_t buffer_size)
{
	size_t i;
	char* data = get_token_data_buffer(token, n, buffer, buffer_size);
	if (data != NULL) {
		for (i=0; i<safe_strlen(data); i++) {
			if ((data[i] == '\\') && (data[i+1] == 'n')) {
				data[i] = '\r';
				data[i+1] = '\n';
			}
		}
	}
	return data;
}

// Parse an update data file and populates a rufus_update structure.
// NB: since this is remote data, and we're running elevated, it *IS* considered
// potentially malicious, even if it comes from a supposedly trusted server.
// len should be the size of the buffer - 1, for the zero terminator
void parse_update(char* buf, size_t len)
{
	size_t i;
	char *data = NULL, *token;
	char allowed_chars[] = " \t\r\nabcdefghijklmnopqrstuvwxyz"
		"ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789!\"$%^&*()-_+=<>(){}[].,:;#@'/?|~";
	rufus_update update;

	if ((buf == NULL) || (len < 2) || (len > 65536) || (buf[len-1] != 0))
		return;
	// Sanitize the data - Of course not a silver bullet, but it helps
	for (i=0; i<len-1; i++) {
		// Do not sanitize \n yet
		// NB: we have a zero terminator, so we can afford a +1 without overflow
		if ((strchr(allowed_chars, buf[i]) == NULL) && (buf[i] != '\\') && (buf[i+1] != 'n')) {
			buf[i] = ' ';
		}
	}

	if ((data = get_sanitized_token_data_buffer("version", 1, buf, len)) != NULL) {
		for (i=0; (i<4) && ((token = strtok((i==0)?data:NULL, ".")) != NULL); i++) {
			update.version[i] = (uint8_t)atoi(token);
		}
		safe_free(data);
	}
	// TODO: use X-Macros?
	update.type = get_sanitized_token_data_buffer("type", 1, buf, len);
	update.platform = get_sanitized_token_data_buffer("platform", 1, buf, len);
	update.platform_arch = get_sanitized_token_data_buffer("platform_arch", 1, buf, len);
	update.platform_min = get_sanitized_token_data_buffer("platform_min", 1, buf, len);
	for (i=0; i<ARRAYSIZE(update.download_url); i++) {
		update.download_url[i] = get_sanitized_token_data_buffer("download_url", (unsigned int)i+1, buf, len);
	}
	update.release_notes = get_sanitized_token_data_buffer("release_notes", 1, buf, len);

	uprintf("UPDATE DATA:\n");
	uprintf("  version: %d.%d.%d.%d\n", update.version[0], update.version[1], update.version[2], update.version[3]);
	uprintf("  platform: %s\r\n  platform_arch: %s\r\n  platform_min: %s\n", update.platform, update.platform_arch, update.platform_min);
	for (i=0; i<ARRAYSIZE(update.download_url); i++) {
		uprintf("  url%d: %s\n", i+1, update.download_url[i]);
	}
	uprintf("RELEASE NOTES:\r\n%s\n", update.release_notes);

	// TODO: free all these strings!
}

// Insert entry 'data' under section 'section' of a config file
// Section must include the relevant delimitors (eg '[', ']') if needed
char* insert_section_data(const char* filename, const char* section, const char* data, BOOL dos2unix)
{
	const wchar_t* outmode[] = { L"w", L"w, ccs=UTF-8", L"w, ccs=UTF-16LE" };
	wchar_t *wsection = NULL, *wfilename = NULL, *wtmpname = NULL, *wdata = NULL, bom = 0;
	wchar_t wspace[] = L" \t";
	wchar_t buf[1024];
	FILE *fd_in = NULL, *fd_out = NULL;
	size_t i, size;
	int mode;
	char *ret = NULL, tmp[2];

	if ((filename == NULL) || (section == NULL) || (data == NULL))
		return NULL;
	if ((filename[0] == 0) || (section[0] == 0) || (data[0] == 0))
		return NULL;

	wfilename = utf8_to_wchar(filename);
	if (wfilename == NULL) {
		uprintf("Could not convert '%s' to UTF-16\n", filename);
		goto out;
	}
	wsection = utf8_to_wchar(section);
	if (wfilename == NULL) {
		uprintf("Could not convert '%s' to UTF-16\n", section);
		goto out;
	}
	wdata = utf8_to_wchar(data);
	if (wdata == NULL) {
		uprintf("Could not convert '%s' to UTF-16\n", data);
		goto out;
	}

	fd_in = _wfopen(wfilename, L"r, ccs=UNICODE");
	if (fd_in == NULL) {
		uprintf("Could not open file '%s'\n", filename);
		goto out;
	}
	// Check the input file's BOM and create an output file with the same
	fread(&bom, sizeof(bom), 1, fd_in);
	switch(bom) {
	case 0xFEFF:
		mode = 2;	// UTF-16 (LE)
		break;
	case 0xBBEF:	// Yeah, the UTF-8 BOM is really 0xEF,0xBB,0xBF, but
		mode = 1;	// find me a non UTF-8 file that actually begins with "<22><>"
		break;
	default:
		mode = 0;	// ANSI
		break;
	}
	fseek(fd_in, 0, SEEK_SET);
//	uprintf("'%s' was detected as %s\n", filename, 
//		(mode==0)?"ANSI/UTF8 (no BOM)":((mode==1)?"UTF8 (with BOM)":"UTF16 (with BOM"));


	wtmpname = (wchar_t*)calloc(wcslen(wfilename)+2, sizeof(wchar_t));
	if (wtmpname == NULL) {
		uprintf("Could not allocate space for temporary output name\n");
		goto out;
	}
	wcscpy(wtmpname, wfilename);
	wtmpname[wcslen(wtmpname)] = '~';

	fd_out = _wfopen(wtmpname, outmode[mode]);
	if (fd_out == NULL) {
		uprintf("Could not open temporary output file %s~\n", filename);
		goto out;
	}

	// Process individual lines. NUL is always appended.
	while (fgetws(buf, ARRAYSIZE(buf), fd_in) != NULL) {

		i = 0;

		// Skip leading spaces
		i += wcsspn(&buf[i], wspace);

		// Our token should begin a line
		if (_wcsnicmp(&buf[i], wsection, wcslen(wsection)) != 0) {
			fputws(buf, fd_out);
			continue;
		}

		// Section was found, output it
		fputws(buf, fd_out);
		// Now output the new data
		fwprintf(fd_out, L"%s\n", wdata);
		ret = (char*)data;
	}

out:
	if (fd_in != NULL) fclose(fd_in);
	if (fd_out != NULL) fclose(fd_out);

	// If an insertion occured, delete existing file and use the new one
	if (ret != NULL) {
		// We're in Windows text mode => Remove CRs if requested
		fd_in = _wfopen(wtmpname, L"rb");
		fd_out = _wfopen(wfilename, L"wb");
		// Don't check fds
		if ((fd_in != NULL) && (fd_out != NULL)) {
			size = (mode==2)?2:1;
			while(fread(tmp, size, 1, fd_in) == 1) {
				if ((!dos2unix) || (tmp[0] != 0x0D))
					fwrite(tmp, size, 1, fd_out);
			}
			fclose(fd_in);
			fclose(fd_out);
		} else {
			uprintf("Could not write %s - original file has been left unmodifiedn", filename);
			ret = NULL;
			if (fd_in != NULL) fclose(fd_in);
			if (fd_out != NULL) fclose(fd_out);
		}
	} 
	_wunlink(wtmpname);
	safe_free(wfilename);
	safe_free(wtmpname);
	safe_free(wsection);
	safe_free(wdata);

	return ret;
}

// Search for a specific 'src' substring the data for all occurences of 'token', and replace
// it with 'rep'. File can be ANSI or UNICODE and is overwritten. Parameters are UTF-8.
// The parsed line is of the form: [ ]token[ ]data
// Returns a pointer to rep if replacement occured, NULL otherwise
char* replace_in_token_data(const char* filename, const char* token, const char* src, const char* rep, BOOL dos2unix)
{
	const wchar_t* outmode[] = { L"w", L"w, ccs=UTF-8", L"w, ccs=UTF-16LE" };
	wchar_t *wtoken = NULL, *wfilename = NULL, *wtmpname = NULL, *wsrc = NULL, *wrep = NULL, bom = 0;
	wchar_t wspace[] = L" \t";
	wchar_t buf[1024], *torep;
	FILE *fd_in = NULL, *fd_out = NULL;
	size_t i, size;
	int mode;
	char *ret = NULL, tmp[2];

	if ((filename == NULL) || (token == NULL) || (src == NULL) || (rep == NULL))
		return NULL;
	if ((filename[0] == 0) || (token[0] == 0) || (src[0] == 0) || (rep[0] == 0))
		return NULL;
	if (strcmp(src, rep) == 0)	// No need for processing is source is same as replacement
		return NULL;

	wfilename = utf8_to_wchar(filename);
	if (wfilename == NULL) {
		uprintf("Could not convert '%s' to UTF-16\n", filename);
		goto out;
	}
	wtoken = utf8_to_wchar(token);
	if (wfilename == NULL) {
		uprintf("Could not convert '%s' to UTF-16\n", token);
		goto out;
	}
	wsrc = utf8_to_wchar(src);
	if (wsrc == NULL) {
		uprintf("Could not convert '%s' to UTF-16\n", src);
		goto out;
	}
	wrep = utf8_to_wchar(rep);
	if (wsrc == NULL) {
		uprintf("Could not convert '%s' to UTF-16\n", rep);
		goto out;
	}

	fd_in = _wfopen(wfilename, L"r, ccs=UNICODE");
	if (fd_in == NULL) {
		uprintf("Could not open file '%s'\n", filename);
		goto out;
	}
	// Check the input file's BOM and create an output file with the same
	fread(&bom, sizeof(bom), 1, fd_in);
	switch(bom) {
	case 0xFEFF:
		mode = 2;	// UTF-16 (LE)
		break;
	case 0xBBEF:	// Yeah, the UTF-8 BOM is really 0xEF,0xBB,0xBF, but
		mode = 1;	// find me a non UTF-8 file that actually begins with "<22><>"
		break;
	default:
		mode = 0;	// ANSI
		break;
	}
	fseek(fd_in, 0, SEEK_SET);
//	uprintf("'%s' was detected as %s\n", filename, 
//		(mode==0)?"ANSI/UTF8 (no BOM)":((mode==1)?"UTF8 (with BOM)":"UTF16 (with BOM"));


	wtmpname = (wchar_t*)calloc(wcslen(wfilename)+2, sizeof(wchar_t));
	if (wtmpname == NULL) {
		uprintf("Could not allocate space for temporary output name\n");
		goto out;
	}
	wcscpy(wtmpname, wfilename);
	wtmpname[wcslen(wtmpname)] = '~';

	fd_out = _wfopen(wtmpname, outmode[mode]);
	if (fd_out == NULL) {
		uprintf("Could not open temporary output file %s~\n", filename);
		goto out;
	}

	// Process individual lines. NUL is always appended.
	while (fgetws(buf, ARRAYSIZE(buf), fd_in) != NULL) {

		i = 0;

		// Skip leading spaces
		i += wcsspn(&buf[i], wspace);

		// Our token should begin a line
		if (_wcsnicmp(&buf[i], wtoken, wcslen(wtoken)) != 0) {
			fputws(buf, fd_out);
			continue;
		}

		// Token was found, move past token
		i += strlen(token);

		// Skip spaces
		i += wcsspn(&buf[i], wspace);

		torep = wcsstr(&buf[i], wsrc);
		if (torep == NULL) {
			fputws(buf, fd_out);
			continue;
		}

		i = (torep-buf) + wcslen(wsrc);
		*torep = 0;
		fwprintf(fd_out, L"%s%s%s", buf, wrep, &buf[i]);
		ret = (char*)rep;
	}

out:
	if (fd_in != NULL) fclose(fd_in);
	if (fd_out != NULL) fclose(fd_out);

	// If a replacement occured, delete existing file and use the new one
	if (ret != NULL) {
		// We're in Windows text mode => Remove CRs if requested
		fd_in = _wfopen(wtmpname, L"rb");
		fd_out = _wfopen(wfilename, L"wb");
		// Don't check fds
		if ((fd_in != NULL) && (fd_out != NULL)) {
			size = (mode==2)?2:1;
			while(fread(tmp, size, 1, fd_in) == 1) {
				if ((!dos2unix) || (tmp[0] != 0x0D))
					fwrite(tmp, size, 1, fd_out);
			}
			fclose(fd_in);
			fclose(fd_out);
		} else {
			uprintf("Could not write %s - original file has been left unmodified.\n", filename);
			ret = NULL;
			if (fd_in != NULL) fclose(fd_in);
			if (fd_out != NULL) fclose(fd_out);
		}
	}
	_wunlink(wtmpname);
	safe_free(wfilename);
	safe_free(wtmpname);
	safe_free(wtoken);
	safe_free(wsrc);
	safe_free(wrep);

	return ret;
}