ds@1952: /**
ds@1952:  * vim: set ts=4 :
ds@1952:  * =============================================================================
ds@1952:  * SourceMod
ds@1952:  * Copyright (C) 2004-2008 AlliedModders LLC.  All rights reserved.
ds@1952:  * =============================================================================
ds@1952:  *
ds@1952:  * This program is free software; you can redistribute it and/or modify it under
ds@1952:  * the terms of the GNU General Public License, version 3.0, as published by the
ds@1952:  * Free Software Foundation.
ds@1952:  * 
ds@1952:  * This program is distributed in the hope that it will be useful, but WITHOUT
ds@1952:  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
ds@1952:  * FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more
ds@1952:  * details.
ds@1952:  *
ds@1952:  * You should have received a copy of the GNU General Public License along with
ds@1952:  * this program.  If not, see <http://www.gnu.org/licenses/>.
ds@1952:  *
ds@1952:  * As a special exception, AlliedModders LLC gives you permission to link the
ds@1952:  * code of this program (as well as its derivative works) to "Half-Life 2," the
ds@1952:  * "Source Engine," the "SourcePawn JIT," and any Game MODs that run on software
ds@1952:  * by the Valve Corporation.  You must obey the GNU General Public License in
ds@1952:  * all respects for all other code used.  Additionally, AlliedModders LLC grants
ds@1952:  * this exception to all derivative works.  AlliedModders LLC defines further
ds@1952:  * exceptions, found in LICENSE.txt (as of this writing, version JULY-31-2007),
ds@1952:  * or <http://www.sourcemod.net/license.php>.
ds@1952:  *
ds@1952:  * Version: $Id$
ds@1952:  */
ds@1952: 
ds@1952: #include <stdio.h>
ds@1952: #include <ctype.h>
ds@1952: #include <wctype.h>
ds@1952: #include <string.h>
ds@1952: #include <stdlib.h>
ds@1952: #include <assert.h>
ds@1952: #include "TextParsers.h"
ds@1952: #include "ShareSys.h"
ds@1952: #include "sm_stringutil.h"
ds@1952: #include "LibrarySys.h"
ds@1952: 
ds@1952: TextParsers g_TextParser;
ds@1952: ITextParsers *textparsers = &g_TextParser;
ds@1952: 
ds@1952: static int g_ini_chartable1[255] = {0};
ds@1952: static int g_ws_chartable[255] = {0};
ds@1952: 
ds@1952: bool TextParsers::IsWhitespace(const char *stream)
ds@1952: {
ds@1952: 	return g_ws_chartable[(unsigned)*stream] == 1;
ds@1952: }
ds@1952: 
ds@1952: TextParsers::TextParsers()
ds@1952: {
ds@1952: 	g_ini_chartable1[(unsigned)'_'] = 1;
ds@1952: 	g_ini_chartable1[(unsigned)'-'] = 1;
ds@1952: 	g_ini_chartable1[(unsigned)','] = 1;
ds@1952: 	g_ini_chartable1[(unsigned)'+'] = 1;
ds@1952: 	g_ini_chartable1[(unsigned)'.'] = 1;
ds@1952: 	g_ini_chartable1[(unsigned)'$'] = 1;
ds@1952: 	g_ini_chartable1[(unsigned)'?'] = 1;
ds@1952: 	g_ini_chartable1[(unsigned)'/'] = 1;
ds@1952: 	g_ws_chartable[(unsigned)'\n'] = 1;
ds@1952: 	g_ws_chartable[(unsigned)'\v'] = 1;
ds@1952: 	g_ws_chartable[(unsigned)'\r'] = 1;
ds@1952: 	g_ws_chartable[(unsigned)'\t'] = 1;
ds@1952: 	g_ws_chartable[(unsigned)'\f'] = 1;
ds@1952: 	g_ws_chartable[(unsigned)' '] = 1;
ds@1952: }
ds@1952: 
ds@1952: void TextParsers::OnSourceModAllInitialized()
ds@1952: {
ds@1952: 	g_ShareSys.AddInterface(NULL, this);
ds@1952: }
ds@1952: 
ds@1952: unsigned int TextParsers::GetUTF8CharBytes(const char *stream)
ds@1952: {
ds@1952: 	return _GetUTF8CharBytes(stream);
ds@1952: }
ds@1952: 
ds@1952: /**
ds@1952:  * Character streams
ds@1952:  */
ds@1952: 
ds@1952: struct CharStream
ds@1952: {
ds@1952: 	const char *curpos;
ds@1952: };
ds@1952: 
ds@1952: bool CharStreamReader(void *stream, char *buffer, size_t maxlength, unsigned int *read)
ds@1952: {
ds@1952: 	CharStream *srdr = (CharStream *)stream;
ds@1952: 
ds@1952: 	const char *ptr = srdr->curpos;
ds@1952: 	for (size_t i=0; i<maxlength; i++)
ds@1952: 	{
ds@1952: 		if (*ptr == '\0')
ds@1952: 		{
ds@1952: 			break;
ds@1952: 		}
ds@1952: 		*buffer++ = *ptr++;
ds@1952: 	}
ds@1952: 
ds@1952: 	*read = ptr - srdr->curpos;
ds@1952: 
ds@1952: 	srdr->curpos = ptr;
ds@1952: 
ds@1952: 	return true;
ds@1952: }
ds@1952: 
ds@1952: SMCError TextParsers::ParseString_SMC(const char *stream, 
ds@1952: 					 ITextListener_SMC *smc,
ds@1952: 					 SMCStates *states)
ds@1952: {
ds@1952: 	CharStream srdr = { stream };
ds@1952: 
ds@1952: 	return ParseStream_SMC(&srdr, CharStreamReader, smc, states);
ds@1952: }
ds@1952: 
ds@1952: /**
ds@1952:  * File streams
ds@1952:  */
ds@1952: 
ds@1952: bool FileStreamReader(void *stream, char *buffer, size_t maxlength, unsigned int *read)
ds@1952: {
ds@1952: 	size_t num = fread(buffer, 1, maxlength, (FILE *)stream);
ds@1952: 
ds@1952: 	*read = static_cast<unsigned int>(num);
ds@1952: 
ds@1952: 	if (num == 0 && feof((FILE *)stream))
ds@1952: 	{
ds@1952: 		return true;
ds@1952: 	}
ds@1952: 
ds@1952: 	return (ferror((FILE *)stream) == 0);
ds@1952: }
ds@1952: 
ds@1952: SMCError TextParsers::ParseFile_SMC(const char *file, ITextListener_SMC *smc, SMCStates *states)
ds@1952: {
ds@1952: 	FILE *fp = fopen(file, "rt");
ds@1952: 
ds@1952: 	if (!fp)
ds@1952: 	{
ds@1952: 		if (states != NULL)
ds@1952: 		{
ds@1952: 			states->line = 0;
ds@1952: 			states->col = 0;
ds@1952: 		}
ds@1952: 		return SMCError_StreamOpen;
ds@1952: 	}
ds@1952: 
ds@1952: 	SMCError result = ParseStream_SMC(fp, FileStreamReader, smc, states);
ds@1952: 
ds@1952: 	fclose(fp);
ds@1952: 
ds@1952: 	return result;
ds@1952: }
ds@1952: 
ds@1952: SMCError TextParsers::ParseSMCFile(const char *file,
ds@1952: 								   ITextListener_SMC *smc_listener,
ds@1952: 								   SMCStates *states,
ds@1952: 								   char *buffer,
ds@1952: 								   size_t maxsize)
ds@1952: {
ds@1952: 	const char *errstr;
ds@1952: 	FILE *fp = fopen(file, "rt");
ds@1952: 
ds@1952: 	if (fp == NULL)
ds@1952: 	{
ds@1952: 		char error[256] = "unknown";
ds@1952: 		if (states != NULL)
ds@1952: 		{
ds@1952: 			states->line = 0;
ds@1952: 			states->col = 0;
ds@1952: 		}
ds@1952: 		g_LibSys.GetPlatformError(error, sizeof(error));
ds@1952: 		UTIL_Format(buffer, maxsize, "File could not be opened: %s", error);
ds@1952: 		return SMCError_StreamOpen;
ds@1952: 	}
ds@1952: 
ds@1952: 	SMCError result = ParseStream_SMC(fp, FileStreamReader, smc_listener, states);
ds@1952: 
ds@1952: 	fclose(fp);
ds@1952: 
ds@1952: 	errstr = GetSMCErrorString(result);
ds@1952: 	UTIL_Format(buffer, maxsize, "%s", errstr != NULL ? errstr : "Unknown error");
ds@1952: 
ds@1952: 	return result;
ds@1952: }
ds@1952: 
ds@1952: /** 
ds@1952:  * Raw parsing of streams with helper functions
ds@1952:  */
ds@1952: 
ds@1952: struct StringInfo
ds@1952: {
ds@1952: 	StringInfo() : quoted(false), ptr(NULL), end(NULL), special(false) { }
ds@1952: 	bool quoted;
ds@1952: 	char *ptr;
ds@1952: 	char *end;
ds@1952: 	bool special;
ds@1952: };
ds@1952: 
ds@1952: const char *FixupString(StringInfo &data)
ds@1952: {
ds@1952: 	if (!data.ptr)
ds@1952: 	{
ds@1952: 		return NULL;
ds@1952: 	}
ds@1952: 
ds@1952: 	if (data.quoted)
ds@1952: 	{
ds@1952: 		data.ptr++;
ds@1952: 	}
ds@1952: #if defined _DEBUG
ds@1952: 	else {
ds@1952: 		/* A string will never have beginning whitespace because we ignore it in the stream.
ds@1952: 		 * Furthermore, if there is trailing whitespace, the end ptr will point to it, so it is valid
ds@1952: 		 * to overwrite!  Lastly, the last character must be whitespace or a comment/invalid character.
ds@1952: 		 */
ds@1952:  	}
ds@1952: #endif
ds@1952: 
ds@1952: 	/* Do some extra work on strings that have special quoted characters. */
ds@1952: 	if (data.special)
ds@1952: 	{
ds@1952: 		char *outptr = data.ptr;
ds@1952: 		size_t len = data.end - data.ptr;
ds@1952: 		if (len >= 2)
ds@1952: 		{
ds@1952: 			for (size_t i=0; i<len; i++)
ds@1952: 			{
ds@1952: 				if (data.ptr[i] == '\\' && i < len - 1)
ds@1952: 				{
ds@1952: 					/* Resolve the next character. */
ds@1952: 					i++;
ds@1952: 					if (data.ptr[i] == 'n')
ds@1952: 					{
ds@1952: 						data.ptr[i] = '\n';
ds@1952: 					} else if (data.ptr[i] == 't') {
ds@1952: 						data.ptr[i] = '\t';
ds@1952: 					} else if (data.ptr[i] == 'r') {
ds@1952: 						data.ptr[i] = '\r';
ds@1952: 					} else if (data.ptr[i] != '\\'
ds@1952: 						&& data.ptr[i] != '"') {
ds@1952: 						/* This character is invalid, so go back one */
ds@1952: 						i--;
ds@1952: 					}
ds@1952: 				}
ds@1952: 				*outptr++ = data.ptr[i];
ds@1952: 			}
ds@1952: 			*outptr = '\0';
ds@1952: 		}
ds@1952: 	}
ds@1952: 
ds@1952: 	if (data.end)
ds@1952: 	{
ds@1952: 		*(data.end) = '\0';
ds@1952: 	}
ds@1952: 
ds@1952: 	return data.ptr;
ds@1952: }
ds@1952: 
ds@1952: const char *rotate(StringInfo info[3])
ds@1952: {
ds@1952: 	if (info[2].ptr != NULL)
ds@1952: 	{
ds@1952: 		return info[2].ptr;
ds@1952: 	}
ds@1952: 
ds@1952: 	if (info[0].ptr != NULL)
ds@1952: 	{
ds@1952: 		info[2] = info[1];
ds@1952: 		info[1] = info[0];
ds@1952: 		info[0] = StringInfo();
ds@1952: 	}
ds@1952: 	
ds@1952: 	return NULL;
ds@1952: }
ds@1952: 
ds@1952: void scrap(StringInfo info[3])
ds@1952: {
ds@1952: 	info[2] = StringInfo();
ds@1952: 	info[1] = StringInfo();
ds@1952: 	info[0] = StringInfo();
ds@1952: }
ds@1952: 
ds@1952: void reloc(StringInfo &data, unsigned int bytes)
ds@1952: {
ds@1952: 	if (data.ptr)
ds@1952: 	{
ds@1952: 		data.ptr -= bytes;
ds@1952: 	}
ds@1952: 	if (data.end)
ds@1952: 	{
ds@1952: 		data.end -= bytes;
ds@1952: 	}
ds@1952: }
ds@1952: 
ds@1952: char *lowstring(StringInfo info[3])
ds@1952: {
ds@1952: 	for (int i=2; i>=0; i--)
ds@1952: 	{
ds@1952: 		if (info[i].ptr)
ds@1952: 		{
ds@1952: 			return info[i].ptr;
ds@1952: 		}
ds@1952: 	}
ds@1952: 
ds@1952: 	return NULL;
ds@1952: }
ds@1952: 
ds@1952: SMCError TextParsers::ParseStream_SMC(void *stream, 
ds@1952: 								   STREAMREADER srdr, 
ds@1952: 								   ITextListener_SMC *smc, 
ds@1952: 								   SMCStates *pStates)
ds@1952: {
ds@1952: 	char *reparse_point = NULL;
ds@1952: 	char in_buf[4096];
ds@1952: 	char *parse_point = in_buf;
ds@1952: 	char *line_begin = in_buf;
ds@1952: 	unsigned int read;
ds@1952: 	unsigned int curlevel = 0;
ds@1952: 	bool in_quote = false;
ds@1952: 	bool ignoring = false;
ds@1952: 	bool eol_comment = false;
ds@1952: 	bool ml_comment = false;
ds@1952: 	unsigned int i;
ds@1952: 	SMCError err = SMCError_Okay;
ds@1952: 	SMCResult res;
ds@1952: 	SMCStates states;
ds@1952: 	char c;
ds@1952: 
ds@1952: 	StringInfo strings[3];
ds@1952: 	StringInfo emptystring;
ds@1952: 
ds@1952: 	states.line = 1;
ds@1952: 	states.col = 0;
ds@1952: 
ds@1952: 	smc->ReadSMC_ParseStart();
ds@1952: 
ds@1952: 	/**
ds@1952: 	 * The stream reader reads in as much as it can fill the buffer with.
ds@1952: 	 * It then processes the buffer.  If the buffer cannot be fully processed, for example, 
ds@1952: 	 * a line is left hanging with no newline, then the contents of the buffer is shifted 
ds@1952: 	 * down, and the buffer is filled from the stream reader again.
ds@1952: 	 *
ds@1952: 	 * What makes this particularly annoying is that we cache pointers everywhere, so when 
ds@1952: 	 * the shifting process takes place, all those pointers must be shifted as well.
ds@1952: 	 */
ds@1952: 	while (srdr(stream, parse_point, sizeof(in_buf) - (parse_point - in_buf) - 1, &read))
ds@1952: 	{
ds@1952: 		if (!read)
ds@1952: 		{
ds@1952: 			break;
ds@1952: 		}
ds@1952: 
ds@1952: 		/* Check for BOM markings, which is only relevant on the first line.
ds@1952: 		 * Not worth it, but it could be moved out of the loop.
ds@1952: 		 */
ds@1952: 		if (states.line == 1 && 
ds@1952: 			in_buf[0] == (char)0xEF && 
ds@1952: 			in_buf[1] == (char)0xBB && 
ds@1952: 			in_buf[2] == (char)0xBF)
ds@1952: 		{
ds@1952: 			/* Move EVERYTHING down :\ */
ds@1952: 			memmove(in_buf, &in_buf[3], read - 3);
ds@1952: 			read -= 3;
ds@1952: 		}
ds@1952: 
ds@1952: 		if (reparse_point)
ds@1952: 		{
ds@1952: 			read += (parse_point - reparse_point);
ds@1952: 			parse_point = reparse_point;
ds@1952: 			reparse_point = NULL;
ds@1952: 		}
ds@1952: 
ds@1952: 		for (i=0; i<read; i++)
ds@1952: 		{
ds@1952: 			c = parse_point[i];
ds@1952: 			if (c == '\n')
ds@1952: 			{
ds@1952: 				/* If we got a newline, there's a lot of things that could have happened in the interim.
ds@1952: 				 * First, let's make sure the staged strings are rotated.
ds@1952: 				 */
ds@1952: 				if (strings[0].ptr)
ds@1952: 				{
ds@1952: 					strings[0].end = &parse_point[i];
ds@1952: 					if (rotate(strings) != NULL)
ds@1952: 					{
ds@1952: 						err = SMCError_InvalidTokens;
ds@1952: 						goto failed;
ds@1952: 					}
ds@1952: 				}
ds@1952: 
ds@1952: 				/* Next, let's clear some line-based values that may no longer have meaning */
ds@1952: 				eol_comment = false;
ds@1952: 				in_quote = false;
ds@1952: 				if (ignoring && !ml_comment)
ds@1952: 				{
ds@1952: 					ignoring = false;
ds@1952: 				}
ds@1952: 
ds@1952: 				/* Pass the raw line onto the listener.  We terminate the line so the receiver 
ds@1952: 				 * doesn't get tons of useless info.  We restore the newline after.
ds@1952: 				 */
ds@1952: 				parse_point[i] = '\0';
ds@1952: 				if ((res=smc->ReadSMC_RawLine(&states, line_begin)) != SMCResult_Continue)
ds@1952: 				{
ds@1952: 					err = (res == SMCResult_HaltFail) ? SMCError_Custom : SMCError_Okay;
ds@1952: 					goto failed;
ds@1952: 				}
ds@1952: 				parse_point[i] = '\n';
ds@1952: 
ds@1952: 				/* Now we check the sanity of our staged strings! */
ds@1952: 				if (strings[2].ptr)
ds@1952: 				{
ds@1952: 					if (!curlevel)
ds@1952: 					{
ds@1952: 						err = SMCError_InvalidProperty1;
ds@1952: 						goto failed;
ds@1952: 					}
ds@1952: 					/* Assume the next string is a property and pass the info on. */
ds@1952: 					if ((res=smc->ReadSMC_KeyValue(
ds@1952: 						&states,
ds@1952: 						FixupString(strings[2]),
ds@1952: 						FixupString(strings[1]))) != SMCResult_Continue)
ds@1952: 					{
ds@1952: 						err = (res == SMCResult_HaltFail) ? SMCError_Custom : SMCError_Okay;
ds@1952: 						goto failed;
ds@1952: 					}
ds@1952: 					scrap(strings);
ds@1952: 				}
ds@1952: 
ds@1952: 				/* Change the states for the next line */
ds@1952: 				states.col = 0;
ds@1952: 				states.line++;
ds@1952: 				line_begin = &parse_point[i+1];		//Note: safe because this gets relocated later
ds@1952: 			} 
ds@1952: 			else if (ignoring) 
ds@1952: 			{
ds@1952: 				if (in_quote)
ds@1952: 				{
ds@1952: 					/* If i was 0, we could have reparsed, so make sure there's no buffer underrun */
ds@1952: 					if ((&parse_point[i] != in_buf) && c == '"' && parse_point[i-1] != '\\')
ds@1952: 					{
ds@1952: 						/* If we reached a quote in an ignore phase,
ds@1952: 						 * we're staging a string and we must rotate it out.
ds@1952: 						 */
ds@1952: 						in_quote = false;
ds@1952: 						ignoring = false;
ds@1952: 						/* Set our info */
ds@1952: 						strings[0].end = &parse_point[i];
ds@1952: 						strings[0].quoted = true;
ds@1952: 						if (rotate(strings) != NULL)
ds@1952: 						{
ds@1952: 							/* If we rotated too many strings, there was too much crap on one line */
ds@1952: 							err = SMCError_InvalidTokens;
ds@1952: 							goto failed;
ds@1952: 						}
ds@1952: 					} 
ds@1952: 					else if (c == '\\') 
ds@1952: 					{
ds@1952: 						strings[0].special = true;
ds@1952: 						if (i == (read - 1))
ds@1952: 						{
ds@1952: 							reparse_point = &parse_point[i];
ds@1952: 							break;
ds@1952: 						}
ds@1952: 					}
ds@1952: 				} 
ds@1952: 				else if (ml_comment) 
ds@1952: 				{
ds@1952: 					if (c == '*')
ds@1952: 					{
ds@1952: 						/* Check if we need to get more input first */
ds@1952: 						if (i == read - 1)
ds@1952: 						{
ds@1952: 							reparse_point = &parse_point[i];
ds@1952: 							break;
ds@1952: 						}
ds@1952: 						if (parse_point[i+1] == '/')
ds@1952: 						{
ds@1952: 							ml_comment = false;
ds@1952: 							ignoring = false;
ds@1952: 							/* We should not be staging anything right now. */
ds@1952: 							assert(strings[0].ptr == NULL);
ds@1952: 							/* Advance the input stream so we don't choke on this token */
ds@1952: 							i++;
ds@1952: 							states.col++;
ds@1952: 						}
ds@1952: 					}
ds@1952: 				}
ds@1952: 			} 
ds@1952: 			else 
ds@1952: 			{
ds@1952: 				/* Check if we're whitespace or not */
ds@1952: 				if (!g_ws_chartable[(unsigned)c])
ds@1952: 				{
ds@1952: 					bool restage = false;
ds@1952: 					/* Check various special tokens:
ds@1952: 					 * ;
ds@1952: 					 * //
ds@1952: 					 * / *
ds@1952: 					 * {
ds@1952: 					 * }
ds@1952: 					 */
ds@1952: 					if (c == ';' || c == '/')
ds@1952: 					{
ds@1952: 						/* If it's a line-based comment (that is, ; or //)
ds@1952: 						 * we will need to scrap everything until the end of the line.
ds@1952: 						 */
ds@1952: 						if (c == '/')
ds@1952: 						{
ds@1952: 							if (i == read - 1)
ds@1952: 							{
ds@1952: 								/* If we reached the end of the look-ahead, we need to re-check our input.
ds@1952: 								 * Breaking out will force this to be the new reparse point!
ds@1952: 								 */
ds@1952: 								reparse_point = &parse_point[i];
ds@1952: 								break;
ds@1952: 							}
ds@1952: 							if (parse_point[i + 1] == '/')
ds@1952: 							{
ds@1952: 								/* standard comment */
ds@1952: 								ignoring = true;
ds@1952: 								eol_comment = true;
ds@1952: 								restage = true;
ds@1952: 							} 
ds@1952: 							else if (parse_point[i+1] == '*') 
ds@1952: 							{
ds@1952: 								/* inline comment - start ignoring */
ds@1952: 								ignoring = true;
ds@1952: 								ml_comment = true;
ds@1952: 								/* yes, we restage, meaning that:
ds@1952: 								 * STR/ *stuff* /ING  (space because ml comments don't nest in C++)
ds@1952: 								 * will not generate 'STRING', but rather 'STR' and 'ING'.
ds@1952: 								 * This should be a rare occurrence and is done here for convenience.
ds@1952: 								 */
ds@1952: 								restage = true;
ds@1952: 							}
ds@1952: 						} 
ds@1952: 						else 
ds@1952: 						{
ds@1952: 							ignoring = true;
ds@1952: 							eol_comment = true;
ds@1952: 							restage = true;
ds@1952: 						}
ds@1952: 					} 
ds@1952: 					else if (c == '{') 
ds@1952: 					{
ds@1952: 						/* If we are staging a string, we must rotate here */
ds@1952: 						if (strings[0].ptr)
ds@1952: 						{
ds@1952: 							/* We have unacceptable tokens on this line */
ds@1952: 							if (rotate(strings) != NULL)
ds@1952: 							{
ds@1952: 								err = SMCError_InvalidSection1;
ds@1952: 								goto failed;
ds@1952: 							}
ds@1952: 						}
ds@1952: 						/* Sections must always be alone */
ds@1952: 						if (strings[2].ptr != NULL)
ds@1952: 						{
ds@1952: 							err = SMCError_InvalidSection1;
ds@1952: 							goto failed;
ds@1952: 						} 
ds@1952: 						else if (strings[1].ptr == NULL)
ds@1952: 						{
ds@1952: 							err = SMCError_InvalidSection2;
ds@1952: 							goto failed;
ds@1952: 						}
ds@1952: 						if ((res=smc->ReadSMC_NewSection(&states, FixupString(strings[1])))
ds@1952: 							!= SMCResult_Continue)
ds@1952: 						{
ds@1952: 							err = (res == SMCResult_HaltFail) ? SMCError_Custom : SMCError_Okay;
ds@1952: 							goto failed;
ds@1952: 						}
ds@1952: 						strings[1] = emptystring;
ds@1952: 						curlevel++;
ds@1952: 					} 
ds@1952: 					else if (c == '}') 
ds@1952: 					{
ds@1952: 						/* Unlike our matching friend, this can be on the same line as something prior */
ds@1952: 						if (rotate(strings) != NULL)
ds@1952: 						{
ds@1952: 							err = SMCError_InvalidSection3;
ds@1952: 							goto failed;
ds@1952: 						}
ds@1952: 						if (strings[2].ptr)
ds@1952: 						{
ds@1952: 							if (!curlevel)
ds@1952: 							{
ds@1952: 								err = SMCError_InvalidProperty1;
ds@1952: 								goto failed;
ds@1952: 							}
ds@1952: 							if ((res=smc->ReadSMC_KeyValue(
ds@1952: 											&states,
ds@1952: 											FixupString(strings[2]),
ds@1952: 											FixupString(strings[1])))
ds@1952: 								!= SMCResult_Continue)
ds@1952: 							{
ds@1952: 								err = (res == SMCResult_HaltFail) ? SMCError_Custom : SMCError_Okay;
ds@1952: 								goto failed;
ds@1952: 							}
ds@1952: 						} 
ds@1952: 						else if (strings[1].ptr) 
ds@1952: 						{
ds@1952: 							err = SMCError_InvalidSection3;
ds@1952: 							goto failed;
ds@1952: 						} 
ds@1952: 						else if (!curlevel) 
ds@1952: 						{
ds@1952: 							err = SMCError_InvalidSection4;
ds@1952: 							goto failed;
ds@1952: 						}
ds@1952: 						/* Now it's safe to leave the section */
ds@1952: 						scrap(strings);
ds@1952: 						if ((res=smc->ReadSMC_LeavingSection(&states)) != SMCResult_Continue)
ds@1952: 						{
ds@1952: 							err = (res == SMCResult_HaltFail) ? SMCError_Custom : SMCError_Okay;
ds@1952: 							goto failed;
ds@1952: 						}
ds@1952: 						curlevel--;
ds@1952: 					} 
ds@1952: 					else if (c == '"') 
ds@1952: 					{
ds@1952: 						/* If we get a quote mark, we always restage, but we need to do it beforehand */
ds@1952: 						if (strings[0].ptr)
ds@1952: 						{
ds@1952: 							strings[0].end = &parse_point[i];
ds@1952: 							if (rotate(strings) != NULL)
ds@1952: 							{
ds@1952: 								err = SMCError_InvalidTokens;
ds@1952: 								goto failed;
ds@1952: 							}
ds@1952: 						}
ds@1952: 						strings[0].ptr = &parse_point[i];
ds@1952: 						in_quote = true;
ds@1952: 						ignoring = true;
ds@1952: 					} 
ds@1952: 					else if (!strings[0].ptr) 
ds@1952: 					{
ds@1952: 						/* If we have no string, we must start one */
ds@1952: 						strings[0].ptr = &parse_point[i];
ds@1952: 					}
ds@1952: 					if (restage && strings[0].ptr)
ds@1952: 					{
ds@1952: 						strings[0].end = &parse_point[i];
ds@1952: 						if (rotate(strings) != NULL)
ds@1952: 						{
ds@1952: 							err = SMCError_InvalidTokens;
ds@1952: 							goto failed;
ds@1952: 						}
ds@1952: 					}
ds@1952: 				} 
ds@1952: 				else 
ds@1952: 				{
ds@1952: 					/* If we're eating a string and get whitespace, we need to restage.
ds@1952: 					 * (Note that if we are quoted, this is being ignored)
ds@1952: 					 */
ds@1952: 					if (strings[0].ptr)
ds@1952: 					{
ds@1952: 						/*
ds@1952: 						 * The specification says the second string in a pair does not need to be quoted.
ds@1952: 						 * Thus, we check if there's already a string on the stack.
ds@1952: 						 * If there's a newline, we always rotate so the newline has an empty starter.
ds@1952: 						 */
ds@1952: 						if (!strings[1].ptr)
ds@1952: 						{
ds@1952: 							/* There's no string, so we must move this one down and eat up another */
ds@1952: 							strings[0].end = &parse_point[i];
ds@1952: 							rotate(strings);
ds@1952: 						} 
ds@1952: 						else if (!strings[1].quoted) 
ds@1952: 						{
ds@1952: 							err = SMCError_InvalidTokens;
ds@1952: 							goto failed;
ds@1952: 						}
ds@1952: 					}
ds@1952: 				}
ds@1952: 			}
ds@1952: 
ds@1952: 			/* Advance which token we're on */
ds@1952: 			states.col++;
ds@1952: 		}
ds@1952: 
ds@1952: 		if (line_begin != in_buf)
ds@1952: 		{
ds@1952: 			/* The line buffer has advanced, so it's safe to copy N bytes back to the beginning.
ds@1952: 			 * What's N?  N is the lowest point we're currently relying on.
ds@1952: 			 */
ds@1952: 			char *stage = lowstring(strings);
ds@1952: 			if (!stage || stage > line_begin)
ds@1952: 			{
ds@1952: 				stage = line_begin;
ds@1952: 			}
ds@1952: 			unsigned int bytes = read - (stage - parse_point);
ds@1952: 
ds@1952: 			/* It is now safe to delete everything before the staged point */
ds@1952: 			memmove(in_buf, stage, bytes);
ds@1952: 
ds@1952: 			/* Calculate the number of bytes in the new buffer */
ds@1952: 			bytes = stage - in_buf;
ds@1952: 			/* Relocate all the cached pointers to our new base */
ds@1952: 			line_begin -= bytes;
ds@1952: 			reloc(strings[0], bytes);
ds@1952: 			reloc(strings[1], bytes);
ds@1952: 			reloc(strings[2], bytes);
ds@1952: 			if (reparse_point)
ds@1952: 			{
ds@1952: 				reparse_point -= bytes;
ds@1952: 			}
ds@1952: 			if (parse_point)
ds@1952: 			{
ds@1952: 				parse_point = &parse_point[read];
ds@1952: 				parse_point -= bytes;
ds@1952: 			}
ds@1952: 		} 
ds@1952: 		else if (read == sizeof(in_buf) - 1) 
ds@1952: 		{
ds@1952: 			err = SMCError_TokenOverflow;
ds@1952: 			goto failed;
ds@1952: 		}
ds@1952: 	}
ds@1952: 
ds@1952: 	/* If we're done parsing and there are tokens left over... */
ds@1952: 	if (curlevel)
ds@1952: 	{
ds@1952: 		err = SMCError_InvalidSection5;
ds@1952: 		goto failed;
ds@1952: 	} 
ds@1952: 	else if (strings[0].ptr || strings[1].ptr) 
ds@1952: 	{
ds@1952: 		err = SMCError_InvalidTokens;
ds@1952: 		goto failed;
ds@1952: 	}
ds@1952: 	
ds@1952: 	smc->ReadSMC_ParseEnd(false, false);
ds@1952: 	
ds@1952: 	if (pStates != NULL)
ds@1952: 	{
ds@1952: 		*pStates = states;
ds@1952: 	}
ds@1952: 
ds@1952: 	return SMCError_Okay;
ds@1952: 
ds@1952: failed:
ds@1952: 	if (pStates != NULL)
ds@1952: 	{
ds@1952: 		*pStates = states;
ds@1952: 	}
ds@1952: 
ds@1952: 	smc->ReadSMC_ParseEnd(true, (err == SMCError_Custom));
ds@1952: 
ds@1952: 	return err;
ds@1952: }
ds@1952: 
ds@1952: 
ds@1952: /**
ds@1952:  * INI parser 
ds@1952:  */
ds@1952: 
ds@1952: bool TextParsers::ParseFile_INI(const char *file, ITextListener_INI *ini_listener, unsigned int *line, unsigned int *col)
ds@1952: {
ds@1952: 	FILE *fp = fopen(file, "rt");
ds@1952: 	unsigned int curline = 0;
ds@1952: 	unsigned int curtok;
ds@1952: 	size_t len;
ds@1952: 
ds@1952: 	if (!fp)
ds@1952: 	{
ds@1952: 		if (line)
ds@1952: 		{
ds@1952: 			*line = 0;
ds@1952: 		}
ds@1952: 
ds@1952: 		return false;
ds@1952: 	}
ds@1952: 
ds@1952: 	char buffer[2048];
ds@1952: 	char *ptr, *save_ptr;
ds@1952: 	bool in_quote;
ds@1952: 
ds@1952: 	while (!feof(fp))
ds@1952: 	{
ds@1952: 		curline++;
ds@1952: 		curtok = 0;
ds@1952: 		buffer[0] = '\0';
ds@1952: 		if (fgets(buffer, sizeof(buffer), fp) == NULL)
ds@1952: 		{
ds@1952: 			break;
ds@1952: 		}
ds@1952: 
ds@1952: 		//:TODO: this will only run once, so find a nice way to move it out of the while loop
ds@1952: 		/* If this is the first line, check the first three bytes for BOM */
ds@1952: 		if (curline == 1 && 
ds@1952: 			buffer[0] == (char)0xEF && 
ds@1952: 			buffer[1] == (char)0xBB && 
ds@1952: 			buffer[2] == (char)0xBF)
ds@1952: 		{
ds@1952: 			/* We have a UTF-8 marked file... skip these bytes */
ds@1952: 			ptr = &buffer[3];
ds@1952: 		} else {
ds@1952: 			ptr = buffer;
ds@1952: 		}
ds@1952: 
ds@1952: 		/***************************************************
ds@1952: 		 * We preprocess the string before parsing tokens! *
ds@1952: 		 ***************************************************/
ds@1952: 
ds@1952: 		/* First strip beginning whitespace */
ds@1952: 		while (*ptr != '\0' && g_ws_chartable[(unsigned)*ptr] != 0)
ds@1952: 		{
ds@1952: 			ptr++;
ds@1952: 		}
ds@1952: 
ds@1952: 		len = strlen(ptr);
ds@1952: 
ds@1952: 		if (!len)
ds@1952: 		{
ds@1952: 			continue;
ds@1952: 		}
ds@1952: 
ds@1952: 		/* Now search for comment characters */
ds@1952: 		in_quote = false;
ds@1952: 		save_ptr = ptr;
ds@1952: 		for (size_t i=0; i<len; i++,ptr++)
ds@1952: 		{
ds@1952: 			if (!in_quote)
ds@1952: 			{
ds@1952: 				switch (*ptr)
ds@1952: 				{
ds@1952: 				case '"':
ds@1952: 					{
ds@1952: 						in_quote = true;
ds@1952: 						break;
ds@1952: 					}
ds@1952: 				case ';':
ds@1952: 					{
ds@1952: 						/* Stop the loop */
ds@1952: 						len = i;
ds@1952: 						/* Terminate the string here */
ds@1952: 						*ptr = '\0';
ds@1952: 						break;
ds@1952: 					}
ds@1952: 				}
ds@1952: 			} else {
ds@1952: 				if (*ptr == '"')
ds@1952: 				{
ds@1952: 					in_quote = false;
ds@1952: 				}
ds@1952: 			}
ds@1952: 		}
ds@1952: 
ds@1952: 		if (!len)
ds@1952: 		{
ds@1952: 			continue;
ds@1952: 		}
ds@1952: 
ds@1952: 		ptr = save_ptr;
ds@1952: 
ds@1952: 		/* Lastly, strip ending whitespace off */
ds@1952: 		for (size_t i=len-1; i>=0 && i<len; i--)
ds@1952: 		{
ds@1952: 			if (g_ws_chartable[(unsigned)ptr[i]])
ds@1952: 			{
ds@1952: 				ptr[i] = '\0';
ds@1952: 				len--;
ds@1952: 			} else {
ds@1952: 				break;
ds@1952: 			}
ds@1952: 		}
ds@1952: 
ds@1952: 		if (!len)
ds@1952: 		{
ds@1952: 			continue;
ds@1952: 		}
ds@1952: 
ds@1952: 		if (!ini_listener->ReadINI_RawLine(ptr, &curtok))
ds@1952: 		{
ds@1952: 			goto event_failed;
ds@1952: 		}
ds@1952: 
ds@1952: 		if (*ptr == '[')
ds@1952: 		{
ds@1952: 			bool invalid_tokens = false;
ds@1952: 			bool got_bracket = false;
ds@1952: 			bool extra_tokens = false;
ds@1952: 			char c;
ds@1952: 			bool alnum;
ds@1952: 			wchar_t wc;
ds@1952: 
ds@1952: 			for (size_t i=1; i<len; i++)
ds@1952: 			{
ds@1952: 				c = ptr[i];
ds@1952: 				alnum = false;
ds@1952: 
ds@1952: 				if (c & (1<<7))
ds@1952: 				{
ds@1952: 					if (mbtowc(&wc, &ptr[i], len-i) != -1)
ds@1952: 					{
ds@1952: 						alnum = (iswalnum(wc) != 0);
ds@1952: 						i += _GetUTF8CharBytes(&ptr[i]) - 1;
ds@1952: 					}
ds@1952: 				} else {
ds@1952: 					alnum = (isalnum(c) != 0) || (g_ini_chartable1[(unsigned)c] != 0);
ds@1952: 				}
ds@1952: 				if (!alnum)
ds@1952: 				{
ds@1952: 					/* First check - is this a bracket? */
ds@1952: 					if (c == ']')
ds@1952: 					{
ds@1952: 						/* Yes! */
ds@1952: 						got_bracket = true;
ds@1952: 						/* If this isn't the last character... */
ds@1952: 						if (i != len - 1)
ds@1952: 						{
ds@1952: 							extra_tokens = true;
ds@1952: 						}
ds@1952: 						/* terminate */
ds@1952: 						ptr[i] = '\0';
ds@1952: 						break;
ds@1952: 					} else {
ds@1952: 						/* n...No! Continue copying. */
ds@1952: 						invalid_tokens = true;
ds@1952: 					}
ds@1952: 				}
ds@1952: 			}
ds@1952: 
ds@1952: 			/* Tell the handler */
ds@1952: 			if (!ini_listener->ReadINI_NewSection(&ptr[1], invalid_tokens, got_bracket, extra_tokens, &curtok))
ds@1952: 			{
ds@1952: 				goto event_failed;
ds@1952: 			}
ds@1952: 		} else {
ds@1952: 			char *key_ptr = ptr;
ds@1952: 			char *val_ptr = NULL;
ds@1952: 			char c;
ds@1952: 			size_t first_space = 0;
ds@1952: 			bool invalid_tokens = false;
ds@1952: 			bool equal_token = false;
ds@1952: 			bool quotes = false;
ds@1952: 			bool alnum;
ds@1952: 			wchar_t wc;
ds@1952: 
ds@1952: 			for (size_t i=0; i<len; i++)
ds@1952: 			{
ds@1952: 				c = ptr[i];
ds@1952: 				alnum = false;
ds@1952: 				/* is this an invalid char? */
ds@1952: 				if (c & (1<<7))
ds@1952: 				{
ds@1952: 					if (mbtowc(&wc, &ptr[i], len-i) != -1)
ds@1952: 					{
ds@1952: 						alnum = (iswalnum(wc) != 0);
ds@1952: 						i += _GetUTF8CharBytes(&ptr[i]) - 1;
ds@1952: 					}
ds@1952: 				} else {
ds@1952: 					alnum = (isalnum(c) != 0) || (g_ini_chartable1[(unsigned)c] != 0);
ds@1952: 				}
ds@1952: 
ds@1952: 				if (!alnum)
ds@1952: 				{
ds@1952: 					if (g_ws_chartable[(unsigned)c])
ds@1952: 					{
ds@1952: 						/* if it's a space, keep track of the first occurring space */
ds@1952: 						if (!first_space)
ds@1952: 						{
ds@1952: 							first_space = i;
ds@1952: 						}
ds@1952: 					} else {
ds@1952: 						if (c == '=')
ds@1952: 						{
ds@1952: 							/* if it's an equal sign, we're done with the key */
ds@1952: 							if (first_space)
ds@1952: 							{
ds@1952: 								/* remove excess whitespace */
ds@1952: 								key_ptr[first_space] = '\0';
ds@1952: 							} else {
ds@1952: 								/* remove the equal sign */
ds@1952: 								key_ptr[i] = '\0';
ds@1952: 							}
ds@1952: 							if (ptr[++i] != '\0')
ds@1952: 							{
ds@1952: 								/* If this isn't the end, set next pointer */
ds@1952: 								val_ptr = &ptr[i];
ds@1952: 							}
ds@1952: 							equal_token = true;
ds@1952: 							break;
ds@1952: 						} else {
ds@1952: 							/* Mark that we got something invalid! */
ds@1952: 							invalid_tokens = true;
ds@1952: 							first_space = 0;
ds@1952: 						}
ds@1952: 					}
ds@1952: 				}
ds@1952: 			}
ds@1952: 
ds@1952: 			/* Now we need to parse the value, if any */
ds@1952: 			if (val_ptr)
ds@1952: 			{
ds@1952: 				/* eat up spaces! there shouldn't be any h*/
ds@1952: 				while ((*val_ptr != '\0') && g_ws_chartable[(unsigned)*val_ptr] != 0)
ds@1952: 				{
ds@1952: 					val_ptr++;
ds@1952: 				}
ds@1952: 				if (*val_ptr == '\0')
ds@1952: 				{
ds@1952: 					val_ptr = NULL;
ds@1952: 					goto skip_value;
ds@1952: 				}
ds@1952: 				/* Do we have an initial quote? If so, the parsing rules change! */
ds@1952: 				if (*val_ptr == '"' && *val_ptr != '\0')
ds@1952: 				{
ds@1952: 					len = strlen(val_ptr);
ds@1952: 					if (val_ptr[len-1] == '"')
ds@1952: 					{
ds@1952: 						/* Strip quotes! */
ds@1952: 						val_ptr[--len] = '\0';
ds@1952: 						val_ptr++;
ds@1952: 						quotes = true;
ds@1952: 					}
ds@1952: 				}
ds@1952: 			}
ds@1952: skip_value:
ds@1952: 			/* We're done! */
ds@1952: 			curtok = val_ptr - buffer;
ds@1952: 			if (!ini_listener->ReadINI_KeyValue(key_ptr, val_ptr, invalid_tokens, equal_token, quotes, &curtok))
ds@1952: 			{
ds@1952: 				curtok = 0;
ds@1952: 				goto event_failed;
ds@1952: 			}
ds@1952: 		}
ds@1952: 	}
ds@1952: 
ds@1952: 	if (line)
ds@1952: 	{
ds@1952: 		*line = curline;
ds@1952: 	}
ds@1952: 
ds@1952: 	fclose(fp);
ds@1952: 
ds@1952: 	return true;
ds@1952: 
ds@1952: event_failed:
ds@1952: 	if (line)
ds@1952: 	{
ds@1952: 		*line = curline;
ds@1952: 	}
ds@1952: 
ds@1952: 	if (col)
ds@1952: 	{
ds@1952: 		*col = curtok;
ds@1952: 	}
ds@1952: 
ds@1952: 	fclose(fp);
ds@1952: 
ds@1952: 	return false;
ds@1952: }
ds@1952: 
ds@1952: const char *TextParsers::GetSMCErrorString(SMCError err)
ds@1952: {
ds@1952: 	static const char *s_errors[] = 
ds@1952: 	{
ds@1952: 		NULL,
ds@1952: 		"Stream failed to open",
ds@1952: 		"Stream returned read error",
ds@1952: 		NULL,
ds@1952: 		"Un-quoted section has invalid tokens",
ds@1952: 		"Section declared without header",
ds@1952: 		"Section declared with unknown tokens",
ds@1952: 		"Section ending without a matching section beginning",
ds@1952: 		"Section beginning without a matching ending",
ds@1952: 		"Line contained too many invalid tokens",
ds@1952: 		"Token buffer overflowed",
ds@1952: 		"A property was declared outside of a section",
ds@1952: 	};
ds@1952: 
ds@1952: 	if (err < SMCError_Okay || err > SMCError_InvalidProperty1)
ds@1952: 	{
ds@1952: 		return NULL;
ds@1952: 	}
ds@1952: 
ds@1952: 	return s_errors[err];
ds@1952: }